I-1

Journal of Acoustic Emission, Volume 28, 2010 VIEWING AN ARTICLE IN CD-ROM You need a software that can read a pdf-file, such as Adobe Acrobat Reader®. Install one first before going further. You can download the latest version from www.adobe.com. Use of Adobe Acrobat® (not provided) allows one to utilize the Search function. This gives the power to search for a name or a word in the entire volume and is highly recommended. PDF Files: Contents are listed below. Individual papers have file names of 28-xxx.pdf, with xxx as their first page number.

CONTENTS 28-001 ESTIMATION OF CRUSTAL STRUCTURE IN HORONOBE AREA, HOKKAIDO, JAPAN, USING MULTIPLET-CLUSTERING ANALYSIS HIROKAZU MORIYA, KOICHI ASAMORI, ITARU KITAMURA, HIKARU HOTTA, HIDEFUMI OHARA and TADAFUMI NIIZATO 1-10 28-011 ACOUSTIC EMISSION TESTING - DEFINING A NEW STANDARD OF ACOUSTIC EMISSION TESTING FOR PRESSURE VESSELS Part 2: Performance analysis of different configurations of real case testing and recommendations for developing a new guide for the application of acoustic emission JOHANN CATTY 11-31 28-032 ACOUSTIC EMISSION MONITORING OF STEEL-FIBER REINFORCED CONCRETE BEAMS UNDER BENDING DIMITRIOS G. AGGELIS, DIMITRA SOULIOTI, NEKTARIA M. BARKOULA, ALKIVIADIS S. PAIPETIS, THEODORE E. MATIKAS and TOMOKI SHIOTANI 32-40 28-041 ON LAMB MODES AS A FUNCTION OF ACOUSTIC EMISSION SOURCE RISE TIME M. A. HAMSTAD 41-58 28-059 WAVEFORM ANALYSIS OF ACOUSTIC EMISSION MONITORING OF TENSILE TESTS ON WELDED WOOD-JOINTS ANDREAS J. BRUNNER, THOMAS TANNERT and TILL VALLÉE 59-67 28-068 USE OF ACOUSTO-ULTRASONIC TECHNIQUESTO DETERMINE PROPERTIES OF REMANUFACTURED PARTICLEBOARDS MADE SOLELY FROM RECYCLED PARTICLES SUMIRE KAWAMOTO 68-75

28-076 ACOUSTIC EMISSION ACTIVITY OF SPRUCE SAPWOOD BECOMES WEAKER AFTER EACH DEHYDRATION-REWETTING CYCLE SABINE ROSNER and SUMIRE KAWAMOTO 76-84

 

I-2

28-085 ACOUSTIC EMISSION SOURCE LOCATION IN PLATE-LIKE STRUCTURES USING A CLOSELY ARRANGED TRIANGULAR SENSOR ARRAY DIRK ALJETS , ALEX CHONG , STEVE WILCOX and KAREN HOLFORD 85-98 28-099 NEURAL NETWORK AE SOURCE LOCATION APART FROM STRUCTURE SIZE AND MATERIAL MILAN CHLADA, ZDENEK PREVOROVSKY and MICHAL BLAHACEK 99-108 28-109 INTELLIGENT AE SIGNAL FILTERING METHODS VERA BARAT, YRIJ BORODIN and ALEXEY KUZMIN 109-119 28-120 DISCRIMINATION OF ACOUSTIC EMISSION HITS FROM DYNAMIC TESTS OF A REINFORCED CONCRETE SLAB ENRIQUE CASTRO, ROSA PIOTRKOWSKI, ANTOLINO GALLEGO and AMADEO BENAVENT CLIMENT 120-128

28-129 USE OF CLUSTER ANALYSIS OF ACOUSTIC EMISSION SIGNALS IN EVALUATING DAMAGE SEVERITY IN CONCRETE STRUCTURES L. CALABRESE, G. CAMPANELLA and E. PROVERBIO 129-141 28-142 SIMULATION OF LAMB WAVE EXCITATION FOR DIFFERENT ELASTIC PROPERTIES AND ACOUSTIC EMISSION SOURCE GEOMETRIES MARKUS G. R. SAUSE and SIEGFRIED HORN 142-154

28-155 ACOUSTIC EMISSION EVENT IDENTIFICATION WITH SIMILAR TRANSFER FUNCTIONS FRANZ RAUSCHER 155-162 28-163 ANALYSIS OF FRACTURE RESISTANCE OF TOOL STEELS BY MEANS OF ACOUSTIC EMISSION EVA MARTINEZ-GONZALEZ, INGRID PICAS, DANIEL CASELLAS and JORDI ROMEU

163-169

28-170 COMPARISON OF ACOUSTIC EMISSION SIGNAL AND X-RAY DIFFRACTION AT INITIAL STAGES OF FATIGUE DAMAGE FRANTISEK VLASIC, PAVEL MAZAL and FILIP HORT 170-178 28-179 AE SIGNALS DURING LASER CUTTING OF DIFFERENT STEEL SHEET THICKNESSES TOMAŽ KEK and JANEZ GRUM 179-187 28-188 ACOUSTIC EMISSION ANALYSIS AND THERMO-HYGRO-MECHANICAL MODEL FOR CONCRETE EXPOSED TO FIRE CHRISTIAN GROSSE, JOŠKO OŽBOLT, RONALD RICHTER and GORAN PERIŠKIĆ 188-203

 

I-3

28-204 AE-SiGMA ANALYSIS IN BRAZILIAN TEST AND ACCELERATED CORROSION TEST OF CONCRETE MASAYASU OHTSU and YUMA KAWASAKI 204-214 28-215 ACOUSTIC EMISSION INSPECTION OF RAIL WHEELS KONSTANTINOS BOLLAS, DIMITRIOS PAPASALOUROS, DIMITRIOS KOUROUSIS and ATHANASIOS ANASTASOPOULOS 215-228 28-229 E07.04 – OVERVIEW OF CURRENT AND DEVELOPING ASTM ACOUSTIC EMISSION (AE) STANDARDS MARK F. CARLOS 229-­‐233 28-234 USE OF AE METHOD FOR DETECTION OF STEEL LAMINATION IN THE INDUSTRIAL PRESSURE VESSEL V.P. GOMERA, V.L. SOKOLOV, V.P. FEDOROV, A.A. OKHOTNIKOV and M.S. SAYKOVA 234-245 28-246 COMPARISON OF ACOUSTIC EMISSION PRODUCED DURING BENDING OF VARIOUS OXIDE CERAMIC AND SHORT FIBER OXIDE CERAMIC MATRIX COMPOSITES S.A. PAPARGYRI-MPENI, D.A. PAPARGYRIS, X. SPILIOTIS and A.D. PAPARGYRIS

246-275

28-256 NEW CHARACTERIZATION METHODS OF AE SENSORS KANJI ONO, HIDEO CHO and TAKUMA MATSUO 256-277 Contents28 Contents of Volume 28 (2010) I-1 – I-3 AUindex28 Authors Index of Volume 28 I-4 AusNotes Policy/Author’s Notes/Meeting Calendar/DVD/ I-5 – I-7 Subscription Information I-7 IAES20 JCAE Kishinoue Awards I-8 AE Literature AELit28 Book on AE by Markus Sause 12th AE conference proceeding in China, 2009: Gongtian Shen I-9 – I-11 Cover illustration See 28-142 by Sause and Horn for details. JAE Index Folder* Cumulative Indices of J. of Acoustic Emission, 1982 – 2010 Contents1-28 Contents Volumes 1-28 Authors Index1-28 Authors Index Volumes 1-28

* indicates the availability in CD-ROM only. Indices are also available for download from www.aewg.org.

 

I-4

AUTHORS INDEX, Volume 28, 2010

DIMITRIOS G. AGGELIS 28-032 DIRK ALJETS 28-085 ATHANASIOS ANASTASOPOULOS 28-215 KOICHI ASAMORI 28-001 VERA BARAT 28-109 NEKTARIA M. BARKOULA 28-032 AMADEO BENAVENT CLIMENT 28-120 MICHAL BLAHACEK 28-099 KONSTANTINOS BOLLAS 28-215 YRIJ BORODIN 28-109 ANDREAS J. BRUNNER 28-059 L. CALABRESE 28-129 G. CAMPANELLA 28-129 MARK F. CARLOS 28-229 DANIEL CASELLAS 28-163 ENRIQUE CASTRO 28-120 JOHANN CATTY 28-011 MILAN CHLADA 28-099 HIDEO CHO 28-256 ALEX CHONG 28-085 V.P. FEDOROV 28-234 ANTOLINO GALLEGO 28-120 V.P. GOMERA 28-234 CHRISTIAN GROSSE 28-188 JANEZ GRUM 28-179 M. A. HAMSTAD 28-041 KAREN HOLFORD 28-085 SIEGFRIED HORN 28-142 FILIP HORT 28-170 HIKARU HOTTA 28-001 SUMIRE KAWAMOTO 28-068, 28-076 YUMA KAWASAKI 28-204 TOMAŽ KEK 28-179 ITARU KITAMURA 28-001 DIMITRIOS KOUROUSIS 28-215 ALEXEY KUZMIN 28-109 EVA MARTINEZ-GONZALEZ 28-163 THEODORE E. MATIKAS 28-032 TAKUMA MATSUO 28-256 PAVEL MAZAL 28-170 HIROKAZU MORIYA 28-001

TADAFUMI NIIZATO 28-001 HIDEFUMI OHARA 28-001 MASAYASU OHTSU 28-204 A.A. OKHOTNIKOV 28-234 KANJI ONO 28-256 JOŠKO OŽBOLT 28-188 ALKIVIADIS S. PAIPETIS 28-032 DIMITRIOS PAPASALOUROS 28-215 S.A. PAPARGYRI-MPENI 28-246 A.D. PAPARGYRIS 28-246 D.A. PAPARGYRIS 28-246 GORAN PERIŠKIĆ 28-188 INGRID PICAS 28-163 ROSA PIOTRKOWSKI 28-120 ZDENEK PREVOROVSKY 28-099 E. PROVERBIO 28-129 FRANZ RAUSCHER 28-155 RONALD RICHTER 28-188 JORDI ROMEU 28-163 SABINE ROSNER 28-076 MARKUS G. R. SAUSE 28-142 M.S. SAYKOVA 28-234 TOMOKI SHIOTANI 28-032 V.L. SOKOLOV 28-234 DIMITRA SOULIOTI 28-032 X. SPILIOTIS 28-246 THOMAS TANNERT 28-059 TILL VALLÉE 28-059 FRANTISEK VLASIC 28-170 STEVE WILCOX 28-085

ISSN 0730-0050 Copyright©2010 Acoustic Emission Group I-5 CODEN: JACEDO All rights reserved.

JOURNAL OF ACOUSTIC EMISSION

Editor: Kanji Ono Associate Editors: A. G. Beattie, T. F. Drouillard, M. Ohtsu and W. H. Prosser

1. Aims and Scope of the Journal Journal of Acoustic Emission is an international journal

designed to be of broad interest and use to both researcher and practitioner of acoustic emission. It will publish original contri-butions of all aspects of research and significant engineering advances in the sciences and applications of acoustic emission. The journal will also publish reviews, the abstracts of papers presented at meetings, technical notes, communications and summaries of reports. Current news of interest to the acoustic emission communities, announcements of future conferences and working group meetings and new products will also be included.

Journal of Acoustic Emission includes the following classes of subject matters;

A. Research Articles: Manuscripts should represent com-pleted original work embodying the results of extensive investi-gation. These will be judged for scientific and technical merit.

B. Applications: Articles must present significant advances in the engineering applications of acoustic emission. Material will be subject to reviews for adequate description of procedures, substantial database and objective interpretation.

C. Technical Notes and Communications: These allow publi-cations of short items of current interest, new or improved experi-mental techniques and procedures, discussion of published articles and relevant applications.

D. AE Program and Data Files: Original program files and data files that can be read by others and analyzed will be distributed in CD-ROM.

Reviews, Tutorial Articles and Special Contributions will address the subjects of general interest. Nontechnical part will cover book reviews, significant personal and technical accom-plishments, current news and new products.

2. Endorsement Acoustic Emission Working Group (AEWG), European

Working Group on Acoustic Emission (EWGAE), have endorsed the publication of Journal of Acoustic Emission.

3. Governing Body The Editor and Associate Editors will implement the editorial

policies described above. The Editorial Board will advise the edi-tors on any major change. The Editor, Professor Kanji Ono, has the general responsibility for all the matters. Associate Editors assist the review processes as lead reviewers. The members of the Editorial Board are selected for their knowledge and experience on AE and will advise and assist the editors on the publication

policies and other aspects. The Board presently includes the following members:

A. Anastasopoulos (Greece) F.C. Beall (USA) J. Bohse (Germany) P. Cole (UK) L. Golaski (Poland) M.A. Hamstad (USA) R. Hay (Canada) K.M. Holford (UK) O.Y. Kwon (Korea) J.C. Lenain (France) G. Manthei (Germany) P. Mazal (Czech Republic) C.R.L. Murthy (India) A.A. Pollock (USA) F. Rauscher (Austria) T. Shiotani (Japan) P. Tschliesnig (Austria) H. Vallen (Germany) M. Wevers (Belgium) B.R.A. Wood (Australia) 4. Publication Journal of Acoustic Emission is published annually in CD-

ROM by Acoustic Emission Group, PMB 409, 4924 Balboa Blvd, Encino, CA 91316. It may also be reached at 2121H, Engr. V, University of California, Los Angeles, California 90095-1595 (USA). tel. 310-825-5233. FAX 310-206-7353. e-mail: aegroup7@gmail.com or ono@ucla.edu

5. Subscription Subscription should be sent to Acoustic Emission Group.

Annual rate for 2011 is US $111.00 including CD-ROM delivery, by priority mail in the U.S. and by air for Canada and elsewhere. For additional print copy, add $40-49. Overseas orders must be paid in US currencies with a check drawn on a US bank. PayPal payment accepted. Inquire for individual (with institutional order) and bookseller discounts. See also page I-7.

6. Advertisement No advertisement will be accepted, but announcements for

books, training courses and future meetings on AE will be included without charge.

I-6

Notes for Contributors

1. General The Journal will publish contributions from all parts of the world and manuscripts for publication should be submitted to the Editor. Send by e-mail or to: Professor Kanji Ono, Editor - JAE Rm. 2121H, Engr. V, MSE Dept. University of California 420 Westwood Plaza, Los Angeles, California 90095-1595 USA e-mail: kanjiono1@gmail.com Authors of any AE related publications are encouraged to send a copy to:

Acoustic Emission Collection Grainger Engineering Library University of Illinois at Urbana-Champaign. c/o: Mary C. Schlembach, Engineering Physics & Astronomy Librarian 154 Grainger Engineering Library Information Center 1301 West Springfield Ave. Urbana, IL 61801 Email: schlemba@illinois.edu

Only papers not previously published will be accepted. Authors must agree to transfer the copyright to the Journal and not to publish elsewhere, the same paper submitted to and accepted by the Journal. A paper is acceptable if it is a revision of a governmental or organizational report, or if it is based on a paper published with limited distribution. The language of the Journal is English. All papers should be written concisely and clearly. 2. Page Charges No page charge is levied. The contents of CD-ROM will be supplied to the authors free of charge via Internet site. 3. Manuscript for Review Manuscripts for review need only to be typed legibly; preferably, double-spaced on only one side of the page with wide margins. The title should be brief. An abstract of 100-200 words is needed for articles. Except for short communications, descriptive heading should be used to divide the paper into its component parts. Use the International System of Units (SI). References to published literature should be quoted in the text citing authors and the year of publication or consecutive numbers. These are to be grouped together

at the end of the paper. Journal references should be arranged as below. Titles for journal or book articles are helpful for readers, but may be omitted. H.L. Dunegan, D.O. Harris and C.A. Tatro (1968), Eng. Fract. Mech., 1, 105-122. Y. Krampfner, A. Kawamoto, K. Ono and A.T. Green (1975), "Acoustic Emission Characteristics of Cu Alloys under Low-Cycle Fatigue Conditions," NASA CR-134766, University of California, Los Angeles and Acoustic Emis-sion Tech. Corp., Sacramento, April. A.E. Lord, Jr. (1975), Physical Acoustics: Principles and Methods, vol. 11, eds. W. P. Mason and R. N. Thurston, Academic Press, New York, pp. 289-353. Abbreviations of journal titles should follow those used in the ASM Metals Abstracts. In every case, authors' initials, appropriate volume and page numbers should be included. The title of the cited journal reference is optional. Illustrations and tables should be planned to fit a single page width (165 mm or 6.5"). For the reviewing processes, these need not be of high quality, but submit glossy prints or equivalent electronic files with the final manuscript. Lines and letters should be legible. 4. Review All manuscripts will be judged by qualified reviewer(s). Each paper is reviewed by one of the editors and may be sent for review by members of the Editorial Board. The Board member may seek another independent review. In case of disputes, the author may request other reviewers. 5. Electronic Media This Journal will be primarily distributed electronically by CD-ROM. In order to expedite processing and minimize errors, the authors are requested to submit electronic files of the paper. On the INTERNET, you can send an MS Word file and the separate figure files (in JPEG or other formats) to "aegroup7@gmail.com". 6. Color Illustration With the new format, authors are encouraged to use them.

I-7

MEETING CALENDAR: World Conference on Acoustic Emission - 2011 Beijing WCAE-2011 is the first international event on AE organized in China. It is organized by the Chinese Society for Non-destructive Testing (ChSNDT) and endorsed by Chinese Mechanical Engineering Soci-ety. WCAE-2011 is undertaken by Technical Committee on Acoustic Emission of ChSNDT. The pri-mary objective of the WCAE 2011 is to exchange research and application information on AE, with par-ticular emphasis on scientific and technical development and cooperation between China and the world. The official language of this event is English. WCAE-2011 is refereed conference and the papers will be peer-reviewed. Abstract deadline is May 15, 2011. Gongtian Shen, Chairman; www.wcae2011.orgMr. Zhanwen Wu, WCAE-2011 Secretariat; E-mail: wcae2011@163.com, wcae2011@vip.csei.org.cn EWGAE30/ICAE7 EWGAE-2012: 30th European Conference on Acoustic Emission Testing will be held September 12-15, 2012, at Granada, Spain. This doubles as the 7th International Conference on Acoustic Emission with the support of the AEWG (Acoustic Emission Working Group), JCAE (Japanese Committee on Acoustic Emission) and GLEA (Grupo Latinoamericano de Emisión Acústica). It is hosted by the University of Granada and Antolino Gallego is the organizer. More information is available at http://www.ewgae.es/en/conference-area/welcome-letters.html JAE DVD: Vol. 1-24 (1982-2006) A DVD contains all the articles from the past 25 years of Journal of Acoustic Emission. Cost is $200 plus shipping of $8. Send an order to Acoustic Emission Group (address below). 2011 SUBSCRIPTION RATES Base rate (CD-ROM) for one year $111.00 CD-ROM + printed copy, US priority mail: $151.00 CD-ROM + printed copy, non-US priority mail: $160.00

Print copy only with priority shipping (US) $140.00 Print copy only with priority/air shipping (non-US) $149.00

Payment must be in U.S. dollars drawn on a U.S. bank. Bank transfer accepted at California Bank and Trust, San Fernando Valley Office (Account No. 080-03416470) 16130 Ventura Blvd, Encino, CA 91436 USA (Swift code: CALBUS66) PAYPAL payment is accepted. Inquire via e-mail. Back issues available in CD only; Inquiry and all orders should be sent to (Fax no longer available): Acoustic Emission Group PMB 409, 4924 Balboa Blvd. Encino, CA 91316 USA For inquiry through Internet, use the following address: aegroup7@gmail.com Editor-Publisher Kanji Ono Tel. (818) 849-9190 Publication Date of This Issue (Volume 28): 30 March 2011.

I-8

20th International Acoustic Emission Symposium hosted by Kumamoto University, held in Kumamoto, Japan, on November 16-19, 2010

JCAE Kishinoue Awards Kishinoue Awards of the Japanese Committee on Acoustic Emission were presented to Professors Gerd Manthei and Kanji Ono for their outstanding contributions to the field of acoustic emission on Nov. 18, 2010 at the 20th IAES at Kumamoto. Two student paper awards were accepted by their mentors, Profs. S. Wakayama and A. Vinogradov.

From L to R: Gerd Manthei, Kanji Ono, S. Wakayama and A. Vinogradov.

I-9

AE LITERATURE

Book on AE M. G. R. Sause, Identification of failure mechanisms in hybrid materials utilizing pattern recognition techniques applied to acoustic emission signals. ISBN: 978-3-86664-889-0, mbv-Verlag, Berlin, 305 pg. (2010).

This new book is a PhD thesis of Markus Sause and can be purchased via Amazon.de. The ".de" does matter, since Amazon.com is not able to find the book. Since most of the international community may struggle to find the book on a german webpage, the author offers to handle the book order. Send an email and transfer funds (59€ + shipping).

Dr. Markus Sause Institute for Physics Augsburg University Universitätsstr. 1 D-86159 Augsburg, Germany markus.sause@physik.uni-augsburg.de

Chinese Acoustic Emission Conference, 2009 Proceedings of 12th National AE Conference of China, July 31-August 3, 2009, Nanjing, China Compiled by Gongtian Shen

Technical Review The Development of Acoustic Emission Testing Technologies in China DAI Guang，SHEN Gong Tian（1） Progress and Status of Research on Acoustic Emission Sensors Calibration HE Longbiao，YANG Ping，LI Guanghai（11） Influences of the Microstructure on the Acoustic Emission of the Metallic Materials Han Zhiyuan，LUO Hongyun（18） Research on Applicability of Acoustic Emission Technology in Complex Environment BAI Lu，ZHANG Hui，LIU Yantao et al.（25） Applications of Acoustic Emission Technology on Composite Materials Damage Detection GONG Lihai，YANG Nengjun（30） Application and Development of Acoustic Emission Technology JIANG Alan，ZHAO Yinghua，ZHANG Liwei（37） Instrument & Signal Processing Development of Leakage Location Testing Instrument For Underground Gas Pipeline SHEN Gong Tian，LIU Shifeng，WANG Wei（47） Development of the AE System Based on the Special USB2.0 Acquisition Card LIN Dongxia，HU Bin，WANG Xudong et al.（54） The Design of A Novel Fiber Optic Acoustic Sensor System XU Zhihong，Fahard Ansari（64） Pattern Recognition of Acoustic Emission Signals from the Flaws and Damages in WPC and DSP Realization YUAN Zhe, LIU Yunfei, YIN Dongmeng et al.（72） Acoustic Emission Monitoring System Based Matlab LIU Lei, PAN Yongdong,LIU Wuxiang et al.（79）Applications of Uitrasionic Detection

I-10

Technology on Geophysical Test About Natural Gas Hydrate YE Yuguang, HU GaoWei, ZHANG Jian et al.（83） Numerical Simulation of Acoustic Emission Signal Produced by Hypervelocity Impact on Aluminum Plate LIU Wugang, HUANG Zehuan, LIU Yantao et al.（88） The Research of Time Delay Estimation Algorithm LIU Weidong, ZHANG Shen, XUN Guangshuai et al.（93） A Method of Comparing Sensitivity of Sensor Using Bottom Plate Testing DUAN Qingshu, MA Juan, ZHU qin et al（99） Research on Simulated Source in the Acoustic Emission Sensor Calibration HUANG Shan, LI Da zi, He Longbiao（107） Theoretical Analysis and Experimental Investigation on Sound Source Location for Thick-wall Pressure Vessel GUO Xiaolian, XU Yanting, LIU Fujun, et al（114） Study on pattern recognition of Crack Acoustic Emission Using Neural Network in Fatigue Testing of Fighter Aircraft GAO Xiaxia, LIU Jian Jun, Wang Wei（123） Acoustic Emission Signal Analysis Based on HHT For GFRP LI Wei, JIANG Peng, DING Lei et al.（130） Applications of Morphology Filter in Acoustic Emission Signal Processing For Rolling Bearing Faults HAO Rujiang MA Bingyu（136） Applications of Correlation Analysis in Locating Defect By Acoustic Emission Testing ZHAO Meiyun, LI Li, LI Xiujuan（144） Applications of Structure Acoustic Emission Based Integrated NDT in Full Scale Aircraft Fatigue test GENG Rongsheng（150） Investigation of Acoustic Emission Technique (AET) Application on Bridge Cranes Testing SHEN Gong Tian, WU Zhan Wen（158） Research on Acoustic Emission Signal of Active Defect in Crane Beam LI li, CHEN XianQian, ZHAO Meiyun, et al.（166） Acoustic Emission Testing of Crane Hanger in Power Plant CHEN XianQian, LI li, DENG Daijun, et al.（171） Study on Acoustic Emission Testing for Atmospheric Storage Tanks LI Guanghai, SHEN Gongtian, YAN He（177） Dependability of Acoustic Emission Inspection of Vertical Storage Tanks XU Yanting, WANG Yadong, XU Jiele, et al.（184） Preliminary Exploration About Acoustic Emission Position of Influencing Factors in Tank Bottom Testing

TIAN Yatuan, JIANG Shiliang（195）

On-line Detection and Safety Assessment of Dangerous Chemical Bottom of Tanks by Acoustic Emission FANG Jiangtao, GUO Chao, XIE Tao（201） Application of Acoustic Emission to Integral Panel Fatigue test LI Zhong, HUANG Huabin, SHI Lei, et al.（207） Research on Acoustic Emission Signal Regularity of Fatigue Damage For Birdge structure KONG De Lian, WANG Wen You, XU Feng Jin（211） Acoustic Emission Monitoring of Birdge Cable Rope in Fatigue Test LUO Wen Hai, XU Feng Jin（217） Research on Acoustic Emission Signal Feature of Fatigue Damage For Birdge structure YANG Jian Feng, WANG Wen You, KONG De Lian, et al.（225） Research on Acoustic Emission Sources Orientation of Jacket Offshore Platforms SUN Haijiao, LIN Zhe, Zhao Deyou（231） Acoustic Emission Testing of Backbone in Large Cable Car WU Zhan Wen, SHEN Gong Tian WANG Yong, et al.（239） A Brief Analysis on the Three Parts in the Acoustic Emission Monitoring of the Stability of Large Intrusion —An Inspiration from the Acoustic Emission monitoring of the Stability of Permanent Navigation Lock’s Side Slope HUO Zhen , CHEN Cui Mei（244） Acoustic Emission On-line Monitoring of Bead Weld Layer Crack For Hydrogenation Reaction Chamber JIANG jun, YI Rong（251）

I-11

Acoustic Emission On-line Testing of Ammonia Reactor LIU Jian Jun, Liu Shi Feng（269） Application of Acoustic Emission Testing to Spherical Tank LI Jianrong, YAO Jianping, DENG Zurong（274） Structure integrity evaluation and safe valuations of multi-layer high-pressure tanks according to TOFD and AE technology SHENG Shui Ping, YU Bing, Wang Bing（279） Research and Application the Method of On-line Acoustic Emission Periodical Test for Adsorption Tower ZHANG Ying, DAI Guang, LiWei（284） The Application of Acoustic Emission Technology on Coke Tower Inspection ZHANG Zhongzheng,Liang Hua, Gong Jianming, et al.（291） Large-sized Vessel Detecting Realized By Connecting AE Units Duan Qingru, Ma Juan, Zhu Qin, et al.（296） Full-life Evaluation of LPG Storage Tank Using Acoustic Emission Testing MAO Guojun, SHEN Jianmin, Sun Lei, et al.（301） Acoustic Emission Testing of Near Surface Crack Flaw on Spheric Tank LI Hong Gang, YE Wei Wen, PENG GO Ping, et al.（306） Acoustic Emission Testing Technique For Composite Pressure Vessels

LIU Zhejun, GE Li, WU Song, et al.（309） Research on Acoustic Emission Inspection of Gas Cylinders For the On-Board Storage of Compressed Natural Gas As a Fuel For Automotive Vehicles LIANG Hu, JIANG Jun, YI Rong, et al.（318） Research on Acoustic Emission Testing Technique of Cylinders For Tube Trailer WANG Yong, ZHENG Hui, LI Guang Hai, et al.（326） Research on On-line Inspection of Gas Cylinders For the On-Board Storage of Compressed Natural Gas As a Fuel For Automotive Vehicles During the Fatigue Test ZHANG Yao Feng, YI Rong, Jiang Jun , et al.（335） Research on Acoustic Emission Location Test For Gas Cylinders LI Li Fei, BO Ke, LI Bang Xian, et al.（341） The Acoustic Emission Testing About the Primary Entering Air Pipeline of Large Structure YAO Li（347） Research on Spectral Characteristics of Leakage Acoustic Emission Signals For Gas Pipeline QIN Xianyong, SHEN Gongtian, ZHANG Zheng, et al.（352） Identification and Analysis of Acoustic Emission Signal on Pressure Pipeline Leakage HAO Yongmei，WANG Kaiquan，ZHANG Changshun（360） Acoustic Emission Source Assessment of Defective In-service Pipeline SHI Ji Qing（366） Experimental Study on Leakage of Simulation Natural Gas Pipeline By Acoustic Emission Technique ZHOU Ning, CHEN Li,WANG Xiao-Yu, et al.（371） Inspection and Repairing of the High-pressure Accumulator LI Jian Rong, YAO Jian Ping (375） Application of Acoustic Emission Testing to Vessel For the 20m3Ammonia Storage ZHENG Yuan Wei, LIU Jian Li（380） Application Status of Acoustic Emission For Bop Testing ZHU XiangJun, LI Ping（384） Material Characterization Experimental Investigation on Feature of Acoustic Emission For 16MnR_0Cr18Ni9Ti Composite Armor Plate in the Damage and Fracture LIANG Zhi Gang, WU Yan Hong , DING Li Wei, et al. （390） The Early Warning of Acoustic Emission Technique in Figure Test FANG Jing, WU Hui Yong（397） Tensile Damage Analysis of Composite Unidirectional Samples Based on Acoustic Emission Inspection HAN Kunfeng, YANG Nengjun, LONG Xianhai（401） Acoustic Emission Real-time Monitoring of Laser Melting process For Nickel-base Alloy LIU Zhe Jun, GE Li, YUAN Jin Ping（406）

I-12

Elementary Experimental Analysis on Evaluation of the Penetration Depth of Electron Beam Weld By Acoustic Emission Testing SUN Chaoming, TANG Guangping, TANG Xing（414） Acoustic Emission Characteristic of Hr-2 Steel Weld joint during tension TANG Xing, ZHANG Wei Bin, XU Lai, et al.（420） Supported AE Analysis By FEM Waveform Simulation For Proportion of Super-elastic Martensitic Transformation Takeshi YASUDA, Baojun PANG, Hideo NISHINO, et al.（425） Random Damage Statistics and Their Modeling Using Maximum Entropy Principle Gang Qi, Steve F. WAYNE（432） Parameter Analysis of Acoustic Emission Signals About Magnesium Alloy Tensile Test LIANG Zhigang, WU Yanhong, JIANG Xu-xin, et al.（439） The AE Signature Analysis of PBX-9003 Satchel Column in the Course of Single Axle Secondary Compression XU Lai, ZAHNG Wei Bin, Tang Xing, et al.（446） Research on Acoustic Emission of Fibre Reinforced Concrete Materials Containing Fatigue Damage XU Feng Jing（450） Assessment of the Damage in Carbon Fiber Reinforced Concrete By acoustic Emission and Resistance Measurement Methods ZHAO Jiao, XU Zhihong（456） AE Characteristic Experiment Research of Process of Whole Stress-strain of Coal and Sandstone YANG Yongjie,

, ZHANG Yangqiang, LI Guibing, et al.（460） Experimental Study on Shear Behavior and AE properties of Concrete Corroded By Acidic Solution ZHANG Liwei, ZHAO Yinghua, JIANG Alan, et al.（465） Study on Acoustic Emission Characteristic of Common Defect in the process of welding SUN Guohao, BAI Mingqing, ZHAO Jiuguo, et al.（470） Brazilian test study on the Acoustic Emission Characteristic of PBX-159 ZHANG Wei-bin, TIAN Yong, TANG Xing, et al.（477）

J. Acoustic Emission, 28 (2010) 1 © 2010 Acoustic Emission Group

ESTIMATION OF CRUSTAL STRUCTURE IN HORONOBE AREA, HOKKAIDO, JAPAN, USING MULTIPLET-CLUSTERING ANALYSIS

HIROKAZU MORIYA1, KOICHI ASAMORI2*, ITARU KITAMURA3, HIKARU

HOTTA3, HIDEFUMI OHARA2** and TADAFUMI NIIZATO2 1) Graduate School of Environmental Studies, Tohoku University, Sendai 980-8579, Japan;

2) Horonobe Underground Research Unit, Japan Atomic Energy Agency, 432-2 Hokushin, Horo-nobe, Hokkaido, 098-3224, Japan; * Now at Tono Geoscientific Research Unit, Japan Atomic Energy Agency, Toki, Gifu, 509-5102, Japan; 3) Construction Project Consultants, Inc., 3-23-1 Takadanobaba, Tokyo, 169-0075, Japan; ** Now at Kumagai Gumi Co., Ltd., Tsukudo 2-1,

Shinjuku, Tokyo 162-8557, Japan

Abstract

Hypocenter locations of shallow earthquakes in Horonobe area, northern Hokkaido, Japan, which is near the convergent boundary between Okhotsk and Amuria plates, were determined to reveal the subsurface structure and the mechanisms of earthquake occurrence. The absolute source locations of 211 earthquakes, which occurred in the period from December 2002 to Sep-tember 2005, were determined; then, those earthquakes with similar waveforms were identified, and the source locations of 26 multiplet groups were relocated by using cross-spectrum and clus-tering analyses. The relocated hypocenters allowed two seismically active areas to be identified, at 10–20 km and 25–30 km depth. The earthquake locations indicate structures trending nearly N-S direction, and the structures causing repeated stick-slips at asperities, thus generating similar earthquakes. The relationships between magnitude and the distance to the next event, and be-tween magnitude and the time interval of event occurrences were investigated. The relationships between them suggest that earthquake swarms would be induced by continuous strain accumula-tion at asperities along faults and its subsequent release as a result of plate tectonic movements. A cutoff line could be also seen in the relationship between magnitude and the distance to the next event, suggesting that the distance between the source locations depends on the event mag-nitude. These results have given us the knowledge that the asperities on the delineated faults in-termittently release strain energy as similar earthquakes with magnitudes of less than about 3.0. Keywords: Subsurface measurement, Multiplet-clustering analysis, Repeating similar earth-quakes Introduction

The eastern edge of the Japan Sea approximately coincides with the convergent boundary be-tween Okhotsk plate and Amuria plates [1], and earthquakes are probably associated with stress accumulation and its release by the fault system in the crust owing to the movement of the plates. GPS and microearthquake data suggest that the plate boundary lies under the land surface in northern Hokkaido, although it is difficult to trace the plate boundary. The Hokkaido Nansei-Oki earthquake (M7.8), which occurred inside the mobile belt, suggests a complicated fault structure, in which the fault planes incline to the east in its northern part and to the west in its southern part. Therefore, it is reasonable to suppose the existence of multiple boundaries in a wide fault zone [e.g., 2]. On the other hand, Horonobe area, northern Hokkaido, is in an area that a recently in-stalled seismic network has shown to be characterized by microearthquake swarms [3]. Although the location accuracy in depth is not still confident, the estimated source depth of the earthquakes is usually shallower than 25 km, and the earthquakes are considered to occur along hidden faults in a stress-concentrated zone. Knowledge of the structures causing these earthquake swarms and understanding of their mechanisms are expected to provide information on the stability of the

2

crust and the relationships between plate tectonics and earthquakes in this region. Similar earth-quake events and high-resolution mapping techniques can be used to reveal these structures and earthquake mechanisms [4 - 12]. In this study, similar earthquakes in Horonobe area were relo-cated using a multiplet-clustering analysis and the subsurface structures and the mechanisms of earthquake swarms were evaluated.

Earthquake Data Collection

The earthquake events studied in this research were collected by the Hi-net system of the Na-tional Research Institute for Earth Science and Disaster Prevention (NIED) and the observatories of the Japan Atomic Energy Agency (JAEA) from 20 December 2002 to 30 September 2005 (Fig. 1). Most of the events had magnitudes of less than 3.0. Among the total of 4217 events located by Hi-net, we searched for those events whose P- and S-wave arrival times were detected at more than four observatories in the JAEA system. A total of 221 events were selected for analy-sis in this study. Similar earthquakes were identified within the selected earthquakes for applica-tion of multiplet-clustering analysis (MCA) (Figs. 2 and 3) [13]. Our aim was to investigate whether seismic clusters and structures could be identified after improvement of the location

Fig. 1. Locations of observatories around Horonobe, showing both those of JAEA and Hi-net. The dotted box denotes the area of the source locations shown in Fig. 3.

3

Fig. 2. Example of the waveforms of a multiplet.

accuracy. The vertical component signal at the JAEA downhole station, –97 m below sea level, was used to identify similar waveforms, because the P-wave waveforms at this observation site were clearer than those at other observation sites. The waveforms of all events were inspected by human eyes, and finally, 188 similar events were classified into 26 groups. Relocation of Similar Earthquakes Using Multiplet-Clustering Analysis

First, the absolute source locations of the selected similar earthquakes were determined using

the hypocenter determination software, Hypomh, based on a three-layered velocity structure shown in Table 1, where the focal mechanism was also estimated [3, 14, 15]. Figure 4 shows the absolute source location determined by Hypomh. The source locations were widely dispersed because of location errors of more than 1 km, and no structures could be identified. After deter-mination of the absolute source locations, MCA was applied to relocate the source locations [8, 13]. The MCA is a method for precise determination of microseismic event locations and is used to identify subsurface fractures and fracture networks. A multiplet is a group of microseismic events with very similar waveforms, despite different origin time, and is likely the expression of stress release on the same structure.

4

Fig. 3. Cross spectrum analysis, and estimated averaged coherence and delay.

Table 1. Velocity structure for estimation of source location [3].

The relative source locations of similar events can be determined with high resolution and accuracy by using the moving-window cross-spectrum analysis technique. Deduced seismic clus-ters, called multiplet clusters, are indicative of seismically activated structures, and the orienta-tions of these structures can be estimated using the seismic clusters even though the absolute lo-cations of the multiplet clusters cannot themselves be determined.

The time-window length for the cross-spectrum estimation was 1 s and that for the fast Fou-

rier transform was 2.56 s, where the frequency resolution was 0.039 Hz. The frequency range for the time-delay estimation was 5–15 Hz, and the mean value of the time delays estimated for 10 different moving time windows was used to estimate differential times.

5

Fig. 4. Source locations of analyzed earthquakes, where the locations are determined by “Hy-pomh”.

After the relocation of the events in each group, typical events were selected from each group, and the similarity of the waveforms was evaluated within the selected representative events. Events with similar waveforms, but with insufficient similarity to have been classified into the same group by the initial classification, were selected, and a clustering analysis was applied to determine the relative positions of the clusters.

The clustering analysis method is used to estimate the relative location of multiplet clusters

by detecting phase differences between similar stacked events. The representative events were selected from each group, and the cross-spectrum analysis was applied to estimate the relative travel time delays. After the determination of relative source location of representative events, the other events in each group were moved to the new locations following the representative events. Through the MCA, the precise relative source location of similar events can be deter-mined within each multiplet group and also determine the relative positions among the multiplet clusters. The frequency band for the cross-spectrum analysis was the same to that in estimation of relative source locations. Figure 5 shows the source locations after the MCA and the fault plane solutions assuming double-couple source mechanism and estimated using the P-wave first motions at stations.

6

Fig. 5. Source locations of analyzed earthquakes, where the cross denotes the locations by “Hy-pomh” and the closed circle and the hexagon denote the source locations after relocation. Discussion

According to the estimated source locations, the depths are less than 30 km (Fig. 5), and two

seismically active areas (areas A and B) were identified. The seismic sources in area A were lo-cated at around 30 km depth, and those in area B were shallower, at 10–25 km depth.

Tamura et al. (2003) [3] imply the existence of earthquakes deeper than 20 km. However, we

think that the absolute source location, especially the depth, should not be discussed in Fig. 5, because the station correction is not considered and the coverage of seismic network is not enough to determine the confident velocity structure for the event locations. We doubted the long period events for the reason of deep events. We have investigated the source locations of long period events, which are observed since 2000 and recorded on catalogue in this area, and con-

7

firmed that the analyzed events in this study are different from the long period events. On the other hand, the result of MCA is confident because the source location is the relative locations, and the influence of velocity structure in determination of source locations is smaller. The aver-age of RMS in the relative locations is from 0.16 ms to 0.09 ms, and which corresponds to 0.8 km and 0.45 km in distance if the wave velocity is assumed to be 5 km/s. Therefore, we do not dis-cuss the absolute location of the sources and the structures.

Fig. 6. Plot of the distance to the next earthquake as a function of the magnitude of the previous event for each consecutive pair of similar earthquakes.

Because similar events were observed, there must exist unstable structures with asperities

causing repeated stick-slips and continuous stress loading enforcing the repeated slips. The nodal planes (Fig. 5) suggest steeply inclined structures oriented NNW to NNE, and the P-axes are ori-ented nearly NNW-SSE and NNE-SSW. These results indicate that compressive stresses ori-ented nearly E-W are acting on faults with steep inclinations. Considering that Horonobe area is at the convergent boundary between the Okhotsk plate and Amuria plates, we infer that the dis-trict is under continuous stress loading in a nearly West to East direction, and strain energy ac-cumulates on the fault planes as a result of the loading. This interpretation is supported by movements of the land surface detected by GPS. Therefore, it is reasonable to infer that the groups of similar earthquakes are expressions of the release of strain energy at faults oriented nearly N-S, and that the faults are subjected to compressive stress in an E-W direction.

The relationship between the magnitude of each event and the distance to next event (Fig. 6)

shows a cutoff magnitude (shown by a broken line), although the range of magnitude is limited (0 to 3.0). The distance to the next source location of similar earthquakes is more than several hundreds of meters, and the distance between event locations is larger than the expected source

8

radius of the earthquakes, suggesting that the distance between source locations depends on the event magnitudes and that the source areas of similar earthquakes do not overlap. It is remark-able that a similar phenomenon has been observed along faults in California, where multiplets have been observed and streaks of earthquakes identified [6]. In the case of creeping faults, lock and release at asperities generate repeated earthquakes. Our results suggest that asperities release strain energy, and no more shear slip is induced at the same asperities. This means that strain en-ergy accumulates at several asperities distributed along the fault because of the regional stress loading, and that the strain energy is released at the different asperities one after the other.

Figure 7 shows the relationship between magnitude and the time interval between the occur-

rence of similar events for the earthquakes in areas A and B. In area A, the maximum magnitude was nearly 2.0, and it did not vary in relation to time intervals of more than 10 hours. On the other hand, in area B, the maximum magnitude increased with the time interval up to magnitude 3.0.

These results suggest that the maximum values of the strain energy release rate and magni-

tude are limited in area A, where the size of asperities and the frictional strength of asperities are limited. On the other hand, the asperities in area B seem to have different sizes or different fric-tional strength, and the range of released energy is extensive in comparison to that of the asperi-ties in area A. This difference between the two areas could not be definitively explained, but we think that it relates to the size and strength of asperities and also to conditions around the faults, such as temperature and rock properties changing with depth. These results give us the key knowledge that the asperities on the delineated faults intermittently release strain energy as simi-lar earthquakes, and that the occurrence of similar earthquake, such as relationship between magnitude and time interval, is different depending on the seismic clusters.

Fig. 7. Magnitude and time interval of event occurrence for each consecutive pairs of similar earthquakes. (a) Area A, (b) Area B.

9

Summary In this paper, similar earthquakes in Horonobe area were relocated by multiplet-clustering

analysis to reveal the mechanisms of earthquake swarms in this district. The seismic clusters re-vealed after the relocation allowed two seismically active areas to be identified, at 10–20 km and around 25–30 km depth. The relationships between magnitude and the distance to the next event, and magnitude and the time interval of event occurrences were investigated. A cutoff line was seen in the relationship between magnitude and distance to next event, suggesting that the dis-tance between the source locations depends on the event magnitude and that the source areas of similar earthquakes do not overlap. The relationships between them have suggested that earth-quake swarms would be induced by continuous strain accumulation along faults and its subse-quent release as a result of plate tectonic movements. These results have given us the key knowledge that the asperities on the delineated faults intermittently release strain energy as simi-lar earthquakes, and that the features of similar earthquake occurrence, such as relationship be-tween magnitude and time interval are different depending on the seismic clusters.

Acknowledgement

The earthquake waveform data was downloaded from the web site of Hi-net owned by Na-

tional Research Institute for Earth Science and Disaster Pretension (NIED), Japan, and the in-formation on the source location was used from the database of Japan Meteorological Agency (JMA). The author (TM) thanks NIED and JMA for providing important waveform and source location data. References 1) G. F. Sella, T. H. Dixon and A. Mao, REVEL: A model for Recent plate velocities from space

geodesy, J. Geophys. Res., 107 (B4)(2002), 10.1029/2000JB000033. 2) M. Kasahara, Chikyu Monthly, 20 (1998), 16-21. (in Japanese) 3) S. Tamura, M. Kasahara and T. Moriya, The micro-seismicity and crustal structure in the

northern part of Hokkaido, inferred from temporal observation, J. Seism. Soc. Japan, 55 (2003), 185-193. (in Japanese)

4) G. Poupinet, W. L. Ellsworth and J. Fréchet, Monitoring velocity variations in crust using

earthquake doublets: an application to the Calaveras fault, California, J. Geophys. Res., 89 (1984), 5719–5731.

5) H. Moriya, K. Nagano and H. Niitsuma, Precise source location of AE doublets by spectral

matrix analysis of triaxial hodogram, Geophysics, 59 (1994), 36–45. 6) D. Schaff, G. C. Beroza and B. E. Shaw, Postseismic response of repeating aftershock, Geo-

phys. Res. Lett., 25 (1998), 4549–4552. 7) A. M. Rubin, G. Dominique and J. Got, Streaks of microseismicities along creeping faults,

Nature, 400 (1999), 635–641. 8) W. S. Phillips, Precise microearthquake locations and fluid flow in the geothermal reservoir at

Soultz-sous-Forêts, France, Bull. Seism. Soc. Am., 90 (2000), 1, 212–228.

10

9) F. Waldhauser and W. L. Ellsworth, A double-difference earthquake location algorithm: Method and application to the North Hayward fault, California, Bull. Seism. Soc. Am., 90 (2000), 6, 1353–1368.

10) C. A. Rowe, R.C. Aster, B. Borchers and C.J. Young, An automatic, adaptive algorithm for

refining phase picks in large seismic data sets, Bull. Seism. Soc. Am., 92 (2002), 1660-1674. 11) T. Matsuzawa, N. Uchida, T. Igarashi, T. Okada and A. Hasegawa, Repeating earthquakes

and quasi-static slip on the plate boundary east off northern Honshu, Japan, Earth Planets Space, 56 (2004), 803–811.

12) N. Uchida, A. Hasegawa, T. Matsuzawa and T. Igarashi, Pre- and post-seismic slow slip on

the plate boundary off Sanriku, NE Japan associated with three interplate earthquakes as es-timated from small repeating earthquake data, Tectonophysics, 385 (2004), 1–15.

13) H. Moriya, H. Niitsuma and R. Baria, Multiplet-clustering analysis reveals structural details

within the seismic cloud at the Soultz geothermal field, Bull. Seism. Soc. Am., 93 (2003), 1606–1620.

14) N. Hirata and M. Matsu’ra, Maximum-likelihood estimation of hypocenter with origin time

eliminated using nonlinear investigation technique, Phys. Earth Planet. Inter., 47 (1987), 50-61.

15) P. Reasenberg and D. Oppenheimer, U.S. Geol. Surv. Open File Rep., 109 (1989), 85-0739.

J. Acoustic Emission, 28 (2010) 11 © 2010 Acoustic Emission Group

ACOUSTIC EMISSION TESTING - DEFINING A NEW STANDARD OF ACOUSTIC EMISSION TESTING FOR PRESSURE VESSELS

Part 2: Performance analysis of different configurations of real case testing and recom-mendations for developing a new guide for the application of acoustic emission

JOHANN CATTY

CETIM, 52 Avenue Félix Louat, 60304 Senlis Cedex, France Abstract

Several regulatory rules, such as the Best Practices Guideline (GBP) [1], exist for acoustic emission (AE) testing of pressure vessels in France and the rest of the world and allow AE test-ing based on two techniques (zonal location and planar source location methods). However, the analysis criteria of data recorded during the testing lack adequate basis. This work highlights inconsistencies in analysis methods defined in these guides or codes based on modelling calcula-tions in Part 1 of this study [2], while this paper shows, from a real case (AE testing of a 2000 m3 spherical storage tank), that the results of an AE test cannot be reproducible due to a lack of strictness in the application rules of AE. Thus, depending on the testing configuration used, some emissive defects can be detected or missed. This study may also be used as a basis for defining a new AE testing standard specifically and quantitatively defining a methodology of analysis based on a different approach from those used currently. Today, the CETIM may apply this new testing methodology based on significant feedback enabling a greater reproducibility and sensitivity of AE testing. Introduction

Acoustic Emission is especially useful in testing of pressure vessels, enabling global and rapid testing of large structures, significantly reducing maintenance time and shutdown of facili-ties. Methods have followed several regulatory rules, codes or standards that have been created, defining the general application rules of this technique. AE testing can be applied according to two techniques (zonal location and planar source location by triangulation or planar method, in short), without any major impact on the analysis criteria of data recorded during the testing. In-deed, the criteria are primarily derived from the zonal testing method, which has limitation in terms of accuracy of analysis, limited to very basic criteria of counting signals collected by the sensors.

In Part 1 of this study [2], the author highlighted, from simulation calculations, that there were significant differences in performance between these two techniques (up to a ratio 5 to 7). The effect of the acquisition threshold has been also quantified. Using planar location method, the use of the amplitude correction has been evaluated, and considered as a good tool for increas-ing the accuracy of the result. The study presented in this article, based on a real industrial case treated by CETIM, aims to highlight the differences in results of testing applied on the same structure with different configurations, all consistent with current rules. We will answer to these questions:

What is the actual performance of the two testing techniques? Which level of accuracy can we affect, or, what uncertainty should we associate to these cri-

teria? What is the effect of operating conditions on the result of the testing?

12

A. Analysis of Testing and Data Processing Methodology: BPG - Real industrial case A.1 Definition of the case studied – context

The performance for both techniques used will be compared in accordance with the recom-mendations from GBP (Best Practices Guideline) as regulated in France. This analysis uses a specific application case; a spherical storage tank with a 45-mm thick unalloyed steel wall, is externally painted. The AE wave attenuation curve is shown in Fig. 1.

Fig. 1: Attenuation curve obtained using the Hsu-Nielsen source on 45-mm thick, unalloyed steel, painted. The frequency of the AE transducers is near 200 kHz.

Based on the GBP recommendations, the maximum allowed distances between sensors for this case are: • For zonal location, the maximum authorized distance between sensors is 1.5 times [Distance

at the assessment threshold = 50 dBAE maximum]; or approximately 1.5 x 6 = 9 m. • For planar location, the maximum authorized distance between sensors of a single mesh

(maximum acquisition threshold of 50 dBAE), is equal to the distance to the acquisition thres-hold + 6 dB; approximately 6 m.

We analyzed real AE data of pressurization testing of the spherical tank, first using the planar configuration, and second by the zonal configuration. Table 1 gives the following main charac-teristics:

Table 1: Main characteristics of the two configurations of testing.

Configuration Number of Sensors Maximum Distance between sensors

Zonal Location 22 About 8.6 m Planar Location 77 About 5.2 m

A.2 Location tests results from Hsu-Nielsen source A.2.1 Test analysis: In order to assess the two testing configurations, it is interesting to analyze the performance in terms of the detectability of Hsu-Nielsen sources (standard AE source). These sources are generated prior to the pressure test on the singularities of the structure. These sources are generated on the vertical and circular welds (location tests 1 and 3), and on the welds of the supports (location test 2). For each case, we calculate the number of detected events (zonal lo-cation), and the number of localized events (planar location). We note that the real number of generated sources (pencil lead breaks) is very close to the number of localized events in planar testing configuration. Results are shown in Figs. 3-5.

13

In the case of location test 2, the difference of performance is significant between the two testing configurations. Only 2 events are located with the zonal testing configuration while 210 events are located with the planar testing configuration. This is because the presence of obstacles to the propagation of the AE waves (support of the tank) disrupts the detection of events by sen-sors close enough in theory. Therefore, only two events were collected by 3 or more sensors. However, the zonal testing configuration meets the minimum requirements for detection, since it receives each Hsu-Nielsen source by at least one sensor (zonal location).

a b Fig. 2a: Overview of planar configuration; 2b: zonal configuration.

a b

Fig. 3a: Result of location test 1, planar test configuration; 3b: zonal test configuration. Results of these tests are summarized in Table 2. Each entry indicates the number of detected or located events. The threshold level used are 50 dB or 65 dB. These tests show vast differences in terms of detection and level of information between the two test configurations. From these tests conducted on a real case, the zonal test configuration will be able to locate only 18% of the Hsu-Nielsen AE sources on average, against 100% for the planar testing configuration.

14

We also observe that in the case of planar configuration, the number of detected events ex-ceeds the number of generated sources. This can be caused either by the detection of pencil-lead break rebounds, or through contacts caused by the operator during testing.

It is, however, important to remember that these two test configurations are in accordance with the rules defined in the ‘Best Practices Guideline’ (GBP).

a b Fig. 4a: Result of location test 2, planar test configuration; 4b: zonal test configuration.

a b

Fig. 5a: Result of location test 3, planar test configuration. 5b: zonal test configuration.

A.2.2 Comparisons with simulation calculations: Referring to simulation calculations developed in Part 1 [2], we find: • A location rate of 43% of the Hsu-Nielsen AE sources in the zonal test configuration, • A location rate of 100% in the case of planar test configuration.

15

Table 2: Number of detected and located events according to test configurations for 3 tests.

Configuration Zonal Configuration Number of Detected Events, A>=50 dBAE

Zonal Configuration Number of Located Events, A>=65 dBAE

Planar Configuration Number of Detected Events, A>=50 dBAE

Planar Configuration Number of Located Events, A>=65 dBAE

Location Test 1 250 85 360 210

Location Test 2 245 2 350 210

Location Test 3 375 50 500 350

To illustrate these calculations, we observe in Fig. 6 the minimum locatable amplitudes in

zonal test configuration (with the distance between sensors of ~8.6 m). When these amplitude values are less than 98 dBAE (approximate magnitude of a Hsu-Nielsen source), this means that the Hsu-Nielsen source is locatable. When this value is greater than 98 dBAE, this means that the Hsu-Nielsen sources are no longer locatable.

We notice that the Hsu-Nielsen sources are locatable at the center of each mesh, and for an ideal configuration where no obstacle disturbs the propagation.

Fig. 6: Simulation of minimum locatable amplitudes, zonal test configuration.

A.2.3 Conclusions: 1. These location tests carried out on a real structure show that the two test configurations pro-duce different level of detection performance. Although the zonal test configuration is able to detect any Hsu-Nielsen source, it was able to locate only 18% of sources against 100% for the planar testing configuration. Nevertheless, these two configurations are in accordance with the rules defined in GBP.

2. There is a good correlation between simulation and tests. Indeed, the simulation, which pre-dicted a location rate of around 40% for zonal test configuration does not take into account any obstacles to the propagation of AE waves. This value of 40% is one actually measured on the test location test 1, the only one for which no obstacle does not disturb the propagation.

16

A.3 Results of AE monitoring during pressurization of the spherical tank The AE test requires pressurization, and its schedule is shown in Fig. 7. The analysis criteria come from GBP. First, the zonal location analysis is performed for the two test configurations. In the second step, we analyze the located AE activity.

Fig. 7: Pressure cycle applied to the spherical tank.

A.3.1 Zonal location analysis: The analysis criteria used in this section are based on GBP [1] and are: • AE events of amplitude greater than or equal to 50 dBAE (N1S). This criterion is calculated

on the whole duration of the test, • AE events of amplitude greater than or equal to 65 dBAE (N2S). This criterion is calculated

on the whole duration of the test, • AE recorded during the constant pressure stages. We consider the AE events of amplitude

greater than or equal to 50 dBAE (N3S). This criterion is calculated from 2 minutes after the beginning of the plateau,

• Changes in activity and intensity during the test. The activity (number of AE events) and intensity (energy) are evaluated throughout the test.

• Felicity ratio calculated on the second stage of pressure rise. Tables 3a (zonal test configuration) and 3b (planar test configuration) include the values of an-alysis criteria for all areas (an area is the part of the structure covered by a sensor). Several observations can be drawn from these two tests: • The total number of recorded events (≥50 dBAE) is about 1300 for zonal test configuration

against about 2500 for the planar test configuration, or about twice. • We can translate that into saying that half of the information disappears from this type of

analysis using a zonal test configuration: It is obvious that the higher the number of sensors, the better the coverage level of the structure. Therefore, we will have more chance to per-ceive AE events.

• If we analyze these results further, we note that some areas classified as category 2 with the planar test configuration disappear from this category with the zonal test configuration: For example, the area covered by the sensor c2 (Table 3b), which reported 292 events over 50 dBAE is classified in category 2.

17

• With zonal test configuration, we can assume that some of this activity is 'recovered' by the sensor c1, perhaps by the sensors c17 and c19, but much of this information is lost (more-over, none of the sensors, c1, c17 and c19 are classified in category 2);

• The classification criteria defined in the GBP are almost identical for both cases. For criteria N1S and N2S, values boundary between Level 1 and Level 2 are higher in the case of the zonal location. It is hardly justifiable, since two opposing effects should be taken into ac-count: - The average distance from a source to the sensor is higher in the case of the zonal con-

figuration (then, the attenuation is higher, therefore less probability to record signal with amplitude exceeding 50 dBAE)

- However, the areas being more extensive in this case, the number of events is potentially higher for each zone.

• The positions of the sensors used in the case of a zonal test configuration more strongly in-fluence the result: In the case studied, if the sensors 2 and 4 were kept in place of the sensors c1 and c3, the sensor c2 area would have been classified in category 2 or more. Chance has a greater impact in the case of a zonal test configuration.

A.3.2 Comparison with simulation calculations: The simulation calculations developed in Part 1 [2], assuming an amplitude distribution of AE sources between 55 and 115 dBAE, showed that 34.6% of sources were detected in the case of a zonal test configuration, against 59% (ratio 1.71) for a planar testing configuration. In the real case studied, we find this order of difference be-tween the two configurations (1300 vs. 2500 events; ratio of 1.92). This shows that the hypoth-esis used in the simulation calculations are quite close to reality.

Table 3a: Classification table, according to GBP, in the case of the zonal test configuration.

18

Table 3b: Classification table, according to GBP, in the case of planar test configuration.

19

Table 3b (continued)

A.3.2 Planar location analysis: The planar location analysis is based on the description of the events that have been located. Only concentrations of localized events (clusters) are taken into account. Concentration thresholds are defined in the GBP [1]. The results of location analysis obtained in the two test configurations are illustrated in Figs. 8a to 8d below. First, we observe that the number of localized events is significantly different with the two configurations:

Fig. 8a: Upper hemisphere, zonal test configuration.

20

Fig. 8b: Lower hemisphere, zonal test configuration.

Fig. 8c: Upper hemisphere, planar test configuration.

• On the lower hemisphere, 37 events for the zonal test configuration vs. 139 with the planar

test configuration, • On the upper hemisphere, 45 events for the zonal test configuration vs 144 with the planar

test configuration. Thus, less than 30% of events located with the planar test configuration are observed with a zonal test configuration. On the other hand, if we observe carefully the concentration areas (clus-ters) identified in the 2 cases, we note:

21

• On the lower hemisphere, 1 cluster for the zonal test configuration vs. 3 with the planar test configuration; the cluster near c2 (12 events), classified in category 2 with the planar test configuration, does not appear with the zonal test configuration.

• On the upper hemisphere, 1 cluster for the zonal test configuration vs. 2 with the planar test configuration.

Fig. 8d: Lower hemisphere, planar test configuration.

A.3.3 Conclusions: If we make a synthesis of both analyses (zonal and planar), we reach the con-clusions summarized in Table 4:

Table 4: Summary of analysis performed in both test configurations.

Configuration Zonal test configuration

Planar test configuration

Areas classified as Category 2

c35, c39, c41, c65 and c77

c2, c21, c25, c36, c40, c41, c65 and c77

Clusters classified as Category 2 - 1 cluster (near c2)

If the differences in this table seem minor, the consequences in terms of further investigat-ions on them are very different: In fact, for any area or cluster in category 2 or more, further in-vestigations are recommended. Therefore, considering the differences in distances between sen-sors for both test configurations, it would be necessary to conduct these investigations on:

• 5 areas of 58 m2 or 290 m2 for the zonal test configuration • 8 areas of 21 m2 or 168 m2 for the planar test configuration (in fact, the surface would be

lower, since the sensors are closer to each other near the poles of the spherical tank). On the other hand, it is important to note that the information from these two tests is not identi-cal: For example, areas c2 and c21, classified in category 2 with the planar test configuration (including a cluster classified in category 2, near c2) will not undergo any further investigation after the AE inspection carried out with the zonal test configuration.

22

A.4 Effect of acquisition threshold on the results of AE monitoring Based on the case discussed above, we adopt, in this section, an acquisition threshold of 40 dBAE (instead of 50 dBAE in the previous sections). The test configuration is the one that uses 77 sensors (planar test configuration). Anxious to achieve the testing of this spherical tank in the best detection conditions, CETIM had adopted this acquisition threshold value, allowing a better detection without being disturbed by environmental conditions. The analysis follows the same pattern as in the previous section, under GBP requirements: First, zonal location analysis, and second, analysis of the located AE activity. A.4.1 Zonal location analysis: The zonal location analysis is identical to that conducted in A.3.1 for the planar test configuration. Indeed, the zonal location analysis, as defined in the GBP, takes into account only the events of magnitude greater than or equal to 50 dBAE. The conclusions of the analysis are identical to those in A.3.1. However, we note that approximately 11,500 events were recorded at acquisition threshold of 40 dBAE, vs. about 2500 for a threshold of 50 dBAE. A.4.2 Planar location analysis: The analysis in planar location provides different results from those described in A.3.2, because the location capability is enhanced by the lower acquisition threshold used. Figures 9a and 9b illustrate the located AE activity on both hemispheres. In this case, the number of located events is: • On the lower hemisphere, 460 vs. 139 AE events observed with a threshold of 50 dBAE, and

37 with the zonal test configuration, • On the upper hemisphere, 440 vs 144 events observed with a threshold of 50 dBAE, and 45

with the zonal test configuration. Therefore, lowering the threshold from 50 to 40 dBAE will have multiplied by 3 the number of located AE events; In this configuration, we get about 10 times more located events than zonal test configuration.

The information obtained is more extensive with a threshold of 40 dBAE: 24 clusters of more than 5 events are identified, vs. 5 with a threshold of 50 dBAE. The information obtained with a higher threshold is included at a lower acquisition threshold, but this information is amplified. For example, the cluster located near c2, which does not appear with the zonal test configuration, is identified as having 12 events with the planar testing configuration (50 dBAE threshold), and becomes a region with 55 events by lowering the acquisition threshold to 40 dBAE.

It is important to note that using a lower acquisition threshold allows more explicit informa-tion: For example, the area of sensor c21, classified in category 2, did not have any cluster identi-fied with a threshold of 50 dBAE. Lowering the threshold has highlighted a cluster in this area. Another example: a located high activity area (15 events) appears between sensors c16, c27 and c28. This active region was not detected with a threshold of 50 dBAE, and it did not belong to a category-2 area. A.4.3 Conclusions: The acquisition threshold is a fundamental parameter of an AE test. The ex-ample of the real case shows that lowering the threshold from 50 to 40 dBAE amplifies the infor-mation picked up by a factor of three. A volume of information three times larger allows greater accuracy in diagnosis, and therefore a smaller area of investigation for possible additional test-ing. The effect of the acquisition threshold is not taken into account in GBP, which imposes a 'maximum' acquisition threshold, not to exceed. The zonal analysis is still limited to AE events

23

Fig. 9a: Upper hemisphere, planar test configuration (Acq. Thr. = 40 dBAE).

Fig. 9b: Lower hemisphere, planar test configuration (Acq. Thr. = 40 dBAE).

of magnitude exceeding 50 dBAE, and analysis of located AE events has the same criteria, re-gardless of the threshold used for acquisition. A.4.4 What is the interest of increasing the sensitivity of a testing? Additional checks (penetrant testing, magnetic particle testing, ultrasonic testing) were made following the AE results (real-ized with the planar test configuration, with a 40 dBAE acquisition threshold), on a majority of the emissive regions. It was found that 8 of the tested areas (clusters) showed significant indica-tions (acceptable for the most part according to codes, unacceptable for the others). None of

24

these regions would have been the subject of further investigations following the AE test carried out with the zonal test configuration (22 sensors). Only one of these regions would have been investigated following the AE test with the planar test configuration (77 sensors), with an acqui-sition threshold of 50 dBAE. This example shows that an application guide such as GBP in France, or other guides sub-stantially equivalent in other countries do not allow reproducibility of AE testing, authorizing very different test configurations, whose differences are not taken into account in the analysis and interpretation of data recorded. So, being in accordance with such a guide does not guarantee the relevance of the result. A.4.5 Comparisons with simulation calculations The simulation calculations developed in Part 1 [2], assuming a distribution amplitude of AE sources between 55 and 115 dBAE, showed that the effect of lowering the acquisition threshold from 50 to 40 dBAE for a planar test configuration, reduces the rate of non-detected sources by 3, and multiplies by 2 the rate of located AE sources. In the cases studied, we observed that the number of AE events (= detected sources) increased from 2500 to 11500, a ratio of about 4.6. The number of located AE sources has been multiplied by 3. Even if the results do not completely fit together, this shows that the assumptions used in the simulation calculations are however quite close to reality. B. Pathways of Progress in the AE Test Implementation - Recommendations for Develop-ing a new Guide for AE application B.1 The necessity of a planar testing configuration The analysis developed from numerical simulations or from real example (see preceding sec-tion 'A') shows that the first parameter influencing the testing is the configuration; that is to say, the number and arrangement of the sensors coupled to the structure. On this point, too much flexibility is allowed under the current guideline because only ‘minimum’ rules are imposed. On the other hand, both techniques (planar location mode and zonal location mode) are allowed, with almost no impact on the methods and analytical criteria, which are the same in both cases.

In order to make guideline meaningful, we should consider two new rules. The first rule needed is to reduce drastically the choice in the configuration of testing. That is to say, the dis-tances between sensors have to be justified by demonstrating and overcoming the impact of the configuration adopted on the 'coverage level' of the structure. Given the large difference in per-formance between the zonal and planar test configurations, it seems reasonable to recommend, if not impose, the second rule of using the planar test configuration. By the addition of these two rules, AE tests will be more reliable, precise and reproducible. B.2 Reference source The term 'planar test configuration', which means that we are capable of carrying a planar location, is not sufficient to establish a strict framework. Indeed, the ability of a test configura-tion to locate is evaluated for a given source type, characterized by power or amplitude. Thus, the Hsu-Nielsen source (0.5 mm-2H) is now taken as a reference.

The comparison of data from simulation (see Part 1 [2]) with tests on real structures (unal-loyed steel structure) tends to show that a distribution amplitude source centered on 85 dBAE is fairly close to experimental data. This shows that taking as reference the Hsu-Nielsen source (0.5

25

mm-2H), amplitude source between 95 and 100 dBAE is a minimum requirement of 'good super-vision' of a trial. A source of lower energy, whose amplitude is closer to 85 dBAE could be taken as new reference. The integration of these requirements in a practice guide then will involve, from the end-users to justify its test configuration, firstly by providing the attenuation curve taken into account, and secondly by demonstrating, either by simulation or by experiment, that the tested structure is completely covered. The advantage of simulation is to highlight and iden-tify the less-well monitored area, and to adopt compensatory measures if necessary.

B.3 Simulation: A tool for evaluating a test configuration The simulation also enable an evaluation of ‘performance’ of a given test configuration by calculating the percentage of detection and location of a reference population of AE sources (e.g., amplitude centered on 85 dBAE, between 55 and 115 dBAE). Simple and synthetics criteria, such as the rate of detected sources, the rate of located sources, the overall error of amplitude measurement could be calculated. By the use of mapping (see e.g., Figs. 11a and b), it allows visualizing the sensitivity level of a testing. This assessment enables: • To compare quantitatively the performance of AE tests. Customers, users of this technology,

and organizations using the results of a testing on one hand can take into account the quality level of a testing, and also ask, following the criticality of the tested device, a minimum re-quirement, more precisely determined by these criteria.

• To adjust classification criteria to the performance level of the test configuration adopted. Indeed, currently, no guide, standard or code does define the classification rule incorporating the 'coverage' testing.

B.4 The advantage of the amplitude correction The calculations developed in Part 1 [2] showed that taking into account the information from the location calculation, correcting the measured amplitude to reach the amplitude at the source allows appreciable gain in terms of accuracy of information: In fact, any located source is measured, in theory, without any error. In the case studied [2], for a planar test configuration, the average error of amplitude measurement on the detected AE sources would decrease from 26.1 to 14.7 dBAE (amplitude distribution of sources centered on 85 dBAE). This implies again that the test configuration should allow planar location as complete as possible, prohibiting the use of a zonal test configuration. If we consider the case studied in Section A, we can observe (Figs. 10a and 10b) that the contribution of the amplitude correction modifies the information: For location tests with Hsu-Nielsen sources (Fig. 10a), the amplitudes are measured between 50 and 90 dBAE (average value of about 70 dBAE). After correction, the amplitudes are between 80 and 107 dBAE (average value of 94 dBAE). We should find in theory the origin amplitude of a Hsu-Nielsen source, which is approximately 100 dBAE. The amplitude correction can result in an average value close to the theory, and concentrates all the sources in the range of 27 dB. Without correction, the amplitudes are spread over an interval of 40 dB, centered at 70 dBAE, which is 30 dB below the real value. If we analyze the AE test results of this spherical tank during the pressure test (Fig. 10b), the amplitude correction changes the perception that we may have observed on the emitting regions: For example, the region 'D' has a maximum amplitude of 76.1 dBAE without amplitude correc-tion, and the region 'K' 75.7 dBAE. After amplitude correction, the region 'D' has a maximum amplitude of 91.5 dBAE, and the region 'K' 104.1 dBAE. Their ranking (in terms of maximum

26

Fig. 10a: Tests of location from Hsu-Nielsen source.

Fig. 10b: (bottom) Results of the pressure test. Comparison of measured amplitudes and cor-rected amplitudes (planar test configuration). amplitude) is reversed. The information is much more relevant, without demanding excessive effort of analysis, in the case where the testing is implemented for a planar location.

27

In order to summarize the study developed in this article, and to show the great disparity in results of AE testing allowed by the existing guides, such as GBP in France, we can take the case of two active regions:

Region K: - In the case of zonal test configuration, it is absolutely not detected and no inspection re-quired after test. - In the case of planar test configuration, with an acquisition threshold of 50 dBAE, it belongs to category 2 (area of the sensor c40), and a cluster of 6 events appears (category 1). - In the case of planar test configuration, with an acquisition threshold of 40 dBAE, it belongs to category 2 (area of the sensor c40), and it is considered as an emissive region (cluster of 17 events), classified in category 2 (with a ranking 4 / 11 of the emissive regions). - In the case of planar test configuration, with an acquisition threshold of 40 dBAE, and the use of amplitude correction, it becomes the 2nd most emissive region. This region, corresponding to a section of a vertical weld, has been inspected by ultrasound after the AE test. This revealed the presence of 3 internal indications, unacceptable under the applicable standards.

Region D: - In the case of zonal test configuration, it is absolutely not detected and no inspection re-quired after test. - In the case of planar test configuration, with an acquisition threshold of 50 dBAE, it is abso-lutely not detected and no inspection required after test. - In the case of planar test configuration, with an acquisition threshold of 40 dBAE, it belongs to an emissive region (cluster of 5 events), classified in category 1 (with a ranking of 3 / 11 of the emissive regions). - In the case of planar test configuration, an acquisition threshold of 40 dBAE, and the use of amplitude correction, its ranking changes from 3 / 11 to 7 / 11. This region, corresponding to a section of a vertical weld, has been inspected by ultrasound after the AE test. This revealed the presence of 5 indications acceptable under the applicable standards.

These conclusions demonstrate first the benefit of using a planar test configuration, and sec-ondly the contribution of applying the amplitude correction in the diagnosis. B.5 A correction factor: the rate of local coverage Although the planar test configuration allows to locate AE sources at least equivalent to a Hsu-Nielsen source, it is however important to be aware that the coverage of a structure is not homogeneous. The simulations developed at CETIM by Catty and Pinto using the CASTOR® software (for the calculation of pressure vessels), as represented in Figs. 11a and 11b illustrate this inhomogeneity. In Fig. 11a, we can observe that in the zonal test configuration, the distances (distance from the sensor reaching the 3rd) are between 3.3 m and 9 m. Figure 11b (planar test configuration) shows that these values fall between 1.4 m and 4.4 m.

28

Fig. 11a: Coverage mapping in the case of zonal test configuration with 22 sensors.

Fig. 11b: Coverage mapping in the case of planar test configuration with 77 sensors.

29

This kind of picture, once interpreted, can prove that a Hsu-Nielsen source can be located anywhere with the planar test configuration, and on the contrary only on a limited part of the spherical tank (represented in blue in Fig. 11a) in the case of the zonal test configuration. We also note that the presence of nozzles (represented by a hole in Fig. 11b) can be taken into ac-count. However, the modelling of the disturbance of acoustic waves caused by such obstacles must be refined.

By recognizing this situation, inherent to the AE technique, we could act on the implementa-tion of AE testing:

- Avoid placing the 'critical' regions in the less-well monitored areas of the vessel. - Refine the diagnosis, for example, by adding an analysis parameter that would ‘evaluate’

the level of AE activity observed in a region compared to the minimum detectable. That is, a cluster with an average amplitude of 90 dBAE observed in an area where the mini-mum detectable amplitude is 60 dBAE has necessarily been well detected and localized. If the same cluster is located in an area whose minimum detectable amplitude is 90 dBAE, it will be necessary to consider it differently.

B.6 Conclusions: Towards an application guide based on planar location The AE test aims to detect emissive phenomena in a structure under examination. Then, the central concern of an application guide should be the detectability of a potential active source. The current guides are, however, constructed from an 'instrumentation' approach. The recom-mendations for implementing this technique and analysis criteria do not take into account (or only partly) what we must detect, but what one is capable of recording. For example, defining a criterion for counting the number of events exceeding a defined amplitude (N1S, N2S in GBP) does not provide relevant and reproducible information from one test to another, if we do not take into account, for example, the surface covered by a sensor. It would be necessary to rebuild a process from what is sought, meaning to ask the question: How can I be sure of detecting a source, with an amplitude at the origin of X dBAE, taking into account its position on the struc-ture? For instance, the implementation guide should establish for each testing, the following phases: • Phase 1: From the geometry of the structure and attenuation of acoustic waves, defining a

grid plan, justified by calculation or modelling. The proposed mesh must meet specific cri-teria: Overall coverage rate, local minimum and maximum coverage rate, taking care to specify the source of reference taken for the calculation of these criteria. After this phase, the mesh being defined in terms of geometry, coverage surfaces of each sensor are known.

• Phase 2: Defining criteria for zonal analysis related to the covered area and attenuation. This step consists in adjusting the criteria for zonal analysis to the specific case studied. We should remember that the adjustment of analysis criteria should allow treating each case the same way.

• Phase 3 (instrumentation of the structure, preliminary verification, pressure test, post-test verification): The different phases are substantially identical to those currently practiced. However, special attention should be paid to the verification of the location. It is especially important to check the conditions of localization in the most penalizing regions of the struc-ture, which are not necessarily located close to sensors (see Sections A.2 and B.5). This step must also verify that the process of amplitude correction permits to find the real source am-plitudes (allowing a margin of error).

• Phase 4 (zonal analysis): The analysis is globally identical to the one currently used. How-ever, some of the criteria are poorly defined, such as, for example, estimating the evolution

30

of the activity or the intensity, or Felicity ratio. This new version of the Guide (see Part 3 under preparation) will lay down rules that will make the calculation of these parameters in-dependent of the operator. On the other hand, more accurate analysis criteria will be defined, replacing the only 3 classes actually used.

• Phase 5 (analysis of the located activity): This step necessarily incorporates the amplitude correction on the observed events. Specific criteria for this analysis will be integrated so that it is not confined to a single count of events, which is currently not identical from one testing to another, since it is related to the distance between sensors. Thus, criteria of density of events per square meter, average or maximum will be calculated to give a more physical sense to this analysis.

The implementation of these changes should provide more reproducibility of AE testing, and a more relevant diagnosis. Conclusions Acoustic emission is a unique method of testing, with great potential because it allows rapid diagnosis of large structures in a minimum time, enabling operators to minimize downtime of their facilities. All regulatory rules, codes and standards, which define the general application rules for this technique, authorize use of AE according to two methods (zonal location and planar location by triangulation). However, no comparative study of their performance, thus enabling their assessment, has been carried out. The study described here from a real case shows that the differences of performance found between these two techniques are substantial, and therefore have a major impact on the results of a testing. Moreover, it shows that an application guide such as GBP does not implement repro-ducibility of AE testing, allowing too diverse test configurations. Moreover their differences are not taken into account in the analysis and interpretation of data recorded. So, being consistent with such a guide does not guarantee the relevance of the result. Many ways of improvement exist, some of which are described in this article: • Reducing drastically the choice in the test configuration, such as the distances between sen-

sors, and better controlling the impact of the configuration adopted on the 'coverage level' of the structure,

• Evaluating the performance of a given test configuration using simple and synthetic criteria, such as the rate of detected sources, the rate of located sources, the overall error of measure-ment of amplitude, etc.

• Systematically applying amplitude correction to make a more relevant diagnosis. • Taking into account the local rate of coverage in order to monitor most critical areas opti-

mally. • Adapting the analysis criteria to the test configuration, making these criteria closer to the

structure. Finally, the results of this study show that using a test configuration for planar location method should be preferred. By increasing sensitivity and allowing the calculation of the real amplitude of an active source, it can greatly minimize the error levels on the measure, and thus the test re-sults.

31

Backed by its experience, CETIM now plans to use these assessment tools and carry out well-controlled AE testing. Nevertheless, the professional guides, standards and codes should change so as to allow the industry to take advantage of the real potential of acoustic emission.

Reference [1] Guide to good practice for AE testing of pressure equipment, 2nd Edition, June 2009. AFIAP (French Association of Pressure Equipment Engineers). Edited by SADAVE. [2] J. Catty, “Acoustic Emission Testing – Defining a new standard of acoustic emission testing for pressure vessels Part 1: Quantitative and comparative performance analysis of zonal location and triangulation methods”, Journal of Acoustic Emission, 27 (2009) 299-313.

J. Acoustic Emission, 28 (2010) 32 © 2010 Acoustic Emission Group

ACOUSTIC EMISSION MONITORING OF STEEL-FIBER REINFORCED CONCRETE BEAMS UNDER BENDING

DIMITRIOS G. AGGELIS 1, DIMITRA SOULIOTI 1, NEKTARIA M. BARKOULA 1,

ALKIVIADIS S. PAIPETIS 1, THEODORE E. MATIKAS 1 and TOMOKI SHIOTANI 2 1 Dept. of Materials Science and Engineering, University of Ioannina, Ioannina 45110, Greece;

2 Dept. of Urban Management, Graduate School of Engineering, Kyoto University, C1-2-236, Kyoto-Daigaku-Katsura, Nishikyo, Kyoto 615-8540, Japan

Abstract

The present paper describes acoustic emission (AE) behavior of concrete under four-point bending. Different contents of steel fiber were included to investigate their influence on the load-bearing capacity, as well as on the fracture mechanisms. The main crack was accompanied by many minor cracks indicating that the fibers increased the width of the fracture process zone. The total AE activity was directly proportional to the fiber content. The subsequent study of the AE waveform characteristics revealed that the increase in the fiber content resulted in a shift from the tensile to the shear mode of failure due to improvement of the weak tensile properties of concrete. Keywords: Bending, concrete, fracture mode, steel fibers. Introduction

The application of fiber reinforcement in cementitious materials continues to expand. Fibers restrain the breakage of the brittle matrix and enhance its weak tensile properties (Stahli and van Mier, 2007). As the content of the fibers increases, the possibility that the crack growth will be hindered through an arrest mechanism also increases. As a result, the material toughness is also increased (Mobasher et al., 1990, Sivakuram and Sathanam, 2007). An unreinforced concrete member fails catastrophically at the maximum load. Fibers mainly improve its post-peak behav-ior, while in many cases the maximum load is also significantly increased (Fischer and Li, 2007, Washer et al. 2004).

To clarify the mechanisms for this behavior, acoustic emission (AE) monitoring has been

carried out during fracture tests. There are numerous applications of the AE technique for dam-age characterization of concrete structures (Shiotani and Aggelis, 2007, Shiotani et al. 2007, 2001, Aggelis et al. 2007, Triantafyllou and Papanikolaou, 2006, Golaski et al., 2006, Shah and Weiss 2006). Study of the AE behavior can lead to the characterization and quantification of the damage level via the use of AE descriptors and thus provides an early warning prior to the mac-roscopically observed fracture. Further study of the transient waveforms provides information about the fracture process. The source of the AE activity is closely connected to the mode of fracture (Shigeishi and Ohtsu, 1999). Nucleation of shear cracks follows tensile crack nucleation. Therefore, the determination of the dominant mode provides an early warning prior catastrophic failure. However, the application of AE to such materials entails certain difficulties. These mainly concern the accurate interpretation of the results due to the different individual processes that contribute to the fracture of concrete. Fracture occurs between different interfaces; cement paste and sand, mortar and aggregates, concrete and fibers. Additionally, failure includes aggre-gate crushing and fiber rupture (Kumar and Gupta, 1996). The fracture process of fiber rein-forced concrete can be divided in three distinct stages. The first is the “micro-cracking stage”.

33

The second stage is the growth of micro-cracks and development of macro-cracks up to a satura-tion level. The final stage includes the rapid expansion of the macro-cracks and the material’s macroscopic fracture (Wu et al. 2000, 2001, Kim and Weiss, 2003, Weiler et al. 1996).

In the present study, steel-fiber reinforced concrete beams were subjected to four-point bend-

ing with simultaneous recording of their AE behavior. An increase of the maximum load and toughness with fiber content was observed, as expected. The improvement of the weak tensile behavior of concrete with the addition of fibers was mirrored in the recorded AE activity, which exhibited events with more “shear” character. Finally, the number of hits correlated well with the measured toughness properties of the reinforced concrete. Experimental Part

The specimens were square beams with dimensions 100x100x400 mm. The mix was typical

for shotcrete applications. The water-to-cement ratio was 0.5 by mass and the aggregates-to-cement ratio was 3.6. The maximum aggregate size was 10 mm. The steel fibers were of wavy shape with diameter 0.75 mm and length 25 mm. Three different volume contents were used, namely 0.5%, 1% and 1.5%. For reference purposes, a mixture of plain concrete was also pro-duced. For each of the four mixtures three specimens were tested in bending with concurrent AE monitoring. The four-point bending experiments for the toughness determination were performed according to the ASTM C1609/C1609M-05 standard. The bottom and top spans were 300 mm and 100 mm, respectively, as shown in Fig. 1. The displacement rate was 0.08 mm/min. More details concerning the mechanical loading procedure can be found in (Soulioti and Matikas, 2008).

For the AE monitoring, two AE sensors resonant at 150 kHz (R15, Physical Acoustics Corp.,

PAC) were attached to the bottom side of the beams at a 50-mm distance from either side of the mid-span (see Fig. 1). The signals were amplified 40 dB and those exceeding 40 dBAE (in refer-ence to 1 µV input) were recorded in a two-channel monitoring system (PCI-2, PAC). The sam-pling rate was 5 MHz.

Fig. 1. Experimental setup. Mechanical Results

Typical load-deflection curves of fiber-reinforced concrete specimens are depicted in Fig. 2. The behavior was typically linear up to the maximum load for all specimens. As expected, the

100 mm

300 mm

Deflection measurement system

AE sensors

100 mm

Concrete specimen

34

specimens reinforced with 1.5% of fibers exhibited higher maximum load. However, the main difference was observed in the post-peak behavior. The plain concrete failed catastrophically in two parts without absorbing any energy after the maximum load was reached. On the other hand, fiber-reinforced concrete specimens, exhibited a sudden drop of load after the initial fracture, but continued to absorb energy while their load was gradually decreased until the end of the test. As can be seen in Table 1, although the increase of fiber content did not necessarily increase the peak load of the material it certainly enhanced its flexural toughness as manifested by the area under the load-deflection curves of Fig. 2 (see also Table 1).

Fig. 2. Load vs. deflection curves for specimens with different fiber contents.

Table 1. Mechanical properties for different fiber contents

Fiber Content (%)

Maximum load (kN)

Flexural toughness, T100,2

* (J) 0 14.9 -

0.5 13.2 7.0 1 15.8 15.3

1.5 19.9 17.3 *The subscripts denote the specimen thickness (100 mm) and the maximum center deflection (2 mm) ac-cording to ASTM C 1609/C 1609M-05.

In Fig. 3(a), the main crack of a plain concrete specimen after failure is shown. The specimen

failed in tension with the crack initiating from the bottom surface (bottom of Fig. 3(a)), which was under tensile load and propagating towards the top splitting the specimen into two parts. On the other hand, fiber-reinforced specimens did not break into two parts even after final failure. As can be seen in Fig. 3(b) the main crack was accompanied by smaller cracks. This phenome-non demonstrates the main reason for including fibers into concrete. The fracture energy is dis-tributed in a larger volume leading to a reduction of the stress intensity. This leads to higher maximum loads of the fiber-reinforced concrete and higher absorbance of fracture energy, which was demonstrated by the increased area under the load-deflection curves of Fig. 2. As was obvi-ous even by visual inspection the inclusion of fibers increased the width of the fracture process zone (FPZ). The same effect is also reported in the literature for large aggregates, which had a similar effect on the FPZ (Mihashi et al., 1991). As the crack propagated to the top side, it split

1.5%

1%

0.5% 0%

35

to different smaller cracks expanding the width of the energy absorbing zone. This bifurcation and changes of direction will be discussed along with the AE data in the next sections.

Fig. 3. (a) Cracks obtained for plain concrete, (b) concrete with 1.5% fibers.

Acoustic Emission Results

Total AE hit activity

Figures 4(a) and (b) show the history of cumulative AE hits along with the loading for typical cases of 1.5% fiber concrete and plain concrete, respectively. In the case of the specimen with fiber, at approximately 70% of the maximum load, a small increase of the AE hits was recorded and the hit rate reached a peak at the macroscopic fracture. This was demonstrated by the vertical increase of the hit line at the moment of peak load. The total number of hits was typically 3000 - 4000 (see Fig. 4(a)). However, in the case of the plain concrete specimens (Fig. 4(b)), a notably smaller number of hits were recorded. From this figure it is also obvious that immediately after the peak load the experiment was terminated due to the complete fracture of the plain specimen.

In Fig. 4(c) the total AE activity is shown for the different fiber contents. Each point is the

average of the total number of hits of the three specimens of each category. The relation between the number of hits and the fiber content is close to linear. The increase of AE hits was attributed to the development of more cracks with increasing fiber content leading to a considerably larger fracture surface area. This trend of AE-hit increase corresponded to the fiber pull-out incidents, which naturally depended on the number of fibers involved in the failure zone. At the moment of the macro-crack formation (main fracture), the maximum rate of AE-hit generation was exhib-ited. In Fig. 4(d) the maximum hit rate (at the moment of main fracture) is shown for the differ-ent types of material. As can be seen, the fibers contributed significantly to the fracture process, even during the critical crack growth stage through the numerous pullout events (approximately 100 hits/s). On the other hand, plain concrete exhibited only 40 hits/s during the catastrophic fracture.

An interesting correlation was noticed between the number of cumulative AE hits and the

toughness of the materials, as depicted in Fig. 5. As the content of fibers increased, so did the

36

toughness and the AE activity. As a result, with proper study, the AE activity can be used as a measure of fracture toughness.

Fig. 4. Load and AE history for specimen with (a) 1.5% fibers, (b) 0% fibers, (c) total AE activ-ity vs. fiber content, (d) maximum AE hit rate vs. fiber content.

Fig. 5. Toughness vs. AE activity for different fiber contents.

Fracture mode

The shape of the AE waveforms is reported to be characteristic of the fracture mode. Shear events are characterized by longer rise time and higher amplitude than tensile events (Shiotani et

37

al., 2001). This is examined by the AE “Grade”, which is defined as the ratio of waveform am-plitude to the rise time. It has been shown that the higher the grade, the more tensile is the nature of the fracture events (Shiotani et al., 2001).

In the present case, the cracking process started at the bottom surface for all specimens due to

the tensile stress. This implied that the initial active failure mode was tensile. In the case of the plain concrete, this failure mode was the dominant mode and led to the catastrophic failure of the specimens. However, the presence of the reinforcement changed the mode of failure of the com-posite. This was manifested by the creation and propagation of multiple cracks and hence a larger zone of influence together with the decrease of the grade of the AE waveforms. The grade change may be either expressed as a mean value for each fiber configuration or as a dynamic phenomenon throughout the loading process.

In Fig. 6(a), the average grade (dB/µs) is depicted for different fiber contents. As can be seen,

the plain concrete specimens exhibited the highest grade from the total of the AE hits recorded throughout the experiments. The inclusion of fibers, even at just 0.5% by volume led to the con-siderable decrease of the grade, which was associated with the change of the failure mode from tensile to shear. Further increase of fibers had a small influence on the grade since a plateau was reached.

Fig. 6. (a) Average grade vs. fiber content, (b) moving average of grade for the three specimens with 1.5% fibers.

Figure 6(b) shows the change of grade vs. time for all three tests in the case of 1.5% fiber content. The curves correspond to the moving average of grade for 50 consecutive hits in order to reduce scatter. It is obvious that as the fracture proceeds, the grade gradually decreased until the end of the experiment.

Similar analysis can be conducted using the average frequency (RILEM, 2008), for which re-

cent results showed that it drops suddenly after a strong fracture incident. This is illustrated in Fig. 7 for two specimens with 1% fibers. It is evident that at the moment the main fracture inci-dent occurs, the average frequency of the signal drops by approximately 200 kHz (marked by arrows). Note that the symbols represent the average frequency of each hit while the black solid line stands for the moving average of the recent 40 hits. It should be highlighted that the data of Fig. 7 were captured using broadband sensors with a sensitivity range between 50 and 800 kHz

(a)

(b) (a)

38

(Pico, PAC). The dynamic change of average frequency should be connected to the failure mode. Before the main fracture, matrix cracking is the most dominant mechanism. After the main fracture, fiber pull-out is added as an active fracture mechanism. The first mode resembles the mode I tensile crack while the latter resembles mode II (shear). This large frequency drop can be captured more adequately in this case than the case of resonant sensors at 150 kHz. However, the broadband sensors record only about one third of the hits a resonant sensor records in a similar experiment. Thus they are not suitable for location purposes since their sensitivity in terms of amplitude response is limited; on the other hand they prove to be more sensitive to the dynamic nature of fracture when they are located near the fracture zone.

Fig. 7. Time history of load and average frequency for specimens with 1% fibers.

From the above it is shown that the nature of events is moving from tensile to shear with the propagation of damage. Although the cracks were initiated by tensile loads, gradually shear stresses dominated the failure process. This has also recently been reported for vinyl-fiber con-crete (Aggelis et al. 2009). Similar behavior has been observed in corrosion cracking of concrete where crack initiation was due to the tensile mode, while as the crack length increased the shear mode became more active (Farid Uddin et al., 2004). As seen in Table 1, the increase of fiber content, which activated the shear mode was connected to the increased fracture toughness of fiber concrete. Therefore, the specimens did not break with a brittle vertical crack but the crack bifurcated to different smaller cracks on different directions, spreading the fracture energy to wider volume. It is mentioned that use of more sensors and application of inversion analysis such as Moment Tensor (Ohtsu et al., 1998) will enable the characterization of different clusters of events according to their mode type, their location in the volume of the specimen and conse-quently the actual width of the FPZ (Grosse and Finck, 2006).

Conclusions

In the present study the influence of steel fibers in concrete under bending is discussed. Mac-

roscopically, increased fiber content resulted in the increase of the maximum load and the frac-ture toughness of the material. At the same time, the number of AE events was proportional to the fiber content and the measured toughness. Analysis of the “AE grade” parameter showed that the tensile mode of fracture was dominant for plain concrete, changing to shear mode as the fiber content increased. This demonstrated the definite reinforcing effect of the fibers against the weak tensile nature of concrete. As the AE activity implied, the mode of fracture changed during the experiment from tensile (initial stage) to shear (final fracture). This was macroscopically mani

39

fested by the crack bifurcation and deflection from parallel to perpendicular direction relatively to the loading axis and the concurrent increase of the zone of influence with increasing fiber con-tent. The average frequency of AE hits is proved to be another parameter sensitive to the damage status especially when broadband sensors are employed.

The identification of the fracture mode, which is feasible via the acoustic emission technique,

is of primary importance as it can lead to more suitable design of the reinforcement in order to withstand the specific stresses. At the same time classification of cracks can be employed for early warning prior to macroscopic failure.

The next step is the use of different types of fibers. This would reveal the best configuration

in terms of volume content, fiber shape and material for the enhancement of the structural com-ponent. Furthermore, source location of the AE events via the use of more sensors can determine the actual expansion of the fracture process zone with increased fiber content.

Acknowledgement

The fibers were supplied by ETAL S.A., Greece. References D.G. Aggelis, T. Shiotani, S. Momoki, M. Terazawa (2007) Advances in Acoustic Emission – 2007, (ed. K. Ono), pp. 390-395. D.G. Aggelis, T. Shiotani, S. Momoki, A. Hirama (2009), “Acoustic emission and ultrasound for damage characterization of concrete elements”, ACI Materials Journal 106(6). 509-514. A.K.M. Farid Uddin, K. Numata, J. Shimasaki, M. Shigeishi, M. Ohtsu (2004), Construction and Building Materials, 18, 181-188. G. Fischer, V.C. Li (2007), Engineering Fracture Mechanics 74, 258-272.

L. Golaski, G. Swit, M. Kalicka, K. Ono (2006), Journal of Acoustic Emission 24, 187-196. C.U. Grosse, F. Finck (2006), Cement and Concrete Composites 28, 330-336.

B. Kim, W.J. Weiss (2003), Cement and Concrete Research 33, 207-214. A. Kumar, A.P. Gupta (1996), Experimental Mechanics 36(3), 258-261.

H. Mihashi, N. Nomura, S. Niiseki (1991), Cement and Concrete Research 21, 737-744. B. Mobasher, H. Stang and S.P. Shah (1990), Cement and Concrete Research 20, 665-676.

M. Ohtsu, T. Okamoto, S. Yuyama (1998), ACI Structural J. 95(2), 87-95. RILEM TC212-ACD (2008) Acoustic emission and related NDE techniques for crack detection and damage evaluation in concrete. Recommendation 3. H. R. Shah, J. Weiss (2006), Materials and Structures 39(9), 887-899.

M. Shigeishi, M. Ohtsu (1999), Acoustic Emission: Standards and Technology Update, ASTM STP 1353, Vahaviolos SJ, ed. American Society for Testing and Materials, pp. 175-188.

T. Shiotani, M. Ohtsu and K. Ikeda (2001), Construction and Building Materials 15(5-6), 235-246.

T. Shiotani, D.G. Aggelis, O. Makishima (2007), Journal of Acoustic Emission 25, 308-315.

40

T. Shiotani, D.G. Aggelis (2007) Journal of Acoustic Emission 25, 69-79. A. Sivakuram, M. Sathanam (2007), Cement and Concrete Composites 29, 603-608.

D. Soulioti, T.E. Matikas (2008), Proc. of the 1st Conference on Structural Materials and Com-ponents, 21-23 May, Athens, Greece, vol. C, 1287-1298, (In Greek)

P. Stahli, J.G.M. van Mier (2007), Engineering Fracture Mechanics 74, 223-242. T.C. Triantafillou and C.G. Papanikolaou (2006), Materials and Structures 39(1), 93-103.

G. Washer, P. Fuchs, B.A. Graybeal, J.L. Hartman (2004): IEEE Transactions on Ultrasonics, Ferroelectrics, and Frequency Control 51(2), 193-201.

B. Weiler, C.U. Grosse, H.W. Reinhardt (1996), Proceedings of the 22nd European Conference on Acoustic Emission Testing, Aberdeen, Scotland, pp. 119-124.

K. Wu, B. Chen, W. Yao (2000), Cement and Concrete Research 30, 1495-1500. K. Wu, B. Chen, W. Yao (2001), Cement and Concrete Research 31, 919-923.

J. Acoustic Emission, 28 (2010) 41 © 2010 Acoustic Emission Group

ON LAMB MODES AS A FUNCTION OF ACOUSTIC EMISSION SOURCE RISE TIME#∞

M. A. HAMSTAD

National Institute of Standards and Technology, Materials Reliability Division (853), 325 Broadway, Boulder, CO 80305-3328 and University of Denver, Department of Mechanical

and Materials Engineering, Denver, CO 80208

Abstract

A study was carried out to examine Lamb-wave modal content as a function of the acoustic emission (AE) source rise-time. The study used a validated finite-element code to model the source operation and subsequent wave propagation up to a distance of 480 mm in a 4.7-mm thick aluminum plate with large transverse dimensions. To obtain the large propagation distances with sufficient transverse dimensions so that plate edge reflections did not arrive during the signifi-cant part of the direct AE signal, an axisymmetric code was used. The buried dipole AE sources were located at three different depths below the top surface of the plate, where the pseudo-AE sensors were located. These sensors provided the out-of-plane displacement as a function of time. The rise-times for the different finite-element runs varied from 0.5 µs to 15 µs. The result-ing data were high-pass filtered at 40 kHz and re-sampled with a time step of 0.1 µs. The intense portions of the Lamb modes were determined by use of a Choi-Williams transform (CWT) for the range of source rise-times, three different source depths and the signal propagation distances. Higher Lamb modes were observed in the CWT results for the shortest rise-times, but fundamen-tal mode frequencies still dominated for all the rise-time cases and different source depths. Thus, only the fundamental modes need to be considered in the determination of accurate signal arrival times.

Keywords: AE source rise-times, Choi-Williams transform, Lamb modes, thin plate

Introduction

In the analysis of burst-type acoustic emission (AE) signals obtained with wideband sensors (nearly flat frequency response), the use of intense Lamb mode/frequency combinations has pro-duced an improvement in the accuracy of source location calculations [1-3]. This result comes from the determination of signal arrival times that are independent of a voltage threshold and that correspond to the same group velocity at each sensor in an array.

A previous paper [4] used finite-element modeling to study the effects of different source

rise-times on the peak amplitudes and frequency content of the far-field AE signals in a plate. The results of that study demonstrated a large drop in the peak amplitude as the source rise-time increased. The purpose of this paper is to extend the previous examination to the modal content of the out-of-plane AE signals that result from sources with different rise-times. Finite-element modeling (FEM) offers several key advantages: i) generation of AE signals from buried self-equilibrated sources at different depths (below the top surface) in a plate, ii) generation of AE signals with specific source rise-times; iii) obtaining the exact out-of-plane displacement versus ______ #Contribution of the U.S. National Institute of Standards and Technology; not subject to copyright in the United States. ∞Trade and company names are included only for complete scientific/technical description; endorsement is neither intended nor implied.

42

time from a pseudo sensor (a perfect, wideband, point-contact sensor with no resonances), iv) obtaining the sensor results at different known propagation distances from the source and v) use of a specimen that did not produce reflections from the specimen edges arriving at the sensors during the major portion of the direct arrival of the AE displacement waves from the source.

Relevant Details on the Specimen Domain and Finite-element Modeling

A circular aluminum plate, 4.7-mm thick with a radius of nearly 1000 mm, served as the sample domain. The large in-plane dimension allowed propagation distances of up to 480 mm such that the direct arrival of the significant portion of the AE waves was not altered by reflec-tions from the plate edge. A total of nine selected source rise-times varied from 0.5 µs to 15 µs. The basis for the selection of this range in rise-times was previously presented [4]. Because the 0.5-µs rise-time generated frequencies well above the frequency sensitivity range of even special AE sensors, shorter rise-times were not examined. The sources in the continuous mesh domain were dipoles (self-equilibrating forces of two oppositely directed single-cell monopoles with one cell between them) by use of the “equivalent body force” concept for displacement discontinui-ties [5]. The forces (each monopole of the dipole had a force of 1 N) were applied with a “cosine bell” temporal time dependence T(t) given by

0 for t < 0, T(t) = (0.5 – 0.5 cos [ t / ] ) for 0 t , and (1) 1 for t > ,

where was the source rise-time. The same very small cell size (54 µm) and time step (7.7 ns) was used for all the FEM runs. These values were chosen to meet the requirements of the vali-dated finite-element code [6] for the shortest rise-time source. An axisymmetric version of the code was used to enable carrying out the modeling in such a large domain with the very small cells and time steps with reasonable computing resources. The finite-element code was run for a small period of time beyond 300 µs from the start of the operation of the AE source. The dipole sources were oriented out of plane, with the center of the dipole located at one of three different depths below the top surface of the plate where the sensors were located. The depths were 2.35 mm (mid-plane), 1.32 mm (mid-depth) and 0.24 mm (near the plate top surface). The entire AE signals were numerically processed with a 40 kHz (eight-pole) high-pass Butterworth filter, fol-lowed by resampling from the original time step to 0.1 µs per point (comparable to the digitiza-tion rates often used in AE measurement systems). The aluminum material properties used for the FEM calculations were bulk longitudinal and shear velocities of 6320 m/s and 3100 m/s, re-spectively, and a density of 2.7 kg/m3. Determination of Frequency/Time Content

Choi-Williams transform (CWT) results were used to enhance the identification of the AE signal Lamb modes, to indicate highly excited frequency regions within the modes, and to obtain the CWT magnitudes (coefficients) in these regions. The CWT results were obtained by use of the AGU-Vallen Wavelet freeware [7] with the key parameter settings being maximum fre-quency up to 2000 kHz, frequency resolution = 2.4 kHz, 112 terms in the damping summation, and an exponential damping parameter of 20. The time-domain setting of the number of samples was up to about 3000 for the furthest propagation distance. In the color CWT figures, red indi-cates the highest-intensity region of the CWT coefficients. By use of the known propagation dis-tances, the appropriate group velocity curves [8] were superimposed on the CWT results to iden-tify the Lamb modes. In some cases, the relevant group-velocity curves for the figures in this

43

Fig. 1. Comparison of CWT results for in-plane versus out-of-plane source orientation. Scaling of 0 to 1000 kHz for vertical axis and 0 to 150 µs for horizontal axis. paper are shown at the bottom of the CWT results, so they do not hide the intense regions in these results. Out-of-plane versus In-plane Source Orientation Based on the CWT results from the three-dimensional FEM calculations shown in Fig. 1, the use of an out-of-plane dipole source orientation provides frequency/time results similar to those from in-plane dipoles for sources at different depths. The results in Fig. 1, for the same material and plate thickness, were for a source rise-time of 1.5 µs and a propagation distance of 180 mm. Again, these CWT results were calculated from 40-kHz high-passed, out-of-plane displacement data. Even with the use of a smaller domain and a longer rise-time (allowing a larger cell size and longer time steps), considerable computer resources were required to make the three-dimensional runs. The results described above indicated that the out-of-plane sources could be used in this study, resulting in a significant reduction in required computer resources.

Modal Content versus Source Rise Time – Mid-plane Source Location

Figures 2 and 3 show the CWT results (first columns) as well as the out-of-plane displace-ment signals (second column of Fig. 2; zero time corresponds to the start of the dipole source function) and their frequency domains (second column of Fig. 3) for a selected set of rise-

44

Fig. 2. CWT (first column) and AE out-of-plane displacement signal (second column) for a se-lection of rise-times for the source at the mid-plane and at a propagation distance of 480 mm. Note: only the group velocity curves are present at the bottom plot of the first column. times for the signals at a propagation distance of 480 mm. Because the FEM code did not apply material attenuation, the displacement results at a propagation distance of 480 mm were used ex-clusively in these figures for all the rise-time cases. This distance provided the best separation of the dispersive modes. Note that the CWT results are duplicated in both figures so that these

45

results can be directly compared by the reader to the AE signals and the frequency-domain re-sults without making the figures so small that they lose their illustrative role. Also, prior to the calculation of the fast Fourier transform (FFT) results, the waveforms were shortened and termi-nated at a zero to remove any effects from the first edge reflection that occurs at about 280 µs for this propagation distance. (Note that the FFTs were calculated with a rectangular window after each terminated signal was extended with zeros to obtain a total of 4096 points). This reflection

Fig. 3. CWT (first column) and corresponding spectra (second column) for a selection of rise-times for the source at the midplane and at a propagation distance of 480 mm. See note in Fig. 2 caption.

46

can be observed most clearly for the longer rise-times. Except for the 15-µs rise-time CWT re-sult, the group-velocity curves were not superimposed on the results. Instead, these curves are shown at the bottom of the CWT results. For the 15-µs rise-time CWT, the S0 curve was super-imposed to show that the intense arrival appears to be both late and extended in time. To exam-ine this situation, the CWT was calculated for this rise-time and propagation distance without applying any frequency filter. In this case, the S0 mode peak intensity was at 41 kHz (9% late in time, based on the thin-plate extensional velocity). For the 40-kHz high-pass data, the S0 mode peak intensity was at 61 kHz (25 % late in time). Thus, the 40-kHz filter contributes to the late arrival. In addition, since the period of the signal in this low frequency range is relatively long, as shown in Fig. 2, it is possible that this influences the CWT calculation and also contributes to the late and extended arrival in the CWT result.

A number of observations for the mid-plane source depth can be made. First, for the 0.5 µs rise-time, a portion of the higher-frequency part of the S1 mode can be seen the CWT result (see arrow in Figs. 2 and 3). However, relative to the CWT magnitude of the most intense portion of the S0 mode at a frequency of 539 kHz, its peak magnitude, which occurs at a frequency of 1430 kHz, is only 18 % of that S0 peak magnitude. At rise-times of 0.75 µs and longer, the S1 mode is no longer clearly visible. Second, at a rise-time of 3 µs and greater, the most intense portion of the CWT moves to the initial arrival portion of the S0 mode as compared to a CWT peak arrival later in the signal for shorter rise-times. Third, as can be observed from the second column of Fig. 2, for all the rise-time cases, the time of arrival of the peak CWT intensity corresponds to the highest or nearly highest-amplitude portion of the signal. Further, in Fig. 2 (second column), this high-signal-amplitude arrival time shifts from late in the signal to early as the rise-time in-creases. Fourth, the most intense region of the CWT corresponds to the highest or near-highest amplitude frequency region of the spectra results, as shown in the second column of Fig. 3. Fi-nally, it is worth noting that for the 3-µs rise-time, the second most intense portion of the S0 mode, with a peak at a frequency of 630 kHz, is directly in a region where the S0 and S1 group velocity curves first intersect.

The above discussion demonstrates that determining signal arrival times at the most intense portions of the CWT corresponds to both the signal amplitudes and signal frequencies that are among the largest in the signals. These are the regions of the AE signals where the best signal-to-noise ratio (based on the electronic preamplifier noise) would be present.

Modal Content versus Source Rise Time – Near Top Surface Source Location

Figures 4 and 5 show the CWT results (first columns) as well as the out-of-plane displace-ment signals versus time (second column of Fig. 4) and the frequency domains (second column of Fig. 5) for a selected set of rise-times for the signals at a propagation distance of 480 mm. The same comments relative to the duplication of the CWT in the figures and the calculation of the FFTs apply to these figures as were made relative to Figs. 2 and 3. Clearly, for this source depth and all the rise-times, there is a strong dominance of the lower-frequency regions of the A0 mode. As the rise-time decreases, higher-frequency portions of the A0 mode are apparent in the CWT, and as the rise-time increases, the most intense portion of this mode gradually decreases in frequency with a significant change in the arrival time of the peak CWT intensity due to the shape of the A0 group velocity curve. This trend of arrival times is observed in the time domains of Fig. 4. Specifically, the change in the most intense frequency of the A0 mode in the CWT re-sult goes from 66 kHz for the 0.5-µs rise-time to 51 kHz at the 15-µs rise-time, while the respec-tive change in the arrival time of the peak intensity changes from about 200 µs to about 230 µs.

47

Over the same range of rise-times, the peak magnitude of the spectra hardly changes from about 50 kHz. At these lower frequencies there is likely some effect of the 40-kHz high-pass filter. As expected, the frequency content (see Fig. 5) in the spectra above 400 kHz gradually fades as the rise-time increases. Fig. 4. CWT (first column) and corresponding out-of-plane displacement AE signal (second col-umn) for a selection of rise-times for the source near the top surface and at a propagation dis-tance of 480 mm. See note in Fig. 2 caption.

48

Fig. 5. CWT (first column) and corresponding spectra (second column) for a selection of rise-times for the source near the top surface and at a propagation distance of 480 mm. See note in Fig. 2 caption. Modal Content versus Source Rise Time – Mid-depth Source Location

Figures 6 and 7 show the CWT results (first columns) as well as the out-of-plane displace-ment versus time (second column of Fig. 6) and frequency domains (second column of Fig. 7)

49

Fig. 6. CWT (first column) and corresponding out-of-plane displacement AE signal (second col-umn) for a selection of rise-times for the source at the mid-depth and at a propagation distance of 480 mm. See note in Fig. 2 caption. for a selected set of rise-times for the signals at a propagation distance of 480 mm. The same comments relative to the duplication of the CWT in both figures and the calculation of the FFT apply to these figures as were made relative to Figs. 2 and 3. Again at the two shortest rise-times, there is evidence in the CWT results of a higher mode (A1, higher-frequency regions of the fun-damental modes and an intense region due to mode intersections (second intersection of the S0

50

and A1 modes). As was the case with the mid-plane source at 0.5-µs rise-time, a higher-frequency region at 542 kHz of the S0 mode is still the most intense. Compared to the magnitude of the S0 mode at that frequency, the magnitude of the CWT coefficients at the intersection of the S0 and A1 modes at 837 kHz is 93 %, and the magnitude of the A1 mode at 993 kHz is 61 %. The Fig. 7. CWT (first column) and corresponding spectra (second column) for a selection of rise-times for the source at the mid-depth and at a propagation distance of 480 mm. See note in Fig. 2 caption.

51

A0 mode also is visible (for this rise-time) in the low frequency region with a comparative mag-nitude of 35 % at its most intense point at 63 kHz.

Clearly, for this source depth, both the axisymmetric and anti-symmetric modes are apparent.

For all the rise-times, the spectra show a low-frequency peak that is the largest in magnitude at the longer rise-times, and it nearly has the largest magnitude at the shorter rise-times. In contrast to the mid-plane source, at the longer rise-times the largest-magnitude region of the CWT of the AE signal is in the low-frequency region of the A0 mode rather than the low-frequency region of the S0 mode at an earlier arrival time for the mid-plane source. At the rise-times of 2.3 µs or longer, the AE signal peak amplitude corresponds to the most intense portion of the A0 mode in the CWT. It is interesting that the spectra for the two shortest rise-times show a high magnitude in a low-frequency region with a peak at approximately 50 kHz, but this frequency is less appar-ent in the CWT and the AE signal. This situation is likely to be the result of the continued pres-ence in time of the A0 mode in the CWT for these two rise-times (see arrows to A0 in Figs. 6 and 7).

Modal Content Versus Propagation Distance

Figure 8 illustrates how the CWT results change as a function of selected propagation dis-tances from 480 mm to 60 mm for two source depths for the sources with a rise-time of 0.5 µs. The first column is for a mid-depth source, and the second column is for a mid-plane source. The cases for the shortest rise-time were selected, since they excite the largest number of modes. In addition, the near top source was not selected for illustration, because only the A0 mode strongly dominated all the rise-time cases at all the propagation distances.

Examination of the first column of Fig. 8 shows that as the propagation distance decreases

for the mid-depth source, similar dominant excited modal features can be identified down to a propagation distance of 180 mm. At 60 mm, the modal results are not clear in this figure. To clarify the CWT results in the first column of Fig. 8 for the shortest propagation distances, Fig. 9 shows in the first column the CWT results for the mid-depth source at distances of 120 mm and 60 mm. In addition, the same results are shown in the second column for the mid-plane source where the time axis is only from zero to 80 µs in both columns. The first column of Fig. 9 shows that the closeness in time of the multiple modes that are excited results in distortion of the CWT results above about 600 kHz for both the 120 mm and 60 mm distances. To determine whether intense arrivals below this frequency corresponded to the correct Lamb wave group velocity, the arrival times of the CWT peak magnitudes were determined at 564 kHz for all the propagation distances. This frequency was that of the peak CWT magnitude for the S0 mode at a propagation distance of 180 mm. A plot (not shown) of propagation distance versus these arrival times had a slope of 1.85 mm/µs, with all the data points on the straight line including those at 120 mm and 60 mm. This velocity is within 1.2 % of the associated Lamb-wave S0 mode group velocity at that frequency. Thus, in spite of the distortion in the CWT results, the correct mode arrival times can be obtained at the most intense mode/frequency combination below 600 kHz.

If the second columns of Figs. 8 and 9 are examined, the same dominant mode and frequency

combinations for the S0 and S1 modes are apparent at each propagation distance from 480 mm down to 60 mm. Thus, when multiple frequency modes are not sufficiently excited over the same frequency range (as was the case in the first column of Fig. 9), the CWT results are not as dis-torted at the shorter propagation distances. Again, as a check on the application of the use of the CWT magnitude peaks at certain frequencies to obtain arrival times, the arrival times at the peak

52

magnitude of the CWT for the S1 mode at 1430 kHz were determined for all the distances. The same type of plot for this data had a slope of 2.20 mm/µs, which differed by less than 1 % from the value obtained from the Lamb-mode group-velocity curve for the S1 mode at that frequency. Thus, even this relatively small intensity mode arrival (for example about 17 % of the CWT peak S0 mode arrival at the propagation distance of 360 mm) provided accurate arrival times.

Fig. 8. CWT results versus propagation distance for 0.5-µs rise-time. First column for mid-depth source and second column for mid-plane source depth.

53

Fig. 9. CWT results versus distance at the two shortest propagation distances. First column for mid-depth source and second column for mid-plane source depth. Fig. 10. CWT results at 480 mm for the indicated two rise-times and three source depths.

54

Modal Content to Possibly Indentify Sources with Different Rise Times

The changes in the intense modal regions as the source depth changes imply that, unless the sources with distinctly different rise-times all originate at the same depth, the modal pattern of intense mode/frequency combinations cannot easily be used to distinguish the different rise-times. To demonstrate this directly, Fig. 10 shows the CWT results at 480 mm for sources at the three different depths with two different rise-times of 1.5 µs and 4 µs (rise-times differ by a fac-tor of 2.7). In this figure, it is clear that if the sources were at the same depth, it would be straightforward to distinguish the different rise-time sources by the differences in modal intensity at different frequency points in the modes excited. For example, in the mid-plane source case, the 4-µs rise-time excites a lower frequency portion of the S0 mode, which is not sufficiently excited in the 1.5-µs rise-time case. In the least straightforward case for the near top source, the calcula-tion of the ratio of the CWT peak magnitudes (for the A0 mode) at the frequency of 60 kHz di-vided by that at 300 kHz would serve to distinguish the two different rise-time sources. In the CWT example in the figure, this ratio for the 1.5-µs source is 4.7, and for the 4-µs source it is 150. However, if the source depths varied, a much more complicated approach would be needed, and to verify such a proposed approach, a more extensive database of possible source depths would be required.

To examine another aspect of the dependence of the modal content on the source rise-time,

the CWT magnitudes of the intense mode/frequency combinations were examined as a function of the source rise-time and the source depth. The normalized results are shown in Fig. 11 for a propagation distance of 480 mm. The selected intense combinations for each depth were as fol-lows: i) mid-depth – S0 at 537 kHz, A0 at 63 kHz and S0 intersection with A1 at 837 kHz, ii) near top surface – A0 at 66 kHz and iii) mid-plane – S0 at 539 kHz. These combinations were selected from the CWT results for the 0.5 µs rise-time cases shown in Figs. 2 through 7. Figure 11 clearly shows that the three high-frequency mode/frequency combinations experience a rapid fall-off of the CWT magnitude as the rise-time increases. The most rapid falloff is the mid-depth source CWT magnitude at 837 kHz followed closely by the mid-plane source result at 539 kHz and the mid-depth result at 537 kHz. It should be noted that these two similar frequency values are those provided by the software, but they clearly are from the same mode/frequency region. These re-sults with rise-time increases are expected, because longer rise-times do not sufficiently excite these higher frequencies. It should be noted that the modal arrivals of the three higher mode/frequency combinations could not be identified in the CWT results for rise-times greater than about two microseconds. The falloff in the CWT magnitude with increasing rise-time of the two lower-frequency combinations was nearly identical, because they involved the same mode, A0, for the two different source depths. The slower fall-off with increasing rise-time for these cases is expected due to the relatively low frequency of this A0 modal region. The potential abil-ity to use these results to identify sources with different rise-times must be combined with the results of CWT-determined mode attenuation with propagation distance.

Figure 12 demonstrates how typical intense mode/frequency combinations attenuate with propagation distance for the 0.5-µs rise-time sources at two different depths. In this figure, the attenuation of the A0 mode at 66 kHz for a near top source and the S0 mode at 539 kHz for a mid-plane source are compared to the amplitude-based geometric attenuation of a wave spread-ing in-plane. The higher-frequency combination has slightly more attenuation than the lower-frequency combination, but both decay significantly more than the geometric attenuation alone. Thus, the CWT magnitudes of the intense mode/frequency arrivals experience an additional source of attenuation beyond the expected geometric attenuation. Kerber et al. [9] recently exam

55

ined Lamb wave modal attenuation with propagation distance by comparing the results from a chirplet transform (CT) to those from a short-time Fourier transform (STFT). The CT allowed them to use an additional degree of freedom “to adjust the window function to the group delay of the signal.” The CT-based algorithm thus allowed them to use the known group-velocity curves. With experimental data for an aluminum plate one millimeter thick, they found significantly smaller mode attenuation with propagation distance with the CT-based algorithm as compared to the STFT results. In their work, the source of the out-of-plane displacement signals was a laser ablation on the aluminum plate surface (out-of-plane surface source). A point-based laser meas-urement system was used to obtain the displacement signals. They also showed further reduc-tions in the attenuation with both the CT and STFT results when synthetic signals were used. Due to the differences in plate thickness and a lack of specific data in this reference, a direct comparison could not be made with the current results for the CWT. However, the results in ref-erence 9 indicate that the added attenuation beyond geometric attenuation is at least in part artifi-cial, as it depends on the signal-processing approach.

Fig. 11. Normalized CWT magnitudes of the intense mode/frequency combinations for the three source depths (at a propagation distance of 480 mm) as a function of the source rise-time.

It is interesting to note that the added attenuation relative to geometric attenuation for these mode/frequency combinations is about 12 to 15 dB for these two cases. This result contrasts with the peak amplitude attenuation of about 3 to 5 dB more than geometric attenuation [4]. To fur-ther illustrate this aspect, Fig. 12 also includes the attenuation of the peak AE signal amplitude with propagation distance for a mid-plane source with a rise-time of 0.5 µs. It is clear that the peak amplitude does not attenuate as rapidly with propagation distance as the intense mode/frequency combinations. It is worth noting, that the peak signal amplitude attenuation with distance is approximately the same for the different source depths and source rise-times [4]. Fur

56

ther, the “wavy” character of the peak amplitude attenuation with distance is due to differences (at different propagation distances) in the superposition of the multiple modes excited by the short rise-time. It was demonstrated previously [4] that the “wavy” character decreases when the source rise-time is longer when fewer modes are generated. Also, when the CWT magnitude of a single mode/frequency combination is considered, the attenuation with propagation distance is smooth, as shown in Fig. 12. Thus, unless the propagation distance to the sensors is approxi-mately the same, the attenuation of the CWT magnitude of the intense mode/frequency combina-tions creates an additional complication to the use of modal information for the identification of sources with different rise-times. Fig. 12. Normalized CWT attenuation of intense mode/frequency combinations and normalized peak amplitude for mid-plane source, 0.5 µs rise-time sources versus geometric attenuation. Comments on Comparisons of Modal Content versus Source Depth In the previous publication [4] which specifically analyzed the frequency spectra dependence on rise-time and source depth, it was pointed out that for rise-times less than 2.3 µs, the mid-depth spectra could be roughly viewed as a superposition of the peak-frequency regions from the mid-plane and near top spectra results at the same rise-times. Examination of Figs. 2, 4 and 6 demonstrates that such a statement cannot be made relative to the CWT results. For example, in the case of a 0.5-µs rise-time, the primary modal content for the source near the top surface is the A0 mode, and for the mid-plane source it is the S0 mode with a small magnitude of the S1 mode. The mid-depth primary modal content is the S0 mode, the A1 mode, the intersection point of these two modes and a small intensity of the A0 mode. Clearly, this does not correspond to the superposition of the primary modal content of the mid-plane and near top depths at this rise-time. At other rise-times, similar inconsistencies were present.

57

Conclusions

Applicable for Choi-Williams transform (CWT) results for a 4.7-mm thick aluminum plate based on rise-times from 0.5 µs to 15 µs and propagation distances from 60 mm to 480 mm: • For sources at the mid-plane and mid-depth, as the rise-time decreases, higher Lamb modes

become visible in the CWT results. • In contrast, for sources near the top surface, as the rise-time decreases, the CWT results are

dominated by the A0 Lamb mode to such an extent that higher Lamb modes are not visible. • For all the source depths, rise-times and propagation distances, the intense portions of the

CWT results are dominated by different frequency portions of only the A0 and S0 fundamen-tal Lamb modes. Thus, only the fundamental modes need to be considered in the determina-tion of accurate signal arrival times which would lead to accurate AE source locations.

• In an experimental situation, to monitor the intense portions of the fundamental modes for the rise-times and source depths considered here would require a sensor response that is nearly flat with frequency (ASTM E1106) from about 40 kHz to about 600 kHz.

• As a function of propagation distance, similar frequency regions of the fundamental modes dominate for a given rise-time and the source depths of mid-plane and near the top.

• For the mid-depth source at the shortest propagation distances (60 mm and 120 mm) and shortest rise-time (0.5 µs), the excitation of both symmetric and anti-symmetric modes re-sults in some distortion of the CWT results at higher frequencies due to the close arrivals in time of the modes. But, by focusing on the lower-frequency intense portions of the funda-mental modes, accurate arrival times can be obtained that correspond to the theoretical Lamb-wave group velocities.

• The CWT magnitudes of higher-frequency intense mode/frequency combinations falloff rap-idly as the source rise-time increases, in contrast to a slower falloff for the lower-frequency intense mode/frequency combinations for this change of rise-times.

• The rate of attenuation with increasing propagation distance of the CWT magnitudes of the intense mode/frequency combinations is greater than that of the peak signal amplitude.

• In AE applications, the above conclusions are applicable relative to the potential to detect AE signals, to obtain accurate source locations and to approaches that use the modal content of AE signals to identify AE source types.

Acknowledgement

The finite-element calculations by Dr. John Gary (retired NIST, Boulder, CO, USA) are gratefully acknowledged. References 1. Hamstad, M.A., K.S. Downs and A. O’Gallagher, “Practical Aspects of Acoustic Emission

Source Location by a Wavelet Transform,” 21, 2003, 70-94.

2. Hamstad, M.A., A. O’Gallagher and J. Gary, “Examination of the Application of a Wavelet Transform to Acoustic Emission Signals: Part 2. Source Location”, Journal of Acoustic Emission, 20, 2002, 62-81.

3. Hamstad, M.A., “Comparison of Wavelet Transform and Choi-Williams Distribution to De-termine Group Velocities For Different Acoustic Emission Sensors,” J. Acoustic Emission, 26, 2008, 40 – 59.

58

4. Hamstad, M.A., “Frequencies and Amplitudes of AE Signals in a Plate as a Function of Source Rise-time,” Proceeding of the 29th European Conference on Acoustic Emission Test-ing, Vienna, Austria, 2010.

5. Burridge, R. and L. Knopoff, “Body force equivalents for seismic dislocations,” Bulletin Seismic Society of America, 54, 1964, 1875-1914.

6. Hamstad, M.A., A. O'Gallagher and J. Gary, "Modeling of buried acoustic emission mo-nopole and dipole sources with a finite-element technique," Journal of Acoustic Emission, 17(3-4), 1999, 97-110.

7. Vallen, J., “AGU-Vallen Wavelet transform software version R2009.1215,” Vallen-Systeme GmbH, Münich, Germany, 2009, Available at http://www.vallen.de/wavelet/index.html.

8. Vallen, J., “Dispersion software version R2009.1215,” Vallen-Systeme GmbH, Münich, Ger-many, 2009, Available at http://www.vallen.de/wavelet/index.html.

9. Kerber, F., H. Sprenger, M. Niethammer, K. Luangvilai and L.J. Jacobs, “Attenuation analy-sis of Lamb waves using the chirplet transform,” EURASIP Journal on Advances in Signal Processing, Vol. 2010, Article ID 375171, 6 pages.

J. Acoustic Emission, 28 (2010) 59 © 2010 Acoustic Emission Group

WAVEFORM ANALYSIS OF ACOUSTIC EMISSION MONITORING OF TENSILE TESTS ON WELDED WOOD-JOINTS

ANDREAS J. BRUNNER 1, THOMAS TANNERT 2 and TILL VALLÉE 3

1 Empa, Swiss Federal Laboratories for Materials Science and Technology, Laboratory of Me-chanical Systems Engineering, Überlandstrasse 129, CH-8600 Dübendorf, Switzerland, 2 Bern

University of Applied Sciences, Timber and Composite Construction, Solothurnstrasse 102, CH-2500 Biel, Switzerland, 3 Ecole d’ingénieurs at d’architectes de Fribourg, Boulevard de Pérolles

80, CH-1705 Fribourg, Switzerland

Abstract

Wood welding is a relatively recent procedure for preparing joints between pieces of wood that shows promising potential for avoiding some of the problems related to conventional adhe-sive jointing of wood. As a first step towards more extensive characterization of their damage behavior and of the failure mechanisms, tensile tests on welded wood joints have been monitored with acoustic emission (AE). This contribution attempts a more detailed analysis of the AE sig-nals from stair step-load tests on welded wood joints. Keywords: Welded wood, tensile test, AE waveform analysis, failure mechanisms Introduction

To connect load bearing timber structures, practitioners have at their disposal a series of methods; a first fraction of them rely on mechanical fasteners, a second type achieves load transmission by means of direct compressive contact between timber members, and a third in-creasingly considered option is adhesively bonding. It is less known that welding of wood also allows for load bearing connections of timber elements.

Wood-to-wood connections by means of welding are an innovative process, which holds

high potential for development. To achieve joints by means of welding, the wooden parts are pressed against each other and a rapid vibration heats up and melts the material at the interface within few seconds. Once the motion stops, and after cooling down, a solid bond is formed [1]. Bonds are completed in less than a minute, and no further preparation of the surfaces is required. First reports related to welding of wood date back to Sutthoff et al. [2], welding of wood based on Linear Vibration Welding (LVW) greatly improves homogeneity and resistance of the result-ing bonds [3]. Two sets of parameters proved to have an influence on the bond strength: firstly, parameters related to the wood [4-5], and secondly parameters related to the LVW device [6].

Since welding of wood is an alternative to adhesive bonding, it is also of interest to compare

the mechanical properties and the AE behavior, i.e., the damage accumulation and failure with that of adhesively bonded joints made from the same type of wood [7]. The AE behavior (AE activity, AE intensity, Felicity-ratio, and linear AE source location) under stair step loading for both types of joints is compared and discussed in [8].

The objectives of this paper are (i) to compare AE waveforms recorded with the two types of

sensors used in the experiments on welded wood joints, and (ii) to look for trends in the fre-quency content of the AE signals with increasing intensity (e.g., AE signal amplitudes observed with increasing loads).

60

Material and Experimental

In order to investigate the strength of wood welded joints, a series of single-lap welded tim-ber joints was produced in which the sole varied parameter was the overlap length, L. The timber species used was spruce (Picea abies) cut from high quality almost defect-free boards. The mate-rial was conditioned to 12% moisture content prior to manufacturing of the specimens, and then again stored in constant climate before and after welding and until testing.

A total of 25 wood welded single lap joints were manufactured by welding two timber boards

(700 mm long, 60 mm wide and 15 mm thick) by means of a Branson M-DT24L linear vibra-tion-welding machine; Fig.1 details a specimen. Subsequently, a grove up to the wood weld was cut in each of the now connected boards, the distance between the two groves defining the over-lap length. The overlap length was varied from 100 mm to 400 mm in steps of 100 mm. Each of the welded lap joints was manufactured and subsequently tested five times. Five additional speci-mens with an overlap length of 100 mm were produced to study the AE behavior.

Fig. 1: Schematic geometry of single lap joint welded specimens (not to scale).

Tensile tests with AE monitoring were performed on a servo-hydraulic test machine (type In-

stron 1251) with hydraulic grips (clamping pressure 30 bar) at 3 mm/min (first specimen) and 0.3 mm/min (all other), respectively. Specimens were stored and tested under laboratory conditions of +23°C and 50% relative humidity (differing from the standard conditions +22°C and 65% relative humidity for wood) after reducing the total thickness from 30 mm (Fig. 1) to 25 mm. Aluminum distance bars were used to prevent extensive compression of the specimens in the grips. A load cell with 200 kN (specimens 1 and 2) and 50 kN range (specimens 3-5) was used.

Acoustic Emission (AE) monitoring of the tensile tests was performed with commercial AE

equipment (type AMSY-5 from Vallen Systeme GmbH) with a total of eight AE sensors (for type SE45-H, and four type SE150-M). Two sensors (SE150-M) were used as guard sensors on the hydraulic grips, the other six were mounted on the specimen with spring clamps using sili-cone-free vacuum grease as couplant. Threshold was set at 40 dBAE and 50 dBAE, for specimens mounted near the top and bottom grip, respectively and at 55 dBAE for the guard sensor on the bottom grip. Both, AE signal parameters and full AE waveforms were recorded, with 5 MHz sampling. Machine load and crosshead displacement was recorded simultaneously with sampling every 100 ms. Data analysis was performed with equipment specific software (VisualAE® and

61

VisualTR® from Vallen Systeme GmbH). Specifically, the Fast Fourier Transform (FFT) feature extractor routine (VisualTR®) has been extensively used in the AE waveform analysis.

Results and Discussion Mechanical Properties

The mechanical properties are reported in detail in [8] and only a few essential points are re-peated here. All wood welded single lap joints tested with AE monitoring exhibited almost per-fectly linear-elastic load-displacement behavior, and failed in a brittle and sudden manner. Fail-ure load and displacement at failure show a fairly linear correlation for all specimens. One specimen was tested quasistatically with AE monitoring (displacement control at 3 mm/min). The remaining four specimens were then tested under quasistatic tensile loading with stepwise increase and unloading under displacement control (0.3 mm/min). The specimen tested qua-sistatically to failure yielded a failure load of 6.7 kN and a crosshead displacement at failure of 0.28 mm; those tested stepwise an average of 5.3 kN (coefficient of variation ± 2.4 kN or 38.1%) with an average crosshead displacement at failure of 0.24 mm (specimen deformation was not directly measured). A closer post-failure observation indicates that the welding process did not always yield in perfectly welded surface, see Fig. 2.

Fig. 2: Failure surfaces of two welded wood specimens tested with AE monitoring (left) and of two others including one specimen without separation of the two joint faces (right), on which the sensor positions are marked with pencil; note the light areas indicating poorly welded surfaces among the dark surface features.

The failure surfaces of the specimens after tensile failure showed a pattern of dark brown and

light yellow color. Qualitatively, specimens 1 and 2 (Fig. 2, left) show a larger amount of light yellow color. This seems to correlate with lower failure loads (6.7 kN, but at higher strain rate, and 3.0 kN, respectively) than the other specimens. It can be noted that on the edges, the fracture surfaces indicate sufficient welding (dark brown areas). Visual inspection before testing hence did not reveal the difference in welding quality.

AE Signal Parameter Analysis

AE activities, AE intensity as a function of load, and linear as well as planar AE signal source location have also been discussed in detail in [8]. Again, the major conclusions are sum-marized here. AE activity and AE intensity both are increasing with increasing level of loads. Under stepwise loading, the joints show a clear Felicity effect with decreasing Felicity-ratio for

62

increasing load steps. Values of the Felicity-ratio around or below 0.90 are indicative of critical damage. Even though AE activity and AE intensity (e.g., measured by the AE signal amplitude) drop when holding the load constant, a surge in AE activity and AE intensity is observed upon unloading from each load level (even if the Felicity-ratio is still >1). This could indicate damage that occurred during loading, but also damage existing in the specimens from the welding proc-ess, such as, e.g., incomplete bonding evident as light yellow surface parts (Fig. 2).

Linear AE signal source location plots projected on the narrow (thickness) side of the speci-

mens yield AE source location clusters (especially when filtered for events with amplitudes ≥60 dBAE), isolated in time roughly in the area of the weld or nearby. The times of occurrence corre-late well with load changes, i.e., the AE activity and AE intensity surges. This highlights the weld zone as a weak area with significant damage accumulation. Damage starts quite early in the tests, i.e., at relatively low loads. This leads to failure by separation of the two adherends in a plane inside the weld-zone. The mechanisms contributing to failure can, however, not be identi-fied with this type of analysis.

AE Waveform Analysis

Analysis of the recorded AE waveforms will be explored for providing more information on the events and mechanisms leading to failure. As a first step, Fast Fourier Transforms (FFT) cal-culated from the recorded AE waveforms are compared for both types of sensor and for different load levels. In a second step, two features of the FFT are compiled (using the FFT Feature Ex-tractor routine provided in the Transient. Recording data analysis package of Vallen Systeme GmbH). This routine determines the frequency at which the peak amplitude in the FFT occurs, as well as that of the center of gravity of the FFT spectrum. These features are compared for the different sensors types and for the different stages of the tests.

From a visual inspection of the FFT calculated from the AE waveforms, there is a slight

trend for a higher share of AE signals with relatively low frequency content, i.e., essentially be-low 100 kHz and 150 kHz, for the sensor type SE45-H and SE150-M, respectively (Fig. 3). It has to be noted that a high-pass frequency filter of 30 kHz has been used in all tests, and that a Hamming window was applied for calculating the FFT.

Visual inspection, therefore, roughly yields two classes of AE waveforms, one with relatively

low frequency content and one with higher frequency contributions (above about 100 kHz and 150 kHz for the two sensor types, respectively). This hints at different AE signal source mecha-nisms, since signal attenuation in the specimens is relatively small and the effect is observed for sensors at comparable locations (near the top of the specimens).

As noted above, a software routine (FFT Feature Extractor, VisualTR® from Vallen Systeme GmbH) has yielded two features from the FFT of the recorded AE waveforms, namely two fre-quencies, i.e., that at which the maximum amplitude in the power spectrum and that at which the center of gravity of the FFT occur. Visual inspection of FFT calculated with different cut-off windows do, in some cases, show significant differences (e.g., between rectangular or Ham-ming), notably resulting in different values for the frequency at which the maximum amplitude of the FFT occurs. Further, in order to reduce the influence of noise at higher frequencies (above 250 to 300 kHz), a cutoff of 5% of the maximum amplitude has been applied. Calculation of the FFT features without a cut-off resulted in a significantly larger spread in the center-of-gravity frequencies, especially for low amplitude AE signals. Figure 5 shows plots of the FFT features for both types of sensors used in the test.

63

595 581

2658 2628

6531 6455 Fig. 3: Fast Fourier transforms (with Hamming window) of AE signals recorded in the early (be-low 3 kN), middle (below 7 kN) and late stage (below 9 kN) of a tensile test with stair step load pattern, (Left) Sensor type SE-45H, (Right) Sensor type SE-150M, showing relatively low fre-quency content (essentially below 150 kHz and 200 kHz, respectively).

From the graphs in Fig. 5 it is clear that there are at least two distinct classes of AE signals

that occur during the load tests to failure. This distinction is evident for both types of sensors, even though the separation between the two clouds is smaller for the sensor type SE-45H than for the resonant sensor type SE-150M. Clearly, there is a larger spread in the frequency at the center-of-gravity of the FFT for the AE signals recorded with the resonant sensor (type SE-150M). For sensor type SE-45H, the plot of frequency at maximum amplitude of the FFT may

64

even indicate a separation into three classes of signals, one ranging from about 30-50 kHz, one from about 70 to 85 kHz, and third from about 90 to 110 kHz.

585 594

3146 2876

6319 6415

Fig. 4: Fast Fourier transforms (with Hamming window) of AE signals recorded in the early (be-low 3 kN), middle (below 7 kN) and late stage (below 9 kN) of a tensile test with stair step load pattern, (Left) Sensor type SE-45H, (Right) Sensor type SE-150M, showing higher frequency content above 150 kHz and 200 kHz, respectively (bottom).

The data sets from both sensor types further indicate that higher amplitude AE signals (above

about 60 dBAE) tend to have a higher frequency at which the maximum amplitude in the FFT occurs. With respect to the frequency at which the center-of-gravity of the FFT occurs (data not

65

shown), the separation into two signal classes is not observed. Figure 6 shows a further analysis of the frequency at which the maximum amplitude of the FFT occurs. This time, the data are plotted as a function of time and the corresponding load signal is shown as well. It can be noted that the separation into two classes is observed almost from the start of the loading (except for the first load step, for which very few signals are recorded). If this separation is interpreted in terms of different AE signal source mechanisms, they are active throughout the test while load is applied (except for the Felicity-effect upon reloading). A closer inspection of the distributions shows a trend for signals with a higher frequency at maximum FFT amplitude to occur mainly during load increase (at least for the first few load steps), with fewer signals observed for in-creasing hold times at constant loads than those with lower frequencies. With increasing load steps, this difference becomes less obvious and at the last load level before failure, both types of signals are observed even during unloading.

Fig. 5: Features of fast Fourier transforms of AE signals recorded in a tensile test with stair step load pattern, (Left) Sensor type SE-45H, (Right) Sensor type SE-150M, (top) frequency of center of gravity versus that of maximum amplitude, (bottom) frequency of maximum amplitude versus AE signal amplitude.

Even with clear evidence for at least two, possibly three classes of AE waveforms that are re-corded during the tests, a direct identification of the responsible AE signal source mechanisms is still not possible. Since mechanical properties of the wood from the welded zone (e.g., stiffness, hardness) are not available it is, however, not clear whether and possibly how much AE signal source mechanisms or AE signal propagation are affected by a possible variation in mechanical properties. The relatively low frequency content of one class of AE signals tentatively hints at

66

friction or friction-like effects. This would also be consistent with the effective shear load that is induced in the welding zone by the tensile load applied to the specimens. It is possible that the poorly welded areas and the observed roughness of the failure surfaces (see Fig. 2) provide a source of these friction-like signals. The AE signals with higher frequency content then quite likely reflect other damage mechanisms. The absence of AE signals with high amplitudes or high signal energy throughout the test (except at failure) would be consistent with defect formation on a “small” scale, excluding, e.g., large, sudden disbonding or delamination.

(a)

(b)

Fig. 6: Frequency of maximum amplitude of fast Fourier transforms of AE signals recorded in a tensile test with stair step load pattern versus time. (a) Sensor type SE-45H, (b) Sensor type SE-150M.

67

Conclusions

AE monitoring of load test on welded wood joints has indicated that AE activity and hence damage occurs mainly in the weld zone (around the center plane of the specimens). AE wave-form analysis has provided evidence for two or three distinct classes of AE signals that occur throughout the test. AE signals with higher frequency content do appear more frequently during loading and less during hold at constant displacement (i.e., slight load decay). The AE signal amplitudes (mostly below 70 dBAE) and the AE signal energies (mostly below 1x105 e.u. from time-integration of the squared voltage signal) hint at damage occurring at micro-or meso-scale, slowly accumulating to larger defects resulting in final failure. Friction or friction-like effects may also play a role, since there are a significant number of signals with “low” frequency con-tent. Visual inspection of the failure surfaces after the test further shows that there is a certain percentage of unbonded or poorly bonded weld area. The size of that roughly correlates with the failure loads (low loads for larger unbonded area). This may provide a possible source of fric-tion-like signals with low frequency content. Acknowledgment

The assistance of Mr. D. Völki for test setup and data acquisition at Empa, and of Dr. Thomas Thenickl (Vallen Systeme GmbH) for support with the data analysis is gratefully acknowledged. References 1. Ganne-Chédeville C, Properzi M, Leban JM, Pizzi A, Pichelin F.: Journal of Adhesion Sci-

ence and Technology, 22 (7), 2008, 761–773.

2. Sutthoff B, Franz U, Hentschel H, Schaaf A (in German) Patentschrift (Patent) DE 196 20 273 C2, 1996, Deutsches Patent- und Markenamt (German Patent and Trade Mark Office).

3. Gfeller B, Zanetti M, Properzi M, Pizzi A, Pichelin F, Lehmann M, Delmotte L.: Journal of Adhesion Science and Technology, 17 (11), 2003, 1573–1589.

4. Properzi M., Leban J.M., Pizzi A., Wieland S., Pichelin F., Lehmann M.: Holzforschung 59 (1), 2005, 23-27.

5. Stamm B, Natterer J, Navi P.: Holz als Roh- und Werkstoff, 63 (5), 2005, 313–320.

6. Ganne-Chédeville C, Duchanois G, Pizzi A, Leban JM, Pichelin F.: Journal of Adhesion Sci-ence and Technology, 22 (12), 2008, 1209-1221.

7. Brunner A.J., Terrasi G.P., Vallée T., Tannert Th.: Proceedings 22nd International Symposium Swiss Bonding, May 11-13, 2009, Rapperswil, Switzerland.

8. Brunner A.J., Tannert Th., Vallée T., Proceedings 29th European Conference on Acoustic Emission, European Working Group (EWGAE), September 8-10, 2010, Vienna, Austria, pa-per No. 4, 7 p. (2010).

J. Acoustic Emission, 28 (2010) 68 © 2010 Acoustic Emission Group

USE OF ACOUSTO-ULTRASONIC TECHNIQUESTO DETERMINE PROPERTIES OF REMANUFACTURED PARTICLEBOARDS MADE

SOLELY FROM RECYCLED PARTICLES

SUMIRE KAWAMOTO

Forestry and Forest Products Research Institute, Ibaraki 305-8687, Japan Abstract

Properties of particleboard manufactured entirely from recycled particleboard were deter-mined and a method for processing three-layered particleboard from all-recycled particles was described. Dynamic MOE (modulus of elasticity) before and after re-manufacturing was tested by a longitudinal stress wave technique and compared with other stress wave techniques. Nonde-structive AU (acousto-ultrasonic) techniques were used to evaluate static, dynamic, and dimen-sional properties of re-manufactured boards. The results showed a large decrease in mechanical and physical properties for remanufactured boards, particularly for those made from high-density boards. AU parameters corresponded with density in the volume between the surfaces of trans-ducers. Correlation between static MOE and dynamic MOE was improved by using the AU cal-culated density. Water absorption and thickness swell corresponded with the AU calculated den-sity.

Keywords: Particleboard, AU (acousto-ultrasonics), recycle, remanufacture, dynamic MOE (modulus of elasticity), MOR (modulus of rapture), density, water absorption

Introduction

For the sake of the environment, raw materials for manufacturing wood-based materials

could be derived from unprocessed forest products such as wood material from thinning opera-tions [1, 2], industrial waste from veneer manufacture or sawmills, and slabs produced by indus-trial operations on dry wood. In addition to such wet and dry wood wastes, fibrous agricultural lignocellulosic natural resources such as flax or kenaf can be used as raw materials [3, 4]. Gener-ally, wood-based composites manufactured from recycled materials use a combination of recy-cled and unused new materials. It is necessary to add new fiber, particles, or strands to compen-sate for the decrease in strength resulting from the reuse of materials. Although new products made from waste wood have been developed in many projects [1, 5], only a few previous works focused solely on recycling particleboard [6, 7].

The objectives of this study are (1) to provide fundamental data regarding the decrease in

mechanical properties resulting from the use of recycled materials, (2) to compare dynamic MOE values of longitudinal stress wave techniques, and (3) to evaluate the mechanical and physical properties of particleboard using acousto-ultrasonic (AU) technique. A long-term goal of this study includes developing better methods for adding new constituents to increase the strength of a recycled structural composite panel.

This study shows an example of using AU techniques to determine the decreased properties

of remanufactured particleboard compared to the original three-layered industrial particleboard. Particleboard is a general term for a panel manufactured from lignocellulosic materials (usually wood), primarily in the form of discrete pieces or particles, as distinguished from fibers, com-bined with synthetic resin or other suitable binder. The particles are bonded together under heat and pressure in a hot press by a process, in which the entire interparticle bond is created by the

69

added binder [8]. Particleboard manufacture is a valuable process for material recycling woody waste material that might otherwise be landfilled or burned.

This study shows an example of using AU techniques to determine the decreased properties of remanufactured particleboard compared to the original three-layered industrial particleboard. Particleboard is a general term for a panel manufactured from lignocellulosic materials (usually wood), primarily in the form of discrete pieces or particles, as distinguished from fibers, com-bined with synthetic resin or other suitable binder. The particles are bonded together under heat and pressure in a hot press by a process, in which the entire interparticle bond is created by the added binder [10]. Particleboard manufacture is a valuable process for material recycling of woody waste material that might otherwise be landfilled or burned.

Properties of particleboard before and after re-manufacturing were compared nondestruc-

tively by both an AU method and a conventional stress wave technique. The acousto-ultrasonic technique, a combination of acoustic emission (AE) signal analysis with an ultrasonic charac-terization method [9], was used to detect and assess damage conditions and variation in me-chanical properties of a test material [10]. Vary used AU evaluation to determine the mechanical properties of composite materials [11]. In regard to wood composite research using AU tech-nique, Beall et al. obtained a patent for bond strength measurement of composite panel products using rotary transducers [12]. Green [13] and Rodgers et al. [14] reported that AU parameters corresponded with IB (internal strength) [13, 14]. These reports described the correlation of AU to IB and were reviewed by Beall in 1993 [15]. Reis investigated AU behavior concerning the dimensional stability of medium-density fiberboard (MDF) [16]. Unlike solid wood, few studies have used AU technique to evaluate wood composites, probably because of the difficulty of measuring properties because of attenuation of AU waves traveling through wood-based com-posites [17].

This study provides fundamental data on a processing technique and on the mechanical

properties of boards made with recycled particles from industrial recycled waste wood. Some new AU techniques are suggested for evaluating the mechanical and dimensional properties of particleboard.

Experimental

Original Materials

Raw materials were commercial three-layered particleboard (Tokyo Board Co. Ltd.). Dimen-sions were 0.9 m in width, 1.8 m in length, and 20 mm in thickness. Boards were manufactured from industrial waste wood such as plywood or slabs that were discarded as a waste in the same company. Species used in the original particleboard were identified as western hemlock (Tsuga heterophylla), Douglas-fir (Pseudotsuga menziesii), and a mixture of tropical species. Three boards, 150 mm in width, were sawn from the edge of the commercial particleboards for remanufacture into three 3-layered particleboards. These three original boards were selected from 50 boards, based on dynamic MOE (modulus of elasticity) values that were measured non-destructively. The dynamic MOE was calculated by a longitudinal stress wave technique using a PVDF (polyvinylidene fluoride) transducer [18]. Density of original boards was 0.675, 0.754, and 0.806 (g/cm3), and dynamic MOE tested by Metrigard equipment was 2.98, 4.09, and 4.60 GPa, respectively.

70

Remanufacturing Method The original particleboards were cut into 50 by 50 mm square pieces on a table saw, then

hammer milled. Particles were classified by size using the screen after drying to 3% moisture content. Particles were then separated using a 16-mesh screen. For the core layer, 16-mesh-on were used. For the surface layers, particles of 16-mesh-off were used. Weight loss from size re-duction process by table saw was 12% to 13% and by hammer mill 3% to 4%, respectively.

Phenolic resin binder (10% of total weight of mat) was sprayed in a rotating drum blender.

Forming (spreading particles in 30- by 30-cm square mat former) was conducted by hand. The weight ratio of the surface layer and core layer was 1.5:7:1.5. Planned densities and thickness were the same as values of the original particleboards (0.675, 0.754, and 0.806 (g/cm3). Hot pressing temperature was 180°C. Pressing time was 20 min. Boards were trimmed to 280 by 280 mm and stored in a conditioning room at 20°C and 65% relative humidity, before and during the property test. For the control, a single-layered particleboard was manufactured using industry sawdust. The species was identified as a mixture of yellow pine and spruce (636 kg/m3).

Testing Properties of Remanufactured Particleboard

Properties of the remanufactured particleboard were measured both by dynamic and static methods. Figure 1 shows locations of each specimen for property tests, which included dynamic and static MOE and modulus of rapture (MOR), density, internal bond strength (IB), thickness swelling (TS), and water absorption (WA). The test was basically performed in accordance with JIS A 5908 [19]. To avoid the effect of water repellant added to original materials, specimens cut from control board were used.

AU Measurements

AU equipment included a pulse generator, transmitting and receiving transducer, and AE equipment. Resonant 1.5-MHz, 20-mm diameter and 150-kHz, 9-mm diameter AE transducers were used. The pulse generator generated a square wave, which excited a transmitting transducer. The output signal was amplified by 40 dB. The maximum amplitude (up to 50-µs from start of signal) and time differential, observed by oscilloscope, were taken as AU parameters.

For determination of density and dimensional properties of whole specimen using AU tech-

nique, 25- by 25-mm specimens were used so that a face of the 1.5-MHz transducer covered most of the surface area of the specimens. AU waves that propagated through the thickness of specimens were observed. Preliminary tests confirmed that AU parameters detected in using a specimen having conventional test size (50 by 50 mm) did not correspond to actual density and dimensional properties. For measurement of AU transmission time traveling in the through lon-gitudinal direction, 150-kHz AE transducers, 9-mm diameter, were used.

For determination of dynamic MOE by AU transmission time, an impact source was a sud-

den stress release generated by pencil-lead breaks (PLB) at the edge of the specimen. The lead length was always about 10 mm. This stress wave signal was very similar to the AE signal and was thus called the artificial AE. Transmission of the longitudinal direction was observed. AU transmission time (the time it took for the signal to travel between the two 150 kHz transducers), was observed by using an oscilloscope. For this AU technique, refer to author’s original report [20].

71

Table 1. Physical and mechanical properties of original and remanufactured particleboards.

Speci-men Number

Density (kg/m3) Original1 Remanu- factured2

Dynamic MOE (GPa) Original3 Remanufactured (a)4 (b) 5 (c) 5

Static MOE5 (GPa)

MOR5 (MPa)

IB6 (MPa)

Ｉ 675 640 2.98 1.59 1.62 1.25 1.36 7.68 0.39 II 754 656 4.09 1.46 1.45 1.35 1.25 6.09 0.22 III 806 677 4.60 1.95 1.90 1.44 1.48 6.98 0.56 Control 664 1.96 1.89 1.52 1.84 11.2 0.27

Notes: 1 Specimen dimensions: 150 mm x 20 mm x 1.8 m. 2 Specimen dimensions: 280 mm x 280 mm x 20 mm. 3 Specimen dimensions: 150 mm x 20 mm x 1.8 m. Longitudinal AU technique using PVDF transducer [19]. 4 PLB on 280 mm x 280 mm remanufactured particleboards before MOR specimens were cut. 5 Specimen dimensions: 345 mm x 50 mm x 20 mm. (cf. Fig. 2). 6 Specimen dimensions: 50 mm x 50 mm x 20 mm. (cf. Fig. 2).

Fig. 1. Geometry for testing properties. Measurement for: D, density; IB, internal bond strength, TS-D, thickness swell and water absorption; MOE, dynamic and static modulus of elasticity; and MOR, modulus of rapture. Determination of Dynamic MOE by Longitudinal Stress Wave Technique 1) Stress Wave Technique by Transmission Time Stress wave transmission time was measured two different ways (AE and PVDF transducers). With AE transducers, an impact was induced by the sudden break of lead of a mechanical pencil. That is, this artificial AE wave method used a mechanical pencil as an AU source and is included as an AU technique. Using PVDF transducers, an impact was induced by tapping an end of each specimen. 2) Stress Wave Technique by Resonant Frequency of Longitudinal Vibration Tapping the end of each specimen generated a stress wave. The signal detected by a PVDF transducer was processed through an FFT (fast Fourier transform) analyzer, and the resonant frequency of the first resonance mode was observed.

72

Result and Discussions Decrease of Properties of Remanufactured Particleboard Table 1 shows the properties of the original and remanufactured boards. The original parti-cleboard with a highest density (0.806 g/cm3) showed a larger decrease of properties than boards made from the lower densities. Property decrease was the largest for the particleboard (III) that had the largest density and dynamic MOE in original boards. The remanufactured boards were intended to have the same density as the original boards, however, their density decreased to 94, 87, and 84% (Table 1, density). This is because larger density raw materials are subjected to larger deformation by pressure and heat during the processing.

Decrease of dynamic MOE obtained by resonant frequency of longitudinal vibration was 40%-65%. Compared to representative values of similar samples from the company that pro-vided original specimens, Static MOE of remanufactured board decreased to 53% to 40%. MOR decreased to 44% to 26%. MOR values showed larger decrease than MOE. Comparison of Stress Wave Techniques

Generally dynamic MOE values obtained by vibration of resonant frequency are smaller than dynamic MOE values obtained by stress wave transmission time propagating in solid wood. However, in the case of particleboard, that relationship is reversed. Comparing the stress wave technique by using PVDF, dynamic MOE values by stress wave transmission time (Table1, (c)) were smaller than values by longitudinal vibration technique calculated by resonant frequency (Table1, (b)). This is probably because the transmission time would be extended by attenuation of waves reaching the further transducer. Attenuation of stress waves in particleboard is much larger than in solid wood. Observing wave signals by oscilloscope, larger attenuation caused dif-ficulty identifying starting points of signal waves. If it attenuated below the noise level, the ac-tual initiation point would be missed. Several cycles of wave would be hidden under the noise level. Thus the initiation point would be extended and longer wave transmission time was ob-served in particleboard.

The stress wave method by artificial AE (Table 1, (a)), showed larger dynamic MOE values than the static MOE values, which was the same as the trend for solid wood and unlike using PVDA transducers. The sudden break of mechanical pencil lead generated a sharp gradient at the start of artificial AE waves. Compared to output signal by PVDF, the artificial AE signal was easy to identify at the starting point.

Comparing three techniques (a), (b) and (c) in Table 1, dynamic MOE values obtained by

resonant frequency (b) corresponded with static MOE the best. MOE values (a) obtained by arti-ficial AE transmission time also showed the same trend and corresponded with static MOE as much as the conventional method (b). Advantage of the artificial AE transmission method is that the measurement is simple. Specimens don’t need to be cut, and can be measured just using a mechanical pencil without using large equipment to induce a pulse. AU and IB Strength

AU amplitudes and AU transmission time of AU signals that propagated through the thick-ness of IB specimens were observed at five locations on IB specimens. Unlike other properties, AU result did not correlated with IB strength. This was because AU values corresponded with density in the volume between the surfaces of transducers (20 mm in diameter), whereas IB strength values were measured in specimens, 50 by 50 (mm). This result also indicates that AU

73

measurement should be taken with considerable attention to ensure using the right AE transduc-ers. Beall [15] and Green [13, 14] reported high correlation of IB strength and AU amplitude of signal propagating through the thickness of IB specimens. In these studies, they used transducers having large diameters that covered almost all the areas of the surface of IB specimens. AU and Density

AU amplitude and transmission time through thickness of specimens correlated with density. Regression coefficient between AU transmission time and density was higher (Fig. 2, r = 0.99) than the one obtained by AU amplitude (r = 0.94). This indicates that the AU technique to de-termine density can be used to estimate property values that correlate with density. In this study, it was used to determine dynamic MOE and the dimensional properties (Fig. 3). The benefit of this method is that specimens don’t require cutting because this AU technique uses AU values in propagation through the thickness of specimens and can be done on a large specimen.

Fig. 2. AU transmission time versus density for particleboard (control). AU Density was deter-mined from AU transmission time measured in propagation through thickness. AU and Dimensional Properties

Water absorption and Thickness swelling (TS) correlated with density (r = 0.97 for 16.5 h, r = 0.96 for 6.4 h). Water absorption corresponded more than TS. Water absorption was calculated from density determined with AU transmission time propagating through thickness (Fig. 2). The regression coefficient was the highest in the relationship between calculated density and water absorption after 16.5 h (Fig. 3). Although only five specimens were used in this work, it shows the possibility for using AU techniques to estimate dimensional pensile proprieties.

Conclusions

(1) Mechanical and physical properties of remanufactured board decreased more with original particleboard having larger density. This indicates that it is important to use lower density mate-rial as a raw material.

(2) The decrease in MOR, caused by the manufacturing, was larger than in static MOE.

(3) The decrease in dynamic MOE obtained by resonant frequency of longitudinal vibration, measured by PVDF transducer, was 40 - 75 %.

74

Fig. 3. Calculated density versus water absorption for particleboard (control). ■: 16.5 h, ●: 6.5 h. (4) Dynamic MOE using PVDF transducer obtained by stress wave transmission time was smaller than the one by frequency of longitudinal vibration. (5) Dynamic MOE obtained by artificial AE wave signals induced by breaking the lead of a mechanical pencil was similar to stress wave MOE values calculated by longitudinal vibration using PVDF transducers. (6) AU behaviors highly corresponded with density in the volume between the surfaces of the transducers. Properties involved with density can be estimated by AU techniques. Scanning of AU signals that propagated through the thickness of particleboards can provide density value nondestructively, thus properties that correspond with density can be estimated by this AU techniques. (7) For the specimens that did not contain water repellent, water absorption correlated with calculated density determined by AU techniques. Acknowledgments This research project at the USDA Forest Products Laboratory, Madison, Wisconsin, was funded by former Science Technology Agency, Government of Japan. The authors wish to thank Mr. Saito, Tokyo Board Co. Ltd. for providing commercial particleboard, Mr. Yasuto Chiba, Japan Forestry and Forest Products Research Institute, retired, for conducting the internal bond strength test, Dr. Regis Miller, (retired), USDA Forest Service, Forest Products Laboratory, for identifying species of particles. First author thanks James H. Muehl (retired), R. Sam Williams (retired), and Robert J. Ross, Forest Products Laboratory, for providing laboratory space at the Forest Products Laboratory.

75

References 1. T. Shibusawa, S. Kawamoto, H. Korai, and T. Fujii. US Patent No. 5814170. 1998.

2. H. Spelter, R. Wang, and P. Ince. FPL-GTR-92. Madison, WI: US Department of Agriculture, Forest Service, Forest Products Laboratory. May 1996, 17 p.

3. R.M. Rowell. Composites from Agri-Based Resources. In: R. Falk, ed. Proceeding No. 7286. Madison, WI: Forest Products Society, 1996, 217-222.

4. S. Kawai, Mokuzai Gakkaishi. 43(8) 1997, 617-622. (in Japanese).

5. Forest Products Laboratory. FPL-GTR-110. Madison, WI: US Department of Agriculture, Forest Service, Forest Products Laboratory. May, 2000, 31 p.

6. T. Shibusawa, Y. Hatano, S. Kawamoto, M. Zhang, Y. Okuno and Y. Chiba. 50th Annual Meeting of Japan Wood Research Society, 2000, 286. (in Japanese)

7. Y. Tominaga, N. Kobayashi, M. Ohmi, Mokuzai Kogyo (Wood Industry). 54 (10) 1999, 473-477.

8. ASTM D 1554-2005, Standard Terminology Relating to Wood-Base Fiber and Particle Panel Materials, Annual Book of ASTM Standards. Philadelphia, PA: American Society for testing and Materials. 2005.

9. ASTM E1316-10a. Terminology for Nondestructive Examinations, Annual Book of ASTM Standards. Philadelphia, PA: American Society for testing and Materials. 2009.

10. ASTM E1495-2007. Standard Guide for Acousto-Ultrasonic Assessment of Composites, Laminates, and Bonded Joints, Annual Book of ASTM Standards. Philadelphia, PA: American Society for testing and Materials. 2007.

11. A. Vary, NASA Technical Memorandum 79180. 1979.

12. D.M. Shearer, R.C. Beetham and F.C. Beall, Bond Strength measurement of composite panel products. Patent No. 4750368. 1998, 7p.

13. A.T. Green, Journal of Acoustic Emission. 8(1/2). 1989, S306 S310. 14. J.M. Rodgers, A. T. Green and S.W. Borup, Materials Evaluation, 5 1991, 556-571. 15. F.C. Beall, Second International Conference on Acousto-Ultrasonics. Atlanta. The American

Society for Nondestructive Testing, Inc. 1993, 153-161. 16. H.L. Reis and D. M. McFarland, Journal of Acoustic Emission. 5(2). 1986, 67-70. 17. S. Kawamoto and R.S. Williams, General Technical Report. FPL-GTR-134. Madison, WI:

US Department of Agriculture, Forest Service, Forest Products Laboratory. May 2002, 16 p. 18. R.J. Ross and R.F. Pellerin, General Technical Report. FPL-GTR-70. Madison, WI: US

Department of Agriculture, Forest Service, Forest Products Laboratory. May, 1994, 40 p. 19. JIS A 5908. Particleboard, Japanese Industrial Standard. Tokyo, Japan. 1994. 20. S. Okumura, S. Kawamoto and M. Toyoda and M. Noguchi. Bulletin of the Kyoto University

Forest. 60, 1988, 299-309. (in Japanese)

J. Acoustic Emission, 28 (2010) 76 © 2010 Acoustic Emission Group

ACOUSTIC EMISSION ACTIVITY OF SPRUCE SAPWOOD BECOMES WEAKER AFTER EACH DEHYDRATION-REWETTING CYCLE

SABINE ROSNER 1 and SUMIRE KAWAMOTO 2

1 Institute of Botany, Department of Integrative Biology, University of Natural Resources and Applied Life Sciences, BOKU Vienna, Gregor Mendel-Str. 33, A-1180 Vienna, Austria;

2 Forestry and Forest Products Research Institute; Tsukuba, Ibaraki 305-8687 Japan Abstract

Acoustic emission (AE) during dehydration at ambient temperature was compared between fully saturated fresh Norway spruce (Picea abies (L.) Karst.) sapwood and sapwood exposed to two dehydration-rewetting cycles in order to get information about the differences in dehydration stress. AE testing was performed within a frequency range of 35-100 kHz (R6) and 100-1000 kHz (WD) at ambient conditions. AE activity became lower and AE energies weaker after each dehydration-rewetting cycle. During the first dehydration run, highest mean AE energies were detected at the beginning of dehydration. In rewetted wood, highest mean AE energies were de-tected towards the end of dehydration. AE of rewetted wood was also characterized by a higher percent of AE with frequencies >70 kHz and 100-175 kHz, as detected by R6 and WD transduc-ers, respectively. It is concluded that fresh never-dried sapwood is more prone to dehydration stresses than pre-dried sapwood. Differences in AE number and AE features appear to be due to micro-mechanical failure that decreased after each dehydration-rewetting run. Dehydration stress seems also to decrease because the membranes of bordered pits, which act as a valve to avoid the breakage of the water columns inside the conduits, become weakened after each dehydration-rewetting run. Keywords: Acoustic emissions, wood drying, dehydration stress Introduction

Dehydration stress can lead to a loss in mechanical strength in re-wetted conifer sapwood due to structural changes of the cell walls and micro cracks [1, 2]. Sapwood is the wetter outer part of a tree trunk, where water is transported from the roots up to the crown within a network of small conduits. The transport from conduit to conduit is achieved through small valves termed bor-dered pits (Fig. 1). Moderate drought stress of trees leads to a reversible closure of the pit mem-branes, in order to avoid the breakage of the water column in the adjacent conduit [3]. High negative pressures that develop during dehydration of isolated sapwood can however induce a weakening or rupture of the membranes of the pits [4] or lead to irreversible pit closure [5]. The dryer inner part of a tree trunk is termed heartwood. Conifer heartwood does not transport water anymore because the pits are permanently closed (aspirated) [6].

Cell wall shrinkage starts when the moisture content drops below fiber saturation; that is

when the conduit contains no longer free water but the cell walls are fully saturated with liquid [7]. Dimensional changes can however be observed long before most of the conduits reach fiber saturation [8, 9]. Similar dimensional changes can be observed as diurnal stem diameter changes in living trees out in the field [e.g. 10]. Due to its low free moisture content, heartwood does not contribute to these diameter changes [8]. In sapwood, negative hydrostatic pressures acting per-pendicular to the conduit walls try to draw the walls inwards. When the negative pressure be

77

comes too high the water column will break, which leads to a sudden stress release. This sudden stress release induces AE with the highest amplitudes in the range of 100-300 kHz [9, 11].

Fig. 1. Four isolated water-conducting elements (tracheids) of Norway spruce earlywood. Ar-rows indicate bordered pits. The reference bar represents 1 mm.

Drying checks develop because fiber saturation is reached far earlier in the drier shell than in the wetter core of a dehydrating specimen and because of the anisotropy of wood concerning cell wall shrinkage [2, 7]. Internal checking can, however, be induced as well by the negative pres-sure of free water [12, 13].

AE testing is a useful tool for optimizing lumber drying conditions. The analysis of the burst

rates, the amplitude or energy distribution of AE signals has been successfully used to pinpoint lumber checking [14 - 16]. The bulk of AE (>15 kHz) during lumber drying is induced by the breakage of the water columns inside the conduits after the sudden release of negative pressures [15]. Crack formation during dehydration is indicated by high AE burst rates and by AE signals with very high amplitudes or energies [12, 16, 17].

The aim of this study was to compare AE during dehydration at ambient temperature be-

tween fresh Norway spruce sapwood and sapwood exposed to two rewetting cycles. Unlike pre-vious work on this topic [5], AE testing was performed within an operating frequency range of 35-100 kHz and 100-1000 kHz. It was, therefore, possible to analyze the average frequency of the AE, which potentially gives additional information about the dehydration stress of fresh and pre-dried sapwood. Material and Methods

Wood specimens came from a 50 year-old healthy Norway spruce (Picea abies (L.) Karst.) harvested in Prinzersdorf (Lower Austria). Wood bole segments, 20 cm in length, were taken immediately after felling at 3 – 4 m height from the ground.

Outer sapwood specimens with a transverse surface of about 0.9 x 0.9 cm were isolated by

splitting the wood along the grain with a chisel. Tangential and radial faces of the beams were planed on a sliding microtome. Transverse sample ends were re-cut using a razor blade. During preparation steps the wood specimens were kept wet. Fresh wood specimens were then soaked in distilled water under partial vacuum for 24 h to refill empty conduits [18] and afterwards stored at 4 °C in degassed water containing 0.01 vol. % Micropur (Katadyn Products Inc., Switzerland) to prevent microbial growth. The final shape of the specimens was 0.6 cm tangential, 0.6 cm radial, and 10.0 cm longitudinal. Fresh and pre-dried wood specimens were dehydrated for 24 h at ambient climatic conditions (25 °C, 30 % r.h.) to the equilibrium moisture content of 9.5 %.

AE (acoustic emission) was monitored with the AMSY-5 High-Speed AE system (Vallen

Systeme GMbH, Munich, Germany). Preamplifiers (40 dB) were used in connection with R6 and

78

WD transducers with an operating frequency range between 35-100 kHz and 100-1000 kHz, respectively (Physical Acoustics Corporation Princeton Jct, PA, USA). Data were recorded with a detection threshold of 40 dB (R6) and 35 dB (WD) (0 dB = 1 µV input). AE features assessed were the peak amplitude, the absolute energy and the average frequency.

AE transducers were positioned on the tangential face of fully saturated standard beams us-

ing an acrylic resin clamp (Fig. 2). Silicone paste (Wacker, Burghausen, Germany) was used as a coupling agent. The specimen was positioned on a sample holder fixed upon a compression spring, which was used to minimize the decrease in contact pressure due to wood shrinkage dur-ing the dehydration processes [9]. Contact pressure was monitored by a load cell (DMS, Type 8416-5500, range 0 - 500 N; Burster, Gernsbach, Germany) positioned between the AE trans-ducer and the screw of the acrylic resin clamp (Fig. 2). Contact pressure between transducer and wood was set to 30 N [14] and the clamp assembly was kept so deep in water that the wood was totally covered till the applied pressure reached a constant value, which was achieved after less than 20 min. Thereafter, superficial water was quickly removed from the assembly and recording of AE and coupling pressure was started.

After the cessation of all AE activity, which was achieved after about 10 hours, specimens

were re-soaked at 4 °C in distilled water containing 0.01 vol. % Micropur under partial vacuum. Specimens were re-soaked till they reached their former saturated weight, which took >96 h. Thereafter, the second AE testing run took place. Before the third AE testing run was started, wood specimens were again re-soaked till the former saturated weight was reached.

Fig. 2. Acrylic resin clamp assemblage for AE testing.

79

Results and Discussion Acoustic activity monitored with the R6 transducer (35-100 kHz)

The total number of AE became increasingly lower with each dehydration cycle. Dehydra-tion of a once rewetted specimen produced 55.9 % and dehydration of a twice rewetted specimen 33.6 % of the AE of fresh sapwood. After the second rewetting cycle, a very low AE rate at the onset of dehydration was detected (Fig. 3). With each rewetting-dehydration cycle, the signals became also weaker. AE energy had much higher mean values in the fresh than in the two pre-dried specimens (Fig. 3). Additionally, the time course of the mean AE energy values differed between treatments. Whereas highest mean energies were measured in fresh sapwood at the be-ginning of dehydration, in pre-dried sapwood highest energies were measured towards the end of dehydration (Fig. 3). Attenuation decreases with increasing moisture loss [14]. Accordingly, the

Fig. 3. Time courses of AE rate/0.1 h clustered in amplitude and frequency steps and time course of AE Energy (eu = aJ) of fresh, pre-dried, and two times pre-tried sapwood detected with the R6 transducer (35-100 kHz). The red line in the AE Energy plot represents the mean AE En-ergy/0.1h.

80

amount of signals >60 dB became higher towards the end of dehydration in pre-dried wood. In fresh wood, however, highest amounts of signals >60 dB were found at the onset of dehydration [9]. All treatments were characterized by very weak signals in the last AE peak. The frequency composition was quite different in fresh and pre-dried sapwood; fresh sapwood showed a higher amount of signals <70 kHz during the first AE peaks, whereas pre-dried wood only during the second AE peak. The third AE peak showed a very high amount of AE >70 kHz in all three specimens.

Fig. 4. Time courses of AE rate/0.1 h clustered in amplitude and frequency steps and time course of AE Energy (eu = aJ) of fresh, pre-dried, and two times pre-tried sapwood detected with the WD transducer (100-1000 kHz). The red line in the AE Energy plot represents the mean AE En-ergy/0.1h. Acoustic activity monitored with the WD transducer (100-1000 kHz)

AE detected with the WD transducer showed a similar amplitude and energy distribution as detected with the R6 transducer. With each rewetting cycle the signals became weaker and high-est mean energies were measured at the onset of dehydration in fresh, whereas towards the end of dehydration in pre-dried sapwood (Fig. 4). Rewetted sapwood showed no first AE peak, which was characterized in fresh wood by a very high amount of signals >60 dB. The amplitude

81

distribution of pre-dried wood resembled the amplitude distribution of the last two AE peaks of fresh wood, which had a very high amount of AE <50 dB. Pre-dried wood lacked also single high energy AE during the first two hours of dehydration, at moderate moisture loss. The first and second AE peak of fresh sapwood had highest amounts of signals between 100 and 175 kHz, whereas the amount of AE within this frequency range was neither higher in once rewetted nor in twice rewetted sapwood during the whole dehydration cycle (Fig. 4). The lowest amount of 100- 175 kHz AE was detected after the second rewetting-dehydration cycle. The last AE peak of fresh dehydrating wood was characterized by a high amount of high frequency AE, as it was typical for pre-dried sapwood throughout the whole dehydration period. Possible reasons for lower AE activity in pre-dried sapwood

During the first dehydration run, irreversible damage must have occurred, which caused no AE during the second and third dehydration run. Events during the first run were characterized by higher AE peak amplitudes and a higher percentage of AE with lower frequency. A lower number of AE detected during the following dehydration runs indicated that fresh sapwood was more prone to dehydration stress than pre-dried wood [19]. Differences in AE rate, peak ampli-tude and AE energy were quite small between the second and third dehydration run [19, 20]. Recently, we [5] reported higher dimensional changes at moderate moisture loss in fresh than in pre-dried wood. It is supposed that dimensional changes of fresh wood during early dehydration stages are merely induced by the negative pressure of free water inside the capillaries [9]. Differ-ences in AE and sensitivity to dimensional changes induced by negative pressure could have their causes in pit functioning and mechanical failure.

Fig. 5. Transverse stained (Toluidine blue) sections of fresh (A) and pre-dried (B) Norway spruce sapwood. Arrows indicate bordered pits, which are open in fresh and permanently closed in pre-dried earlywood. Reference bars represent 40 µm [5].

AE generated during dehydration of conifer sapwood generally exceeds AE generated during

heartwood dehydration [20]. In spruce sapwood, bordered pits of the first formed earlywood cell rows become irreversibly closed when sapwood is dried at ambient conditions (Fig. 5). The hy-draulic permeability at full saturation is therefore much lower in pre-dried than in fresh sapwood [5]. If pit closure was a rapid process, it could be an AE source in fresh wood. Permanently closed pits in pre-dried wood will then produce no AE during the following dehydration cycles.

82

Pre-dried sapwood resembles heartwood, where conduits lost their function of water conductance due to permanent pit membrane closure [6]. Heartwood does not contribute to the dimensional changes induced by negative pressures [8] because the conduits contain no or only a low amount of free water. However, even if conduits would refill through the cell walls, pit closure might be advantageous in order to avoid internal checking in living trees by reducing the transmission of negative pressure from conduit to conduit. Reference [13] supposed that sapwood within-ring internal checking is caused by deformation due to negative pressure and not by differential cell wall shrinkage, because checking occurred already at an early stage of kiln drying. Internal sap-wood checks caused by negative pressure may also occur in living trees during periods of sum-mer drought, and are a severe problem of short rotation forests [21].

Some pit membranes can also be ripped out of their sealing position due to high negative wa-

ter pressures [4], which might be an AE source in fresh wood but not in pre-dried sapwood. Drought stress leads as well to a phenomenon termed “cavitation fatigue”. Cavitation fatigue is induced by the rupture or loosening of the cellulose mesh of inter-conduit pit membranes, which results in a weakened response of the pits to drought stress [3]. Pre-dried spruce sapwood is, therefore, more vulnerable to the breakage of the water columns than fresh wood, which means that less negative pressure is necessary in pre-dried than in fresh wood to result in the breakage of the water columns [5]. As a consequence, pre-died rewetted wood is less prone to deformation under negative pressure, because the stress is more easily released due to the breakage of the water columns (cavitation fatigue).

Fresh wood emitted much stronger signals than pre-dried wood. The sudden breakage of the

water column in a conduit of the same size should give AE with lower frequencies and higher peak amplitudes when it occurs at more negative pressure, because more elastic energy (more deformation) is stored before the breakage occurs. Minimizing any kind of dimensional changes is advantageous for heartwood concerning crack formation in stems of living trees.

High AE burst rates or AE with high peak amplitudes indicate mechanical failure in fresh

wood at moderate water losses [12, 15 - 17]. Single high energy AE measured during the first dehydration run suggests that micro-cracks occurred in the sapwood specimens. In fresh sap-wood, highest peak amplitudes were measured during the first hours of dehydration, whereas in pre-dried sapwood during the last hours. Moreover, in some pre-dried specimens, the first AE peak was totally missing. This might indicate that some AE arose from irreversible damage in-duced by negative water pressures because it was detected already at an early stage of dehydra-tion [13]. By means of X-ray CT scanning (resolution 4.5 µm) no mechanical failure on a set of comparable Norway spruce specimens could be detected [5]. Micro-cracks in dried specimens are, however, sometimes hard to detect, because they close again after dehydration [2]. For de-tection of matrix-cracks in the conduit cell walls, higher resolution techniques were required, such as scanning electron microscopy [1]. Changes in the frequency composition might be due to micro-structural changes in the cell walls, which influenced wave propagation. Towards the end of dehydration, a high amount of high frequency AE was detected in fresh and pre-dried speci-mens. Fresh specimens showed, however, a much higher amount of lower frequency signals at the onset of dehydration than pre-dried specimens (Figs. 3 & 4). At this stage of dehydration, we detected a higher amount of signals <70 kHz by means of the resonant R6 transducer and a higher amount of AE <175 kHz by means of the broad-band WD transducer. The average fre-quency composition can give relevant information about critical dehydration stages, which should be topic of further studies.

83

Conclusions

AE testing results suggest that fresh sapwood is more prone to dehydration stresses than pre-dried sapwood. The dehydration stress became lower after the first dehydration cycle; thereafter it showed only a slight decrease. Differences in AE might be due to changes in pit functioning and mechanical failure. Permanently closed pits, pit membrane rupture and micro-cracks in the cell walls, which developed during a previous dehydration cycle, will produce no AE during the next dehydration cycle. Additionally, the weakening of the pit membranes might lower dehydra-tion stress because the breakage of the water columns occurs at less negative pressures. There-fore, a higher number of AE with higher AE energies and lower average frequencies were meas-ured in fresh than in pre-dried wood. If it is assumed that pre-dried sapwood resembles heart-wood, changes in pit functioning could be advantageous for avoiding internal checking in living trees by reducing dehydration stress.

Our results can also be relevant for analyzing checking processes during industrial lumber

drying, especially when wood boards contain a high amount of sapwood. Critical stages of dry-ing could be pinpointed by the analyzing the frequency composition and the AE energy. Ex-tremely high energy AE events could be used in future studies on source location of lumber checking processes. Acknowledgements

This study was financed by the Austrian Science Fund FWF (Project V146-B16). Peter

Kritsch is thanked for providing Norway spruce sample material.

References 1. Müller U., Joscak T., Teischinger A.: Holz als Roh- und Werkstoff 61, 2004, 439-443.

2. Sakagami H., Tsuda K., Matsumura J., Oda K. IAWA J. 30, 2009, 179-187.

3. Hacke U.G., Stiller V., Sperry J.S., Pitterman J., McCulloh K.A.: Plant Physiol. 125, 2001, 779-786.

4. Domec J.-C., Lachenbruch B., Meinzer F.C.: Amer. J. Bot. 93, 2006, 1588-1600.

5. Rosner S., Konnerth J., Plank B., Salaberger D., Hansmann C.: Trees 24, 2010, 931-940.

6. Hansmann C., Gindl W., Wimmer R., Teischinger A.: Wood Research 47, 2002, 1-16.

7. Skaar C.: Wood-Water Relations, 1988, Springer-Verlag, Berlin, Germany.

8. Irvine J., Grace J.: Planta 202, 1997, 455-461.

9. Rosner S., Karlsson B., Konnerth J., Hansmann C.: Tree Physiol. 29, 2009, 1419-1431.

10. Offenthaler I., Hietz P., Richter H.: Trees 15, 2001, 215-221.

11. Tyree M.T., Zimmermann M.H.: Xylem Structure and the Ascent of Sap, 2002, Springer, Berlin.

12. Quarles S.L.: Wood Fiber Sci. 24, 1992, 2-12.

13. Booker R.E.: Proceedings of the 4th IUFRO Wood Drying Conference: Improving Wood Drying Technology, 1994, Forest Research Institute, Rotura, NZ, pp 133-140.

84

14. Beall F.C.: Wood Sci. Techn., 36, 2002, 197-212.

15. Kawamoto S., Williams R.S.: Gen Techn Rep FPL-GTR-134, 2002, U.S. Department of Ag-riculture, Forest Service, Forest Products Laboratory, Madison, WI, pp 1-16.

16. Beall F.C., Breiner T.A., Wang J.: For. Prod. J. 55, 2005, 167-174.

17. Niemz P., Emmler R., Pridöhl E., Fröhlich J., Lühmann A.: Holz als Roh- und Werkstoff 52, 1994:162-168.

18. Hietz P., Rosner S., Sorz J., Mayr S.: Ann. For. Sci. 65, 2008, 502p7.

19. Okumura S., Kiyotaki T., Noguchi M.: Bull. Kyoto University Forests. 59, 1987,283-291.

20. Kuroiwa M., Okumura S., Fujii Y.: Bull. Kyoto University Forests. 68, 1996, 151-160.

21. Grabner M., Cherubini P., Rozenberg P., Hannrup B.: Scand. J. For. Res. 21, 2006, 151-157.

J. Acoustic Emission, 28 (2010) 85 © 2010 Acoustic Emission Group

ACOUSTIC EMISSION SOURCE LOCATION IN PLATE-LIKE STRUCTURES USING A CLOSELY ARRANGED TRIANGULAR

SENSOR ARRAY

DIRK ALJETS 1, ALEX CHONG

1, STEVE WILCOX 1 and KAREN HOLFORD

2

1 University of Glamorgan, Faculty of Advanced Technology, Department of Engineering, Pontypridd, Wales, UK CF37 1DL; 2 Cardiff University, Cardiff School of Engineering,

Cardiff, UK CF24 3AA

Abstract In order to identify the location of acoustic emission (AE) sources in large plate-like struc-

tures it typically requires the use of at least three widely spaced sensors. The distance between these sensors is defined by, for example, expected AE intensity and attenuation of the signals. This work presents a novel configuration of the three sensors, which are installed in a closely arranged triangular array with the sensors just a few centimetres apart. The algorithm locates AE sources by determining the direction from which the wave is approaching the array using the time of arrival and the distance the wave has travelled using the wave mode separation. Tests were conducted on a composite (CFRP) plate with anisotropic lay-up. In this work it is shown that the technique is able to accurately identify the source location. The technique is particularly suitable for Non-Destructive Testing (NDT) and Structural Health Monitoring (SHM) applica-tions where the close positioning of the sensors allows the array to be installed in one housing to simplify mounting and wiring. It is expected that this sensor arrangement could reduce the num-ber of sensors needed to monitor large plate-like structures compared to more conventional AE source location methods.

Keywords: Source location, Wave mode analysis, sensor array, composite material Introduction

Source location of acoustic emission events has become an important tool for Non-

Destructive Testing (NDT) and Structural Health Monitoring (SHM) research and field applica-tions. Localising the exact origin of an AE wave in a structure can help determine the source type, for example, crack propagation from a drill hole or rivet, impact damage or even just fric-tion between different parts of the structure. This information can be used to evaluate the severity of the damage for the structure and can also help understand the damage mechanism and propa-gation.

Many different source location methods have been developed in the past with a variety of

applications and accuracies. The simplest method is the “first hit” method or zone location where the sensor with the first arrival or the highest amplitude of the wave is said to be the closest to the source [1]. This is adequate when the sensor spacing or the area to be monitored is small or the likely damage initiation point is known (for instance a bolt or notch). Other source location methods use the time of arrival of the AE wave at the different sensors. This method is often re-ferred to as Time of Arrival (TOA) [2]. Two sensors can be used for linear source location be-tween the sensors. At least three sensors are necessary to pinpoint a source on a plate. The source must be located inside or near the sensor array to give an adequate result and the wave velocity must be known.

86

Recent papers are increasingly using the modal nature of AE waves for source location pur-poses [3 - 5]. Acoustic emission in thin structures propagates in two fundamental wave packages. The first, usually faster wave package is a compressional wave where the material oscillates in the same direction as the wave propagates. The wave mode is referred to as symmetrical mode (S0 mode). The velocity of this mode depends on the material properties (tensile stiffness), which can result in different propagation velocities in different direction in anisotropic materials [6]. The second, usually much slower propagating mode is a flexural wave where the material oscil-lates perpendicular to the wave propagation. This mode is referred to as anti-symmetrical mode (A0 mode) and is mainly caused by shear stress inside the material [7]. This mode is dispersive, depends on the thickness of the structure (flexural stiffness) and is not affected by the fibre orien-tation [6]. The amplitude ratio between both modes can vary depending on the source orientation in the specimen and the amount of shear stress and tensile stress affecting the material [8]. Since both modes travel with different velocities the wave packages separate during propagation and the distance between the origin of the wave and the sensor can be calculated when both velocities are known. Using this method source location is not limited to the area enclosed within an array of sensors. Other authors use the modal analysis to determine the exact wave velocity depending on which mode triggered the AE system [9].

Sachse [10] patented a source location method with a small array of four sensors. This

method monitors the area exterior to the array by determining the distance and direction to the source without prior system calibration and group velocity measurements.

Another method proposed in the literature where the source location is not calculated but the

arrival time of an event at different sensors is compared with an arrival time map. This map has to be generated for the area to be investigated beforehand using artificial AE [2, 5]. Proposed Source Location Method

The proposed source location method in this paper uses modal analysis to evaluate the origin

of the AE event with respect to the centre of the sensor array. The sensor arrangement is a close triangular setup with a sensor spacing of just 45 mm. The algorithm in the method first deter-mines the direction from which the wave travels to the array and secondly, the distance between sensor array and source using the wave mode separation. Experimental setup

Artificial AE was generated using pencil-lead fractures (Hsu-Neilsen source or H-N source) on a large carbon fibre reinforced epoxy composite plate (CFRP). The plate was manufactured by Carbon-Composite Technology with a carbon fibre twill weave, an approximate fibre content of 60% and fibre alignments in 0° and 90°. The dimensions of the plate were 1300 x 900 x 2 mm (1.17 m2 area). Bubble wrap was placed between the composite plate and the table to reduce wave dispersion effects into the table. Three AE sensors (Micro-80S, Physical Acoustics Corpo-ration, 100-1200 kHz, sensor diameter: 10 mm) were installed in the centre of the plate in a tri-angular arrangement with a sensor to sensor distance of 45 mm (centre point to centre point) re-spectively. The signal was pre-amplified by a Vallen Systeme AEP4 (40 dB) and recorded by a Gage Octopus CompuScope CS-8280 data acquisition card (12 bits, 5MS/s, 100 MHz band-width) driven by LabView. The recording was threshold triggered, where a pre-trigger setting of the data acquisition card allowed the capturing of data prior the triggering event to ensure that the whole event was recorded. Afterwards the raw data was digitally filtered by a band-pass filter of 100-1000 kHz and processed using LabView and Scilab. The plate was divided into a 15 x 23

87

grid starting 100 mm from the four edges of the plate respectively to avoid signal corruption by reflections. The grid spacing was 50 mm in horizontal and vertical directions. Three pencil-lead breaks were performed on every grid point (a total of 1035 events) and the arrival time of the S0 mode and the A0 mode was determined for all events and for all three sensor signals respectively.

Wave mode analysis

An algorithm was written in LabView to automate the arrival time measurement of both wave modes for every event. First of all, the Gabor wavelet transform (WT) was calculated to generate a time–frequency representation of the signal. Fig. 2a) shows an AE signal generated by a pencil-lead break. This event was located in a distance of about 540 mm from the sensor array. Both wave modes are clearly visible in the Gabor WT, as shown in Fig. 2b). Since the wave ve-locity is dependent on material, wave mode and frequency [7], a narrow frequency band was chosen to measure the mode arrival time for both modes respectively. For the automated wave mode detection a frequency was chosen where both modes were present, but the earlier arriving S0 mode had a much smaller amplitude in comparison to the A0 mode. In the test setup used here, the frequency of 100 kHz was best suited and an example of the WT coefficient of the signal is shown in Fig. 2c). A threshold was set just above the noise level to detect the beginning of the event and a peak detection tool in LabView was used to find the first peak after the first thres-hold crossing. The peak detection should lower the influence of the actual threshold setting. Since the S0 mode is the faster propagation wave, the first peak is said to be the S0 mode. A sec-ond threshold is set to 90% of the maximum amplitude of each event respectively and again a

Fig. 1: Test rig.

88

peak detection tool is used to find the first peak after the crossing. This second peak is said to be the A0 mode.

Fig. 2: H-N source 540mm away from sensor: a) sensor signal, b) Gabor WT, c) Gabor WT co-efficient at 100 kHz.

In comparison, Fig. 3 shows the sensor signal and the Gabor transform of a pencil-lead frac-ture event generated in close approximation to the sensor. The same wave feature triggered both the S0 and A0 threshold. In general, both modes are generated at the same time and separate due to their velocity difference. In this example travelled distance of the wave was not sufficient en-ough to develop a noticeable separation and hence the wave mode separation is 0 µs.

89

Fig. 3: H-N source close to sensor: a) sensor signal, b) Gabor WT, c) Gabor WT coefficient at 100 kHz.

Misreading of the mode arrival times were identified when one result differs significantly from the result of neighbouring grid points or from the result of the other two sensor signals of the same event. These events were then manually inspected. A total of 47 of the 3105 AE signals (1035 events x 3 sensors) had to be reassessed. This is an error of 1.51%. In most cases just one of the three sensor signals were corrupted.

This automated wave mode detection was used for H-N sources, which are quite repeatable.

However the wave mode amplitude ratio varies significantly for real AE data dependent on the

90

source orientation and failure mechanism [8]. This will affect the accuracy of this method for “real” AE testing. Also noisy environmental could make the detection of a weak S0 mode at low frequencies difficult. In that case it is recommended to measure this mode at a higher frequency where it is more pronounced.

Evaluating direction from which the wave approaches the array

In the first step to locate an AE source the direction from which the wave travels to the sen-sor array has to be determined. In anisotropic materials, such as CFRP, this has to be undertaken before calculating the distance because the S0 wave velocity is dependent on the fibre orienta-tion. The A0 mode is chosen to determine the direction since its velocity is not dependent on the fibre orientation.

Fig. 4: A0 mode arrival time differences (ΔtA0) for all three sensor pairs in µs.

Figure 4 shows the arrival time differences of the A0 mode (ΔtA0) for all three sensor pairs at every grid point. Values between grid points were interpolated. The maximum absolute differ-ence in the arrival time of the A0 mode (ΔTmax) between two sensors would be measured when the AE source is in line with the sensor pair (dark red and dark blue colour shades). This time is calculated using the distance between the sensor and the wave velocity. When the source is lo-cated perpendicular to a strait line through the sensors the wave would arrive at exactly the same time at both sensors (green colour shades), thus ΔtA0 is zero.

Fig. 5: Theoretical ΔtA0 hyperbola.

[µs]

91

The theoretical ΔtA0 hyperbola example is shown in Fig. 5 to clarify the ΔtA0 pattern found in Fig. 4. The direction in which the hyperbola opens is dependent on which sensor picks up the signal first. The opening angle is dependent on the time delay between both sensors. Each hyper-bola represents the position of all points with the same ΔtA0. Strictly speaking, the algorithm pre-sented in the following paragraph calculates the angle of the asymptotes of the hyperbola (dashed lines).

Fig. 6: Angle of arrival calculation.

The angle of arrival referenced to the centre point of one sensor pair can be calculated using the inverse cosine function (Arcus-Cosine) (Fig. 6). However this leads to two possible results: One angle between 0° and 180° (denoted as α) and the other one is the angle α mirrored about the 0°-180° axis (360°– α). This ambiguity can be removed by introducing the third sensor to the array. The First Hit method is used to break down the search area in three segments (Fig. 7). The sensor at which the AE wave arrived first gives the indication in which segment the source is lo-cated. The angle is calculated with the smallest absolute arrival time difference of all three sensor pairs in the array. Thereby the two results of the Arcus-Cosine function are relatively far apart (First result: between 30° and 150°; second result between 210° and 330° with respect to the sen-sor pair). Only one result can be within the area found by the first hit method.

Fig. 7: First hit method.

This result is relative to the sensors pair chosen for the calculation. Dependent on the orienta-tion of this sensor pair, the angle needs to be rotated to coincide with the global coordinates. For example: the coordinate system for the array should be 0° pointing to the right and 90° pointing upwards, the angle α for sensor pair 1-2 is α – 60°, for sensor pair 1-3 α = α – 120° and sensor pair 2-3 coincide already with the global coordinate system.

92

Evaluating Sensor to Source Distance The distance between each sensor and the AE source can be calculated by the formula [11]:

If the velocities of both modes (VA0 and VS0) are known and the arrival time of both modes is

bA0 and bS0, the distance between sensor and source is L. If the material is isotropic or quasi-isotropic, the waves propagate with the same velocity in

all directions. However in anisotropic materials the wave velocity depends on the material prop-erties and its orientation. In the case of CFRP, the wave speed of the compression wave mode (S0) varies depending on the fibre orientation. This problem must be addressed for an accurate source location.

Fig. 8: Wave mode separation in µs at sensor 1.

The composite plate used in this experiment had a fibre lay-up in 0° and 90°. The S0 mode has a velocity of 5.80 mm/µs in fibre direction and 4.80 mm/µs in 45° direction. The A0 mode had a constant velocity of 1.68 mm/µs in all directions. Hence the modes are slightly faster sepa-rating in fibre direction compared to 45° (and 135°) direction. The wave mode separation for the whole plate is shown in Fig. 8. The separation gradually increases with distance from the sensor. The effect of the fibres is barely visible in this plot but had noticeable effect on the source lo-cation result if not considered.

[µs]

93

Fig. 9: Wave mode velocity dependent on angle.

The wave propagation pattern in the plate is illustrated in Fig. 9. This pattern can be repre-sented sufficiently by the cosine function with an offset and an amplitude dependent on the wave velocities of both modes. The actual wave velocity (VS0) can be calculated as a function of the angle of arrival (α) using the following formula:

The maximum S0 velocity (VSmax) is in the 0° and 90° direction, VSmin is diagonal to the fibre orientation (±45°). If the propagation pattern is elliptical in unidirectional composites, the factor of 4α needs to be replaced by a factor of 2α.

Fig. 10: Sensor-to-Source distance.

The source-sensor distance was identified for all sensors and then the average of all three re-sults was calculated to get a result, which is relative to the array centre point. This causes an er-ror for sources inside or close to the array but decreases rapidly with distance (Error <1.7 mm at a radius 100 mm). This error is tolerated since it is very small in comparison to the error caused by mode arrival measurements. For example sources inside of the array are too close to the sen-sors to detect a separation of modes and hence the calculated distance is zero. The Standard Deviation of wave mode separation measurements for H-N sources at the same location was about 4µs which is correspondent to an error of ±10 mm.

a) b)

94

Source Location Block Diagram Figure 11 illustrates the source location algorithm. It gives a stepwise guideline to reproduce

the algorithm but does not include the actual programming code.

Fig. 11: Block diagram of source location algorithm.

95

The input is the A0 mode arrival and mode separation at all sensors. Furthermore the S0 and A0 wave velocities are needed and the maximum theoretical time delay of the A0 mode arrival at different sensors (ΔTmax), which is a function of the sensor-to-sensor distance and the wave ve-locity.

The algorithm could also be enhanced to issue warnings when an error in the measured mode

arrival is likely. This could be, for example, when the time delay of the mode arrival from one sensor to another is larger than the theoretical maximum time delay (ΔTA0 > ΔTMAX) or when there is a significant deviation between the measured mode separation at different sensors. Results

The result of the source location of each event can either be expressed as polar coordinates (distance, angle) or as Cartesian coordinates (x, y) relative to the array centre point. Figure 12 shows the source location result of one AE event for each grid point respectively (randomly se-lected). “x” represents the actual source location and “Δ” the calculated location. Each calculated location is connected to its real counterpart by a line.

It can be seen that this method works quite accurate on close range, but its accuracy de-

creases with distance. The Standard Deviation in a radius of 350 mm from the array was 13 mm, whilst for the whole plate was 33 mm.

It also can be seen that the radial distance between array and source is usually determined

quite accurately but the angle is more prone to errors. This is probably due to errors when meas-uring the exact arrival times especially of highly attenuated signals. The Standard Deviation of the angle was 4.3° in a radius of 350 mm and 6.0° for the whole plate. The Standard Deviation of the distance was 7 mm and 11mm respectively.

Benefits and Limitations of the Proposed Source Location Method

The biggest challenge for this method is to get the exact mode arrival times. There are situa-

tions when one mode is very weak and hardly detectable. This source location technique only works when both modes are detected in the same event. Tests with real damage had shown that usually a large number of events are emitted long before the specimen finally fails and even if not all events can be used for source location, areas with high AE activity can still be located.

The range of this array is limited by the actual attenuation in the material. Theoretically the

accuracy of the direction decreases with distance but the main error is thought to be associated with the uncertainty in detecting the mode arrival times.

The main benefit of a closely arranged sensor array is the possibility to install all sensors in

one housing. This would simplify mounting and wiring up of the AE system. Furthermore it is suitable for wireless application where the three sensors, the power supply and the transmitter could be installed as one unit.

Since the sensors are installed very close to each other, the signal of an event should look

fairly similar at all sensors. This makes it easier to identify exactly the same wave feature in all three sensor signals. In a wide spread sensor array, the signal changes quite significantly due to

96

attenuation and dispersion. As a result wave features could alter, be hidden or completely disap-pear in the different sensor signals. This could lead to measuring errors.

Fig. 12: Source location results (x = actual location; Δ = calculated location; ● = sensor).

Fig. 13: Comparison of covered area of TOA (left) and local triangular sensor array (right) method.

97

Another advantage of this source location method is the potential for reducing the sensor number needed to monitor a large area. A propagating damage releases energy in form of mo-tion, heat and acoustic emission. The energy of the acoustic wave is proportional to the overall energy released during the event [12]; hence is related to the damage size and propagation speed. It also depends on other different factors such as the material involved. The AE wave propagates from the source and attenuates due to wave spreading, reflections and friction in the material. The maximum distance between sensors depends on the minimum damage size to be detected and the intrinsic attenuation characteristics in the material. The circles in Fig. 13 represent this minimum distance. The left figure represents a conventional Time of Arrival (TOA) source lo-cation method. Source location is only possible in the red area in the middle where all three circles are overlapping. Acoustic emission will also be detected in the other areas, but the exact position cannot be evaluated. The sensor array presented here covers a bigger area as long as both wave modes can be detected. Conclusion

A new source location method has been introduced which uses a closely arranged sensor ar-ray. The location of AE sources is determined by a combination of time of flight and modal an-alysis. Its benefits and limitations have been highlighted as well. References 1. K. Ono, "Acoustic Emission," in Handbook of acoustics, M.J. Crocker, Ed. New York:

Wiley-Interscience Publication, 1998, ch. 53, p. 643.

2. M.G. Baxter, R. Pullin, K.M. Holford, and S.L. Evans, "Delta T source location for acous-tic emission," Mechanical Systems and Signal Processing, 21, 1512-1520, 2007.

3. M. Surgeon and M. Wevers, "One sensor linear location of acoustic emission events using plate wave theories," Materials Science and Engineering, A265, 254-261, 1999.

4. K.M. Holford and D.C. Carter, "Acoustic Emission source location," Key Engineering Ma-terials, 167-168, 162-171, 1999.

5. J.J. Scholey, P.D. Wilcox, M.R. Wisnom, and M.I. Friswell, "Two dimentional source lo-cation techniques for large composite plates," in European Conference on Acoustic Emis-sion Testing, Krakow, 2008, pp. 160-165.

6. A.A. Pollock, "Classical Wave Theory in Practical AE Testing," in 8th International Acoustic Emission Symposium, Tokyo, 1986, pp. 708-721.

7. J. Rose, Ultrasonic waves in solid media. Cambridge: Cambridge University Press, 1999.

8. W.H. Prosser, "Applications of advanced, waveform based AE techniques for testing Composite Materials," in SPIE Conference on Nondestructive Evaluation Techniques for Aging Infrastructure and Manufacturing, Scottsdale, 1996, pp. 146-153.

98

9. J.J. Scholey, P.D. Wilcox, M.R. Wisnom and M.I. Friswell "Quantitative experimantal measurements of matrix cracking and delamination using acoustic emission," Composites: Part A, 41, 612-623, 2010.

10. W.H. Sachse, "Acoustic Emission source location on plate-like structures using a small ar-ray of transducers," 4592034, May 27, 1986.

11. J. Jion, B. Wu, and C. He, "Acoustic emission source location methods using mode and frequency analysis," Structural Control and Health Monitoring, 15, 642-651, 2008.

12. ASTM, "Standard Definition of terms relating to Acoustic Emission," ANSI/ASTM E619-77, 1977.

J. Acoustic Emission, 28 (2010) 99 © 2010 Acoustic Emission Group

NEURAL NETWORK AE SOURCE LOCATION APART FROM STRUCTURE SIZE AND MATERIAL

MILAN CHLADA, ZDENEK PREVOROVSKY and MICHAL BLAHACEK Institute of Thermomechanics AS CR, v. v. i., Dolejskova 1402/5, 182 00 Prague 8,

Czech Republic

Abstract

AE localization procedures using artificial neural networks (ANN) represent extremely effec-tive alternative to classical triangulation methods. Nevertheless, their application always requires full-scale, time consuming ANN training on each specific structure. Disadvantage of particularly trained ANN algorithm is in its non-transferability to any other object. A new ANN-based AE source location approach is proposed in this paper to overcome such limitation. The method re-places standard arrival time differences at the ANN inputs by so called signal arrival time pro-files, independent on material and scale changes. The ANN training can be also performed theo-retically on geometrical models (i.e. without any experimental errors) and learned ANN is then applied on real structures of different dimensions and materials. Such approach enables consid-erable extension of ANN application possibilities. The use of new AE source location method is illustrated on experimental data obtained during aircraft structure part testing. Keywords: AE source location, artificial neural networks, arrival time profiles. Introduction

Acoustic emission (AE) method is widely applied in non-destructive testing of various tech-nical structures. After the signal detection, accurate location of AE source is the primary goal of the defect analysis and the basic requirement for further damage mechanism characterization. AE source coordinates are mostly determined by common triangulation algorithm based on arrival time differences of AE signals recorded by several transducers [1]. The algorithm computes the polar coordinates (r, α) of AE sources using analytical formulas. Time differences Δt of AE sig-nal arrival times to different transducers along with elastic wave velocity are necessary input data for triangulation source location algorithm (see Fig. 1).

Analytical formulas are known for isotropic plates. However, there are many practical situa-

tions, in which the triangulation algorithm fails, especially if the more complex structure is tested. Procedures based on ANN are used in such cases as an alternative approach to triangula-tion algorithm. Contrary to the classical methods, the ANN-based location procedures have two important advantages: they are suitable for AE source location in highly anisotropic media, and elastic wave velocity is not necessary input parameter of the algorithms [1 - 3].

The basic ANN-based algorithm uses time differences as input parameters similarly to com-

mon triangulation algorithm. However, the correct arrival time (signal onset) determination is the crucial factor for the algorithms using time differences and, therefore, the methods can give erro-neous results, especially in cases of high noise background level. To solve the problem, certain modification of ANN-based method for AE source location was proposed [4]. The method uses common AE signal parameters, such as RMS, rise time, maximum signal amplitude, etc, which represent redundant input data set for ANN location algorithm. The algorithm is flexible, be-cause it enables selection of the optimal parameters set in each specific situation. Nevertheless,

100

the sensitivity analysis of trained neural networks has extracted RMS parameter as the most sig-nificant signal feature suitable for AE source location, which corresponds with the basic features of dispersive elastic-wave propagation. The information about the source location, which is hid-den in energy parameters, can be better extracted when we use certain modifications of RMS signal parameter, as it was shown and successfully tested in [4].

Contrary to the computational power of ANN's, their application possibility is strongly re-

stricted due to several reasons. The first problem is collecting sufficiently extensive training and testing data sets, which may be excessively time-consuming, expensive or even impossible in many practical cases. The next important reason is the non-portability of particular trained net-work to any other situation. ANN should be learned and applied in exactly the same task only, i.e. the similar computation in the same input-output parameter space is required. Hence, the above-mentioned method using RMS signal parameter for AE source localization need not repre-sent suitable solution.

To solve both above problems of ANN-based AE source localization, a new approach was

proposed. This innovative localization method uses new way of signal arrival time parameteriza-tion (called signal arrival time profiles), independent on the material and scale changes. Under some non-restrictive conditions, such approach employs the ANN training on numerical model data (i.e., without experimental errors) and allows the application of learned ANN on real struc-tures with diverse scales and materials.

Fig. 1. Demonstration scheme of AE signal measurement (2D model).

Definition of Arrival Time Profile (ATP)

The new way of signal arrival time treatment is inspired by the preliminary expert analysis of

AE source location. The whole monitored structure is separated into several zones according to signal first arrival to each sensor (see Fig. 2). In this way, the AE source can be roughly localized using the information which sensor is the nearest one. To describe the signal detection chronol-ogy more precisely, so‑called arrival time profiles are proposed.

101

Fig. 2. Illustrative splitting of a complex structure by first AE signal arrival to each sensor Si. Let's suppose hypothetical configuration of several AE sensors S1, ... ,SN placed on planar

material (in a case of four sensors see Fig. 1), detecting elastic waves emitted by AE sources in various locations. Signal propagation time from the source to sensor i is denoted Ti (arrival time period). Then, the arrival time profile (ATP) is a vector of numbers pi defined as follows:

(1) No AE analyzer measures that arrival time periods Ti before the precise AE source location is

performed. Only the AE signal arrival times ti are available. Nevertheless, if we assume ts as the time of AE source initiation, it is easy to revise the original formula, while Ti = ti - ts :

(2)

Fig. 3. Arrival time information processing

102

Computation of arrival time profiles is similar to general data standardization. In a first step, the mean value of arrival times is subtracted from each ti. Such data is then normalized by the mean of its absolute values. Figure 3 illustrates described arrival time information processing. Independence of ATP on Wave Velocity and Scale Changes

Let us denote di the distance between AE source and sensor i, and v the elastic wave velocity.

By substitution of relation Ti = di /v to (1) we obtain: (3) Arrival time profiles can be also calculated using the distances di, which enables the learning

of neural networks with numerically computed data from the distances measured on proportional model of considered structure (i.e. without experimental errors) and, afterwards, testing by proc-essed real arrival times. The elastic wave velocity is cancelled out in eq. 3, so the arrival time profiles are independent on elastic wave velocity. Analogical cancellation of any common multi-ple of distances di proves the independence to structure scale changes.

Theoretical Aspects

The potentialities of newly proposed parameterization using arrival time profiles were at first

examined numerically. Theoretical case of 2D plate of any isotropic material and scale was taken into account. Due to the complex topology of ATP parametric space, it was necessary to find sensor configurations warranting numerical stability of the method. Possible division by small numbers in eq. 3 may generate rapidly changing data that neural network cannot learn. Typically, this occurs in regions where the distances of AE sources to all considered sensors are similar.

Fig. 4. Numerical testing of sensor configurations.

103

Three different configurations of four sensors (black dots) are shown in Fig. 4. To compute the training data, the rectangular domain from 5 to 150 and from 5 to 60 units of the scale was taken into account for x- and y-coordinates, respectively. Equidistantly spaced points having 2.5 units of the scale between each other cover the whole area. For all such virtual AE sources, the distances to considered sensors were computed and, consequently, eq. 3 was applied. Before neural network learning, data standardization is realized. The two-hidden-layer back-propagation networks (see Fig. 5) were trained to estimate the original coordinates of AE source using corre-sponding arrival time profile. Finally, all training points (their arrival time profiles) were pro-jected by neural network onto the original area (see crosslets in Fig. 4).

Fig. 5. Localization ANN scheme.

Many theoretical experiments proved that for ATP defined by eq. 1, it is needed to configure

sensors and select the training data so as to avoid learning with signal source locations having nearly the same distances to all sensors. The problem is well illustrated by the configuration 3 in Fig. 4. Nevertheless, it can be solved by another definition of ATP, where the mean of absolute values of centered arrival times in denominator is substituted by normalization period T repre-senting the time interval, which AE signal needs to propagate between two appropriately selected points. However, compared to original definition, such method demands additional measurement of normalization period on real structure, which can be a source of additional experimental er-rors.

Regarding the comparison of the new ATP-based algorithm with the ANN approach exploit-

ing the time differences, both methods give similar results in accordance with the network learn-ing speed, input sensitivity and final mean-square error (MSE) of network outputs. Possible little variances of outputs are caused rather by different versions of trained networks, than by arrival time parameterization method. However, the parametric space of time profiles provides many advantages. Beyond the mentioned portability to another scale and material, it is more robust with regard to input errors. Due to the subtraction of arrival time mean value, the error of par-ticular signal edge determination is consequently spread out to all inputs.

It is well known that 2-D location of AE sources by triangulation algorithm using three trans-

ducers is ambiguous [5]. In Fig. 6 the two triangle configurations (rectangular and equiangular) of transducers are drawn. Areas of ambiguity are marked out for both configurations. Any AE source in dark grey area has the same time-differences corresponding to source in light grey area and vice versa. The problem is usually solved by exploitation of four transducers, while two dif-ferent triplets of transducers are used in triangulation algorithm. However, theoretically it is still

104

not a correct solution, because the area of ambiguity is only reduced to insular points. Neverthe-less, the probability of such AE source location is close to zero in the real world.

Fig. 6. Ambiguity of 2-D AE source location using 3 AE transducers. Now, if we take into account the theoretical ambiguity "behind the three sensors" in Fig. 6, it

is possible to interpret the erroneous accumulation of crosslets in analogous problematic areas in Fig. 4. As in the case of time differences, if one arrival time profile corresponds to different AE source locations, neural network cannot interpret such relation, since it is possible only for func-tional dependencies. However, the problem can be solved using more dimensional time profiles, i.e., by additional suitably placed sensors. The higher errors in some subsets of training area (e.g. near the sensors) are caused by closeness of points in time profiles parametric space.

Practical Results

To demonstrate the new AE source location method in practice, an experiment on a small air-

craft part has been performed. The task was to localize AE sources on a steering actuator bracket (SAB), which is a part of the aircraft nose landing gear. AE signals were monitored and recorded by DAKEL XEDO AE system. For signal onset assessment, an "expert edge detection" algorithm of the elastic wave arrival was applied [6]. Although the method is relatively precise, its accuracy is much lower than the approximation error of neural network, especially for longer distances from AE source to sensor. Therefore, the two training areas of ANN were considered (left and right part of critical zone of SAB) instead of the “global” one.

The scheme in Fig. 7 shows the configuration of four transducers in case of the left training

zone. The learning points were selected to equidistantly cover the zone among the sensors S2, S3, S4 and S9. To compute the arrival time profiles, eq. 3 was applied using the distances evaluated by the algorithm finding the shortest ways in raster (digital) picture of the structure. The length of the real elastic wave path through the material is approximated by the shortest po-lygonal line, while the connecting line segments between the nodes must come through the pix-els representing the body. The third dimension of the structure could be neglected because of the small thickness of SAB.

Similarly to numerical testing of the method, neural network with two hidden layers having

39 and 17 neurons was trained by virtual AE sources corresponding to points equidistantly spaced 5 pixels between each other (see Fig. 7), covering the whole training area. Distances of

105

all such virtual AE sources to four sensors considered were computed, and eq. 3 was applied. During the learning process, weights and biases of the network were iteratively adjusted by fast resilient back-propagation training algorithm and generalization-improving regularization to achieve sufficiently low MSE (less than 0.001). For illustration of approximation error, all train-ing points were projected by neural network back onto the SAB (see Fig. 8).

Fig. 7. Configuration of AE sensors and training points for ANN.

Fig. 8. ANN projection of training points.

106

Trained neural networks were finally tested with real experimental data. Three sets of 25 AE signals recorded during pencil-lead break (PLB) tests were obtained. The original locations of PLB tests correspond to cross points of the dashed lines in Fig. 9. The arrival times of each group of 25 signals were determined by expert edge detection algorithm and eq. 2 was used to compute the arrival time profiles as ANN inputs. The outputs of the network were coordinates of esti-mated PLB test locations illustrated as crosslets in Fig. 9.

Fig. 9. ANN PLB tests localization results.

While analyzing the localization results, the inaccuracies of experimental measurement must

be considered. These are mostly caused by the signal onset determination errors. Due to disper-sion and reflections, the shape of signal envelope undergoes substantial changes with distance between the source and sensor. Hence, it is possible to expect higher signal edge detection error due to signal dissimilarity for sources near single sensor and simultaneously far from other trans-ducers. Such situation is evident in PLB test location #3 (see Fig. 9), where the localization error is worse than in wider structure areas. For location #3, the PLB tests were done not too close to sensors. Next undesirable effect is that the centers of crosslet groups are not placed at original PLB test locations. This can be interpreted by different magnitude of the signal edge detection error in each of four channels, which can result in the shift of arrival time profiles. However, the results for locations #1 and #2 can be rated as very good, since the error is comparable with the aperture of sensors (9 mm) and better results are not expected. We can say that the majority of location errors is caused by experimental measurement inaccuracies, not by the low arrival time profile robustness, or neural network approximation variance.

After testing of complex methodology for AE sources with known localization, let us intro-

duce the results measured during the laboratory fatigue tests of SAB. The bracket was loaded under stress cycles (R = 0, frequency 1 Hz) on Instron-Schenck 100-kN uniaxial loading ma-chine to the maximal load level of 43 kN causing the maximum stress of about of 870 MPa in the critical points. AE was monitored during the loading cycles by DAKEL XEDO AE system. All

107

detected AE signals were stored, and after certain loading periods, the detailed AE source loca-tion analysis was performed so as to detect crack initiation. Most AE events arose around the mounting holes of the bracket. Figure 10 shows the typical localization results of sources in the left part of SAB. Most of the AE sources originated from the clearly visible damages on the inte-rior surface of the hole caused by the friction with mounting shank (shown in insert of Fig. 10). Fig. 10. Results of real AE source localization. Insert shows damages inside the hole, producing most of AE. Conclusions

A new approach to the AE source location is introduced in this paper. Described algorithm is based on neural networks exploiting newly defined signal arrival time profiles (ATP) as inputs. This innovative way of AE signal arrival time relativism facilitates considerable extension of ANN applications in practice. Processing of the real experimental data measured on a small air-craft part confirmed advantages of such approach. Main aspects of the method can be summa-rized as follow: • The new method is inspired by the preliminary expert analysis of AE source location exploit-

ing the signal detection chronology. To utilize such information more precisely, so-called ar-rival time profiles are introduced.

• Arrival time profiles enable numerical generation of sufficient ANN training data sets with optimal precision and no measurement errors. Afterwards, trained networks can be success-fully tested on data coming from real experiment.

• Arrival time profiles are independent on elastic wave velocity and structure scale changes. Hence, the trained neural networks are applicable to all proportionally identical structures of any isotropic material.

108

• Theoretical experiments showed that it is needed to configure sensors and select the training data so as to avoid learning with signal source locations having nearly the same distances to all sensors.

• The time profiles parametric space is featured by robustness regarding the input errors. The subtraction of arrival time mean distributes one particular signal edge determination error to all inputs, which reduces the global input error.

• The new AE source location approach based on neural network with arrival time profiles was demonstrated by experiments on a small aircraft part of complicated shape.

• Resulting real localization error is influenced mostly by experimental measurement inaccura-cies, i.e. it is not caused by neural network approximation variance.

• It has been proved that the arrival time profiles enable portability, robustness and easy appli-cability of ANN-based algorithm for AE source location.

Acknowledgements

The work was supported by the Grant Agency of the Czech Republic under project no. 104/10/1430 and by the Czech Ministry of Trade and Industry under project no. FR-TI1/274.

References 1. M Blahacek, 'Acoustic Emission Source Location Using Artificial Neural Networks', PhD.

Thesis, Institute of Thermomechanics AS CR, Prague 2000 (in Czech).

2. M Blahacek and Z Prevorovsky, 'Probabilistic Definition of AE Events', Proc. 25th European conference on Acoustic Emission Testing, (1): Prague 2002, pp. 63-68.

3. I Grabec and W Sachse, 'Synergetics of Measurements, Prediction and Control', Springer-Verlag, Berlin, 1995.

4. M Blahacek, M Chlada and Z Prevorovsky, 'Acoustic Emission Source Location Based on Signal Features', Proc. 27th European Conference on Acoustic Emission Testing, Cardiff, 20-22 September 2006, ed. by R. Pullin, K. M. Holford, S. L. Evans, J. M. Dulieu-Barton, Trans Tech Publications, pp. 77-82.

5. M Blahacek, 'Time Differences Uncertainty Influence on Acoustic Emission Source Location Accuracy', Proc. 4th workshop NDT in progress, Prague, 2007, ISBN 978-80-214-3505-6, pp. 15-22.

6. M Chlada, 'Expert AE Signal Arrival Detection', Proc. 4th Workshop NDT in Progress, Pra-gue, 2007, pp. 81-88.

J. Acoustic Emission, 28 (2010) 109 © 2010 Acoustic Emission Group

INTELLIGENT AE SIGNAL FILTERING METHODS

VERA BARAT, YRIJ BORODIN and ALEXEY KUZMIN Interunis Ltd., bld. 24 corp 3-4, Myasnitskaya str., Moscow, 101000, Russia

Abstract

One problem of the acoustic emission (AE) testing is a high level of noise affecting the diag-nosis results. Electric noise, electromagnetic interference, background acoustic noise, and rub-bing noise are far from the full list of noise present during measurements. With the high level of noise, the operator has to increase the recording threshold of the AE impulses through reducing the testing sensitivity at the risk of missing a dangerous defect. Lack of the data filtering can re-sult in an incorrect location and erroneous definition of the danger level of AE source. To im-prove the noise immunity of the AE system, the data filtering algorithms are to be used and ex-amined in this paper. Keywords: AE data filtering, AE signal processing, wavelet analysis Introduction The noise recorded during the acoustic emission (AE) testing is highly varied. Noise can be caused by various physical reasons, such as sensor noise, imperfection of a measuring path, and technological noise of the testing object. The noise can be essentially different in the signal waveform. They can be stochastic and deterministic, stationary and non-stationary, broadband and narrow-band. To solve effectively the problem of filtering of the AE testing data, it is possi-ble to suggest one method of noise classification – according to the filtering complexity. In Table 1 the noise of different types arising during the AE testing are divided into three groups. The sig-nals from the first group have a low filtering complexity, the signals from the second group have an average filtering complexity, while filtering of the third group noise is a complex nontrivial problem.

The types of noise that can be successfully removed by means of traditional filtering methods

are impulse and harmonic noise belonging to the first group, as well as the low-frequency and high-frequency noise lying outside the informative range of frequencies. Such noise can be easily removed by means of the frequency or median filtering.

Now, only one type of noise is related to the second group – a stationary white noise. The fil-

tering complexity lies in the fact that both the white noise and the impulse signal are broadband processes. This complicates the separation by means of the traditional frequency filtering. To detect AE impulses against the background of the stationary white noise, we utilize the algorithm of filtering based on a discrete wavelet-transform [1, 2].

The noise, which is similar to the AE signals by both the shape and the spectrum, is related to

the third group. These are generally technological noise of the test object, such as rubbing, vibra-tion, various hydrodynamic noise, etc. For filtering noise of this type, it is recommended to re-cord long, from 5 to 30 minutes, waveform of AE signals. Increase in data recording time allows to make the total time of process observation exceed the stationary interval of noise. By means of long observation we can observe the short impulse’s quasi-deterministic components correspond-ing to the AE impulses on the background of stationary noise being present continuously.

110

Table 1. Type of noise signal and filtering methods. Type of noise signal Cause of appearance Filtering methods Filtering

complexity 1 Spikes (short impulses),

harmonic noise, high-frequency and low-frequency noise

Electric and electromagnetic noise

FIR filters, IIR fil-ters, median filter

low

2 White stationary noise AES noise, noise of electronic components

Wavelet-filtering average

3 Impulse, Non-stationary noise

Rubbing noise, hydro-dynamic disturbances, and cavitation

Analysis of long realizations of AE signals

high

Filtering of Harmonic and Impulse Noise

The traditional frequency filtering is applied when the signal and noise lie in the different

frequency ranges. Uses of the low-frequency, high-frequency, band-pass and band-rejection fil-ters allows for allocating the ranges of frequencies relevant to the AE diagnostic signal.

On the basis of digital filtering it is possible to perform, if need be, integration and differen-

tiation of the signals. The possibility to define an arbitrary shape of the filter transfer characteris-tic allows for carrying out a deconvolution operation or inverse filtering that results in compensa-tion of distortions contributed into the signal by the measuring path.

Fig. 1. Some types of noise signals. a) electromagnetic noise; b) electric noise; c) narrow-band harmonic noise.

The impulse noise is widely found, such as various electric and electromagnetic noise (Fig.

1a, b), arising in case of problems with grounding or violation of principles of electromagnetic compatibility.

Application of the frequency filtering for removing the impulse noise does not result in an

acceptable solution, because the noise-like impulse signals have wide spectrum overlapped with spectrum of AE signal. The median filtering is applied for suppressing the impulse noise. A one-dimensional median filter is a sliding window including an odd quantity of signal readings. The

111

central reading is replaced with the median of all readings in the window. When selecting the filtering window that exceeds duration of impulse components of noise, the impulse will be sup-pressed, while energy of the AE signal will decrease insignificantly.

The median filtering can also help in filtering of narrow-band amplitude-modulated noise

(Fig. 1c); in this case the signal spectrum undergoes the median filtering. The spectrum of such signal has a well-defined single peak, which is removed during the median filtering of the signal spectrum. Energy of the narrow-band quasi-harmonic noise decreases by hundreds of times dur-ing the spectrum median filtering, while energy of the AE signal, which has no defined single peaks in the spectrum, varies insignificantly. Additive White Noise Filtering

After allocating the signal in the desired frequency range and excluding the impulse noise,

the problem of exclusion of an additive white noise from AE signals remains undetermined. For filtering of additive random noise in the system, it is recommended to use the algorithm for wavelet-filtering.

The wavelet-thresholding algorithm is performed on the basis of a discrete wavelet-

transformation [3]. The AE signal s(n) is represented as a wavelet-decomposition, as a set of de-tailing and approximation coefficients, Dm,k and Am',k, which correspond to different scale values.

s(n) = Am',k ϕm',k(n) + Dm,k ψm,k(n), (1)

where ϕ(n) is basis function of sequence of orthogonal embedded subspaces, while ψ(n) is its orthogonal complement.

Representation as a wavelet-decomposition is natural for a multicomponent AE signal, which is a superposition of different wave modes. Characteristics of separate modes of the wave packet are described by the detailing and approximation coefficients.

Fig. 2. Wavelet thresholding algorithm.

Figure 2 shows the algorithm diagram. Upon calculation of the wavelet-decomposition, a

threshold restriction of the detailing coefficients is performed followed by a signal reverse re-covery [4]. There are two main methods of thresholding; that is, soft and hard thresholding. Un-der the hard thresholding the coefficients below the threshold values are set to zero, while the soft thresholding, which is recommended herein, suggests linear approximation of the sub-threshold values.

One of key questions is the correct selection of a threshold. The criterion of Stein's unbiased

risk estimation (SURE) is the most effective and widespread [2, 5]. Assume that the informative AE signal is observed against the background of the additive noise. When using the Stein crite-rion, it is expected that the coefficients resulting from the wavelet-transformation of signals are in conformity with the model,

112

, (2)

where ri,j are empirical coefficients of the wavelet transformation of noisy signal, di,j are true co-efficients, free from noise, value σj characterizes a noise dispersion, while ξi,j corresponds to the normally distributed noise with zero mean and single dispersion. As a result of the wavelet-filtering, on the basis of ri,j evaluation of the true values of wave-let-coefficients, is obtained, and the creation rule is described by equation (3), where

is an operator of thresholding of empirical coefficients of wavelet-transformation with a threshold value τj.

(3)

(4)

(5)

For selecting the optimum threshold value, the risk function Rij given by equation (4) is minimized. For soft thresholding minimization of equation (4) results in equation (5), in confor-mity with which the threshold value is adaptively selected from the range of τi,j, with provision for values of empirical coefficients of the wavelet-transformation and the noise dispersion esti-mation.

Application of the threshold algorithm is equivalent to an adaptive smoothing. When readings

of the detailing coefficients are put to zero, which correspond to the signal high-frequency com-ponents, the signal low-frequency filtering is carried out. On the other hand, only the readings, whose fraction of the high-frequency components is relatively small, below the threshold, are put to zero; in this case no distortion of the signal fragments, in which the high-frequency component prevails, occurs. Thus, a local selective removal of the high-frequency noise components takes place.

Figure 3 shows an example of the wavelet-filtering of AE signal. The stationary white noise is

removed practically completely; the ratio of signal energy and noise for the smaller amplitude impulse is increased fivefold. Filtering of Noise Similar in Waveform to Acoustic Emission Signals

The noise similar in waveform to AE signals are of particular complexity for filtering. These

are the noise generated by various mechanical reasons - impact, impact of foreign objects, and precipitations (such as rain and hail). The noise of this kind also occurs at heating a testing ob-ject and at various hydrodynamic phenomena, for example, resulting from cavitation.

Such noise neither in waveform, nor in spectrum, does not differ from AE signals. Thus, nei-

ther of the above-mentioned methods is suitable for their filtering. The fundamental difference of the noise process and the plastic deformation process is a distinct regularity of generation of im-pulses characterizing one or another process. Times of recording of impulses relevant to the noise action are spread quasi-uniformly (as in case of impact or precipitations) or in accordance with the Poisson's law (in case of cavitation or heating effect). Impulses characterizing an active source of AE (defect) follow the complicated distribution law, of which parameters vary as

113

defect develops. The defect impulses are recorded, as a rule, infrequently, and times of their re-cording bring a disharmony in the form of a certain trend to the distribution law of the back-ground noise impulses.

To research statistical regularities of various processes, it is recommended to increase the su-

pervision time in such a way that the observation length exceeds the stationary interval of noise signal.

Fig. 3 Result of wavelet-filtering: a) initial AE signal; b) AE signal after wavelet-filtering; c) wavelet decomposition of initial signal; d) wavelet coefficient after thresholding reconstructed to the initial length

Fig. 4 Signals characterizing a) single rubbing event; b) rubbing process of 4-second duration.

114

Figure 4 shows signals received under simulation of rubbing. In Fig. 4a one can see a signal characterizing a single rubbing event. This signal is of the impulse and non-stationary type. Fig-ure 4b shows a plot of the 4 second-long signal, being a continuous stationary process, in which a periodic component is distinctly present. According to this example it is possible to make a conclusion that various scales of process supervision create various representations of it, the longer is the time of process supervision, the more definitely we can define its properties.

Wavelet-filtering of long durations

Record of long-waveform AE signals creates additional capabilities for algorithms of filter-ing. For example, the wavelet-filtering can more effectively be performed for the long duration. At the greater value of supervision time the dispersion of stochastic noise, whose value governs the threshold value, is determined more precisely. Further, for the long supervision time the pro-cedure of adaptive wavelet-thresholding can be realized. This procedure assumes representation of the analyzed signal as a set of different-length segments, the length and quantity of segmenta-tion intervals are selected with provision for the signal shape in such a way that the maximum ratio of signal energy and noise after filtering is provided.

Fig. 5 Wavelet-filtering of long-duration AE signal. a) initial signal; b) filtering result.

Optimum filtering of long-waveform AE signals

Effective optimum and adaptive methods of filtering [6] can be applied only in the case of record of the long-waveform AE signals. Synthesis of optimum filters supposes application of priory information, both on the useful AE signals, and on the noise. When employing a tradi-tional threshold procedure for the detection of AE impulses, the limited supervision time does not allow for estimating the statistical parameters required. On the contrary, when recording for long durations, the probability distribution, spectral, correlation and cross-correlation characteris-tics required for optimum and adaptive filtering can be estimated.

The simplest and widespread method of optimum filtering is a Wiener filter. The Wiener fil-

ter is an optimum filter for the detection of the useful signal, which contains in the initial AE sig-nal along with noise. Priory information on spectral density of signal (or noise) is required for its use. As a criterion of its optimization used is the mean-square deviation of signal at the filter output from the specified waveform of signal (or noise).

With the use of this filter, it is supposed that the noise has an additive character, given in

equation (6). The filter coefficients w are calculated in compliance with the optimization crite-rion on the basis of equation (7), where Rff and Rnn are autocorrelation matrices of the AE signal and the noise signal, respectively.

115

(6) (7)

(8)

The filter frequency characteristic is set by equation (8), where Pff and Pnn are the power spectra of f(n) and noise(n). Figures 6 to 8 give the results of Wiener filter application. Figure 6 shows the initial signals corresponding to the AE impulse and rubbing noise. Both signals are received in laboratory conditions as a result of simulation. Its duration is approximately 0.5 second. By applying numerical amplification of the noise signal, we can achieve various values of signal-to-noise ratio (SNR). Figures 7 and 8 show the result of the Wiener filter application for the SNR ratio 0.5 and 0.9, respectively. After filtering we can detect with confidence the AE impulse against the noise background, and SNR has increased approximately tenfold.

Fig. 6 a) AE impulse, b) friction noise; received under simulation.

Fig. 7 a) AE impulse against the rubbing noise background SNR<0.9; b) result of the Wiener fil-ter application SNR>10.

Fig. 8 a) AE impulse against the rubbing noise background SNR<0.5; b) result of the Wiener fil-ter application SNR>5.

116

Signal compression To reduce the data volume, it is recommended to operate not the signal itself, but its enve-

lope. Such replacement does not result in diminution of algorithm accuracy, since the signal spectrum, in this case, is not informative. The envelope characterizes a low-frequency portion of the signal, thus it can be effectively compressed through wavelet-transformation. The data com-pression ratio depends on the contraction ratio of frequency range. If under detection of the sig-nal envelope the value of upper frequency of spectrum is decreased by 28 (or 256) times, the en-velope duration can be reduced in comparison with the initial signal length by 28 times without significant losses of information.

Noisy AE signal is shown in Fig. 9. The noise source is friction, which is artificially modeled

with help of abrasive paper. Besides noise we can observe impulse AE signal generated by Hsu-Nielsen source. Figure 9b shows the signals envelope compressed thousand times.

Fig. 9. Noisy AE signal and its envelope.

Principle components method for noise filtration

To identify processes with a different nature of periodicity, it is possible to employ one of methods of time series analysis – method of principal components. The method of principal components allows for dividing the AE signal represented as the time series, into several ele-ments (components) – periodic components, chance variations and a trend. In the problem of in-terpretation of AE signals, the periodic components characterize, as a rule, deterministic noise, chance variations – random noise, and the trend characterizes the AE source activity.

The method of principal components is a method of data multi-dimensional analysis, but for

the time series analysis, the scientists from the St.-Petersburg State University designed its spe-cial modification – singular spectrum analysis (or Caterpillar method) [7-10]. This method sug-gests a transformation of the univariate time series (Caterpillar), Xi, to the set of time series, si, which represent a great number of fragments of the original signal (Caterpillar links), cut with a sliding time window, as given in equation (9). The time window duration shall be selected so that all processes (the noise processes and AE process) are able to become apparent.

Xi = (si-1,…,si+L-2), where L is window length (9)

When the parameters are selected correctly as a result of application of the method of princi-

pal components, the signal is divided into several components, which characterize the various processes generated by various sources, both the noise sources and the AE sources.

To divide the time series into components, the covariance matrix V is calculated by (10), and

its eigenvalues and eigenvectors are calculated by (11); under expansion of the covariance matrix in terms of eigencomponents, a set of principal components of time implementation is obtained.

117

V = (1/k) XT X (10)

V = P L PT, (11)

where P is the orthogonal matrix of eigenvectors. The eigenvectors by themselves are impulse components of the filters yi, as given in (12), detecting various components of the process.

X P = Y = (y1, y2,...,yM) (12)

In particular, among the principal components one may select a trend, slowly varying com-ponents, periodic and impulse components. Expansion of the AE signal into principal compo-nents is informative, because it allows for getting an insight of the processes occurring in the testing object. Thus, for example, the presence of an increasing trend in the signal can give evi-dence of a progressive defect, while the periodic component characterizes noise

Fig. 10. Filtered AE signal without friction noise.

Figure 10 shows an example of successful application of the method of principal components

for the filtering of AE signal (see Fig. 9a). As a result of signal decomposition, equation (13), it turns out that the periodic noise and the useful impulse signal comply with various components of decomposition. The signal shown in Fig. 9 is restored without considering the periodic noise components.

Fig. 11 Long-waveform AE signals corresponding to a) corrosion process; b) rain and hail noise; c) sum of a) and b) processes.

A more interesting example is shown in Fig. 11. Figure 11a shows the AE signal characteriz-ing development of a corrosion defect, while the process shown in Fig. 11b is noise received as a result of simulation of precipitations effect (rain, hail). Figure 11c gives the sum of these proc-esses. Figure 12 shows the decomposition of the signal presented in Fig. 11c into the main com-ponents. As a result of grouping of the correlated components, we have deduced 9 groups of the components of different types. The first 5 groups showed no indication of the corrosion signals. Next three groups exhibited the presence of the corrosion signals, but still inadequate for clear

118

identification. The highest-order components allow for restoring the useful signal characterizing a corrosion attack, as separately shown in Fig. 13. The strongest 5 AE signals are clearly recov-ered from Fig. 11c.

Fig. 12 Main component decomposition for the detection of useful corrosion signal. Top left is the lowest order and the bottom right is the highest order components.

Fig. 13 The highest order component for the detection of useful corrosion signal, recovered from noisy signals (Fig. 11c). Enlargement of the bottom right in Fig. 12.

119

In the near future, Interunis Ltd. plans to release new generation of A-Line family AE sys-tems. The main advantage of new systems will be the non-threshold principle of AE signal detec-tion and adaptive noise filtration, whose principles are stated in this article. Conclusion

The paper is concerned with various methods of filtering of the AE data. The algorithm of

impulse noise filtering by means of the median filter is described. The efficiency of wavelet thresholding for removal of white stochastic noise is shown. Also the method of noise removal is described by means of the time series analysis using the method of principal components and optimal filtration algorithm.

Acknowledgement Authors would like to acknowledge and extend their heartfelt gratitude to Professor Kanji Ono for his encouragement and valuable remarks. References

1. Daubechies, Ingrid, Ten Lectures on Wavelets, SIAM, Philadelphia, 1992.

2. Donoho, D.L., De-Noising by Soft-Thresholding. IEEE Transaction on Information Theory. 41: 613-627, 1995.

3. Mallat S., A Wavelet Tour of Signal Processing, 2nd ed., Academic Press, 1999, 673 p.

4. Ajvazyan S.A., Buhshtaber V.M., Enjukov I.S., Meshalkin L.D., Applied Statistics in 3 parts, 1989. (in Russian).

5. Stein, C.M. "Estimation of the Mean of a Multivariate Normal Distribution". The Annals of Statistics 9 (6): 1135–1151, 1981.

6. Vaseghi, S.V., Advanced Digital Signal Processing and Noise Reduction, John Wiley & Sons Ltd, 2000.

7. Golyandina, N., Usevich, K., Principal Components of Time Series: the Caterpillar Method, 1997 (in Russian).

8. Golyandina, N., Nekrutkin, V., Zhigljavsky, A., Analysis of Time Series Structure. SSA and Related Techniques, Chapman and Hall/CRC, 2001, 320 p.

9. Special Issue on "Theory and Practice in Singular Spectrum Analysis of Time Series", Statis-tics and Its Interface, 3 (3), 2010.

10. Golyandina N. and Osipov, E., “The “Caterpillar”-SSA method for analysis of time series with missing values”. Journal of Statistical Planning and Inference, 137 (8), 2642-2653, 2007.

J. Acoustic Emission, 28 (2010) 120 © 2010 Acoustic Emission Group

DISCRIMINATION OF ACOUSTIC EMISSION HITS FROM DYNAMIC TESTS OF A REINFORCED CONCRETE SLAB

ENRIQUE CASTRO

1, ROSA PIOTRKOWSKI 2, ANTOLINO GALLEGO

3 and AMADEO BENAVENT CLIMENT

4

1Politechnic University of Cartagena, Department of Applied Physics, Cartagena, Spain, 2 National University of San Martín, ECyT, Buenos Aires, Argentina,

3 University of Granada, Department of Applied Physics, Granada, Spain, 4 University of Granada, Department of Mechanics of Structures, Granada, Spain

Abstract

In the evaluation with acoustic emission (AE) of the damage and state of a concrete structure under a dynamic force, the most relevant information come from the AE hits produced by crack-ing processes. But, unfortunately, there are other sources of AE, which are unrelated with the damage state of the specimen, as the closing of cracks, friction between different elements, or noise from the testing equipment. The discrimination between AE hits from cracking processes and the other sources is useful in order to have an accurate evaluation. In this paper the classifi-cation of AE hits with several signal-processing techniques is investigated. Several dynamic tests were carried out with a reinforced concrete slab attached to a 3-m square MTS shaking table, and during these experiments the AE hits were recorded. The specimen represents, at the 1/3 scale, a flat slab supported on four box-type steel columns, and it was submitted to a simulation of the Campano-Lucano earthquake recorded at Calitri (Italy). After the test, the AE transients were extracted and classified according to several signal parameters. The autocorrelation, wavelet power, kurtosis, RMS in different parts of the signal and approximate entropy were calculated for each signal. A comparison and evaluation of the different classifications according to each parameter is presented. Keywords: concrete structures, transient classification, dynamic test. Introduction

One considerable source of damage in reinforced concrete (RC) structures when they are lo-cated in earthquake-prone areas is cyclic loading induced by ground acceleration during seismic events. These structures are commonly designed to sustain, essentially, two levels of seismic action: low-to-moderate intensity earthquakes (level I), and strong earthquakes (level II). Under level I earthquakes, the RC structure must remain basically elastic. In moderate or high seismic-ity regions, an RC structure can experience several tens of low-to-moderate intensity earthquakes during its lifetime. When an RC structure experiences a high number of low and moderate inten-sity earthquakes there is a concrete degradation of the RC structure that can even produces the slip between reinforcing steel and concrete, which is an important damage for this kind of struc-tures [1]. As visual inspection is not possible in most cases and does not allow knowing in which state the internal part of the structure is, it would be very useful to have a non-destructive tech-nique, which can evaluate the state of the RC structure during its lifetime. The AE technique is very suitable of this purpose because it renders the possibility of recording information of the cracking process during an earthquake, even when it is very low [2].

In order to investigate the health monitoring of an RC structure submitted to low-to-moderate

intensity earthquakes using AE, a prototype structure consisting of an RC slab supported on four

121

box-type steel columns was placed on a shake table. This structure underwent several simula-tions of the Campano-Lucano earthquake recorded at Calitri (Italy) and the generated AE hits were recorded by eight AE low-frequency sensors (type VS30 set in the range 20-100 kHz) dis-tributed near the most critical parts of the structure.

As in a dynamic test like this there are many sources of noise, a lot of precautions were taken

in order to prevent spurious hits. So rubber layers and teflon films were inserted between any surfaces whose contact could generate friction noise, and four guard sensors were placed near the bottom end of the columns to prevent noise generated by the contact between the base plate of the columns and the shake table.

In spite of all the preventive measures adopted to avoid undesired noise, during the analysis

of the results it was observed that the AE sensors registered spurious signals coming from fric-tion or electromagnetic noise. As this noise could disturb the AE analysis and could also appear in a real situation when a building is monitored, it was necessary to investigate the way of filter-ing out the hits, which are not coming from concrete cracking processes. The investigation that has been carried out in order to classify the AE hits is presented in this paper. Different parame-ters calculated from the AE transients have been tested in order to know their capacity of filter-ing out the hits coming from friction or noise. These parameters were: the autocorrelation of the signal, its wavelet power, the kurtosis, its approximate entropy and the root mean square (RMS) of different parts of the signal.

The objective of this paper is to obtain a filter using signal parameters of the AE hits. The de-

scription of the experiment and the obtaining and analysis of the data can be found in [3]. Never-theless, the experiment is also briefly presented here for clarity. Experiment

A prototype structure consisting of a RC slab supported on four box-type steel columns was designed according to Spanish codes. The prototype structure has one story 2.8 m in height and 4.8×4.8 m2 in plan. It is assumed to be located in the moderate-seismicity Mediterranean area. Accordingly, from the prototype structure, the corresponding test model was derived by applying the following scaling factors for geometry, the acceleration and the stress, respectively: λl = 1/2, λa = 1 and λσ = 1. Figure 1 shows the geometry and reinforcing details of the test model. The slab measures 125 mm in depth and it is reinforced with two steel meshes, one on the top made with 6mm diameter bars spaced 100 mm, and another on the bottom consisting of 10 mm diameter bars spaced 75 mm. The slab was reinforced at the corners by shearheads consisting of steel U-shapes 60 mm in depth in order to prevent punching shear failure. The average yield stress fs of the reinforcing steel was 467 MPa, and the average concrete strength, fc = 23.5 MPa.

The test model was placed on the uniaxial MTS 3×3 m2 shake table as indicated in Fig. 2. To

satisfy the similitude requirements between the prototype and the test model, additional steel blocks were attached on the top of the RC slab (total mass m = 7,390 kg). The shake table move-ments were patterned after the Calitri 1980 NS earthquake (Campano-Lucano, Italy) with the time scale compressed by a factor of λl = (1/2)0.5 = 0.707. Two series of seismic simulations were conducted using the Calitri 1980 NS accelerogram scaled to different amplitudes.

122

Fig. 1. Test model: (a) elevation; (b) plan (bottom view).

The Vallen Systeme ASMY-5 was used to measure AE during the tests. Eight low-frequency

AE sensors (type VS30 set in the range 20-100 kHz) were placed on the test model as shown in Fig. 1b. The threshold detection of the AE sensors was set at 45 dB. To prevent undesired noise generated by the contact between the base plate of the columns and the shake table, four guard sensors were placed near the bottom end of the columns as indicated in Fig. 1a (one sensor at each column). Moreover, in order to remove or reduce the sources of spurious friction noise, rubber layers and teflon films were placed between the added steel blocks and the slab. Teflon films were also inserted between any metallic surfaces whose contact could generate spurious friction noise, such as screws, steel plates for fixing the accelerometers to the slab, cables, etc.

123

Fig. 2: Experimental set-up.

Noise Problem

All the preventive measures were not enough to avoid the presence of hits coming from noise sources. Three principal noise sources were indentified:

a) The friction between different elements of the experimental structure. b) The closing of cracks c) Electromagnetic noise

The noise source a) was reduced by the rubber layers and teflon films, but it was impossible to have it completely eliminated due to the fast movements produced during the experiment. The noise sources b) and c) could not be suppressed in any way, the first one because the same specimen has to be used during different seismic simulations, and the second one because it was produced by the shake table, at which the test model were attached.

During a cracking process a high amount of elastic energy is released fast and it propagates as elastic waves with high amplitude at its beginning and then decaying exponentially. So the temporal signal of the AE hits is non-stationary. In this article we will name them as “cracking hits”. However, in signals coming from friction or electromagnetic noise the amplitude must be approximately constant over all the duration of the hit, as the energy is released continuously during a longer period of time. So, the temporal signal of these AE hits is typically stationary. We will call them as “noisy hits”. In Fig. 3 two examples of these kinds of AE hits from our ex-periment are shown.

The objective of the paper is to find a way of filtering out the noisy hits. To do this, a first at-tempt was made noticing that hits with long duration and low amplitude were noisy hits. So the hits that do not meet the following two conditions were not taken into account:

- The duration in µs must be less than 3000 µs + 80 µs/dB*(Amplitude (in dB)-45 dB). - The amplitude must be higher than 60 dB.

It was found, however, that this filtering was not enough as many noisy hits meet the conditions and pass the filter, so it seems that a filter based on the classical AE parameters could not work

124

properly. Thus, it was necessary to find another filtering condition to eliminate the remaining noisy hits based on signal processing of the hits waveform.

Fig. 3. Examples of a cracking hit (left) and noisy hit (right).

Filtering Parameters

In order to build a proper filter, it is necessary to look for a signal parameter, which always takes different values for cracking and noisy hits. If this ideal parameter is found, setting the right threshold will allow us to filter out all the noisy hits. To find this ideal parameter, several signal parameters were investigated and their discrimination capacity tested. These signal pa-rameters were: • The autocorrelation [4]. As the cracking hits are non-stationary and have the maximum am-

plitude at the beginning of the hit, their autocorrelation must decay faster than the autocorre-lation of the noisy hits, which are stationary. We measure this difference by calculating the distance between the two symmetric points of the autocorrelation whose value is 66% of the maximum value. We call this parameter “Autocorrelation Width”

• The wavelet power. In previous investigations in other applications of AE, it was found that the density of wavelet power (DWP) in different frequency bands can be used to determinate the source of AE hits. See [5] for more details. In this paper, the quotient of the DWP in the band [40-100] kHz divided by the DWP in the band [10-40] kHz is used as a signal parame-ter for the discrimination of AE hits. We call it “DWP quotient”.

• The kurtosis. The kurtosis [6] is a measure of the number of extreme values that are in a sig-nal or in a distribution. A higher kurtosis implies that more of the variance is the result of ex-treme values, so their frequency is higher than in a signal with lower kurtosis.

• The approximate entropy. The approximate entropy (ApEn) is a measure of the complexity of a signal [7]. A higher value of ApEn means a higher degree of complexity. As the relevant part of the transient signal of the cracking hits is concentrated at the beginning of the hit, they are less complex and disordered than the noisy hits.

125

• The root mean square (RMS) [6]. As the cracking hits are non-stationary, the amplitude dis-tribution is concentrated at the beginning of the hit, and the RMS of the amplitude in the first part of the signal is higher than the RMS in the final part. The quotient of the RMS in the band [0-400 µs] divided by the RMS in the band [400-1300 µs] is used for discriminating signal parameter and we named it as “RMS rate”.

Results

All the considered parameters have very different values for pure cracking and noisy hits, as can be seen in Table 1 for the hits shown in Fig. 3, so it seems that all of them must filter the noisy hits and allow the cracking hits to pass. However, the problem is that the system under study is very complex and dispersive, so the waveform can change very much in the path from the source to the sensors. It shows that the noisy and cracking hits can have similar values of the parameters.

Table 1. Values of the filtering parameters for the hits shown in Fig. 3.

Autocorrelation Width

DWP quo-tient

Kurtosis ApEn RMS rate

Cracking hit 12 2.16 16.87 0.7379 3.5955 Noisy hit 114 0.056 2.18 0.8392 1.2001 For these reasons, it is necessary to test the filtering capacity of each parameter. We have

visually classified a set of 200 AE hits in cracking or noisy hits and the values of the filtering parameters have been calculated. In Fig. 4 these values can be seen for the cracking and noisy hits. The ideal situation would be to have two sets or clusters of cracking and noisy hits perfectly separated, but, unfortunately, the two sets are mixed for all of the filtering parameters, although not in the same degree. In the case of the Autocorrelation Width and the Approximate Entropy the cracking and noisy hits have similar values. With the DWP quotient, the kurtosis and the RMS rate, the noisy hits have in general a lower value than the cracking hits, so the separation between the two groups and the filtering of the noisy hits is possible. Some cracking hits have a low value of these parameters and are mixed with the noisy hits, and some noisy hits have a high value of the parameters and are mixed with the cracking hits. So, any separation of the two groups of hits based on the value of the filtering parameters will have some hits wrongly classi-fied because they have a parameter value more similar to the other group.

The automatic classifications of the hits using the filtering parameters is made by taking a threshold and place the hits with a parameter value higher than the threshold in a group and the hits with the parameter value lower in the other group. The value of the threshold is very impor-tant to achieve a correct discrimination between the cracking and noisy hits and its influence in the filtering process has to be checked in order to choose the best possible threshold for each filtering parameter. So that, a study of how adequate the filtering process is for each parameter and different thresholds has been carried out.

The results can be seen in Fig. 5, where the rate of cracking and noisy hits that passed the fil-

ter for each parameter depending on the threshold is represented. It can be seen that it is not pos-sible to make a perfect filter, because when all the cracking hits pass the filter there is a high

126

Fig. 4. Values of the filtering parameters for the cracking and noisy hits: a) Autocorrelation Width. b) Kurtosis. c) Approximate Entropy. d) DWP Quotient. e) RMS Rate.

Fig. 5. Percentage of cracking and noisy hits that passed the filter depending on the threshold for each parameter.

127

percentage of noisy hits that pass the filter too. When all the noisy hits are eliminated, almost all of the cracking hits are eliminated too. In general, the percentage of cracking and noisy hits that passes the filter increase and decrease together with the variation of the threshold, although the behavior is different for each parameter. In the case of the Autocorrelation Width, the percentage of noisy hits that pass the filter is higher than the percentage of cracking hits for most of the val-ues of the threshold. The other parameters have a better performance because the percentage of cracking hits is always higher than the noisy ones, but, as it have been said before, neither of them is perfect.

At this point, it is necessary to choose the parameter and threshold that we are going to use.

With Fig. 5 it is possible to choose the parameter and the threshold that can be used depending of our needs. If it is necessary to have all the cracking hits and it does not matter if a certain per-centage of noisy hits is not filtered, then a low threshold of the kurtosis or the DWP quotient can be used. In our case, it is desirable to have the highest percentage of cracking hits together with the lower percentage of noisy hits, so we have chosen to use the parameter and threshold with the maximum difference between the percentage of cracking and noisy hits that pass the filter. With this choice we make sure that most of the hits that pass the filter are cracking hits. Table 2. Percentage of cracking and noisy hits that passed the filter when the best threshold is used, their difference and best threshold for each parameter.

Parameter Autocorrelation Width

DWP quotient

Kurtosis Approximate Entropy

RMS rate

Cracking hits that pass the

filter (%)

91.01 % 76.40 % 74.16 % 59.55 % 79.78 %

Noisy hits that pass the filter

(%)

84.68 % 18.92 % 14.41 % 33.33 % 17.12 %

Maximum Difference:

6.33 % 57.48 % 59.75 % 26.22 % 62.66 %

Threshold: 58.1 0.49 4.5 0.72 1.69

In Table 2 the maximum difference between the percentage of cracking and noisy hits that pass the filter for each filtering parameter can be seen. Using this criterion, the best filtering pa-rameters are RMS rate, Kurtosis and DWP quotient. As the maximum difference is achieved by using RMS rate, it has been used to filter the AE hits in [3]. The AE hits that have been used to obtain the filtering results are a subset of the AE hits, which have been recorded in the experi-ment, so it is expected that very similar percentages of cracking and noisy hits passed the filter when we applied it to the whole set of AE hits. By this reason, after the filter we must have about a 79.89 % of cracking hits and about 17.12 % of noisy hits. In the AE hits, which are visually classified we have a 55.5 % of noisy hits and a 44.5 % of cracking hits, so we have a little more noisy hits than cracking hits, but the difference is very low. It is expected to have similar rates in the whole set of recorded AE hits, so we have that after the filtering process we have 3.83 times more cracking hits than noisy hits. This means a very good improvement over the initial situa-tion.

The quality of the filter can be checked in the analysis of the results. After the filtering proc-

ess the AE hits are analyzed in [3], where it can be seen that the cumulative AE energy is related

128

with the hysteretic strain energy. In [3] it is shown that the results are consistent with the experi-ment and the hysteretic strain energy, so we concluded that the filtering process is adequate for this kind of experiment. Conclusions

We have investigated the way of filtering noisy AE hits, which come from friction or elec-tromagnetic noise. To do so, several signal parameters have been tested in order to find which one provides very different values for the noisy and cracking hits. It was been found that the complexity of the specimen and the dispersion properties of the material make it difficult to separate the noisy and cracking hits because sometimes parameters have similar values.

In order to find which parameter is the most appropriate the percentage of cracking and noisy

hits that pass the filter for each parameter were obtained considering a wide interval of threshold values. It was found that RMS rate, Kurtosis and DWP quotient are the most effective parame-ters because they have the highest difference between the rate of cracking and noisy hits that pass the filter.

This filtering process makes the analysis of the AE results clearer and more precise, and al-

lows us to carry out conclusions with more confidence as we are sure that most of the AE hits that we are analyzing came from a cracking process and not from a noise source. The results of the AE analysis are shown in [3]. References 1. Yuyama S., Li Z.W., Yoshizawa M., Tomokiyo T., Uomoto T.: Evaluation of fatigue damage

in reinforced concrete slab by acoustic emission. NDT & E International, 34, 2001, 381-387.

2. Benavent-Climent A., Castro E. Gallego, A.: AE Monitoring for Damage Assessment of RC Exterior Beam-Column Subassemblages Subjected to Cyclic Loading. Structural Health Monitoring, 8, 2009, 175-189.

3. Gallego A., Benavent-Climent A., Vico J.M., Infantes C., Castro E.: Health monitoring of reinforced concrete slabs during seismic tests using acoustic emission. 29th European Con-ference on Acoustic Emission Testing, Vienna, 2010, Austria.

4. Shiavi, Richard: Introduction to Applied Statistical Signal Analysis. Academic Press, 2007.

5. Piotrkowski R., Castro E., Gallego A.: Wavelet power, entropy and bispectrum applied to AE signals for damage identification and evaluation of corroded galvanized steel. Mechanical Systems and Signal Processing, 23 (2), 2009, 432-445.

6. Lutes L.D., Sarkani S.: Random Vibrations. Butterworth-Heinemann, 2004.

7. Pincus S.M.: Approximate Entropy as a measure of a system complexity. Proceedings of the National Academy of Sciences of United States of America, 88, 1991, 2297-2301.

J. Acoustic Emission, 28 (2010) 129 © 2010 Acoustic Emission Group

USE OF CLUSTER ANALYSIS OF ACOUSTIC EMISSION SIGNALS IN EVALUATING DAMAGE SEVERITY IN CONCRETE STRUCTURES

L. CALABRESE, G. CAMPANELLA and E. PROVERBIO

University of Messina, Dept. Industrial Chemistry and Material Engineering, Messina, Italy

Abstract

AE technique gained increasing interest in the last decade in the field of civil engineering as a monitoring methodology and as an assessment tool for safety and reliability evaluation of reinforced concrete structures, historic and masonry buildings. There are several established statistical methods (Z, RA, b and Ib value), which can be used in analyzing AE data to evaluate damage status of a structure subjected to a particular loading condition. Artificial neural networks (ANN) have recently been applied as a tool to reduce data redundancy and to optimize feature set of AE signals. Cluster analysis was generally used to separate a set of parameters into several classes reflecting dataset internal structure. In this paper such analytical procedure was applied in evaluating acoustic emission data obtained during 4-point bending tests on concrete beams under cycling and constant load condition and at increasing loads. Two kinds of unsupervised clustering methods were used: the principal component analysis (PCA) and the self-organized map (Kohonen map). Combining both methods, it has been possible to quantify the damage severity and to identify the evolution of the damage itself during the test.

Keywords: Civil engineering, concrete, damage analysis, cluster analysis, principal component analysis, Kohonen map, SOM

Introduction

The correlation between the propagation of acoustic waves due to cracking phenomena inside a concrete structure and the seismic wave propagation during an earthquake activity is a well -known concept. Based on these similarities, starting from seismology studies, some statistical techniques to evaluate the “health condition” of concrete structure in a particular load condition have been developed. The b value analysis, from Gutenberg and Richter formula (Gutenberg and Richter, 1954), originally outlines the relation between the magnitude of earthquake with the number of events connected to them. In an AE structure monitoring test, this method and its improved evolution (Ib value) becomes a powerful tool for the identification of fracture modes. This kind of investigation does not require the identification of sources location (Kurz et al., 2006).

Seismicity rate changes have been used in great number of studies as a significant tool in

order to explain the stress distribution in a specific area of the Earth’s crust. Some studies on precursors of past earthquakes suggest that particular space-time seismicity patterns, including the phenomenon of precursory quiescence, can be related to the seismotectonics process that lead to earthquakes. The quiescence hypothesis, as formulated by Wiss and Haberman (1988), postulates that the quiet volume overlaps the main shock source volume. The seismic quiescence hypothesis assumes that some main shocks are preceded by seismic quiescence that is by a significant decrease of the mean seismicity rate (number of events of magnitude exceeding a given threshold, per unit time), as compared to the preceding background rate in the same crustal volume. This consideration has been transferred to cracking phenomena in concrete structure comparing the AE rate within a temporal window with a background rate (Proverbio et al. 2009).

130

As far as fracture mode evaluation is concerned, different approach can be used, the most common one being the combination of two particular AE parameters (RA value and average frequency) (Shiotani et al., 2001a). Trying to define a common rule to identify the different damage mechanisms, particular AE descriptors like amplitude or duration can indeed be used. Sometimes, the analysis of a single parameter, or a combination of a few of them is sufficient; but sometimes this procedure does not provide immediately clear and coherent results. As a consequence, a discrimination of the AE signatures based only on one parameter is debatable, and a multi-variate technique appears an essential instrument for analyzing data not immediately linked to each others (Rippengill et al., 2003; Johnson, 2002).

When the type of damage mechanisms is known in advance the supervised pattern

recognition is used as in the K-nearest neighbors method (K-NN method) (Godin et al., 2003; Hattori and Takahashi, 1999). The term unsupervised pattern recognition is on the other hand used to describe the complete methodology consisting of procedures for descriptors selection, cluster analysis and cluster validity, when no information on attended clusters are available. A popular clustering method is the k-means algorithm (Likas et al., 2003, Ng, 2000). The dimension reduction of large data set can be instead obtained by means of the principal component analysis (PCA), which is a classical method of multivariate statistics (Jolliffe, 2002). PCA involves a mathematical procedure that transforms a number of possibly correlated variables into a smaller number of uncorrelated variables called principal components. The first principal component accounts for as much of the variability in the data as possible, and each succeeding component accounts for as much of the remaining variability as possible.

Neural network procedures also have been successfully adopted to separate numerically

different classes of data, among other the Kohonen's self-organizing map (SOM) (Kohonen, 1990). The self-organizing map is a sheet-like artificial network, the cells of which become specifically tuned to various input signal patterns or classes of patterns through an unsupervised learning process. The main characteristic of the Kohonen algorithm is its ability to develop feature maps corresponding to the distributions of vectors in the input set and to organize such maps in a topologically coherent manner (Godin et al., 2004; Huguet et al., 2002). Thus, given a set of data, the Kohonen algorithm would be expected to organize the output layer as a map, on which similar shapes are detected by clusters of neurons close to each other. The combination of AE multi-parameter analysis and neural networks, in the form of a Kohonen's SOM, was successfully employed to discriminate signals originating from different types of damage (de Oliveira and Marques, 2008).

In the following sections the application of different analytical procedures applied to AE data

collected during concrete beam loading test will be described. Univariate analysis procedures like Ib, Z and RA values were adopted to identify damage occurred during the different loading tests. Furthermore, multivariate analysis based on clustering procedures like PCA and SOM were also applied. Aim of this work was to develop a methodology investigation procedure based on the use of different unsupervised algorithms that could allow not only to handle a great amount of data to obtain a generic information about the structure monitored, but also, in synergy with other classical methodologies, to allow us to extract specific consideration about the structural integrity of the tested structure.

Univariate Analysis Methodologies

131

Ib and Z value Starting from the original Gutenberg and Richter formula (Gutenberg and Richter, 1954), it

could be possible to distinguish the different fracture condition by calculating the slope (b value) of the function:

log N = a – b(AdB /20) (1)

where N is the number of AE events with an amplitude higher than AdB. The b-value analysis of AE is in general applied to groups of subsequent events, a group of 300 events was used in this work. While testing materials undergoing brittle failure, the b-value is found to range from 1.5 to 2.5 in the initial stages (Rao, 2005), it then decreases with increase in stress to attain values about 1.00 and less, showing temporal fluctuations as the impending failure approaches in the material. A high b-value arises due to a large number of small AE hits (or events) representing new crack formation and slow crack growth, whereas a low b-value indicates faster or unstable crack growth accompanied by relatively high amplitude AE in large numbers (Colombo et al., 2003).

By implementing the formula in order to avoid the problem to define amplitude range and the

number of AE data to obtain the b-value, an improved b-value (Ib-value) was proposed by Shiotani (Shiotani et al., 2001b);

Ib =

log10N(w1) − log10N(w2 )(a1 + a2 )σ

(2)

where N(w1) is the accumulated number of AE events, in which the amplitude is more than µ – α1σ, and N(w2) is the accumulated number of AE events, in which the amplitude is more than µ + α2σ, σ is the standard deviation of the magnitude distribution of one group of events, µ is the mean value of the magnitude distribution of the same group of events, α1 and α2 are constants.

A parameter to describe changes in seismicity activity is the seismic quiescence. A test commonly used to compare seismicity rate changes is the Z-value test for a difference between two means. The Z value measures the significance of the difference between the mean seismicity rates µ1 and µ2 within two time intervals (Monterosso, 2003).

Z =µ1 − µ2

σ12

n1

+σ 2

2

n2

(3)

here σ and n are, respectively, the standard deviation of the rate and the number of events in the group. In particular, the long term average (LTA) is the Z value resulting from the comparison of the rate within a window, µWin and the background rate, µAll defined as the overall mean rate in the volume. Negative Z values indicate rate increases while a positive Z corresponds to rate decrease. In this work, groups of events detected in a temporal window of 2000 seconds during each load condition were considered.

RA values analysis

During structure damage, it is well known that the cracking fracture mode evolves from mode I type (nucleation and growth of tensile crack) to mode II type (shear propagation of crack) with progression of fracture. RA value and average frequency (AF) are classical parameters to classify active cracks (Japan Construction Material Standards, 2003). RA value is defined as the ratio between rise time and the peak amplitude, while the AF of a waveform is calculated as a ratio between the AE counts and duration. Small RA values and the large average frequencies are

132

attended under flexural load, where tensile cracking was prominent, while under shear loading, large RA values and lower average frequency can be observed.

Multivariate Analysis: Clustering Algorithms Principal components analysis (PCA)

PCA is a mathematical algorithm used to reduce the dimensionality of a data set for compression, pattern recognition and data interpretation. The algorithm projects, by a linear transformation, a p-dimensional data vector X into a new q-dimensional data vector Z, containing what is referred to as the data’s ‘principal components’ (Johnson, 2002). Given the data Xi = (x1i, x2i,..., xpi) with i = 1,...,N, the new data vector is Zi = (z1i, z2i,...., zqi) where z1 is the linear combination of the original xj (j = 1,....,p) with maximal variance, z2 is the linear combination, which explains most of the remaining variance and so on, i.e. the new data vector, referred to as the principal component, is a linear combination of the original data and is both uncorrelated and orthogonal to all other principal components. The first principal component accounts for the maximum variance in the data, the second principal component accounts for the maximum of the remaining variance and so on. This method is based on achieving a specified cumulative percentage of total variance extracted by successive factors. The purpose was to ensure practical significance for the derived factors by ensuring that they explain at least a specified amount of variance. In our case 67% of total variance was evaluated and this led to realize PCA plots by using only three principal components.

Self-organizing neural networks

The self-organizing map (SOM) is one of the most prominent artificial neural network models adhering to the unsupervised learning paradigm (Kohonen, 1990). The model consists of a number of neural processing elements, i.e. units. Each of the units i is assigned an n-dimensional weight vector mi, mi Є Rm. The training process of self-organizing map may be described in terms of input pattern presentation and weight vector adaptation. Each training iteration t starts with the random selection of one input pattern x(t). This input pattern is presented to the self-organizing map and each unit determines its activation. Euclidean distance between the weight vector and the input pattern was used to calculate a unit’s activation. In this particular case, the unit with the lowest activation is referred to as the winner, c, of the training iteration, as given in equation 4;

c: mc (t) = mini x(t) – mi (t) . (4)

Subsequently, the weight vector of the winner as well as the weight vectors of selected units

in the vicinity of the winner are adapted. This adaptation is implemented as a gradual reduction of the difference between corresponding components of the input pattern and the weight vector, as shown in equation 5;

mi(t+1) = mi(t) +α(t)*hci(t)*[x(t) – mi(t)] (5) Geometrically speaking, the weight vectors of the adapted units are moved slightly toward

the input pattern (Fig. 1). The amount of weight vector movement is guided by a so-called learning rate, α, decreasing with time. The number of units that are affected by adaptation as well as the strength of adaptation is determined by a so-called neighborhood function, hci. This number of units decreases with time. Typically, the neighborhood function is a unimodal function, which is symmetric around the location of the winner and monotonically decreasing with increasing distance from the winner.

133

Fig. 1. Scheme of SOM operation.

A Gaussian is used to model the neighborhood function. At the beginning of training a wide area of output space is subject to adaptation. The special width of units affected by adaptation is reduced gradually during the training process. Such a strategy allows the formation of large clusters at the beginning and a fine-grained input discrimination towards the end of training process. The results of the training process are the so-called U-matrix. Additional information can be obtained plotting the hit histogram U-matrix. This map shows the projection of data samples into map. The data are calculated by finding the BMU (best matching unit) of each data sample from the map, and increasing a counter in the map unit each time it is the BMU. The colors are related to a specific level of a variable and hexagon size is related to the numbers of AE hits related to that cluster point.

Experimental

Concrete beams, 500x140x140 mm, made with 15 MPa flexural strength ordinary Portland cement, have been tested in a four-point bending test configuration (Fig. 2). AE signals were recorded by ten-channel Vallen AMSY-5 equipment. A total of 8 piezoelectric transducers (4 for each lateral side), VS30-V type, with a flat response between 23-80 kHz were used. Threshold value after pencil-lead break calibration was set at 40 dB: that allowed to eliminate the background noise and to record only the emissions due to cracking of the concrete. Load tests including 4 main loading cycles were performed with increasing loads. Every cycle, moreover, was set up by 3 different load conditions (Table 1 and Fig. 3).

Every main loading test starts with a 10-min. constant rate load condition and 10-min.

constant rate unloading condition for two reasons: first our aim was to pre-crack the samples before applying a long time load, and the second aim was to evaluate the differences in energetic release during unloading step from the stored energy during loading. Every sub-cycle of the alternate loading step was at a frequency of 0.1 Hz. Constant load condition at the maximum load had duration of about 300 minutes. Each AE signals was described by 12 parameters, including Amplitude, Counts, Duration, Rise-time and Energy that were calculated by the

134

acquisition system, while Historic Index, Severity, RA value and Average Frequency were derived from standard parameters. Time, sensor number and “load cycle” were also here considered as variables. To allow data comparison with different scale, logarithmic values were used instead of their natural values.

Fig. 2- Four point bending test set-up

Results

Ib and Z values vs. test number (i.e. loading history) is reported in Fig. 4. It is possible to observe that values were characterized by a floating trend. Before the 8th load test, the curve has low values. A crescent trend can be identified in each group of test characterized by same maximum load.

135

Table. 1. Load cycle sequences.

Test no Cycle no Maximum load (kN)

Load condition Duration (min)

1 15 Slow loading and slow unloading 21

2 15 Alternating loading 340

1

3 15 Constant loading 320

4 18 Slow loading and slow unloading 21

5 18 Alternating loading 390

2 6 18 Constant loading 320

7 19 Slow loading and slow unloading 21

8 19 Alternating loading 370

3

9 19 Constant loading 230

10 20 Slow loading and slow unloading 21

11 20 Alternating loading 410

4

12 20 Constant loading 253

Fig. 3. Load cycle sequences. As expressed in literature, low Ib values correspond to the macrocrack opening. Once the

generated macrocracks opened up, most of the energy has been released. It implies the creation of many weak events leading to an increase of Ib value. In our results, during cycling loading tests low values of Ib are obtained, as a consequence of the activation of macrocracks. In the constant load tests, the Ib value increases due to reduced energy released for crack propagation. It could be related with microcracking events. The constant load conditions (3rd and 6th load tests) influence the progression of Ib value more than the alternating loading ones (2nd and 5th load tests). With the 8th load test (alternating load test), it is possible to observe the dominant generation of new microcracks (evidenced by relevant increase of Ib) that propagate and become macrocracks during the constant load condition (9th load test). This behavior clarify that test sequence at 19 kN (Test No. 3) critically compromised the mechanical integrity of the concrete structure.

136

Fig. 4. Ib and Z values vs typology load test.

The Z and Ib statistics show a similar pattern. A significantly high Z-value was found at the

8th load test. Paraphrasing the seismologic terminology, this particular load condition could be considered like the beginning of a quiescent period preceding a severe damage condition of the structure. However, it should be pointed out that Z trend does not show efficiently the local maxima and minima, evidenced instead in the Ib pattern. This suggests that the Ib analysis is more sensitive to local stress variations, while the pattern of Z can be useful in understanding the global behavior of the investigated structure

Fig. 5. Cracking changing from tensile type of fracture to shear type during load step cycles.

137

Figure 5 shows relationship between RA value and average frequency of concrete specimen under all loadings. The data are clustered with a centroid representative of all data for all sensors and then classified for load cycle typology. Analyzing the figure, it’s possible to observe two different behaviors: before the 8th load test, the plotted data are distributed in the area where the average frequencies are medium/high and RA values are low. In particular, at increasing cycle, the magnitude of AF parameter increases. This region is representative of mode I crack behavior. In the 9th load test an inversion point can be found; in fact at very high load cycles, the AE events are observed in the region where the average frequencies are low. This region is typically connected with mode II crack behavior. Thus, 8th and 9th load tests represent the transition from tensile type to shear type of fracture mode.

Fig. 6. Variable distribution on components; dotted circles refer to main variable groups; dotted arrow indicates trend of data clusters against time.

These results confirm that “traditional methods” can be useful to investigate the damage evolution of the structure, even if information is sometimes incomplete and difficult to interpret. The support by “unconventional methods”, specifically applied to multivariable systems (such as PCA and SOM) can be considered a valid tool for a better comprehension of the damage phenomena. In Fig. 6 the PCA plot is shown. In the graph, the vertical axis represents the unrotated factor II (component 2), and the horizontal axis represents the unrotated factor I (component 1). The axes are labeled with 0 at the origin and extend outward to ±0.5. The numbers on the axes represent the factor loadings. In this graph, all the variables are labeled. From visual inspection, it shows the presence of 2 main groups of variables: Test, Load and Time (variables that we can define generically as time dependent) are grouped together and

138

factor loading is fairly high. The variables strictly connected to the wave properties (Rise-time, Duration, Counts and Energy) generates a second group high on the unrotated first factor. Moreover, analyzing the particular data distribution, we can consider them as organized in three clusters strictly connected with time dependent variables trend. This indicates the degree of contribution of these variables as data discriminating factors. Irregular are the distribution of other variables like RA, Historical Index and AF. Severity is equally correlated to time and wave dependent variables, as shown by its high factor loading in all principal components and therefore included in the following SOM analysis. Analyzing the variables distribution on first and third component graph, Amplitude loads high on the unrotated third factor. A little more difficult is the comprehension of groups analyzing the distribution on first and second component graph. These plots are not included in this work because the variable classification and their clustering are less interesting for the interpretation of experimental data.

For better comprehension of PCA results, factor loadings grater than ± 0.3 are considered to

meet the minimal level of significance; loadings greater than ± 0.4 are considered practically significant. Comparing PCA analysis results with correlation matrix of variables, Load and Test variables can be removed, because these correlated with Time variable. Rise-time, Counts and Energy can be removed because of their equivalence to Duration. Using theses residual variables and the ungrouped ones, the SOM algorithm was adopted.

a b c Fig. 7 a) U-matrix resulting from the application of the SOM; b) Hit histogram U-matrix for the Load variable; c) Hits U-matrix considering the cycling load conditions as variables.

In Fig. 7a, the U-matrix map related to SOM analysis is reported. This map shows distances

from maps units and their nearest neighborhood, evaluated by Euclidean method. High values of U-matrix map (red and yellow pixels) mean large distances between neighboring map units. Elements belonging to the same cluster are therefore identified by uniform areas of low value

139

(blue pixels). In this specific case, we identified 3 main depression areas associated with their cluster areas.

In Fig. 7b, the hit histogram U-matrix related to the load variable is shown. Cluster 1 is

mainly related to 19 kN load condition (blue hexagons) and as discussed before, this region represents the condition, in which weak events increase Ib value, as a consequence of microcrack development, where Z value revealed a quiescence period. Cluster 2 is related to 15 kN load (red hexagons) and 18 kN (green hexagons) conditions. This area was connected to nucleation and growth of macrocracks (mode I fracture behavior). Cluster 3 is related to 20 kN load condition (yellow hexagons) and as reported before, related to existing crack propagation in shearing mode (mode II or III fracture behavior).

In Fig. 7c, the hits U-matrix related to the specified test conditions is represented. The 19 kN

alternating load condition, identified as 8th load cycle, outlines the transition from tensile type to shear type of fracture mode. Through the Hits U-matrix, it was possible to localize, inside the map, hits connected to this particular load condition and to compare these positions with the hits representative of two other alternating load conditions; that is, the hits connected to 18kN alternating load condition (5th load cycle) and 20 kN alternating load condition (11th load cycle). As can be seen, the 8th load cycle (red hexagons) is mainly present in Cluster 1 region and we can identify this area as quiescent seismic region; the 5th load cycle (blue hexagons), is present in Cluster 2 region (mode I cracks area) while the 11th load cycle (green hexagons) is mainly present in Cluster 3 region (mode II cracks area).

Fig. 8. Variable maps resulted from application of Kohonen self-organizing map algorithm.

It is interesting to compare U-matrix with topological maps of single variables (Fig. 8). The

cluster regions observed in U matrix are quite similar to Time Map, outlining the strong weight of this variable compared with others. Similar considerations came out in PCA analysis about time dependent variables. Sensor Map is characterized by homogeneous distribution of results. This correspondence confirms the high coherence in data acquisition by each sensor during all

140

loading tests. Furthermore, analyzing Severity Map, the highest severity conditions were connected to Cluster 3 region. In this region we identified hits connected to 20 kN load condition and, referring to Z value results in this area exceeding quiescence condition, the structural integrity of specimen starts to be compromised. Minimum values correspond to data inside Cluster 1 region (intermediate test loading condition).

Conclusions

Clustering algorithms were used and compared with traditional procedures for damage

evaluation (Ib, Z and RA values). Principal components analysis allowed us to correlate clusters to groups of variables and to identify related variables. In such a way a reduction of original data matrix was obtained. The Kohonen’s self-organizing map (SOM) algorithm was adopted to identify clusters and relate them to the AE patterns. Although the use of these new methodologies appears promising as a practical tool of evaluation of many different parameters in a restricted number of graphs, it requires the development of validation procedure to optimize a correct interpretation of great amounts of data, even requiring integration with classical well-established procedures. References Brock G., Pihur V., Datta S. Datta S. (2008), “clValid: an R package for cluster validation”, J Stat Softw, 25 (4), 1-22.

Colombo S., Main I.G., Forde M.C.J. (2003), “Assessing damage of reinforced concrete beam using “b-value” analysis of acoustic emission signals”, J. Mater. Civil Eng. 15, 280-286.

Cox T.F. (2005) An Introduction to Multivariate Data Analysis, Hodder Arnold Publication, London.

de Oliveira R., Marques A.T., (2008), “Health monitoring of FRP using acoustic emission and artificial neural networks”, Comput Struct, 86 (3-5), 367-373.

Duch W., (1997) “Neural minimal distance methods”, in Proc. 3rd Conference on Neural Networks and Their Applications , Kule, Poland, Oct. 1997, p. 183–188.

Emamian V., Kaveh M., Tewfik A., Shi Z., Jacobs L.J., Jarzynski J., (2003) “Robust clustering of acoustic emission signals using neural networks and signal subspace projections”, EURASIP J Applied Signal Processing, 3, 276-286.

Godin N., Huguet S., Gaertner R., Salmon L., (2004), “Clustering of Acoustic Emission Signals Collected during Tensile Tests on Unidirectional Glass/Polyester Composite Using Supervised and Unsupervised Classifiers” NDT&E Intl, 37, 253-264.

Golaski L., Gebski P., Ono K. (2002), “Diagnostics of Reinforced Concrete Bridges by Acoustic Emission”, Journal of Acoustic Emission, 20, 83-98.

Gutenberg B., Richter F., (1954), Seismicity of the Earth and Associated Phenomena, Princeton University, Princeton.

Hair J.F., Anderson R.E., Tatham R.L., Black W.C. (1995), Multivariate Data Analysis, Prentice-Hall, Upper Saddle River, NJ.

Hattori K. and Takahashi M., (1999), “A new nearest-neighbor rule in the pattern classification problem”, Pattern Recogn, 32 (3), 425-432.

141

Huguet S., Godin N., Gaertner R., Salmon L., Villard D., (2002), “Use of acoustic emission to identify damage modes in glass fibre reinforced polyester”, Compos Sci Technol, 62 (10), 1433-1444.

Japan Construction Material Standards (2003), “Monitoring Method for Active Cracks in Concrete by Acoustic Emission”, JCMS-IIIB5706 The Federation of Construction Material Industries; Tokyo, Japan.

Johnson M., (2002), “Waveform Based Clustering and Classification of AE Transients in Composite Laminates Using Principal Component Analysis”, NDT&E Int, 35, 367-376.

Jolliffe I.T., (2002), Principal Component Analysis, 2nd ed., Springer, NY.

Kohonen T. (1990), "The Self-organizing Map." Proceedings of IEEE, 78 (9), 1464-1480.

Kurz J.H., Finck F., Grosse C.U., Reinhardt H.W. (2006) “Stress Drop and Stress Redistribution in Concrete Quantified Over Time by the b-value Analysis”, Structural Health Monitoring, 5; 69-81.

Likas A., Vlassis N., Verbeek J.J., (2003), “The global k-means clustering algorithm” Pattern Recogn, 36 (2), 451-461.

Monterosso Juarez A.D., (2003), “Statistical Seismology Studies in Central America, b-value, seismic hazard and seismic quiescence”, Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology, Acta Universitatis Upsaliensis, Uppsala, 897, 1-27.

Ng M.K., (2000), “A note on constrained k-means algorithms” Pattern Recogn, 33 (3), 515-519.

Proverbio E., Venturi V., Campanella G. (2009) “Damage assessment in post-tensioned concrete viaduct by b- and Ib-value analysis of AE signal”, in Proc. of NDTCE’09 Non-Destructive Testing in Civil Engineering ,Nantes, France, June 30th -July 3rd.

Rao M.V.M.S., Prasanna Lakshmi K.J., (2005) “Analysis of b-value and improved b-value of acoustic emissions accompanying rock fracture”, Curr. Sci. India, 89, 1577-1582.

Rippengill S., Worden K., Holford K.M., Pullin R., (2003), “Automatic Classification of Acoustic Emission Patterns”, Strain, 39 (1), 31-41.

Shiotani T., Ohtsu M., Ikeda K., (2001a) “Detection and evaluation of AE waves due to rock deformation”, Constr. Build.Mater, 15, 235-246.

Shiotani, T., Yuyuma, S., Li, Z. and Ohtsu, M. (2001b) “Application of AE improved b-value to quantitative evaluation of fracture process in concrete materials” Journal of Acoustic Emission, 18, 118-133.

Wyss M., Habermann R. E. (1988) “ Precursory seismic quiescence”; Pageoph, 126, 319-332.

J. Acoustic Emission, 28 (2010) 142 © 2010 Acoustic Emission Group

SIMULATION OF LAMB WAVE EXCITATION FOR DIFFERENT ELAS-TIC PROPERTIES AND ACOUSTIC EMISSION SOURCE GEOMETRIES

MARKUS G. R. SAUSE and SIEGFRIED HORN

University of Augsburg, Institute for Physics, Experimentalphysik II, D-86135 Augsburg, Germany

Abstract

In the present study the influence of microscopic elastic properties and the geometry of the

acoustic emission (AE) source are investigated by finite element simulation. We investigate the formation process of Lamb waves in isotropic and anisotropic plate specimen by simulation of five different AE source configurations. As reference a microscopically and macroscopically homogeneous, isotropic aluminum specimen with point force couples is used. For fiber-reinforced materials, the elastic properties are typically anisotropic. A comparison is made be-tween the Lamb wave formation of the anisotropic, homogeneous model specimen and those of an anisotropic, microscopically inhomogeneous model. It is demonstrated, that the microscopic elastic properties of the AE source have significant influence on the excitation of distinct Lamb wave modes. This can be used to distinguish between failure mechanisms like fiber breakage or resin fracture in fiber reinforced materials. 1. Introduction

Analytical descriptions of AE sources due to crack formation and propagation and the subse-quent propagation of ultrasonic signals are well established in literature [Ohtsu1984, Gior-dano1999, Lysak1996]. Based on the generalized theory of AE, the microscopic crack surface displacement can be linked to the macroscopic displacement at the surface of the solid by suit-able Green’s functions [Ohtsu1986]. While the Green’s functions for partially infinite media are easy to obtain, the situation grows more difficult, when dealing with complex geometries of fi-nite extent. In addition, another major difficulty is the application of Green’s functions to inho-mogeneous, anisotropic media. Although such cases can be investigated by analytical descrip-tions [Green1995, Green1998], they are typically only suited for the specific geometry and mate-rial under investigation.

In recent years various authors have applied the finite element method (FEM) to the simula-tion of AE formation and AE signal propagation for the case of plate specimens [Dietz-hausen1998, Prosser1999, Hamstad2002]. The AE source was typically modeled as point force couple. Due to the increase in computational capacities such simulations are becoming an impor-tant part to improve the understanding of AE signal formation and propagation.

The finite element method allows an intuitive approach to subdivide modeling and simulation

of AE signals. Utilizing a multi-scale approach the modeling of AE can be split into three dis-tinct steps:

• Modeling of source mechanism • Modeling of signal propagation • Modeling of signal detection The first step addresses the description of the AE source. Typically buried dipole sources re-

alized as point force couples are used. This requires knowledge of the source radiation direction

143

and the associated magnitudes of the modeled source mechanism. In addition, various source-time functions are employed in literature, which model the source kinematics [Ohtsu1986, Gior-dano1999, Hamstad1999]. In a previous publication, we presented a new source model to simu-late the microscopic failure in carbon-fiber reinforced plastics (CFRP). This source model takes into account the finite extent of the crack and the microscopic geometry and elastic properties of the AE source [Sause2010]. Thus in the following the term “source model” refers to the finite dimensions of the crack model and the geometry and elastic properties in the vicinity of the crack model.

Subsequent to signal excitation, the elastic wave propagates into the solid and is reflected at

the respective boundaries of the geometry. In plate-like geometries with thickness d the signal propagation is analytically described by Lamb waves for wavelengths λ < 3d [Lamb1917]. In finite element simulations the propagation of such signals is modeled based on time dependent solutions of differential equations of the equilibrium states [Castaings2004, Nieuwenhuis2005, Prosser1999, Hamstad1999, Sause2010]. As pointed out for isotropic media by Hamstad et al., different source radiation directions can excite different ratios of symmetric and antisymmetric Lamb wave modes [Hamstad2002]. This in turn can lead to different frequency compositions of the detected AE signals, as the propagation of the zero-order symmetric (S0) mode occurs at higher frequency than the zero-order antisymmetric (A0) mode [Hamstad2002]. Due to the mi-croscopically inhomogeneous nature of fiber-reinforced composites, strong differences in the source radiation direction are expected depending on the failure mechanism. In a previous publi-cation it was demonstrated, that such differences of the source radiation direction and of the mi-croscopic elastic properties cause distinct differences in the excited Lamb wave modes, which compare well to experimental data recorded during failure of CFRP specimens [Sause2010, Sause2010b]. In addition, influences of the displacement magnitude and the source excitation time were investigated [Sause2010].

Signal propagation in the previously used specimen types suffered drastically from boundary

reflections [Sause2010, Sause2010b], since interference of signals with their boundary reflec-tions makes identification of distinct Lamb wave modes more difficult. Therefore, this approach is extended to larger specimen geometries. In addition, the case of isotropic media is treated as a reference.

Finally, the importance of simulation of the detection process of AE signals cannot be over-emphasized. While in [Sause2010] an abstract representation of a typical broadband piezoelectric sensor was used, this approach was extended to a full-scale simulation in [Sause2010b]. How-ever, since the present study will focus on basic investigation of the influence of the source mi-crostructure the detection process will be limited to the simulation of surface displacements at distinct positions at the model specimen surface. 2. Description of Model

The present simulations were performed for rectangular plate specimens with 200 mm x 400 mm edge length and 1.4 mm thickness. Utilizing symmetric boundary conditions at the yz- and xz-plane only one quarter of the total plate volume was modeled as shown in Fig. 1. The AE source is located at the medial plane at the center of the plate specimen (x,y,z) = (0,0,0) mm. All elastic properties describing the propagation medium and the source model are summarized in Table 1. The respective AE signals are obtained from the surface displacements at the positions marked in Fig. 1 (red crosses) under an angle .

144

Fig. 1: Schematic drawing of plate model including dimensions and symmetry planes.

Table 1: Summary of elastic properties of the materials used within the simulation.

Aluminum

6063-T83 CFRP T800/913

Resin HexPly 913

Carbon Fiber T800S

Density [kg/m³] 2700 1550 1230 1810

Poisson-Ratio 0.33 - 0.35 0.20

Elastic coefficients

[GPa]

E = 69.0 C11 = 154.0 C12 = 3.7 C13 = 3.7 C22 = 9.5 C23 = 5.2 C33 = 9.5 C44 = 2.5 C55 = 4.2 C66 = 4.2

E = 3.39 E = 294.0

A quarter-volume representation of the various microscopic source models is shown in Fig.

2-a-f. Five different configurations of source models were used in the following:

(A) Isotropic aluminum plate and homogeneous source (B) Anisotropic CFRP plate and homogeneous source (C) Anisotropic CFRP plate and inhomogeneous source (resin properties) (D) Anisotropic CFRP plate and inhomogeneous source (carbon fiber properties) (E) Point force couple The first configuration is shown in Fig. 2-a. The crack is modeled as a three-axes cross,

which is cut out of the homogeneous medium around the source. Each of the three axes of the cross is composed of a bar of 30-µm length, and 10 µm x 10 µm cross-section, respectively. The remaining domains visible in Fig. 2-a are not relevant for this case and will be explained below. For the present configuration, all domains were assigned elastic properties of aluminum. The second configuration is also represented by Fig. 2-a, although all domains were now assigned

145

homogeneous, anisotropic elastic properties of T800/913. The third configuration is shown in Fig. 2-b. Here the microscopic elastic properties of a cube with edge length rcube = 200 µm were those of the HexPly 913 resin as given in Table 1. The elastic properties of the macroscopic plate were those of T800/913. While the dimensions of the crack model and the remaining domains were all comparable to the configurations considered before, now a smooth transition between microscopic and macroscopic elastic properties is required. This is achieved utilizing a cube with 2rcube = 400 µm edge length which encloses the resin cube. This so-called perfectly matched layer (PML) gradually changes the elastic properties of the enclosed medium Cij, 0 to those of the surrounding medium Cij, 1 as a function of distance r as follows:

(1)

The material density was gradually changed accordingly. This smooth transition is exemplarily visualized in Fig. 2-d in a pseudo-color diagram. The fourth source model configuration differs from the third configuration only by a different assignment of elastic properties of the domain marked in red in Fig. 2-c. In this case the cuboid with edge length of 85 µm and 10 µm x 10 µm cross-section has elastic properties of T800S carbon fibers instead of HexPly913 resin. The re-spective transition to the macroscopic properties of the T800/913 CFRP plate is realized by a respective PML as shown in Fig. 2-e. For the purpose of comparison, point force couples as shown in Fig. 2-f were also simulated. In the present configuration, the total distance of the two points acting as buried dipole source was 300 µm with 1 N force magnitude.

In order to excite an AE signal, in the following a linear source-time function was used. Since the influence of various source radiation directions was already addressed in [Sause2010], in the present study only in-plane crack surface displacements dx as marked in Fig. 2-b were con-sidered. Following the previous publications, a linear source-time function was used to deflect the respective crack surface by a magnitude d0,x within an excitation time Te,x [Sause2010, Sause2010b]:

(2)

After the excitation time Te,x, the boundaries of the source model are free of constraints. The

subscript x marks the respective direction of crack surface deflection. It is worth to note that due to the symmetry conditions of the model the crack surface displacement is also symmetrical with respect to the yz-plane. For better comparison of the obtained results, the magnitude d0,x and ex-citation time Te,x was kept constant for all models at d0,x = 100 nm and Te,x = 100 ns.

Since the present approach introduces a multi-scale problem, the resolution of the finite ele-ment mesh was gradually refined from 2.0 mm down to 0.01 mm when approaching the model source. The higher mesh resolution close to the source is necessary to resolve the geometric de-tails of the source model, while the coarse macroscopic resolution is still sufficient to describe signal propagation within the plate specimen up to frequencies of 1 MHz. Since exact symmetry and quality of arbitrarily generated 3D tetrahedral meshes is typically difficult to achieve [Fre-itag1997, Parthasarathy1994], the source model could not be meshed symmetrically with respect to the xy-plane.

146

Consequently, there is a break of symmetry with respect to the xy-plane, which can cause differences in the calculation of signal propagation along the positive and negative direction of the z-axis, respectively. As a result antisymmetric Lamb wave modes can occur. The sensitivity of the Lamb wave formation based on this effect cannot be overemphasized. Within reasonable steps of refinement of the mesh density around the source model the excitation of antisymmetric Lamb waves could not be prevented. Based on physical considerations such microscopic asym-metry within the material is always expected and therefore the present configuration is meant to reflect realistic conditions better than perfect symmetric mesh conditions. To ensure comparabil-ity of the results of different source model configurations, in the following all simulations were performed using an identical mesh.

The temporal resolution was chosen to be 10 ns for < 1 µs, which is required to resolve the

excitation process. For 1 µs ≤ ≤ 50 µs the temporal resolution was decreased to 100 ns, which is sufficient to temporally resolve the signal propagation process.

Fig. 2: Quarter-volume representation of AE source model configurations. (a) Homogeneous case. (b) model of matrix cracking. (c) model of fiber breakage. (d, e) Pseudo-color diagram of material density visualizes transition of elastic properties utilizing perfectly matched layer and (f) model of point force couple.

147

3. Results The results of the simulations are presented below for the five different source model con-figurations. To visualize the modal composition of the AE signals the Choi-Williams transforma-tion is calculated using the software package AGU-Vallen Wavelet [Choi1989, Vallen2010]. The results are compared to analytical solutions of dispersion curves calculated for the respective propagation medium and distance [Vallen2010, Zeyde2010]. 3.1 Homogeneous, isotropic medium The Lamb wave propagation excited by source model configuration A is shown in Fig. 3 at six distinct times. The in-plane displacement located at the medial plane of the plate excites a strong S0 Lamb wave mode with preferential orientation along the x-axis. As visible in Fig. 3, for 35 µs 50 µs minor reflections occur at the boundary in y-direction. But for times 40 µs no boundary reflections arrive at the designated detection positions.

Fig. 3: Propagation of S0 Lamb wave mode in aluminum plate at six distinct times.

As already mentioned in the introduction section, point force couples are often used in litera-

ture to model AE sources. Following the approach of Hamstad et al. [Hamstad1999] a respective aluminum plate with dimensions as given in Fig. 1 was modeled, using an in-plane point force couple. The resulting CWD-diagram of the z-displacement of the surface at (x,y,z) = (50,0,0.7) mm is shown in Fig. 4-a. For comparison the CWD-diagram of the signal at the re-spective position using the source model configuration A is shown in Fig. 4-b. In addition both CWD-diagrams show dispersion curves for the S0 and A0 Lamb wave modes calculated for the propagation distance of 50 mm. For aluminum with elastic properties as given in Table 1, the

148

longitudinal and transversal wave velocities were calculated as vL = 6153 m/s and vT = 3100 m/s, respectively.

Clearly, the signals of both simulations are dominated by the contribution of the S0 Lamb wave mode as seen in Fig. 3. In direct comparison the source model configuration of finite extent shows in addition an excitation of the A0 Lamb wave mode. This contribution is hardly visible in the CWD-diagram in Fig. 4, since the magnitude of the coefficients is dominated by the intensity of the S0 Lamb wave mode. Instead, the minor contribution of the A0 Lamb wave mode is ob-served better as low frequency oscillation of the signal in time domain after t = 13 µs.

Fig. 4: Comparison of surface displacement signals at (x,y,z) = (50,0,0.7) mm for point force couple model (a) and model of finite extent (b) with superimposed dispersion curves.

Since the source model configuration A uses an axially oriented surface displacement, a de-pendency on the orientation of the source relative to the sensor can be expected. As depicted in Fig. 1 the surface displacement signals were detected at three distinct points, reflecting three different source-sensor angles of θ = 0°, 45° and 90°, respectively, between the crack surface normal (x-axis) and the investigated propagation direction.

The CWD-diagrams in Fig. 5 show the respective simulated signals for the source model

configuration A with additionally calculated dispersion curves. Clearly, the modal composition of the detected Lamb waves depends on the angle. While for the source-sensor angle 0° the S0 mode dominates, the S0 mode contribution decreases for 45° and vanishes for 90°. In contrast, the contribution of the A0 Lamb wave mode is almost constant for all three angles, although barely visible in the color-range of Fig. 5-a. Due to the orientation of the in-plane crack surface displacement, this preferential spatial distribution of the S0 mode intensity is expected [Ea-ton2008].

3.2 Homogeneous, anisotropic medium

The Lamb wave propagation excited by source model configuration B is visible in Fig. 6. Generally, the propagation of Lamb waves in anisotropic media like CFRP is more complex than for isotropic media. Due to the elastic anisotropy, the propagation of distinct Lamb wave modes is strictly asymmetric. Due to the high elastic modulus in x-direction (154 GPa) the propagation velocity of the S0 Lamb wave mode along the x-axis is very high and thus the reflection at the x-edge of the specimen interferes at the observation position (x,y,z) = (50,0,0.7) mm for 35 µs.

149

Fig. 5: Influence of source-sensor angle for source model configuration A: (a) at (x,y,z) = (50,0,0.7) mm, (b) (x,y,z) = (35,35,0.7) mm, (c) (x,y,z) = (0,50,0.7) mm. However, the other observation positions marked in Fig. 6 show no interference with incident reflections for the investigated time 50 µs.

In analogy to the case of isotropic media considered before, a comparison is made between signals excited by point force couples in an anisotropic plate and those of source model configuration B. The resulting CWD-diagrams of the z-displacement signal of the surface at (x,y,z) = (50,0,0.7) mm are shown in Fig. 7-a and 7-b. In addition, dispersion curves of the S0 and A0 Lamb wave modes are shown. Both were calculated for propagation in x-direction based on the anisotropic elastic properties given in Table 1 using the software package of R. Zeyde [Zeyde2010]. As visible in Fig. 7, both signals are clearly dominated by the S0 Lamb wave mode. 3.3 Inhomogeneous, anisotropic medium

After introducing an anisotropic propagation medium and AE sources of finite dimensions in source models A and B, the source model configurations C and D introduce microscopic inho-mogeneous elastic properties. These are used to model the microscopic presence of resin and carbon fibers as shown in Fig. 2-b and 2-c. A comparison of CWD-diagrams of both source model configurations is shown in Fig. 8 for each source-sensor angle investigated. In addition, the dispersion curves of the fundamental Lamb wave modes are shown in each CWD-diagram.

150

These were calculated independently for each source-sensor angle and the respective propagation distance using the software package of Zeyde [Zeyde2010].

Fig. 6: Propagation of several Lamb wave modes in CFRP plate at six distinct times.

Fig. 7: Comparison of surface displacement signals at (x,y,z) = (50,0,0.7) mm for point force couple model (a) and model of finite extent (b) with superimposed dispersion curves.

For source model configuration C (model of matrix cracking), the signals are dominated by the A0 mode for all source-sensor angles (see Fig. 8a-c). Only minor contributions of the S0 Lamb wave modes are observed. These are barely visible in the CWD-diagrams, but are easier to

151

Fig. 8: Influence of source-sensor angle for source model configuration C (a) at (x,y,z) = (50,0,0.7) mm, (b) (x,y,z) = (35,35,0.7) mm, (c) (x,y,z) = (0,50,0.7) mm and (d – f) for source model configuration D, respectively. identify in the time domain at the beginning of the signals, in particular for observation position (x,y,z) = (50,0,0.7).

152

Different CWD-diagrams are observed for source model configuration D (model of fiber breakage) at the respective observation positions. Here a strong contribution of the S0 Lamb wave mode is found for propagation direction parallel to the crack surface normal (see Fig. 8-d). With increasing source-sensor angle, the contribution of the S0 mode decreases, since it is di-rected mostly along the x-axis as seen in Fig. 6. Thus the dominant contribution for (x,y,z) = (35,35,0.7) mm and (x,y,z) = (0,50,0.7) mm originates from the A0 mode.

It is worth pointing out that the same source excitation time was used in the current simula-

tions for both source model configurations. Based on the transversal sound velocity, the source excitation time of fiber breakage is expected to be shorter than for resin fracture. As already pre-sented in [Sause2010] this results in a stronger contribution of the intensity of the S0 mode for the case of fiber breakage and, consequently, emphasizes the characteristic difference between the associated source mechanisms further. 4. Conclusions

For isotropic media the results of conventional point force couples and of the source model of finite extent as introduced in [Sause2010] agree reasonably well. For the source model of fi-nite extent a weak additional contribution of the A0 Lamb wave mode is found, which is attrib-uted to the influence of asymmetry of the used tetrahedral mesh on a microscopic scale. Al-though the asymmetry was introduced accidentally, it is expected to reflect real experimental conditions better than symmetric meshes, since minor deviations from perfect symmetry in the form of voids and flaws will always be present. Due to the fact that the dimensions of these in-homogeneities are within the range of the wavelengths occurring close to the source they will strongly affect the symmetry of the source radiation patterns.

Independent of the chosen mesh, for the investigated source model configurations in anisot-

ropic media a strong influence of the microscopic elastic properties on the modal composition of the excited Lamb waves was found. In this context it was demonstrated, that just the change of local elastic properties from those of resin to those of carbon fiber cause a different excitation ratio of symmetric and antisymmetric Lamb wave modes.

In all specimens a strong dependency of the signal propagation on the orientation between

the axis normal to crack surface displacement and the source-sensor axis was found. For iso-tropic media the orientation of the in-plane source causes a preferential orientation of the S0 mode propagation along the direction of the crack surface normal. A superposition of the calcu-lated fundamental Lamb wave modes shows good agreement to simulated signals.

Since for anisotropic media the dispersive propagation of Lamb waves depends on the direc-

tion, different dispersion curves were calculated for the investigated propagation angles of 0°, 45° and 90°. Similar to the isotropic propagation medium, the S0 mode propagates with preferen-tial orientation along the direction of the crack surface normal. For the current model this direc-tion coincidences with the fiber axes and thus results in a fast propagation of the S0 mode. For the purpose of source identification procedures, this dependency of Lamb wave propagation on the source-sensor angle introduces additional difficulties. For a valid identification of the micro-scopic source mechanism this effect has to be taken into account.

153

For further investigations of the effect of the source microstructure the geometry of the pre-sented source model should be extended. In order to model typical failure mechanisms in CFRP, the influence of the vicinity of the source should be investigated in more detail. In particular, a more realistic crack geometry should be considered. The influence of inhomogeneities, like addi-tional fibers and voids in the surrounding of the source should be investigated. In addition, the influence of fiber-matrix interfacial strength is not taken into account in the current model. In future attempts, this interaction should be treated accordingly.

For signals detected at distances below 50 mm, geometric spreading is the dominating con-

tribution to attenuation. To interpret signals at longer propagation distances additional contribu-tions to Lamb wave attenuation should be taken into account by modeling of dispersive media.

In summary, the presented simulations demonstrate that the microscopic conditions close to the source influence the excitation of distinct Lamb wave modes significantly. For experimental attempts regarding source identification procedures, such influences should be taken into ac-count, since they can dominate the signal characteristics and thus superimpose effects of disper-sive signal propagation. Acknowledgments

We like to thank Professor M. Hamstad for valuable suggestions regarding simulation of acoustic emission. References [Castaings, 2004] M. Castaings, C. Bacon, B. Hosten and M.V. Predoi, “Finite element predic-tions for the dynamic response of thermo-viscoelastic material structures”, Journal of the Acous-tic Society of America, 115 (3), 1125.1133 (2004).

[Choi, 1989] H.-I. Choi and W. Williams, “Improved Time-Frequency Representation of Mul-ticomponent Signals Using Exponential Kernels”, IEEE Transactions on Acoustics, Speech and Signal Processing, 37 (6), 862-872 (1989). [Dietzhausen, 1998] H. Dietzhausen, M. Dong and S. Schmauder, “Numerical simulation of acoustic emission in fiber reinforced polymers”, Computational Materials Science, 13, 23-30 (1998).

[Eaton, 2008] M.J. Eaton, R. Oullin, K.M. Holford and C.A. Featherston, “AE wave propaga-tion and novel source location in composite plates”, 28th European Conference on AE Testing, Krakow, Poland (2008) [Freitag, 1997] L.A. Freitag and C. Ollivier-Gooch, “Tetrahedral mesh improvement using swapping and smoothing”, Numerical Methods in Engineering, 40 (21), 3979-4002 (1997) [Giordano, 1999] M. Giordano, L. Condelli and L. Nicolais, “Acoustic emission wave propaga-tion in a viscoelastic plate”, Composites Science and Technology, 59, 1735-1743 (1999). [Green, 1995] E. R. Green, “Acoustic emission sources in a cross-ply laminated plate”, Compos-ites Engineering, 5, 1453-1469 (1995). [Green, 1998] E.R. Green, “Acoustic Emission in Composite Laminates”, Journal of Nondestruc-tive Evaluation, 17 (3), 117-127 (1998).

154

[Hamstad, 2002] M.A. Hamstad, A. O’Gallagher and J. Gary, “A Wavelet Transform Applied To Acoustic Emission Signals: Part 1: Source Identification”, Journal of Acoustic Emission, 20, 39-61 (2002). [Hamstad, 1999] M.A. Hamstad, A. O’Gallagher and J. Gary, "Modeling of Buried Acoustic Emission Monopole and Dipole Sources With a Finite Element Technique", Journal of Acoustic Emission, 17 (3-4), 97-110 (1999).

[Lamb, 1917] H. Lamb. “On Waves in an Elastic Plate”, Proceedings of the Royal Society of London Series A, Containing Papers of a Mathematical and Physical Character, 93, 114-128 (1917). [Lysak, 1996] M. Lysak. “Development of the theory of acoustic emission by propagating cracks in terms of fracture mechanics”, Engineering Fracture Mechanics, 55 (3), 443-452 (1996). [Nieuwenhuis, 2005] J.H. Nieuwenhuis, J. Neumann, D.W. Greve and I.J. Oppenheim, “Genera-tion and detection of guided waves using PZT wafer transducers”, IEEE Transactions Ultrason-ics, Ferroelectrics and Frequency Control, 52, 2103-2111 (2005).

[Ohtsu, 1984] M. Ohtsu and K. Ono, “A generalized theory of acoustic emission and Green's function in a half space”, Journal of Acoustic Emission, 3, 27-40 (1984).

[Ohtsu, 1986] M. Ohtsu and K. Ono, “The generalized theory and source representation of acoustic emission”, Journal of Acoustic Emission, 5, 124-133 (1986).

[Parthasarathy, 1994] V.N. Parthasarathy, C.M. Graichen and A.F. Hathaway, “A comparison of tetrahedron quality measures”, Finite Elements in Analysis and Design, 15 (3), 255-261 (1994).

[Prosser, 1999] W.H. Prosser, M.A. Hamstad, J. Gary and A. O’Gallagher, “Finite Element and Plate Theory Modeling of Acoustic Emission Waveforms”, Journal of Nondestructive Evalua-tion, 18 (3), 83-90 (1999). [Sause, 2010] M.G.R. Sause and S. Horn, “Simulation of acoustic emission in planar carbon fi-ber reinforced plastic specimens”, Journal of Nondestructive Evaluation, 29 (2), 123-142 (2010). [Sause, 2010b] Markus G.R. Sause and Siegfried Horn, “Influence of Specimen Geometry on Acoustic Emission Signals in Fiber Reinforced Composites: FEM-Simulations and Experi-ments”, Conference Proceedings: 29th European Conference on Acoustic Emission Testing, Vi-enna, Austria (2010). [Vallen, 2010] Vallen Systeme GmbH (Munich, Germany), Aoyama Gakuin University (Tokyo, Japan). AGU-Vallen Wavelet (2010). [Zeyde, 2010] R. Zeyde, “Notes on orthotropic Lamb waves”, Technion - Israel Institute of Technology, http://mercurial.intuxication.org/hg/elasticsim/ (2010).

J. Acoustic Emission, 28 (2010) 155 © 2010 Acoustic Emission Group

ACOUSTIC EMISSION EVENT IDENTIFICATION WITH SIMILAR TRANSFER FUNCTIONS

FRANZ RAUSCHER

Vienna University of Technology, Institute for Engineering Design and Logistics Engineering, Getreidemarkt 9, 1060 Vienna, Austria

Abstract

A special similarity measurement for acoustic emission (AE) events based on sampled sig-

nals from multiple channels was developed. Data from pairs of channels is processed in a way that events with different source functions, but the same source mechanism and transfer func-tions show large correlation. The evaluated X22-correlation can be used for clustering and sub-clustering. Additionally, a double time difference is evaluated, which can be used for double difference location [1] also called relative location [2]. An additional application of this similar-ity measurement is the comparison of measured signals to the ones calculated by simulations.

Keywords: Cross-correlation, clustering, relative location, double difference location

Introduction

Many methods exist for the evaluation of AE data [3], and clustering of events is an impor-

tant task. Techniques applied in seismology [2, 4] are also interesting for the evaluation of AE-data, but one has to consider that dispersive plate waves are important in AE. We consider here a similarity measurement based on sampled signal data, which can be used for clustering, but also for relative location processing, as used in seismology.

Mathematical Model

The model (Fig. 1) consists of a body with a number of AE (or seismic) sensors (SENj)

mounted at fixed positions. Transient AE signals from micro-fracture or other natural sources (Ei) are acquired by the AE equipment with sensors, preamplifiers and filters.

The wave initiation, wave propagation, and the measurement of the surface movement are

mathematically described as follows:

(1)

source function for the event Ei transfer function from the source function si(t) to the sensor SENj transfer function of sensor (SENj), amplifiers, filters and acquisition equipment signal function at sensor SENj from event Ei

* convolution operator.

The model is based on the following assumptions: • Acquired signal data aij can be separated in a way that every aij is caused by one single

event. This means that the time difference between events Ei must be large enough that signals from different events do not overlap.

156

Fig. 1: Basic model for wave transfer. • The signals aij acquired at different sensors SENj caused by one event Ei can be grouped. • The signals aij acquired at different sensors SENj caused by one event Ei have a common

source function . • The relation of the source function to the acquired signals can be described by lin-

ear differential equation and can, therefore, be described by convolution with transfer functions ( ).

Events with similar transfer functions

A pair of events (E1 and E2) and the measured signals from these events at two sensors (SEN1 and SEN2) are investigated (Fig. 2).

We search for a check to find out whether or not the transfer functions to the considered sen-

sors (u1j and u2j) are (almost) the same for the considered two events - The source functions may be different. Such a similarity is expected if the source orientation and mechanism of both events are the same, and the events are located close to one another ( ). To allow for local shifts dx12, differences in the time delays and should not destroy the similarity:

(2)

(3)

The time difference is the difference of the delays in the transfer functions from the two source functions to the first sensor and is the same difference to the second sensor.

In AE measurements only the acquired signals aij, not the source functions, are available for

checking similarity. Using the signals from one sensor only does not allow to “cancel” the source function. Ratios of the acquired signals at one channel to the one at another channel appear to be promising, because in a fraction the parts, which are identical for the numerator and the denomi-nator are cancelled. The problem with fractions is that here instead of division, deconvolution operation have to be used, which is impractical. Therefore, another approach, based on convolu-tions of signal functions, is chosen as follows:

157

Fig. 2: Two similar events.

(4)

In this equation, * is convolution and (t – dt1-2, 1-2). at the right side means that the function resulting from convolution (a12 * a21) is delayed by a time dt1-2, 1-2. Equation 4 is fulfilled if Eq. 2, Eq. 3 and a few simple conditions are fulfilled. To show these conditions, the model for the wave transfer given in Eq. 1 is inserted in Eq. 4:

(5)

(6) Rearranging Eq. 5 leads to Eq. 6, where we can see the following:

• The convolution (s1 * s2) appears on both sides so that the similarity even holds for differ-ent source functions s1 and s2. The important condition is that the signals acquired from one event at different sensors have a common source function.

• The transfer functions of sensor and acquisition equipment arise also on both sides. If the same sensors and equipment are used for both events, sensor and equipment transfer functions of the two channels may be different without destroying similarity in Eq. 4.

Another case, which is not covered by the notation used here, may be useful when measured

signal data has to be compared to simulated ones: If channels 1 and 2 have the same transfer function e1 = e2 (equivalent sensors, filters, and equipments), different transfer functions could be used for event 1 and for event 2. The data a11 and a12 (event 1) could be a measured measurement signal and a21 and a22 (event 2) a calculated surface movement.

Quantification of similarity (X22-correlation)

When applying Eq. 4, the left and the right hand sides are never exactly equal. Therefore, a quantitative measurement of similarity is necessary. In this investigation the largest success was reached by taking the maximum of a normalized cross-correlation function. Therefore, the sam-pled signals aij

k (k indicating sampled signals) are used, and the left and the right sides of Eq. 4 were evaluated:

(7)

(8)

158

The normalized cross-correlation function is evaluated in the following way:

(9)

For the normalisation in Eq. 9, it is assumed that the mean values of the convolution func-

tions c2211k and c1221

k are zero. The set K of integers, for which the sums are evaluated, has to include the beginning of the transient signal. If the whole time window with nonzero signal caused by the corresponding event is acquired and usable, Eqs. 7 to 9 can easily be evaluated in the frequency space. If this is not the case (which was the case for the data used in this paper), for each specific Δk, the set K must be the same for the numerator as for the dominator. For in-complete signals, consistent evaluation of the equations was done in time space. This time con-suming option was chosen here, being aware that further development is necessary to reduce calculation time.

The similarity is now quantified as the maximum of this normalized cross-correlation func-

tion: (10)

(11)

Because of the crosswise convolution of the signals from two events and two channels, this

type of similarity measurement is called X22-correlation. Additionally, the time shift for the maximum the cross-correlation function , which corresponds to the double time difference , is evaluated. This double time difference can be used for double difference location calculations [1], also called relative location [2].

Simple cross-correlation between measured signals

A simpler way of comparing the signals from different events is to build the maximum of the normalized cross-correlation function (Eqs. 9-11) directly for the signals acquired at one channel j (a1j and a2j) instead of applying it on the convolutions c2211

k and c1221k. In this way, one gets a

similarity measurement for each channel (instead of each channel combination) and differential travel times for earthquake relocation are sometimes evaluated [5]. The maxima of these normal-ized cross-correlation functions are called simple correlation and used for comparison in this paper.

Applying the Similarity Measurement on Different Artificial Sources

For testing the introduced similarity measurements, artificial sources with equal direction of

stimulation force were applied on a simple rectangular steel plate (Fig. 3a). Four resonant AE-sensors (Vallen VS150) with a resonant frequency of 150 kHz were attached on the plate and connected to a Vallen AMSY5 equipment. For the first two artificial source types (see Table 1), a VS150 sensor was electrically stimulated by two different signal forms. Additionally, pencil-lead breaks (PLB) of 0.3 mm and hardness 2H were used directly and via a center punch (Fig. 3b).

159

a) b) Fig. 3: a) Arrangement for simple tests; steel plate with thickness of 6 mm; b) Pencil lead break via centre punch.

Table 1: Tested event types. Description

PU1 Piezoelectric transducer VS150 with broadband voltage pulse (Vallen internal pulser) PU2 Piezoelectric trans. VS150 with 110 kHz narrow-band voltage pulse (function genera-

tor - sine-function windowed by a sine-function with 10 oscillations) PL1 0.3-mm pencil-lead break, hardness H2 on surface PLK 0.3-mm pencil-lead break, hardness H2 via a centre punch

First, all the artificial sources were applied at point P1 (Fig. 3a). The signals were acquired

with a sample frequency of 5 MHz, and 2056 samples were used (Fig. 4). Selected X22-correlations and simple cross-correlations are shown in (Table 2). Here the X22-correlations ( ) are given for all sensor combinations. At the left side mean values ( ) are given. In the line below the evaluated double time differences ( corresponding to ) are listed. For comparison simple cross-correlations ( ), are given for each sensor.

In the case of the pulsar, a common source function for each event exists, so perfect X22-correlation is shown for different electrical sources at the same pulsar. Also large X22-correlation was found for the other combinations of source types. The X22-correlation of PU1 vs. PL1 at sensor combination 1-3 shows a double time difference , which is invalid, in-dicating that a check of the evaluated time differences is necessary. The simple cross-correlations ( ) are small in these cases because events with different source functions are compared.

The mean values of these correlations are summarized in Table 3. Also correlations of the

same type of source (repeated experiments) are given. Additionally, one source 70 mm from P1 is included. The simple cross-correlation shows only large values (>0.9) if source location and source type are the same. For different source types or larger distances, the values are between 0.2 and 0.5 (0.59). In the case of X22-correlation, large values are seen also if the source func-tions are different. Sources with larger distances (different transfer function) from one another have X22-correlations from 0.6 to 0.83.

160

Fig. 4: Signals (voltage in [mV] vs. time in [µs]) of tested event types at sensor 2.

Dependence of the Similarity Measurement on the Distance between Sources

For relative location processing based on the evaluated double time differences, it is impor-

tant to know the behavior of the X22-correlation if small distances arise between the sources. Therefore, some PLB were performed in different distances from P1 (direction to sensor 4 – Fig. 3a). In Table 4, the X22-correlations between sensors 2 and 4 were listed, showing high values and appropriate double time differences up to a distance of 5 mm. This corresponds to a double difference in the propagation distances of 10 mm.

Table 2: X22-correlation for different combinations of selected event types at point P1.

Sensor combination k-l 1-2 1-3 1-4 2-3 2-4 3-4

Mean value

Symbol mean

0.999 0.997 0.997 0.9995 0.9993 0.9992 0.999

[µs] 0 0 0 0.2 0.2 0

PU1 vs.

PU2 SEN1: 0.30; SEN2: 0.24; SEN3: 0.13; SEN4: 0.18 0.21

0.98 0.94 0.97 0.82 0.86 0.97 0.92

[µs] 0 -6.4 0.6 0.8 0.8 -0.2

PU1 vs.

PL1 SEN1: 0.74; SEN2: 0.65; SEN3: 0.61; SEN4: 0.68 0.67

0.98 0.97 0.95 0.95 0.98 0.97 0.97

[µs] -0.2 -0.4 -0.4 -0.6 -0.6 0.2

PL1 vs.

PLK SEN1: 0.64; SEN2: 0.60; SEN3: 0.56; SEN4: 0.54 0.59

161

Table 3: Summery of mean values X22-correlations and simple cross-correlations for different event types at point P1 and 70 mm from this point.

PU1 PU2 PL1 PLK PL1 – 70mm

PU1

0.999 0.21

0.92 0.67

0.95 0.44

0.64 0.34

PU2

0.99 0.52

0.995 0.51

0.83 0.43

PL1

0.99; 0.97 0.88; 0.96

0.97 0.59

0.58 0.45

PLK

0.995; 0.998 0.96; 0.992

0.69 0.38

PL1 – 70 mm

0.98 0.95

Table 4: X22-correlation for OP1 to different event types at different distances.

2 mm 5 mm 10 mm 15 mm 0.99 0.92 0.76 0.73 Window 1

[µs] -2.0 -4.2 35 -5.2

Applying Relative Location with PLB (double difference location)

Pencil lead breaks were performed on the plate to form the letters “AE” (Figs. 3a and 5a).

For the measured signals, X22-corrrelations and the corresponding double time differences were evaluated. Based on this data, a simple least-square fitting algorithm, with the center of the clus-ter as input value, was used for the evaluation of relative location. When using the signals from one event at each point no recognizable pattern was received. Extending the set by a second event at each point leads to a recognizable pattern (Fig. 5b).

a) b) Fig. 5: Relative location of pencil lead breaks; a) pattern of pencil lead breaks; b) results from relative location processing

162

Conclusions A similarity measurement called X22-correlation, which is able to group source according to

there transfer function, was developed. With artificial sources, it could be shown that events at the same location and with the same direction of stimulation force, but with different source functions, show large X22-correlation. This X22-correlation can be used for clustering and sub-clustering. The evaluated double time differences were used for double difference (relative) loca-tion processing.

At present, it is not known which type of source mechanism show this type of similarity. For

some applications, development is necessary to increase the maximum local distance between events for recognizing similarity. When large numbers of events have to be processed, develop-ment for decreasing computer load is necessary.

References

1. Waldhauser F., Ellsworth W. L.: A Double-Difference Earthquake Location Algorithm:

Method and Application to the Northern Hayward Fault, California. Bulletin of the Seismol-ogical Society of America, 90 (6), (2000) 1353–1368.

2. Young R. P., Collins D. S., Reyes-Montes J.M., Baker C.: Quantification and interpretation of seismicity. Int. J. of Rock Mechanics & Mining Sciences 41 (2004) 1317-1327.

3. Ono K.: Structural Integrity Evaluation Using Acoustic Emission. J. Acoustic Emission, 25, (2007) 1-20.

4. Brückl E., Mertl S.: Seismic Monitoring of Deep-Seated Mass Movements. Vortrag: Inter-praevent, Niigata, Japan; 25.09.2006 - 29.09.2006; in: Disaster Mitigation of Debris Flows, Slope Failures and Landslides, Universal Academic Press, Inc. Tokyo, Japan, (2006), S. 571 - 580.

5. Menke W.: Using waveform similarity to constrain earthquake locations. Bulletin of the Seis-mological Society of America, 89 (6), (1999) 1143–1146.

J. Acoustic Emission, 28 (2010) 163 © 2010 Acoustic Emission Group

ANALYSIS OF FRACTURE RESISTANCE OF TOOL STEELS BY MEANS OF ACOUSTIC EMISSION

EVA MARTINEZ-GONZALEZ

1, INGRID PICAS 2, DANIEL CASELLAS

2,3 and JORDI ROMEU

1

1 Laboratory of Mechanical and Acoustical Engineering (LEAM), Universitat Politècnica de Catalunya (UPC), EUETIB c/ Urgell 187, 08036 Barcelona, Spain; 2 Department of Material Technology, CTM-Technological Centre, Av. Bases de Manresa 1, 08242 Manresa, Spain;

3 Department of Materials Science and Engineering Metallurgy, Universitata Poltècnica de Cata-lunya (UPC), EPSEM, Avda Bases de Manresa 61, 08242 Manresa, Spain

Abstract

The automotive manufacturers currently use advanced high-strength steels (AHSS) to pro-

duce lightweight vehicles with increased safety performance. While showing excellent strength-to-weight ratios, AHSS have several limitations due to the high loads required in cold forming and cutting processes, which lead to accelerated wear and premature fracture of tools. Thus, new tool materials with improved mechanical behavior ought to be developed with regard to the tool failure mechanisms. The aim of this work is to shed light on the fracture mechanisms acting in tools (i.e. crack nucleation and propagation) applying the acoustic emission (AE) technique. Bending tests using two different tool steels were monitored in order to establish a relationship between AE signals and fracture events. Keywords: Carbides, microstructure, advanced high-strength steels, AHSS Introduction

The utilization of advanced high-strength steels (AHSS) in structural automotive components

has been broadened in the past few years to satisfy the stronger regulation towards low CO2 emissions and high safety performance of vehicles. The high fracture strength of AHSS (from 600 to around 1500 MPa) enables to manufacture parts with high crash-resistance and light-weight. Conversely, the high yield stress of AHSS is a main inconvenience during cold forming and cutting processes, since the high loads, which are required cause accelerated wear and pre-mature fracture of tools [1]. Thus, the development of new tool materials combining high wear resistance and fracture toughness is required to fully exploit the potential of AHSS. In this framework, proper tool-steel microstructural design is needed, taking the influence of the micro constituents on the fracture event into account. The size, shape and distribution of the primary alloy carbides embedded in the matrix, as well as their chemical composition and mechanical properties, are being studied to optimize the mechanical response of tool steels [2].

A detailed examination of conventional AHSS forming tools (i.e. tools employed in cutting,

drawing, stamping, bending and profile rolling) has permitted to identify that fatigue is one of the most common mechanisms involved in tools failure. Fatigue initiation sites are usually re-lated to the presence of broken primary carbides, from which cracks start to propagate under the applied stress fields [3, 4]. Thus, an increase of carbides mechanical properties is expected to raise fracture and fatigue resistances of tool steels. This is why Casellas et al. [2] used nano-indentation techniques to determine the mechanical properties (i.e., hardness, Young’s modulus and fracture toughness) of primary carbides embedded in several tool steels. However, as shown by Picas et al. [3], the fracture strength of carbides not only depends on their mechanical proper-

164

ties, but also on their shape, arrangement and presence of internal defects. Accurate data on the primary carbides fracture strength in tool steels will bring valuable knowledge to light, concern-ing the global mechanical response of tools.

The obtained results were promising, showing the applicability of AE to the characterization

of fracture mechanisms in tool steels. It is well known that AE can be used as a non-destructive method for testing material degradation. AE is based on real-time detection of elastic waves, generated in a volume of stressed material as a result of the micro energy emissions induced in different internal dynamic micro mechanisms (i.e. plastic deformation, movements in edge dislo-cations, sliding, grain rotation, micro fractures of inclusions, appearance or propagation of crack-ing, etc.). The first work showing the application of AE in tool steels was carried out by Fukaura and Ono [5]. They used AE to determine the stress level at which carbides start to break. Fukaura and Ono showed the relationship between the characteristic AE wave parameters and material damage evolution. However, as these authors indicated, the use of relatively low-frequency sen-sors (375 kHz) might have had introduced some limitations that could have been avoided by the application of higher resonance frequency detectors. That is why in the present work, 700 kHz sensors are employed. Yamada and Wakayama [6] related the signal types registered in a bend-ing test to their predominant frequency, and correlated continuous low-frequency signals to plas-tic deformation events, while sudden high-frequency signals to fracture events.

Aimed at better understanding the role of the microstructural constituents (primary carbides

and metallic matrix) in the fracture mechanisms of tool steels, this work is focused on the appli-cation of AE in monotonic bending tests to determine the stress levels, at which carbides start to break. Experimental procedure A. Specimens

Two commercially available tool steels were selected: DIN 1.2379 (equivalent to AISI D2) and a tool steel named as UNIVERSAL (developed by ROVALMA S.A.). The chemical compo-sition of both can be found in [2]. The DIN 1.2379 tool steel shows a ledeburitic microstructure with primary carbides formed during solidification by the eutectic reaction ⇒ γ + M7C3 [7]. In UNIVERSAL steel two types of primary carbides are present, one type is M7C3 but shows higher vanadium and tungsten content than those found in DIN 1.2379, and the second type is vana-dium-rich MC and shows increased hardness and toughness values compared to the other types. Properties of these carbides can be found in reference 2. Carbides in DIN 1.2379 have elongated shapes and are dispersed in the matrix forming bands along the forging direction. In the mean-while, primary carbides of UNIVERSAL steel are more equiaxed and homogeneously distributed in the matrix [3].

Prismatic specimens with dimensions 50 x 8 x 6 mm were extracted from forged billets, with

the longitudinal axis parallel to the forging direction. The obtained samples were heat treated by quenching in oil and tempering up to a hardness of 60 - 61 HRC (heat treatment schedule can be found in [2]). B. Monitored AE-bending test.

A three-point bending test was used to evaluate the mechanical behavior of specimens. The surface exposed to tensile stress was ground and polished, and the edges were rounded in order

165

to avoid stress concentration effects. Tests were carried out in a universal testing machine in air at room temperature and the load rate was fixed at 1 mm/s.

Two small magnetic AE sensors were attached at both ends of the specimens in order to de-

tect the AE signals. The sensors had a fixed resonance frequency of 700 kHz (VS700D, Vallen System Gmbh). Two pre-amplifiers with a 34-dB gain of the same brand were used (AEP4). Signals were recorded and analyzed using the Vallen Systeme Gmbh AMSY5 analyzer. During the measurements, digital filters of 95-850 kHz were applied.

Fig. 1: Diagram of the experimental setup.

C. Test procedure.

For each type of tool steel, several bending tests were first carried out to the breakage of the specimen in order to define the typical AE signal pattern in the fully run test. Later, the tests were paused upon significant changes of the AE signals. However, a maximum load of 70 % of the material fracture strength was applied so as to prevent the sensors from being damaged as a consequence of the sudden specimen break. DIN 1.2379 and UNIVERSAL characteristic frac-ture strength values were determined by Picas et al. and can be found in [3]. At each halt, the surface subjected to tensile loading was examined by means of optical and confocal microscopy (OM and CM) to identify the potential sources in the microstructure responsible for the detected changes in AE signals. Results and Discussion

The results obtained in the fully run tests are displayed in Figs. 2 and 3 for DIN 1.2379 and UNIVERSAL respectively. They show the correlation between the applied stress and the AE signal intensity as a function of time. Both DIN 1.2379 and UNIVERSAL steels show a similar pattern with regard to the captured AE signals. Three different zones can be distinguished: First zone

During the first stage of the test, AE signals are almost absent. The reason is that the material is under elastic deformation and neither the carbides nor the matrix show fracture or plastic de-formation. In other words, once the noise is filtered out, the absence of AE signals can be associ-ated with material linear behavior. Nonetheless, the detection of a few isolated AE signals could presumably be caused by the fracture of some carbides or a material chip off due to polishing; even though no certainty can be given at this point.

166

Fig. 2: Bending test results for steel DIN 1.239. Blue line: stress vs. time; red points: AE signal amplitude vs. time.

Fig. 3: Bending test results for steel UNIVERSAL. Blue line: stress vs. time; red points: AE sig-nal amplitude vs. time. Second zone

In this zone appear the first AE signals. In the DIN 1.2379 they start at approximately 11 s and in the UNIVERSAL at 27 s, which correspond to a stress level of 640 MPa and 1700 MPa, respectively. The signals increase gradually in intensity and abruptly in number. During this stage, several broken carbides can be discerned in the steel microstructures. As shown in Figs. 4 and 5 for DIN 1.2379 and UNIVERSAL, respectively, the amount of broken carbides increases considerably with the rising stress. In some cases a small area of plastic deformation appears.

Third zone

The amounts of detected signals, as well as the cumulative energy, considerably increase. The intensity does not increase but, in contrast, the average value shows a slight decrease. At these stress levels, a high number of carbides are found to be fractured, the plastically deformed areas have spread and deepened and even small cracks are observed in the matrix, as shown in Fig. 6. At this point, the test has been stopped to prevent the sensors from damage due to the sudden specimen breakdown. As shown by Picas et al. [3], the number of broken carbides and matrix cracks is expected to keep growing up to coalesce the ones with each others, leading to the final fracture.

167

Fig. 4: Example of broken carbides observed by means of CM in the tensile stressed surface of DIN 1.2379 specimens.

Fig. 5: Example of broken carbides observed by means of CM in the tensile stressed surface of UNIVERSAL specimens.

In this study, the stress level at which carbides are found to start breaking in DIN 1.2379 is 640 MPa. This is around 20 % of the fracture strength determined for this material [3]. In the case of UNIVERSAL, the signals that point out breakage of carbides appears from 1700 MPa, which corresponds to approximately 45 % of the fracture strength of this material [3]. Such re-sults can be explained by the lower fracture toughness of primary carbides in DIN 1.2379, as has been previously evaluated by Casellas et al. [2], and the elongated shape and arrangement in

168

bands (in which local stresses can be higher) of carbides in DIN 1.2379, as has been stated by Picas et al [3].  

               

Fig. 6: Broken carbides, plastically deformed areas and small cracks in the matrix can be identi-fied by means of CM in the tensile stressed surface of specimens: a) DIN 1.2379 at 2100 MPa and b) UNIVERSAL at 2800 MPa.

This investigation is currently run in order to study other AE parameters that could bring valuable information regarding the physical phenomena involved in the fracture event. The analysis of waveforms to identify the sources, using their frequencies, is of special interest to develop an efficient diagnostic system. Some of the results presented here show agreement with those obtained by Yamada and Wakayama with Ti(C,N)-based cermets [6]. Basically, two types of AE signals can be identified: a high-frequency burst-type signal and a lower-frequency con-tinuous type. The first type could be attributed to the carbide micro-cracking processes while the continuous type could be related to plastic deformation in the matrix. Although this theory is consistent with the results obtained in this study, additional research is required to ascertain these correlations. Conclusions

According to the experimental results obtained by means of coupling AE to bending tests, and the microstructural observation of two different tool steels, the following conclusions can be drawn: 1. AE is an appropriate experimental tool to detect carbide cracking in tool steels microstructure. 2. An experimental relationship has been established between the fracture of carbides and the

obtained AE signal. The stress levels, at which carbides were broken, were determined by AE and experimentally confirmed by microstructural inspection.

3. The tool steel with the highest broken carbide content induces the highest quantity of AE sig-

nals with more temporal dispersion. Acknowledgments

This work has been supported partially by Department of Education and Science (Spanish Government) by means of the program of helps to Projects of Scientific Investigation and Tech-

169

nological Development, 2007-2010, project DPI2007-66688-C02-01, and by the ACC10 under the grant CT09-1-0006. References 1. Advanced High Strength Steel (AHSS) Application Guidelines Version 3. International Iron

& Steel Institute. Committee On Automotive Applications, 2006.

2. Casellas, D., Caro, J., Molas, S., Valls, I., Prado, J.M., “Fracture toughness of carbides in tool steels evaluated by nanoindentation”, Acta Materialia, 55, 4277, 2007.

3. Picas, I., Cuadrado, N., Casellas, D., Goez, A., Llanes, L., “Microestructural effects on the fatigue crack nucleation in cold work tool steels”, Procedia Engineering, 2, 1777, 2010.

4. Berns, H., Broeckmann, C., “Fracture of Hot Formed Ledeburitic Tool Steels”, Engineering Fracture Mechanics, 58, 311, 1997.

5. Fukaura, K., Ono, K., “Acoustic emission analysis of carbide cracking in tool steels”, J. Acoustic Emission, 19, 91, 2001.

6. Yamada, K., Wakayama, S., “AE monitoring of microdamage during flexural fracture of cer-mets”, Proceedings EURO PM2009- Hardmetals & Cermets, 2009, pp. 247-252.

7. Metals Handbook – Metallography, Structures and Phase Diagrams, ASM International, Vol. 8, p. 402, 1978.

J. Acoustic Emission, 28 (2010) 170 © 2010 Acoustic Emission Group

COMPARISON OF ACOUSTIC EMISSION SIGNAL AND X-RAY DIF-FRACTION AT INITIAL STAGES OF FATIGUE DAMAGE

FRANTISEK VLASIC, PAVEL MAZAL and FILIP HORT

Brno University of Technology, FME, Technicka 2, CZ 616 69 Brno, Czech Republic

Abstract

This contribution treats the evaluation of degradation processes appearing during the cyclic loading of the samples from Al alloys EN AW-2017/T4 and EN AW-6082/T6. Two NDT meth-ods have been applied in this case, namely the acoustic emission (AE) and the X-ray diffraction used to detect the microstructural changes in the material during the fatigue damages as well as to report the microcracks appearing afterwards. These measurements have been completed with reports of changes to the loading (resonance) frequency coming from fatigue loading equipment. Moreover, the article describes the records of AE signal coming from a newly developed AE analyser. The main advantage of this analyser (compared with current AE measuring systems) is continuous sampling and full recording of the entire AE signals. The AE detects the changes in the material during the damage accumulation. It is a problem to identify such changes. The struc-ture monitoring by means of X-ray diffraction assists in identifying the AE sources. The results imply a correlation between the AE signal changes and the changes of so-called mosaic blocks of crystal structure detected by means of X-ray diffraction.

Keywords: Fatigue; aluminium alloys; fatigue crack; damage; X-ray diffraction Introduction

The fatigue damage is divided into several stages. It is the dislocation structure that develops changes by cyclic hardening/softening at the beginning of the cyclic loading. During the next stage, no material changes appear visually, but new processes proceed by continuous changes inside the structure. This stage is usually designated as the damage accumulation. At the end of this stage, the damage becomes localized at predisposed sites. Microcracks appear and some of them start to join neighboring ones producing short cracks and finally the main crack forms.

In fatigue damage research, the biggest attention is given to the description of crack initia-

tion. It is, however, quite important to follow the material property evolution immediately pre-ceding the appearance of the fatigue crack. The material property changes take place on the mi-crostructural level and the current investigation methods identify them only with great difficulty.

Cyclic property degradation process of aluminum alloys is quite different from that of iron al-

loys. The main difference is in identifying fatigue threshold. For example, fatigue damage of AlMg alloys proceeds even at low loading amplitudes, while fatigue threshold of ferrous materi-als can typically be defined. That is why a conventional fatigue limit is defined for ferrous mate-rials, which is on the level of 108 loading cycles at least [1].

An Al alloy EN AW-2017/T4 has been used as an experimental material as test specimens.

Similar to the materials tested previously (i.e., EN AW-6082, EN AW-7075), it has structural anisotropy due to the extruding technology [2, 3] (see Fig. 1). Chemical composition of the al-loys is shown in Table 1.

171

Fig. 1 Structural inhomogeneity (orientation effects) of EN AW-2017/T4 alloy.

Table 1 Chemical composition of used alloys EN AW-2017 and 6082 (wt %). material

element Si Fe Cu Mn Mg Cr Zn Ti Al

EN AW-2017/T4 0.8 0.7 2.5 0.2 0.25 0.1 0.25 - rest

EN AW-6082/T6 1.01 0.17 0.067 0.66 0.84 0.16 0.030 0.032 rest

The test specimens for fatigue tests are shown with their geometry and dimensions in Fig. 2. These have directions LS, LT (in the extrusion direction) and TS and TL (transversal directions), see Fig. 1. Shallow notches are used for localization of the crack starting point.

Fig. 2 Specimen geometry for the fatigue tests.

The fatigue tests have been conducted using an electro-resonance RUMUL Cracktronic

8204/160 machine working on the principle of electromagnetic resonance and loading the speci-men with four point bending. The resonance frequency of the specimens was 70~90 Hz with the (bending) tensile stress of 210 MPa (strength limit = 340 Mpa).

Two piezoelectric AE sensors made by DAKEL (type MIDI, see Fig. 3 on right) have been

used for AE detection during fatigue cycling process. One has been placed near the notch by means of mechanical fixing lug and the other one has been glued to the specimen front. The sig-nal coming from AE sensor is amplified by a preamplifier and sent to the measuring systems DAKEL-XEDO and DAKEL-IPL. Methods and Test Operating Conditions

The test procedures have been divided into 3 phases. The first one concerned the test speci-mens in all directions up to the fracture without AE measurements. These specimens served to

172

monitor the changes in structure (as produced by cyclic loading) using X-ray diffraction. The second phase detected the AE signals throughout the whole fatigue test to the fracture of the specimen. The system DAKEL-XEDO has been used for measurements. The main aim here was to measure the AE signal activity during the cyclic loading when the microstructural changes and the microcracks occur. The last phase has again measured the AE signals; however, this time we used a measuring system DAKEL-IPL with fully continuous signal recording. All measurements have been completed with noticed resonance frequency changes of the tested sample coming from the fatigue load apparatus RUMUL Cracktronic.

The microstructural changes were monitored by means of X-ray diffraction at an interval of

5000 cycles for EN AW-2017 and 10,000 cycles for EN AW-6082 between each measurement [4, 5].

Fig. 3 Basic parts of the Cracktronic 8204/160 machine.

Overview of Experimental Results

Figures 4 and 5 show examples of AE activity entries, completed with the diagram of loading frequencies for both materials. Particular entries show the cyclic softening (or hardening) stage, rise and development of the first microcracks and the course of main crack propagation. The plot of the loading frequency (see Figs. 4a, 5a) indicates some changes in the period of damage ac-cumulation but it is not very conclusive.

The creation and propagation of fatigue crack is registered unambiguously. The higher AE

activity confirms this result (see Fig. 4b, for example, between 22nd and 32nd minute). It is, how-ever, evident that the AE signal changes also in the period between these basic phases. The changes of AE signal before the final period gives the chance to see the beginning rise of the first microcracks and other changes of microstructure. The record of AE signal indicates well in Fig. 4b the particularly period before the course of main crack propagation (35th to 45th minute). The period of main crack propagation we can see well in Fig. 5b (from 55th minute).

The change of AE activity is also observed before crack initiation (for example in Fig. 4b at 6

and 18 minute of loading). Classical techniques of optical observation of surface do not provide any obvious reason for this phenomenon. On the basis of analogy with other observations, it is possible to expect that the reason of AE activity can be for example the accumulation of energy sufficient for loosening of dislocations captured in structural obstacles. The reason of increased AE activity during the so-called stable mode can be also the changes of material substructure –

173

emphasizing of sub-seeds, mutual rotation of suitable areas in frame of individual seeds of mate-rial, etc. can appear [4, 5].

a)

b) Fig. 4 a) Changes of loading frequency during a fatigue test of EN AW-6082/T6 alloy (symmet-ric cycle, σ a = 210 MPa, Nf = 315,500 cycles), b) AE counts rate (in log scale – multi-color), RMS (blue), cumulative AE events (black) during a fatigue loading test of the same sample.

The changes of basic parameters of AE events (i.e., peak amplitude and rise time) during the fatigue test of aluminum alloy EN AW-2017/T6 are shown in Fig. 6.

Results of AE Signal Measured by DAKEL – IPL Analyzer

The AE signal activity has been measured during the fatigue tests of the Al alloys with ana-lyser DAKEL-IPL. The same sensors used for DAKEL-XEDO measuring system are used. The examples of results of AE signal measurement in the LS direction during the fatigue test under

174

a)

b) Fig. 5 a) Plot of loading frequency changes, b) AE counts rate, RMS, cumulative AE events for alloy EN AW-2017/T6 (symmetric cycle, σa = 210 MPa, Nf = 442,000 cycles).

the bending tension 210 MPa are plotted in Fig. 7. Here, time changes of AE signal frequency in the interval 75-400 kHz (Fig. 7b) and 460-750 kHz (Fig. 7c) are compared – these are horizontal cuts by the 3D plot from Fig. 7a. From the record of lower frequencies (Fig. 7b), it is possible to identify creation and propagation of fatigue crack.

Changes of structure in the period before the creation of this crack are, however, not well visible. On the record of higher frequencies in Fig. 7c, we can see an evident change of the char-acter of AE signal source in the period between 6th and 15th minute of loading. We suppose that in this period the above-mentioned changes of microstructure occur. It is difficult to reliably identify the essence of these changes from AE signal and therefore it is necessary to use addi-tional NDT procedures (at least at this stage of research).

175

Fig. 6 Changes of basic parameters of AE events (peak amplitude and rise time) during the fa-tigue test of EN AW-2017/T6 (symmetric, σa = 210 MPa, Nf = 442,000 cycles).

Fig. 7 Records of the amplitude changes on selected frequencies in time of Al alloy EN AW-6082/T6 (horizontal cut of the 3D spectral map), b) 75 – 400 kHz, c) 460 – 750 kHz.

176

Figure 8 shows the changes of frequency spectra (left) and the scan of frequency domain at 5 test times chosen. From the sensor on the sample notch (see Fig. 8 right), it is possible to see the frequency maxima moved in broader dissipation area, namely from 130 up to 400 kHz. We can also notice the signal power exists in the higher frequency domains (650 kHz) mostly toward the test end.

Fig. 8 Continuously sampled AE signal obtained by new analyzer DAKEL-IPL (AE sensor MIDI – at the notch (channel 1, 10 dB).

Overview of X-ray Diffraction Analysis

X-ray diffraction analysis of loaded sample structure (Fig. 9) is based on the knowledge that in case of loading of wrought materials the redistribution of deformation strengthening occurs, which is caused by re-arrangement of dislocations.

By X-ray examination of the aluminum alloy EN AW-6082/T6 (results of EN AW-2017/T6

are not at disposition at this paper deadline), we have found that its microstructure, as character-ized by the proportion R of the large coherently scattering regions (CSR´s) (i.e., mosaic blocks or

a) b)

Fig. 9 X-ray diffraction analysis: a) test specimen in the initial state - azimuth profile of the (200) Al diffraction line, b) results of the same test specimen after 60,000 loading cycles [4].

177

structural cells) greater than 10 µm, does change much under repeated stress [4]. In the first stage of cyclic loading, the CSR´s due to the introduced plastic strain disintegrate. As a result, the boundary surface of the cell aggregate extends, and, consequently, its energy finally increases enough to initiate its coarsening. The energy necessary to activate the growth of the larger blocks at the expense of the smaller ones is supplied by further cycling. During this growth, paracrystal-line distortions emerge, which gradually accumulate in the interior of the widening cells, raising their volume energy. After next fatigue loading, the energy increases to such degree that the blocks begin to disintegrate anew. Due to this disintegration, the paracrystalline distortions relax and the alloy energy drops [7, 8].

Fig. 10 Record of real change of x-ray K factor vs. time of fatigue loading for EN AW-6082/T6 alloy [6]

In a cyclically loaded Al alloy we found correlation between the increased AE activity and

the growth of CSR’s (so-called mosaic blocks of the crystal structure) independently reflecting X-rays (see Fig. 12) [7, 8]. Cyclic deformation, unlike unidirectional deformation, does not spread uniformly through a grain but tends instead to concentrate cycle after cycle in preferred zones. These dislocation sinks, which eventually turn into microcracks, reduce the grain into a disoriented cell structure. This is readily proved by the so-called „diffraction imaging“ or „grain by grain“ X-ray diffraction technique [7, 8]. The interior of a cell, being relatively free of dislo-cations, scatters the x-rays coherently giving rise to a diffraction spot.

From Fig. 10 it is seen that the cellular structure changes during the cyclic loading considera-

bly. Figure 10 also shows the records of the X-ray “K coefficient” changes, which were meas-ured during fatigue tests of four samples made from alloy EN AW-6082 (with various orientation relative to forming direction) with same level of cyclic loading.

This K coefficient is the width of the peak of azimuth profile of diffraction line (200) of the

alloy matrix at the half height of this peak; this coefficient represents the amount of disorienta-tion of mosaic block structure. From Fig. 10 are evident the cyclical changes of mosaic blocks, which appear at all the mentioned samples – on the given level of loading approximately in the area 60 and 110~120 × 103 loading cycles. These changes of microstructure can be the cause for local amplification of AE signal in the stage of damage accumulation.

178

Conclusion

The AE signal measurement during the fatigue loading of Al alloys and monitoring the struc-tural changes in the material by means of the X-ray diffraction topography have been conducted. These show the application of these non-destructive methods on research of the fatigue processes in the material greatly contribute to basic knowledge. The AE methods identified the changes in the material during the fatigue load, namely the process of accumulation of damages and the start of microcrack propagation. Detailed information of AE parameters (rise time, maximum ampli-tude, etc.) was obtained in this research. Unfortunately, the correlation between these parameters cannot yet describe the changes occurring inside the material or process of damage accumula-tion. X-ray diffraction measurements of the evolution in tested material structure under cyclic load were made, showing changes (growth and shrinkage) in the disorientation of mosaic blocks during the cyclic loading. The resonance frequency observances acquired by loading instrument Cracktonic appeared also to be useful parameter on the fatigue degradation of the material.

AE brings advantageous information on macroscopic volumes of structurally heterogeneous

material under cyclic loading. Combining the observations of AE and X-ray diffraction enables to extend our knowledge of fatigue processes.

Works presented in this paper have been carried out in the framework of the project of MSMT CR nr. 1M2560471601 „Ekocenter of applied research of nonferrous metals“ - work-place Brno University of Technology, FME. References [1] Michna S. et al.: Aluminium materials and technologies from A to Z, Adin s.r.o., Presov 2007, Slovakia, ISBN 987-80-89244-18-8.

[2] Cerny I., Ocenasek V., Hnilica F.: Problems of fatigue crack growth in strongly anisotropic Al-alloys, Materials Science Forum, 251-252, 2003, 61-72.

[3] Garratt M.D., Bray G.H., Koss D.A.: Influence of texture on fatigue crack growth behavior. In: Proc. of Materials Solution Conf., Indianapolis, ASM International, 2001, pp. 151-159.

[4] Fiala J., Mazal P., Kolega M.: Cycle induced microstructural changes. In: Int. Conf. NDE for Safety, 2007, Prague, CNDT, pp. 73-80, ISBN 978-80214-3506-3.

[5] Mazal P., Pazdera L., Fiala J.: Contribution to identification of cyclic damage development of AlMg alloy. In: NDE for Safety, 2007, Prague, CNDT, pp. 169-174, ISBN 978-80214-3506-3.

[6] Mazal P., Vlasic F.: Applications of selected NDT procedures for more detailed identification of crack initiation stage at fatigue tests of aluminium alloys. In Fatigue design 2009. Senlis, France, CETIM Senlis. 2009. pp. 68 - 75. ISBN 978-2-85400-908-8.

[7] Fewster P.J., Andrew N.L.: Reciprocal space mapping and ultrahigh resolution diffraction of polycrystalline materials. In: Defect and microstructure analysis by diffraction, Snyder R.L., Fiala J., Bunge H.J. (eds.), pp. 346-364, Oxford University Press, New York, 1999.

[8] Fiala J., Kolega M.: Application of two-dimensional detectors in x-ray diffraction materials structure analysis, Particle and Particle Systems Characterization 22, 2005, 397-400.

[9] Mathis K., Chmelik F., TrojanovA Z., Lukac P., Lendvai J.: Investigation of some magne-sium alloys by use of the acoustic emission technique, Materials Science and Engineering A387-389 (2004), 331–335.

J. Acoustic Emission, 28 (2010) 179 © 2010 Acoustic Emission Group

AE SIGNALS DURING LASER CUTTING OF DIFFERENT STEEL SHEET THICKNESSES

TOMAŽ KEK and JANEZ GRUM

University of Ljubljana, Faculty of Mechanical Engineering, Askerceva 6, 1000 Ljubljana, Slovenia

Abstract

Laser cutting is an established industrial method for cutting of various steels in the produc-tion process. However, each laser cut poses a question of quality estimation during and immedi-ately after the laser cutting process. The capturing and evaluating of acoustic emission (AE) sig-nals reveals a great potential for the laser cut quality estimation. This paper presents the results of the AE bursts measurement after the laser cutting of steel sheets of various thicknesses using oxygen or nitrogen as cutting gas. The method of the laser cut quality estimation described here was applied to the austenitic stainless steel, mild steel, and structural steel. The laser cut quality, which is related to the size of the dross and the waviness of the cut surface can be successfully forecast based on the captured AE signals. Keywords: Burst acoustic emission, laser cutting, PZT sensor, dross, unalloyed steel, austenitic stainless steel Introduction

The price of the equipment and operation costs are considerable when it comes to laser sys-tems, but the higher cost is well justified by exceedingly faster treatment processes as well as good dimensional accuracy and repeatability. Laser cutting quality depends mainly on the selec-tion of the process parameters that were established with trial and error method. Changes in the beam quality, gas flow and surface of materials can lead to reduced laser cut quality. In order to eliminate these weaknesses, the monitoring and adequate adjustments of the treatment need to be ensured.

Laser cut quality is defined primarily by the condition of the laser cut surface and the pres-

ence of the re-solidified material occurring as dross at the lower cut edge. The condition of the laser cut surface is defined to a great extent by the occurrence of striations. There are a number of interpretations of mechanisms of striation formation. The three more important models are presented below.

(1) Schuöcker attributed the periodic striation formation to a pulsation of the melt layer on cutting front. The fluctuations in the laser beam energy absorption and reactive gas flow causes thickness and temperature oscillations of the melt layer (Schuöcker 1987). The fluctuations in the laser beam energy absorption can be influenced by the periodic nature of the laser beam or the changes in the material’s absorptivity. The changes in the heat generated by the oxidation reac-tion can be the result of an oscillation in the cutting gas flow due to the turbulences in the laser cut.

(2) The second model describing striation formations on the laser cut surfaces was introduced by Arata, Miyamoto and Maruo and is generally more established. They defined the laser cutting mechanism as ignition and extinction of oxidation reaction (Arata et al. 1979). The model is convincing in explaining the formation of striations on the laser cut surface. However, certain researchers consider it as a quality model with lacking mathematical definition (Di Pietro and

180

Yao 1995). Despite this incomplete definition of the cutting process, it nevertheless offers the most widely adopted definition of striation formation occurring on the laser cut surfaces during the cutting processes performed on structural steels.

(3) Arata´s model was upgraded by Iverson and Powell by describing more accurately the cyclic laser cutting oxidation processes (Iverson et al. 1994). The speed of the oxidation reaction depends on oxygen diffusion in the cutting front, which changes with time. At the beginning the oxidation velocity is high, but it rapidly reduces as the thickness of the oxide layer increases. This means that the oxide layer thickens intensely in the initial phase and then gradually slower. If the oxide layer is sufficiently thick, the oxygen flow can blow it off the cutting front surface and a new oxidation cycle will begin. The most important contribution by Iverson and Powell is an explanation of the termination of the oxidation process during laser cutting.

The waviness of the laser cut surface depends on the laser cutting parameters. Striving for a

more cost-efficient laser cutting by accelerating the cutting speed increases the waviness of the laser cut surface and even causes deterioration in striation pattern, mainly at the lower cut edge. This in turn reduces laser cut quality. Accelerating the cutting speed also leads to an accumula-tion of the re-solidified material at the lower cut edge. A lower energy input at the interaction zone caused by increased cutting speed results in lower temperature and increased viscosity of the melt. The portion of the oxidized melt also decreases. When the melt flows out from the cut-ting front, droplets form at the lower cut edge. When the forces due to gas flow cannot exceed the adhesion forces of the melt to the sheet surface, the melt’s leftover solidifies in the form of droplets at the lower cut edge, i.e. dross will form. The solidification of the molten material per-sisting at the lower edge and the cracking of the oxide layer produce AE bursts in the continuous signal.

With the AE signals captured during and immediately after the termination of laser cutting

process, the deterioration of the laser cut surface and occurrence of dross formation at the lower cut edge can be detected. It has been determined that captured AE signals offers good informa-tion for controlling the laser cutting process in order to ensure high laser cut quality. Experimental Procedure

The laser cutting of steel sheets was performed using a CO2 laser system with a TEM00 laser beam power distribution. The steels used in the process were unalloyed deep-drawing steel DC04 of 1.5-mm thickness and austenitic stainless deep-drawing steel X5CrNi18-10 of 1.5-mm thick-ness. Both types are commonly used in the production of automotive body parts. Oxygen was used as cutting gas. In addition to the thinner steel sheets, structural steel St37 of 8-mm thickness and austenitic stainless steel X5CrNi18-10 of 6-mm thickness were laser cut, using oxygen and nitrogen, respectively.

During the laser cutting process, the steel sheets were positioned on a soft rubber support to

eliminate noise. For the detection of AE, a contact resonant PZT sensor was used. The sensor measures ultrasonic waves in a frequency range between 100 and 450 kHz. The sensor was con-nected to an AE measuring device, via a pre-amplifier.

Laser Cutting of Unalloyed Steel with Different Thicknesses

During the laser cutting process, a turbulent flow of the cutting gas produces continuous AE signals, in which changes of the signal amplitude and frequency can be detected (Kek et al.

181

2009). Laser cut quality depending on the laser cutting parameters can be determined from both the continuous signals and AE bursts after the termination of laser cutting. Figure 1 shows the influence of the position of the laser-beam focal point regarding the sheet’s surface on the AE signals captured during laser cutting. Due to the bent steel sheet and the straight movement of the laser head the laser-beam focal point moves away from its ideal position, which reduces the en-ergy density in the cutting front leads to poorer cut quality and dross formation. The solidifica-tion of the molten material persisting at the lower cut edge and the cracking of the oxide layer produce AE bursts in the continuous signal and an increase of the amplitude value in the con-tinuous signal.

Fig. 1. Influence of the position of laser-beam focal point during laser cutting on the AE signals. DC04, δ = 1.5 mm, P = 430 W, O2, v = 1500 mm/min.

Figure 2 shows the changes in laser cut quality that occur with the changing of the cutting

speed with constant power. A pronounced dross growth can be detected at the lower cut edge with deviation from ideal parameters of the laser cutting. There is no formation of secondary striations on the laser cut surface in the cutting of the unalloyed steel sheet DC04 (δ = 1.5 mm). Only a change in striation form at the lower cut edge can be seen with greater cutting speeds. The reduced laser cut quality in the cutting of thinner steel sheets is attributed mainly to dross formation at the lower edge.

Striations can be divided to primary and secondary. Primary striations are generally detected

in the laser cutting of thin low-carbon steel sheets using oxygen. In the cutting of steel sheets thicker than 2 mm, primary striations occur only in the upper part of the cut, while in the lower part usually more random pattern of secondary striations is formed. As a rule, it can be said that the lower cut edge is more closely linked to the forces causing the ejection of the melt due to shear and the pressure gradient, which are connected to the gas flow (Iverson et al. 1994, Shariff et al. 1999). Therefore, the highest quality of the lower surface can be obtained when the local maximum of the gas mass flow rate has been reached, which is connected with a turbulence of gas stream flowing to the cutting front.

182

P = 430 W v [mm/min]

2500

2000

1500

1000

Fig. 2. a) Images of dross at lower cut edge at cutting of DC04 with different cutting speeds (δ = 1.5 mm, O2). b) laser cut surface at v = 1000 mm/min and c) laser cut surface at v = 2500 mm/min.

(a)

v = 1000 mm/min (good quality cut)

v = 2500 mm/min (bad quality cut)

(b) (c)

v [mm/min]

1900

2000

2100

2200

2300

Fig. 3. Laser cut surfaces of the structural steel St37 for different laser cutting parameters.

Incr

ease

d su

rfac

e ro

ughn

ess

Mor

e in

tens

e st

riatio

n pa

ttern

on

a lo

wer

pa

rt of

the

lase

r cut

0.5 mm 0.5 mm

183

Figure 3 shows the influence of different cutting speeds on the quality of the laser cut surface in the structural steel sheets St37. This figure also shows characteristic pattern of primary and secondary striations. Degradation of the secondary striation pattern at the lower part of laser cut, and a greater surface waviness, are important indicators of laser cut quality. In this case, reduced laser cut quality is not to be attributed to greater dross formation.

On the basis of the captured AE signals, laser cut quality level can successfully be deter-

mined in instances of dross formations at the lower cut edge and changes (degradation) of stria-tion patterns.

When cutting is stopped, AE in the form of bursts with appertaining signal duration can be

captured. The signal duration is a time interval between the first and last transition of the abso-lute signal voltage value across the amplitude threshold set, i.e. 0.1 mV (40 dB). Figure 4 shows the relation between AE bursts and laser cut quality after termination of laser cutting of the DC04 steel sheet (δ = 1.5 mm). The distance of the exponential trend line from the origin of the coordinate system indicates laser cut quality. In comparison, Figure 5 shows AE bursts obtained in the considerably thicker structural steel St37 (δ = 8 mm). A distinct Fe2O3 oxide layer can be noticed on the mild steel and structural steel laser cut surfaces. Different coefficient of thermal expansion and the resultant shear stresses between the base material and the oxide layer cause the oxide layer to crack and peel during cooling, which provokes intensive burst AE. The forma-tion of dross or intensive striation pattern on the surface of the laser cut results in an increased area of the oxide layer, which is reflected in the measured AE. Thicker steel sheets therefore reveal a greater number of AE bursts. Although no significant dross formation occurs at the lower cut edge during the cutting of the thicker structural steel sheet, its quality can nevertheless be evaluated.

Fig. 4. Amplitude distribution after termination of laser cutting of DC04 steel sheet, using oxy-gen.

Laser Cutting of Austenitic Stainless Steel Sheet with Different Thicknesses and Different Cutting Gas Jets

The laser cuts obtained with austenitic stainless steel differ from those with unalloyed steels. With stainless steels a laser-cut surface does not show pronounced striations. Surface roughness

184

Fig. 5. Amplitude distribution after termination of laser cutting of St37 steel sheet, using oxygen. of a laser cut on stainless steels results primarily from rapid solidification of a thin film of the melt flowing out during cutting. The laser-cut surface is covered by the oxide layer consisting of a mixture of iron (Fe2O3) and chromium oxides (Cr2O3). The portions of the iron and chromium oxides in the oxide layer are approximately equal, which indicates higher affinity of chromium to oxygen atoms in comparison to iron atoms (Powell 1993). During laser cutting the hot melt in the cutting front is exposed to the oxygen jet. The iron and chromium atoms enter an oxidation reaction. Greater affinity of chromium results in a higher chromium content at the outer layer of the melt flowing out and lower chromium content below the oxide layer. Below the oxide layer there is a quickly solidified layer of the unoxidized substrate. This results in a re-solidified melt holding to the substrate and the presence of dross at the lower cut edge (Fig. 6). Larger amounts of the solidified oxides and of the substrate at the cut surface and in the form of dross at the lower cut edge in comparison with unalloyed steel were confirmed also by the results of the analysis of the AE bursts after the termination of cutting.

High-quality cuts can be produced by using pressure inert-gas jet rather than the more usual oxygen jet. The quality is superior to laser-oxygen cutting, but production costs are up to three times higher. Cutting mechanism is melt shearing. The resulting cut surface on austenitic stain-less steel is unoxidized. The absence of oxidation means that the cut edge will have the same corrosion resistance as the bulk material. Also the cut edges may be welded without any post-cutting preparation. Greater thicknesses of stainless steel can be cut by high-pressure inert-gas cutting than are possible by oxygen cutting, but speeds are very low. Figure 7 shows the laser-cut surfaces obtained in the cutting of the austenitic stainless steel X5CrNi18-10 of 6-mm thickness, using nitrogen. Although the nitrogen is not completely inert gas, this is also used in place of inert gas. The reason nitrogen was chosen over the completely inert argon is the cost.

185

Fig. 6. a) Images of dross at lower cut edge by cutting with different cutting speeds. b) Laser-cut surface at v = 2000 mm/min. (X5CrNi18-10, δ = 1.5 mm, P = 430 W, O2).

v [mm/min]

2750

2800

2850

2900

Fig. 7. Austenitic stainless steel laser surfaces for different process parameters.

(X5CrNi18-10, δ = 6 mm, P = 2.4 kW, N2).

Figure 8 shows the amplitude distributions of AE burst signals in a period of 30 s after the termination of laser cutting of the flat X5CrNi18-10 steel sheet. The results shown refer to cut-ting with a power P = 430 W and with various cutting speeds. Similarly as with DC04 steel, an

v [mm/min]

2500

2000

1500

1000

a) b)

Big

ger a

mou

nt o

f dro

ss a

t the

low

er la

ser c

ut e

dge

186

exponential trend line can be fitted to the amplitude distributions. In cutting with different cut-ting speeds but with the same power, i.e. 430 W, different qualities of the cut are obtained. The latter are indicated also by the appertaining reference images of dross at the lower cut edge. Similarly as with unalloyed steel, poorer cut quality will show in a larger number of bursts after the termination of cutting. A reduced cutting speed with the power chosen permits a higher en-ergy input into the cutting front. This results in scarcer occurrence of dross and a cut of higher quality, which is confirmed by a displacement of the exponential trend line towards the origin of the coordinate system in the diagram of the amplitude distribution of AE signals.

Fig. 8. Amplitude distribution after termination of laser cutting of the austenitic stainless steel sheet X5CrNi18-10, δ = 1.5 mm, using oxygen.

Fig. 9. Amplitude distribution after termination of laser cutting of the austenitic stainless steel sheet X5CrNi18-10, δ = 6 mm, using nitrogen.

187

Greater differences in the captured AE signals are found when laser cutting of austenitic stainless steels is performed with nitrogen (Fig. 9). Using nitrogen as cutting gas significantly reduces the oxidation of the laser cut surface. The whirling of the gas at the lower part of the cutting edge induces the oxidation of a smaller section of the heated material, which results in AE bursts even in excellent cuts. In case of dross formation, the surface of the oxide layer in-creases and causes the exponential trend line to move away from the origin of the coordinate system. Comparing Figs. 8 and 9, a marked difference can be noted in the measured AE bursts between the cutting processes using oxygen and nitrogen, even with four times greater thick-nesses. Conclusions

The following conclusions can be drawn from the analysis of the AE signals captured during laser cutting process: - The AE signals captured during the laser cutting process and immediately after its termina-

tion allow the estimation of laser cut quality defined by the state of the laser cut surface and the occurrence of dross at the lower cut edge.

- The measurements confirm that this method of laser cut quality determination is applicable to unalloyed and alloy steels with different thicknesses.

- The method of the laser cut quality estimation was performed for laser cutting using oxygen and nitrogen as cutting gases.

- The laser cut quality estimation requires a prior system calibration for various types of mate-rials and cutting gases.

The new AE monitoring technique described in this paper was developed at the Faculty of

Mechanical Engineering, University of Ljubljana, and proves to be a promising method of the laser cutting production monitoring system. References Arata Y., Maruo H., Miyamoto I., Takeuchi S. (1979), Dynamic Behavior in Laser Cutting of Mild Steel; Transactions of Japanese Welding Research Institute, 8 (2), 15 - 26.

Di Pietro P., Yao Y.L. (1995), A New Technique to Characterize and Predict Laser Cut Stria-tions; Internationa Journal of Mach. Tools Manufacturing, 35 (7), 993 - 1002.

Iverson A., Powell J., Kamalu J., Magnusson C. (1994), The oxidation dynamics of laser cutting of mild steel and the generation of striations on the cut edge; Journal of Materials Processing Technology, 40, 359-374.

Kek T., Grum J. (2009), AE Signals as Laser Cutting Quality Indicators; Insight 51 (3), 124 - 128.

Powell J. (1993), CO2 Laser Cutting; Springer-Verlag, London, pp.16 - 22.

Schuöcker D. (1987), The Physical Mechanism and Theory of Laser Cutting; The Industrial La-ser Annual Handbook, Penn Well Books, Tulsa, Oklahoma, pp. 65-79.

Shariff S. M., Sundararajan G., Joshi S. V. (1999), Parametric Influence on Cut Quality Attrib-utes and Generation of Processing Maps for Laser Cutting; Journal of Laser Applications, 11 (2), 54 - 63.

J. Acoustic Emission, 28 (2010) 188 © 2010 Acoustic Emission Group

ACOUSTIC EMISSION ANALYSIS AND THERMO-HYGRO-MECHANICAL MODEL FOR CONCRETE EXPOSED TO FIRE

CHRISTIAN GROSSE1, JOŠKO OŽBOLT2, RONALD RICHTER1

and GORAN PERIŠKIĆ2

1 Technische Universität München, Non-destructive Testing Laboratory, cbm, Baumbachstr. 7, 81245 Munich, Germany; 2 Universität Stuttgart, Institut für Werkstoffe im Bauwesen, Pfaffen-

waldring 4, 70569 Stuttgart, Germany Abstract

High performance concrete (HPC) is often used in tunneling, for columns of high raise build-

ings and similar structures, which require high compressive strength. However, when exposed to high temperature (fire) there is a strong degradation of mechanical properties of HPC and it shows unfavorable behavior i.e., explosive spalling of concrete cover. In the present experiments the behavior of normal strength concrete that was exposed to fire is monitored by acoustic emis-sion (AE) technique. By the use of the AE technique damage processes in concrete can be ob-served during the entire fire history and therefore the detection of initiation of explosive spalling can also be detected. The method enables the location and characterization of the micro-cracking before failure. The paper describes the proposed concept and preliminary results of fire experi-ments on concrete specimens made of normal strength concrete. To support experimental results it is also important to have a numerical model, which is able to realistically predict behavior of concrete at high temperatures. Therefore, in the paper is also briefly discussed fully coupled thermo-hygro-mechanical model for concrete that is implemented into a 3D finite element code. The application of the model is illustrated on one numerical example, which demonstrates that the pore pressure in combination with thermally induced stresses can lead to explosive spalling of concrete cover. Keywords: Concrete, high temperature, micro-cracking, spalling, numerical analysis Introduction

When temperature increases for a couple of hundred of degrees Celsius, behavior of concrete

changes significantly. The concrete mechanical properties, such as strength, Young’s modulus and fracture energy, are at high temperatures rather different from the concrete at normal tem-perature. At high temperatures, large temperature gradients subject concrete structures to tem-perature-induced stresses, which cause damage. Furthermore, creep and relaxation of concrete that is due to high temperature play also an important role. One of the reasons for the complexity of the behavior of concrete at high temperatures is due to the fact that concrete contains water, which at high temperature changes its state of aggregation and can generate significant pore pressure. Furthermore, the microstructure of concrete is extremely complex and at high tempera-tures there are chemical changes, which significantly influence overall properties of concrete. Moreover, at high temperatures the aggregate can change its structure or it can lose its weight through the emission of CO2, as in calcium-based stones. Consequently, the behavior of concrete at high temperatures is strongly dependent on concrete type.

Although the behavior of concrete at high temperatures is in the literature well documented

(Thelandersson 1983, Khoury et al. 1985, Schneider 1986, Bažant and Kaplan 1996, Zhang and Bićanić 2002, Khoury 2006, Zieml et al. 2006) further experimental and theoretical studies are

189

needed to clarify the interaction between hygro-thermal and mechanical properties; for instance, explosive type of failure due to spalling of concrete cover, which is typical for high performance concrete (HPC). The main problem in the experimental investigations is that such experiments are rather demanding, i.e. one has to perform loading and measurement at extremely high tem-peratures. Furthermore, such experiments can be carried out only on relatively small structures.

To better understand the behavior of concrete at high temperatures and to support the ex-

periments, numerical analysis can be useful. However, one needs models, which can realistically predict behavior of concrete at high temperatures. To obtain realistic predictions the calibration of the models is needed. This can be done using acoustic emission and other non-destructive testing techniques. In the present paper, an experimental setup is proposed in order to observe the initial micro-cracking process and possibly explosive spalling during fire loading of concrete. The measured data will be in future compared with numerical results in order to improve the model. Acoustic emission (AE) technique is a well-established method for the detection of dam-age in concrete structures that can be used to verify and improve numerical models. In the paper are shown some preliminary experimental results obtained by using AE technique on concrete specimens exposed to high temperatures. Moreover, a thermo-hygro-mechanical model for con-crete is discussed, which should be in future verified and improved based on the experimental results on HPC using AE technique. Setup and Equipment

For the preliminary fire tests one of the controlled furnaces at the material testing institute,

University of Stuttgart (MPA), is used (see Fig. 1). The furnace is equipped with test specimen at two sides as well as at the top center of the oven as shown in Fig. 2. In the figure are marked the positions of the AE sensors. The burner cannot be seen because it is covered by the concrete specimen that have a size of 540 x 220 x 480 mm (side wall) and 800 x 630 x 220 mm (top wall), respectively. In the experiments the normal strength concrete was used with uniaxial compres-sive strength of approximately 30 MPa.

Fig. 1: Furnace stands at the MPA University of Stuttgart.

190

For the preliminary tests, a set of 24 different sensors in total (see Fig. 3a) was used includ-ing resonant, multi-resonant and broadband transducers (Grosse et al. 2003; Grosse and Ohtsu 2008; Richter 2009). After pre-amplification AE signals were recorded using three multi-channel transient recorders. In parallel to the AE data the temperature in the furnace as well as on differ-ent positions of the surface was recorded. Additionally the specimens were monitored using an infrared video camera observing the specimen from the outside and with a camcorder inspecting the inner furnace chamber (Fig. 3b).

Fig. 2: 3D model of the experimental setup including the positions of the AE sensors.

(a) (b)

Fig. 3: (a) Some AE sensors and one of the transient recorders and (b) view into the furnace chamber.

191

Measurements

Concerning AE recording, the experimental setup is critical due to the noise emitted from the furnace. Accordingly, the number of noise signals was unusually high. However, several hundred events were recorded showing a good signal-to-noise ratio enabling localization of events. An example of such signals showing an event recorded with eight channels is presented in Fig. 4.

Fig. 4: Acoustic emission event recorded by an eight-channel transient recorded. The vertical lines mark the signals' onset at each channel.

The AE hit rate is an equivalent for the number of recorded AE signals crossing the trigger level previously set. A comparison of the hit rate with the temperature in the furnace is displayed in Fig. 5, which gives a first overview of the deteriorations. According to AE signals, the crack-ing process (damage) starts at 500°C and the number of AE hits develop with a high slope up to a temperature of approximately 750°C. After reaching this temperature the hit rate gradient is getting lower. In this temperature range strong degradation of hydration products usually occur in concrete (Schneider 1986).

192

Fig. 5: Hit rate and temperature evolution inside the furnace chamber.

Fig. 6: 3D localization of the AE epicenters and projection to the planes.

The AE signals were classified in a way that only events recorded by six or more channels

with localization accuracy better than 20 mm were recorded. An example of the localized events of one of the experiments is shown in Fig. 6. Most of the source locations are detected in the center of the slab that is the region of the highest temperature gradient. The accuracy of the AE locations given by the deviations in x and z direction is shown in Fig. 7.

193

Fig. 7: Projection to the x/z plane and error bars of localization.

A reliable localization is based on the knowledge of the compression wave velocity that is

usually determined prior to the experiment. However, the temperature gradients in the slabs dur-ing the experiment caused velocity anomalies prohibiting the use of uniform velocity values. Therefore, infrared thermography (Fig. 8) will be used in further experiments to overcome this problem. Moreover, the fracturing of the material at higher temperatures leads to further ambi-guities due to micro-cracks that change the travel path of the waves between source and receiver. A clustered AE event analysis using events clustered in a timely manner will be performed in future experiments to eliminate this effect. Travel time tomography methods will enable for a better determination of the velocity model.

Fig. 8: Infrared thermography showing the temperature gradient in a slab during the experiment.

194

The present experiments were conducted on normal strength concrete. Due to the fact that normal strength concrete is rather porous and has relatively high permeability, no explosive spalling was observed in the experiments. Most of damage detected by AE is due to the degrada-tion of concrete properties at high temperatures and thermally induced stresses in concrete specimen. However, in future experiments the technique presented here will be used to detect damage processes in HPC before explosive spalling that takes place 4 to 10 minutes after the start of heating (Ozbolt et al. 2008). Thermo-hygro-mechanical Model

The present phenomenological model for concrete is a thermo-hygro-mechanical one. It is formulated in the framework of continuum mechanics under the assumption of validity of irre-versible thermodynamic. The following unknowns control the response of the model: tempera-ture, pore pressure (moisture), stresses and strains. In the numerical model, temperature, mois-ture and pore pressure are coupled with stresses and strains, i.e. thermo-hygral part of the model depends on damage of concrete. Moreover, the relevant macroscopic mechanical properties of concrete (Young’s modulus, tensile strength, compressive strength and fracture energy) are tem-perature dependent. Coupled Heat and Moisture Transfer in Concrete

The general approach for the solution of the problem of coupled heat and mass transfer in a porous solid, such as concrete, is well known within the framework of irreversible thermody-namic. However, there are a number of complex details; therefore, for the practical application the problem must be simplified (Bažant and Thonguthai 1978).

After introducing simplifications and assuming for a moment that the moisture flux (J) and

heat flux (q) in concrete are independent of the stress and strain, the following is valid (Bažant and Thonguthai 1978):

(1)

(2)

where p = pore pressure, T = temperature, b = heat conductivity, ap/g = permeability, which is in the present model taken as a function of temperature according to the proposal of Bažant and Thonguthai (1978), with g = gravity constant.

The governing equation for mass conservation is written in Eq. (3), where w = water content, t = time and wd = total mass of water released into the pore by dehydration. In the present model, dehydration is not accounted for. The balance of heat is written in Eq. (4), where C = mass den-sity and isobaric heat capacity of concrete, Ca = heat sorption of free water and Cw = heat capac-ity of water, which is in the present model neglected.

(3)

(4)

Boundary conditions at concrete surface can be defined as:

195

(5) (6)

where αw = surface emissivity of water, αG = surface emissivity of heat, T0 and p0 are tempera-ture and pore pressure at concrete surface and TE and pE are temperature and pore pressure of environment.

The main difficulty in the above equations is the determination of the material properties. Assuming for a moment no stress-dependency, the constitutive laws for p, w and T follow sim-plified suggestions proposed by Bažant and Thonguthai (1978). To describe the state of pore water in concrete, one has to distinguish between three different states: (i) non-saturated con-crete, (ii) saturated concrete and (iii) transition from non-saturated to saturated concrete. For non-saturated concrete, the pore pressure is mainly influenced by the total amount of free water w and by the saturation water content ws, which is a function of temperature. In the present model, a semi-empirical expression proposed by Bažant and Thonguthai (1978) is adopted.

For saturated concrete one can theoretically calculate p for given w and T by using steam ta-

bles, accounting for the porosity as the given volume available to pore water. This would, how-ever, already for relative low temperatures yield to extremely high pore pressure, which is not realistic. Therefore, one has to take into account the increase of the available pore space (poros-ity), which is mostly due to the decrease of the adsorbed water portion; i.e. the release of some water molecules, which were chemically bound at room temperature (Bažant and Thonguthai 1978). In the presented model, the pore pressure for saturated concrete is mainly controlled by porosity, which is defined as a function of temperature. Following the weight loss experiments by Harmathy and Allen (1973), it is assumed that the amount of mass loss of concrete at certain temperature equals to the mass of evaporated water. Assuming constant water weight density, it is possible to calculate the change of water content in the available volume.

Except for an extremely slow change in pore pressure, the transition from non-saturated to

saturated concrete can be abrupt. For most practical situations the transition is most likely smooth. Furthermore, an abrupt transition would cause numerical difficulties. Therefore, for rela-tive humidity between 0.96 and 1.04, linear increase of free water content is assumed (Bažant and Thonguthai 1978). Thermo-hygro-mechanical Coupling

To account for the influence of temperature on the strain development in concrete, the total strain tensor ε for stressed concrete exposed to high temperature is decomposed as (Khoury et al. 1985, Nielsen et al. 2002):

(7) where εm = mechanical strain tensor, ε ft = free thermal strain tensor, ε tm = thermo-mechanical strain tensor and ε c are strains due to the temperature dependent creep of concrete. For more de-tails about the mentioned strain components, see Ožbolt et al. (2008).

The mechanical strain tensor that comes into the 3D constitutive law for concrete (mi-croplane model) is calculated as = . The mechanical strains are than used to calculate the effective stresses increments (stress in solid phase of concrete matrix) and

196

macroscopic stresses increments from the microplane constitutive law (Ožbolt et al. 2001, 2005, 2008):

(8) where D = tangent material stiffness tensor obtained from the microplane model, = increment of the mechanical strain tensor and = increment of pore pressure, which is calculated from the increment of volumetric pore pressure , with = increment of pore pressure. Note that according to definition pore pressure p is negative.

In general, the mechanical strain component can be decomposed into elastic, plastic and damage part. The source of mechanical strain is external load, thermal strain induced stress and pore pressure in concrete. The parameters of the microplane model are modified such that the macroscopic response of the model fits temperature dependent mechanical properties of concrete (Ožbolt et al. 2005). To account for finite strains the co-rotational stress tensor together with Gree-Lagrange strain tensor (Ožbolt et al. 2008) are used in the formulation of microplane model. The finite strain formulation is needed in order to investigate the influence of geometrical instabilities of concrete layer (buckling) on the explosive type of spalling of concrete cover.

Free thermal strain is stress independent and it is experimentally obtained by measurements

on a load-free specimen. The total free thermal strain consists of temperature dilatation and shrinkage of concrete, which can be experimentally isolated (Khoury 2006). In the present model, the temperature dependent shrinkage is a part of the free thermal strain.

The thermo-mechanical strain is stress and temperature dependent. It appears only during

first heating and not during subsequent cooling and heating cycles (Khoury et al. 1985, Khoury 2006). This strain is irrecoverable and leads to severe tensile stresses in concrete structures dur-ing cooling. Temperature dependent creep strain is of the same nature as the thermo-mechanical strain except that it is partly recoverable. In an experiment it is not possible to isolate this strain. For low temperature rate, which is normal case in the experiments, this strain component com-pared to the thermo-mechanical strain is small. Therefore, temperature dependent creep strain is neglected in the present model.

(a)

(b)

Fig. 9. Porosity (a) and normalized permeability (b) as function of crack width.

197

The thermo-hygral processes are coupled with mechanical properties of concrete. It is known that permeability and porosity of concrete are relevant parameters that control transport proc-esses in concrete. On the other hand, both porosity and permeability are strongly influenced by damage, i.e. for higher level of damage, porosity and permeability increase. To account for this, permeability and porosity of concrete are assumed to be strain dependent. Following suggestions from the literature (Wang et al. 1997, Aldea et al. 2000), the relations plotted in Figs. 9 are adopted in the present model. Damage (crack opening) is calculated using temperature dependent microplane model (Ožbolt et al. 2008).

Besides the constitutive law in the mechanical part of the model the equilibrium must be full-

field: (9)

where V represents volume forces (gravity loads and pore pressure). Numerical Analysis

The strong finite element (FE) formulation described above is first rewritten into a weak FE form and then implemented into a 3D FE code. The FE analysis is incremental and based on the direct integration method. To assure stability of the time integration, a backward difference method is used. Since the controlling parameters are coupled, the linear differential equations 3 and 4 have to be solved iteratively. In solving the non-mechanical part of the model, the me-chanical properties of concrete (damage) are assumed to be constant. Similarly, in solving equi-librium equation, non-mechanical properties of concrete (temperature and pore pressure) are assumed to be constant. For more detail see Ožbolt et al. (2001, 2005, 2008). Numerical Example – Explosive spalling of concrete cover

The proposed thermo-hygro-mechanical model for concrete is employed to investigate explo-sive spalling of concrete cover. In the numerical example a concrete slab of infinite length is locally heated at the free surface (see Fig. 10a). The height of the segment of the slab is 150 mm and the considered width of the slab is 480 mm (only symmetric part is analyzed). Varied are concrete strength, porosity, permeability and moisture content. Moreover, the role of geometrical nonlinearity (instability) for the problem of explosive spalling is studied as well. To assure mesh objective results, in the mechanical part of the model the crack band method (Ožbolt et al. 2005) is used. In the analysis eight-node solid elements are used (see Fig. 10b) assuming plane strain condition. Except at the free concrete surface, all boundaries of the specimen are restrained in all three directions. Note that the analysis is static, i.e. structural inertia forces are not accounted for. The linear increase of air temperature in time is applied with a temperature gradient of 80°C/min. Pore pressure at the surface is taken to be 1.0 kN/m2. The analysis is performed for the time pe-riod of 12 minutes (duration of heating). The initial thermo-hygro-mechanical properties of con-crete are summarized in Table 1.

The concrete properties used in the present example of explosive spalling of concrete cover were: initial permeability a0 = 10-12 m/s, humidity RH0 = 75%, initial temperature of concrete T0 = 20°C, initial porosity n0 = 0.10. The remaining parameters were the same as listed in Ta-ble 1. The failure mode, which is typical for explosive spalling, is shown in Fig. 11. Figure 11a shows the initiation of spalling. The dark zone (maximal mechanical principal strain) shows lo-calization of damage. After initiation, complete spalling takes place very quickly. The corre-

198

sponding failure mode is shown in Fig. 11b. To understand the failure mechanism, the distribu-tion of macroscopic stresses parallel (σy) and perpendicular (σx) to the free surface, at initiation and after spalling, are shown in Fig. 12. From Fig. 12a it can be seen that because of equilibrium reason the stresses in direction perpendicular to the free surface are almost zero. On the other hand, stresses parallel to the free surface are relatively high compressive stresses, which after spalling reduce almost to zero. It turns out that these stresses are not high enough to cause buck-ling of concrete cover. Namely, the companion numerical analysis, in which pore pressure was not considered, predicts almost the same macroscopic stresses but no spalling. In absence of lat-eral stresses, which are generated by pore pressure, there is no spalling. Therefore, it can be con-cluded that in the present example the main driving force for initiation of spalling is pore pres-sure and not strain induced stresses. (a)

(b)

Fig. 10: Geometry of the heated concrete slab and the corresponding FE model.

Table 1 Properties of concrete used in the FE analysis.

Young’s modulus E [MPa] 30000 Mass density ρ [kg/m3] 2400 Poisson’s ratio ν 0.18 Water/cement ratio 0.5 Tensile strength ft [MPa] 2.0 Saturation water content [kg/m3] 100 Uniaxial comp. strength fc [MPa] 30.0 Parameter αa 0.0 Fracture energy GF [N/mm] 0.1 Parameter αn 0.375 Conductivity b [J/(msK)] 1.67 Surface emissivity of water αw max. Heat capacity C [J/(kgK)] 900 Surface emissivity of heat αG max.

Figure 13a shows the time evolution of pore pressure in the element, in which spalling is ini-tiated. The relative high pore pressure predicted by the model is supported by a simple engineer-ing model according to which the critical pore pressure, i.e. pore pressure that corresponds to the tensile strength of bulk material (concrete), reads: pcrit = ft (1-n)/n. For initial porosity n = 0.1, pore pressure at initiation of spalling is pcrit = 9ft. After accounting for the reduction of tensile strength due to high temperature, tensile strength of concrete is in the range of 1.0 to 1.5 MPa. For this strength the critical pore pressure is in the range of 9.0 to 13.5 MPa, which is in good agreement with the outcome of the present model.

In the present study, pore pressure, and therefore initialization of spalling, is mainly con-trolled by permeability of concrete. To illustrate this, the evolution of pore pressure for the initial

199

permeability, which is 10 times larger than the original one, is also shown in Fig. 13a. Compared to the original case, the maximal pore pressure is reduced by the factor of approximately 5 and therefore no spalling takes place. (a)

(b)

Fig. 11: a) Localization of damage (max. mechanical principal strains) at initialization of spalling and b) crack pattern due to spalling. (a)

(b)

Fig. 12: Distribution of stresses caused by deformation of concrete before and after spalling: a) in direction perpendicular to concrete surface (σx) and b) in direction parallel to concrete surface (σy).

Figures 13a, b also show the evolution of volumetric stress, temperature, humidity and satu-ration pressure. In Figs. 13c, d, the results are plotted for the analyses without geometrical non-linearity. It can be seen that the evolution of relevant quantities in the critical section (initializa-tion element) is similar as shown in Figs. 13a, b (geometrical nonlinearity), except that in the analysis with geometrical nonlinearity spalling initiates earlier and closer to the concrete surface. This clearly indicates that geometrical nonlinearity increases the risk for explosive spalling.

Typical distribution of pore pressure, volumetric stresses, moisture and temperature as a

function of the depth is shown in Fig. 14. The results are plotted for time steps before and after

200

spalling (dotted lines). It can be seen that the maximal moisture (over saturation) is localized in front of maximal pore pressure. The highest pore pressure and volumetric stresses are generated at crack initiation. After crack propagation (spalling) pore pressure and volumetric stresses abruptly decrease. The pore pressure decreases from approximately 20 MPa at initiation of spall-ing to 4 MPa at spalling. The volumetric stress drops from 8 MPa to 1 MPa. (a)

(b)

(c)

(d)

Fig. 13: Time evolution of: a) pore pressure, volumetric stress and saturated pore pressure and b) temperature and humidity in the finite element in which spalling initiates for the case with geometrical non-linearity; c) and d) the same as a) and b) but without geometrical non-linearity (with initial parameters – a0 = 10-12 m/s, RH0 = 75%, T0 = 20°C and n0 = 0.1).

Although the present results are not directly compared with predictions of similar models,

based on the literature review (Gawin et al. 1999, 2006, Tenchev et al. 2001), it should be men-tioned that the majority of models principally predict similar results. The largest discrepancy in the predictions of various models is due to the prediction of pore pressure. For the same or simi-lar material properties and boundary conditions as used in the present example, the predicted maximal pore pressure in the literature varies from 1 MPa to 10 MPa (Gawin et al. 1999, 2006, Tenchev et al. 2001). Therefore, the results of the present model must be considered as a pre-liminary, i.e. further work is needed to verify the model and to come up with results, which can bring more light in the mechanisms that govern explosive spalling of concrete cover. This is es-pecially the case for the loading with higher heating rates where besides pore pressure the

201

compressive stresses parallel to free concrete surface together with geometrical nonlinearity may have influence on explosive spalling.

(a)

(b)

(c)

(d)

Fig. 14: Distribution of: a) pore pressure, b) volumetric stresses, c) temperature and d) humidity, before and after spalling, measured from the concrete surface (with initial parameters – a0 = 10–12 m/s, RH0 = 75%, T0 = 20°C and n0 = 0.1). Conclusions and Outlook

The preliminary experiments presented demonstrated that AE method can successfully be

used to investigate deterioration of concrete under fire load. Signal-based AE techniques can produce detailed 3D picture of the fracture process during fire loading. Problems related to this method of the analysis are due to the large temperature gradient and the micro-cracking phenom-ena that both cause ambiguities of the velocity model. In future these problems will be handled and a series of experiments testing different concrete mixes will be conducted. The main goal of future research is to better understand explosive spalling of concrete in order to improve per-formance of the HPC under fire load. Moreover, future experiments based on the AE method will be used to verify and improve the thermo-hygro-mechanical model of concrete presented here.

202

Acknowledgements

The authors are grateful for the collaboration with unit 51320 "Fire resistance" of the Materi-alprüfungsanstalt Universität Stuttgart (MPA) permitting the use of the furnace stand and con-ducting the described experiments. Further the advice of Dr. Jürgen Frick, Dr. Markus Krüger and Mr. Gerhard Bahr is acknowledged. The research project started January 2010 and is spon-sored by the Deutsche Forschungsgemeinschaft DFG via grants OZ 10/8-1 and GR1664/7-1. References 1. Aldea, C.M. Ghandehari, M. Shah, S.P. Karr, A. Estimation of Water Flow through Cracked

Concrete under Load. ACI Materials Journal, 97, 567-575, 2000.

2. Bažant, Z.P. and Kaplan, M.F. (1996), Concrete at High Temperatures: Material Properties and Mathematical Models, Harlow, Longman.

3. Bažant, Z.P. and Thonguthai W. (1978). Pore pressure and drying concrete at high tempera-ture, ASCE, J. of Eng. Mech., 104, 1059-1079.

4. Gawin, D., Majorana, C.E. and Schrefler, B.A. (1999). Numerical analysis of hygro-thermal behavior and damage of concrete at high temperature, Mechanics of cohesive frictional mate-rials, 4, 37-74.

5. Gawin, D., Pesavento, F. and Schrefler B.A. (2006). Towards prediction of the thermal spall-ing risk through a multi phase porous media model of concrete, Comput. Methods Appl. Mech. Eng., 195, 5707-5729.

6. Grosse C., Reinhardt H.-W., Finck F. (2003). Signal-based acoustic emission techniques in civil engineering. J. of Mat. in Civ. Eng. 15 (3), 274-279.

7. Grosse, C. and Ohtsu M. (2008). Acoustic Emission Testing in Engineering – Basics and Applications. Springer publ., Heidelberg, ISBN: 978-3-540-69895-1, 415 p.

8. Harmathy, T.Z. and Allen, L.W. (1973). Thermal properties of selected masonry unit con-cretes, J. of Amer. Inst., 70, 132-144.

9. Khoury, G.A., Grainger, B.N. and Sullivan, P.J.E. (1985). Transient thermal strain of con-crete: literature review, conditions within specimens and behavior of individual constituents, Mag. of Conc. Res., 37 (132), 131-144.

10. Khoury, G.A. (2006). Strain of heated concrete during two thermal cycles – Parts 1,2 and 3, Mag. of Conc. Res., 58 (6), 367-385, 58 (6), 387-400 and 58 (7), 421-435.

11. Nielsen, C.V., Pearce, C.J. and Bićanić, N. (2002). Theoretical model of high temperature effects on uniaxial concrete member under elastic restraint, Mag. of Conc. Res., 54 (4), 239-249.

12. Ožbolt, J., Li, Y.-J. and Kožar, I. (2001). Microplane model for concrete with relaxed kine-matic constraint, International Journal of Solids and Structures, 38, 2683-2711.

13. Ožbolt, J., Kožar, I., Eligehausen, R. und Periškić, G. (2005). Instationäres 3D Thermo-mechanisches Modell für Beton. Beton- und Stahlbetonbau, 100 (2005), Heft 1, 39-51.

14. Ožbolt, J., Periškić, G. Reinhardt, H.W. and Eligehausen, R. (2008). Numerical analysis of spalling of concrete cover at high temperature, Comp. & Concrete, 5 (4), 279-294.

203

15. Richter, R. (2009). Einsatz der Schallemissionsanalyse zur Detektion des Riss- und Ab-platzungs verhaltens von Beton unter Brandeinwirkung, Diplomarbeit, Materialprüfungs-anstalt Universität Stuttgart.

16. Schneider, U. (1986). Properties of Materials at High Temperatures, Concrete, 2nd. Edition, RILEM Technical Committee 44-PHT, Technical University of Kassel, Kassel.

17. Tenchev, R.T., Li, L.Y. and Purkiss, J.A. (2001). Finite element analysis of coupled heat and moisture transfer in concrete subjected to fire, Numerical Heat Transfer, 39, 685-710.

18. Thelandersson, S. (1983). On the multiaxial behaviour of concrete exposed to high tempera-ture, Nucl. Eng. and Design, 75 (2), 271-282.

19. Wang, K., Jansen, D.C. and Shah, P.S. (1997). Permeability study of cracked concrete”, Cem. and Conc. Res., 27 (3), 381-393.

20. Zhang, B. and Bićanić, N. (2002). Residual fracture toughness of normal- and high-strength gravel concrete after heating to 600°C, ACI Mat. J., 99 (3), 217-226.

21. Zieml, M., Leithner, D., Lackner, R. and Mang, H.A. (2006). How do polypropylene fibers improve the spalling behavior of in-situ concrete, Cem. and Conc. Res., 36 (5), 929-942.

J. Acoustic Emission, 28 (2010) 204 © 2010 Acoustic Emission Group

AE-SiGMA ANALYSIS IN BRAZILIAN TEST AND ACCELERATED CORROSION TEST OF CONCRETE

MASAYASU OHTSU and YUMA KAWASAKI

Graduate School of Science and Technology, Kumamoto University 2-39-1 Kurokami, Kumamoto 860-8555, Japan

Abstract

The moment tensor analysis of AE waveforms has been developed and is available for iden-tifying crack kinematics of a location, a crack-type and a crack orientation in a material as the SiGMA (Simplified Green’s functions for Moment tensor Analysis) analysis. Mechanisms of cracking can be visually and quantitatively studied at the meso-scale in concrete. Since the ten-sile strength of concrete is normally evaluated by the Brazilian test, mechanisms of macro-scale tensile failure in concrete are examined as the cracking process at the meso-scale using the SiGMA analysis. Evolution of the fracture process zone under the combination of tensile and compressive stresses is discussed. In the corrosion process of reinforced concrete, high AE ac-tivities are observed twice during the corrosion process at the onset of corrosion in rebar and the nucleation of cracking in concrete. The SiGMA analysis is applied to an accelerated corrosion test of a reinforced concrete beam. Kinematics of corrosion-induced cracks in concrete is iden-tified and applicability to early warning of corrosion damage in reinforced concrete structures is discussed. Keywords: AE-SiGMA analysis, concrete, Brazilian test, corrosion-induced cracking in rein-forced concrete Introduction

AE source mechanisms in engineering materials can be kinematically identified by applying the moment tensor analysis of AE signal-based methods (Grosse and Ohtsu, 2008). One pow-erful technique for the moment tensor analysis was developed as SiGMA (Simplified Green’s functions for Moment tensor Analysis) analysis, by which crack kinematics of locations, crack types and orientations are quantitatively determined (Ohtsu, 1991). The SiGMA analysis was successfully applied to fracture tests of reinforced concrete specimens to visualize the cracking mechanisms (Ohtsu, et al., 1998; Ohno and Ohtsu, 2007). Because the SiGMA analysis is closely associated with AE source modeling, a relation with such force models as a dipole force and a couple force is discussed. In addition, another relationship with linear fracture mechanics in concrete is briefly summarized.

As applications of the SiGMA analysis, results of two experimental studies are discussed.

In the first application, the fracture process of the Brazilian test was observed. In concrete, the tensile strength of concrete is normally evaluated by this test, where a cylindrical specimen is compressed in the diametral direction, and macro-scale tensile failure is observed. A relation between the generation of macro-scale tensile cracks and the accumulation of meso-scale cracks is studied. In the second application, continuous AE measurement is conducted to monitor the corrosion process in reinforced concrete specimens. During cyclic wet-dry tests, kinematics of corrosion-induced cracks in concrete is identified, and an early warning for the corrosion damage in concrete is discussed.

205

AE-SiGMA Analysis

Although the generalized theory of AE was published (Ohtsu and Ono, 1984), AE source for a tensile crack is still mistakenly modeled by a single dipole force. In principle, AE source can be modeled by the dislocations, which are not referred to as crystalline motions, but discontinui-ties of displacements in a material (Eshelby, 1973). As is well known, the moment tensor is defined,

Mpq = SCpqjknkbj (y,t)∫ dS = Cpqjknkl j S b(y,t)∫ dS = Cpqjknkl jΔV . (1)

Here, vector l is the unit vector of crack motion and ΔV is the crack volume, corresponding to the integration of crack motion b(y,t) over the crack surface S. Since nkbj

corresponds to the eigen-

strain in micromechanics (Mura, 1982), the moment tensor as the product of the elastic constants with strain is equivalent to stress as the second-rank tensor. It is noted that the physical unit is the moment as the elastic constants [N/m2] times the crack volume [m3]. From Eq. 1, it is de-rived that AE source for a tensile crack is modeled by three normal components of the stresses. These components are equivalent to three dipole forces, which are decomposed into the com-pensated-linear vector dipoles (CLVD) defined by Knopoff and Randall (1970) and the hydro-static components. In the case of a shear crack, AE source for a shear crack is modeled by shear stresses, which are equivalent to double-couple forces in the equilibrium state. In order to classify crack type, the eigenvalue analysis of the moment tensor was developed (Ohtsu, 1991). The eigenvalues of the moment tensor are to be composed of the combination of the shear crack and the tensile crack. The decomposition leads to the shear component (X), the deviatoric tensile component (Y), and the hydrostatic component (Z) of the tensile crack. The last two components correspond to CLVD and the equivalent stresses to the crack volume ΔV in Eq. 1. In the present SiGMA code, AE source with the shear ratio X > 60 % is classified as the shear crack, one with X < 40 % as a tensile crack and one with 40 % < X < 60 % as a mixed-mode crack. The crack-motion vector l and the normal vector n in Eq. 1 are determined from three eigenvectors. Results are visually displayed, by using a graphic software (Light-Wave 3D, NewTek). Three models of these cracks are illustrated in Fig. 1. Cracks are classi-fied into three types of shear, mixed-mode and tensile, and their crack planes normal to vectors n are illustrated by circles and directions of crack motions l are shown by arrows.

Fig. 1 Models for tensile crack, mixed-mode crack and shear crack.

It is worth noting a relationship with the modes in linear fracture mechanics. Although the

crack classification of the tensile crack and the shear crack in the SiGMA analysis was referred to as identical to mode I and mode II in fiber-reinforced concrete (Carpinteri et al., 2007), the

206

treatment is not rational. Rigorously speaking, the crack classification by the moment tensor is based on crack displacements (motions) on the crack surface at the meso-scale. In contrast, the modes in fracture mechanics represent crack propagation from the crack-tip at the macro-scale in composite materials. In case that crack orientation θ is taken into account for crack propagation, the modes can be calculated from the criterion on the crack extension (Erdogan and Sih, 1963),

KI sinθ + KII (3cosθ −1) = 0 , and (2)

cosθ2(KI cos

2 θ2−32KII sinθ) = KIC . (3)

Here KI and KII are the stress intensity factors of mode I and mode II, respectively. KIC is the critical stress intensity factor of mode I. Referring to the propagation angle θ as the angle be-tween consecutive AE sources determined from normal vectors n, the extension of tensile cracks due to volumetric expansion in concrete was studied, applying Eqs. 2 and 3 (Ohstu and Uddin, 2008). It was found that the extension of the tensile cracks at the macro-scale is predominantly governed by mode I failure, even though all the types of tensile, mixed-mode and shear cracks were observed at the meso-scale in the SiGMA analysis.

In the SiGMA analysis, the determination of the two parameters of the arrival time and the amplitude of the first motion is essential and previously carried out by visual and hand-picking methods. To improve these enormous efforts for the analysis, we have currently developed an automated detection method based on AIC (Akaike Information Criteria) (Ohno and Ohtsu, 2010). Brazilian Tests (1) Tensile strength of concrete The Brazilian test is also known as a split-tensile test for the tensile strength of concrete. Later, a direct tensile test was alternatively applied, because the test is conducted under com-pression. Currently, it is found that the difference between the strengths determined by the Brazilian test and those by the direct-tensile test is within 10 %. As a result, the Brazilian test is usually applied to estimate the tensile strength of concrete.

As is well-known in elasto-statics, stresses at the center of a disk of diameter d and unit thickness under compressive load P in the diametral direction is known (Timoshenko and Good-ier, 1970) as,

and ,

where σxx is the tensile stress in the horizontal direction and σyy is the compressive stress in the vertical direction. σxx is the maximum tensile stress, and stresses near loading plates are even more compressive than σyy in the cross-section. This implies that a tensile crack at the macro- scale could start from the center of the disk, but experimental observations suggest that the crack starts from the areas near the loading plates.

Recently, fracture mechanics has been applied to the tensile test of concrete (Akita, 1998). It is realized that the fracture process zone is nucleated prior to final failure in the direct tensile test. So, observation of the fracture process in the Brazilian test could provide a new insight for cracking mechanisms at the meso-scale in concrete.

207

(2) Experiments Cylindrical specimens of 150 mm diameter and 100 mm height were made. The compres-

sive strength of concrete at 28 days under the standard curing was 37.2 MPa and the velocity of P wave was 4410 m/s. AE measurement was performed by employing AEWin SAMOS (PAC). Eight sensors of 150 kHz resonance (R15I-AST) were attached to the specimen. The frequency range was 10 kHz to 2 MHz and total amplification was 60 dB gain. Three specimens were tested to confirm the reproducibility of results.

Prior to the test, the sensor array was determined by a simulation analysis. As shown in Fig. 2, AE sources are assumed at lattice points with 5 mm grid at five cross-sections, C1 to C5, at 5 mm apart. The velocity of P wave was set to 4000 m/s and arrival times were estimated with 1 µs sampling. It was found that computational errors of the source locations were the minimum within 1.6 mm in the case of the sensor array shown in Fig. 3 and their coordinates given in Ta-ble 1.

Table 1 Optimal sensor array determined.

Fig. 2 Source locations assumed in a simulation analysis.

(3) Results and discussion

Because all results of the SiGMA analysis are similar among three specimens, those of two specimens A and B are shown in Figs. 4 and 5. At the beginning (1st) stage, AE sources are

Sensor coordinates x (m) y (m) z (m) Channel 1 0.065 0.025 0.113 Channel 2 0.065 0.075 0.038 Channel 3 -0.065 0.075 0.113 Channel 4 -0.065 0.025 0.038 Channel 5 -0.040 0.000 0.120 Channel 6 0.040 0.000 0.030 Channel 7 0.040 0.100 0.120 Channel 8 -0.040 0.100 0.030

208

Fig. 3 Cylindrical specimen and AE sensor array for Brazilian test (unit: mm).

Fig. 4 Results of SiGMA analysis in the Brazilian test (Specimen A).

observed only near two loading plates in the diametral direction. Although a few events are identified, dominant cracking-modes are tensile and mixed mode. This suggests that tensile cracks in the meso-scale are generated due to contact between the loading plates and the cylin-drical specimen under compressive loading in the diagonal direction. In all three specimens, it was demonstrated that AE sources were observed near the loading plates at the 1st stage.

At the 2nd stage, AE sources of all kinds are distributed near the final tensile failure-plane.

In the elevation view, AE sources are concentrated near the failure plane at the macro-scale, while they are distributed widely at the cross-section in the side view. It is noted that no visual

209

Fig. 5 Results of SiGMA analysis in the Brazilian test (Specimen B).

cracks were found at this stage in the test. In total, all events are plotted and compared with the macro-scale tensile cracks finally observed at the surface. Agreement between AE clusters and the final failure surfaces is reasonable. At the macro-scale, tensile-type cracks are observed, while all kinds of tensile, mixed-mode and shear cracks are identified at the meso-scale. These results suggest that the fracture process zone is nucleated around the diagonal cross-section, and these meso-scale cracks are coalesced to create macro-scale tensile cracks. This fact is in good agreement with detailed observation in the direct tensile test (Akita, 1998). Thus, in the both Brazilian test and direct-tensile test, the fracture process zone is, in advance, nucleated around the final failure surface. Coalescing process of meso-scale cracks identified by the SiGMA analysis results in the fact that the tensile strengths estimated by both tests become comparable. Accelerated Corrosion Test of Reinforced Concrete (1) Corrosion process

In concrete structures, reinforcing steel-bars (rebars) normally do not corrode because of a passive film nucleated on the surface of rebar in concrete of high pH. Corrosion of embedded rebars arises when a reinforced concrete structure is located in marine or salt environment. When the chloride concentration at rebar exceeds the critical value (Peterson, 1992), a passive film on the surface of rebar is destroyed and the corrosion is started. The electrochemical reac-tion continues with a supply of oxygen and water. Due to the expansion of corrosion products, corrosion-induced cracks are generated in concrete. In order to avoid these harmful cracks, an early warning by nondestructive evaluation (NDE) is desirable. It was reported that by applying AE technique, micro-cracks were readily detected (Yoon et al., 2000).

According to a phenomenological model of reinforcement corrosion in marine environments (Melchers and Li, 2006), a typical corrosion loss is illustrated as shown in Fig. 6. At stage 1,

210

Fig. 6 Typical corrosion loss for steel in seawater immersion (Melchers and Li, 2006). the corrosion is initiated. The rate of the corrosion process is controlled by the rate of transport of oxygen. As the corrosion products build up on the corroding surface of rebar, the flow of oxygen is eventually inhibited and the rate of the corrosion loss decreases at stage 2. The cor-rosion process involves further corrosion loss as stages 3 and 4 due to anaerobic corrosion. Thus, two-step corrosion losses are modeled. As physical phenomena, the periods of the onset of corrosion in reinforcement and that of the nucleation of cracking in concrete are to be identi-fied for inspection. So far, these stages are normally defined by the chloride threshold (Ny-gaard and Geiker, 2005). This is because NDE techniques presently available are marginally successful for identifying the two periods.

Recently, we have demonstrated that high AE activities are observed twice during the corro-

sion process (Ohtsu and Tomoda, 2008). It is found that a curve of total AE hits (counts) is in remarkable agreement with the curve shown in Fig. 6. Thus, in reinforced concrete, the first high AE activity at stage 1 reasonably corresponds to the onset of corrosion in reinforcement. During stages 3 and 4, corrosion-induced cracks in concrete could be generated due to expansion of corrosion products in reinforced concrete, and the second high AE activity is observed.

Based on these findings, continuous AE measurement was conducted to monitor the corro-

sion process in a reinforced concrete specimen in laboratory. During cyclic wet-dry tests, the SiGMA analysis is applied, and kinematics of corrosion-induced cracks in concrete are investi-gated.

(2) Experiment

For the accelerated corrosion test, reinforced concrete specimens of dimensions 100 mm × 75 mm × 400 mm were made. One deformed rebar of 13-mm diameter was embedded with 20 mm cover-thickness from concrete surface. Configuration of the specimen is illustrated in Fig. 7. The rebar was coated by epoxy except for a target area. NaCl solution was employed as mixed-water. After the standard curing, chloride content was measured and found to be 0.175 kg/m3 in concrete volume. The compressive strength of concrete at 28 days of the standard curing was 43.9 MPa, and the velocity of P wave was 4330 m/s.

To simulate the corrosion process in a typical seawater environment, a cyclic wet-dry test was carried out. The specimens were submerged into 3%-NaCl solution up to the height of the

211

rebar in the container for a week, and subsequently taken out of the solution to dry under ambient temperature for another week. In one specimen, AE measurement was continuously conducted, by using AE measurement system (DiSP, PAC) and six AE sensors (R15, PAC). The sensor array is also shown in Fig. 7. These six sensors were arranged so that the target area was rea-sonably covered, because the simulation analysis was difficult in reinforced concrete. The fre-quency range of the measurement was also10 kHz to 2 MHz and total amplification was 60 dB gain. For event counting, the dead-time was set to 2 ms and the threshold level was set to 40 dBAE.

Fig. 7 Reinforced concrete specimen and AE sensor array for corrosion test.

Fig. 8 Distribution of ferrous ions on rebar surface. (3) Results and discussion

During the cyclic wet-dry test, a numerical simulation predicted that chloride concentration at rebar reached the lower-bound threshold for corrosion of 0.3 kg/m3 after 28 days. At ap-proximately 70 days elapsed, the chloride concentration at rebar became higher than 1.2 kg/m3, which is known as the nominal corrosion-trigger level of chloride concentration. Accordingly, at 28 days elapsed and 70 days elapsed, rebars were removed from the specimens and inspected by scanning electron microscopy (SEM).

Figure 8 shows SEM photographs; At left, no corrosion is identified after 28 days from the

cross-sectional view, although some exfoliation of the oxide film is observed at the surface. At 28 days, only the surface of rebar is slightly corroded due to penetration of chloride ions. At 70

212

days (shown at right), rust and loss of the oxide film are clearly observed. The corrosion started to penetrate inside rebar, nucleating corrosion-induced cracks in concrete due to expansion of corrosion products. Thus, the growth of corrosion products is confirmed after 70 days.

In the SiGMA analysis, AE event definition time (EDT) is set to 100 µs. EDT is applied to recognize AE waves occurring within the specified time from the first-hit wave and to classify them as part of the current event. Results of the SiGMA analysis at stage 1 and at stage 2 are shown in Fig. 9. At the stage 1 (at 28 days elapsed), only 6 AE events are determined. These events are located mostly near the top of the specimen, as it is realized that these events are not directly related to the onset of corrosion. This is because only large AE sources can be ana-lyzed in the SiGMA analysis. Some shrinkage cracks might be responsible for these events in an early age of concrete.

Fig. 9 Results of SiGMA analysis in reinforced concrete beam.

At stage 2 of 70 days elapsed, 49 AE events are analyzed by the SiGMA analysis. These

events are located, surrounding the rebar, especially at the left portion of the specimen. Be-cause expansion of corrosion products is suggested and generation of corrosion-induced cracks in concrete is expected, these results suggest that the final crack could progress mostly at the left portion of the specimen. In addition, some cracks (AE sources) are located from the rebar

213

toward the bottom of the specimen, of which AE sources are mostly classified into tensile cracks and mixed-mode cracks. These results clearly suggest that corrosion-induced cracks extend outward from the rebar. Shear cracks were dominantly observed in the later part of the stage. Thus, the mechanisms in the corrosion process are indicated as follow: tensile and mixed-mode cracks start to be first generated around the rebar. Next, coalescing and connecting of these cracks lead to the nucleation of shear cracks. This finding is useful for early warning of corro-sion damage in reinforced concrete structures. Conclusions

The SiGMA analysis is applied to the Brazilian test of concrete cylinders, and corro-sion-induced cracks in a reinforced concrete beam during the cyclic wet-dry test. The follow-ing conclusions are derived. (1) A relation between macro-scale tensile failure and nucleation of AE sources in the meso-scale is clarified in the Brazilian test for the tensile strength of concrete. In the macro-scale, tensile-type cracks are only observed, while all kinds of tensile, mixed-mode and shear cracks are identified at the meso-scale. During propagation of tensile cracks at the macro-scale, other types of AE sources of mixed-mode and shear cracks were actively identified. Thus, nucleation of the fracture process zone is confirmed around the final failure surface. This is a reason why the tensile strengths estimated by the Brazilian test are comparable to those by the direct-tensile test, although stress distributions are quite different. (2) It is found that high AE activities are observed twice during the corrosion process, which correspond to the onset of corrosion in rebar and the nucleation of cracking in concrete due to expansion of corrosion products. In relation to the 2nd AE activity, the generating mechanisms of corrosion-induced cracks are studied in reinforced concrete. Concerning the mechanisms in the corrosion process, it can be summarized that tensile and mixed-mode cracks start to be gen-erated around the rebar, and coalescing and connecting these cracks, shear cracks are nucleated at the meso-scale. This implies that the SiGMA analysis is useful for early warning of corro-sion damage in reinforced concrete structures. Acknowledgement

The research conducted was supported by Kumamoto University Global COE (Center of Ex-cellence) Program: Global Initiative Center for Pulsed Power Engineering. To perform experi-ments and analyses, the assistance of technical associate, Dr. Yuichi Tomoda was valuable. The authors wish to deeply thank the program and his support. References Akita, H., Koide, H. and Tomon, M. (1998), ”Uniaxial Tensile Test of Unnotched Specimens under Correcting Flexure,” Fracture Mechanics of Concrete Structures, Aedificatio Publishers, Freiburg, Vol. I, 367- 377.

Carpinteri, A., Lacidogna, G. and Manuello, A. (2007), ”An Experimental Study on Retrofitted Fiber-Reinforced Concrete Beams using AE,” Fracture Mechanics of Concrete and Concrete Structures, Taylor & Francis, London, Vol. 2, 1061-1068.

Erdogan, F. and Sih, G. C. (1963), ”On the Crack Extension in Plates under Plane Loading and Transverse Shear,” J. Basic Eng., 12, 519-527.

214

Eshelby, J. D. (1973), ”Dislocation Theory for Geophysical Applications,” Phil. Trans. R. Soc. London, A274, 331-338.

Grosse, C. U. and Ohtsu, M. (2008), Acoustic Emission Testing, Springer-Verlag, Berlin.

Knopoff, L. and Randall, M. J. (1970), ”The Compensated Liner-Vector Dipole,” J. Geoohys. Res., 75(26), 4957-4963.

Melchers, R. E. and Li, C. Q., (2006) Phenomenological Modeling of Reinforcement Corrosion in Marine Environments, ACI Materials Journal, 103(1), 25-32.

Mura, T. (1982), Micromechanics of Defects in Solids, Martinus Nijhoff Pub. The Hague.

Nygaard, P. V. and Geiker, M. R. (2005), ”A Method for Measuring the Chloride Threshold Level required to initiate Reinforcement Corrosion in Concrete,” Materials and Structures, 38, 489-494.

Ohno, K. and Ohtsu, M. (2007), ”Cracking Mechanisms of Diagonal-Shear Failure monitored band identified by AE-SiGMA Analysis,” Fracture Mechanics of Concrete and Concrete Struc-tures, Taylor & Francis, London, Vol. 2, 991-998.

Ohno, K. and Ohtsu, M. (2010), ”Crcak Classification in Concrete based on AE,” Construction and Building Materials, 24(12), 2339-2346.

Ohtsu, M. and Ono, K. (1984), ”A Generalized Theory of AE and Green’s Functions in a Half Space,” Journal of AE, 3(1), 124-133.

Ohtsu, M. (1991), ”Simplified Moment Tensor Analysis and Unified Decomposition of AE Source,” J. Geophys. Res., 96(B4), 6211-6221.

Ohtsu M, Okamoto T, Yuyama S. (1998), ”Moment Tensor Analysis of AE for Cracking Mechanisms in Concrete,” ACI Structural Journal, 95(2), 87-95.

Ohtsu, M. and Tomoda, Y. (2008) Phenomenological Model of Corrosion Process in Reinforced Concrete identified by AE, ACI Materials Journal, 105 (2), 194-199.

Ohtsu, M. and Uddin, F. A. K. M. (2008),”Mechanisms of Corrosion-Induced Cracks in Con-crete at Meso- and Macro-Scales,” Journal of ACT, JCI, 6(3), 419-429.

Timoshenko, S. P. and Goodier, J. N. (1970), Theory of Elasticity, McGraw-Hill Book Company, New York.

Peterson K. (1992),”Corrosion Threshold Value and Corrosion Rate in Reinforced Concrete,” Swedish Cement and Concrete Research Institute, CIB report 2.

Sause, M. and Horn, S. (2010), ”Influence of Specimen Geometry on AE Signals in Fiber Rein-forced Composite: FEM Simulations and Experiments,” Proc. The 29th European Conf. on AE Testing, [CD-ROM].

Yoon D. J., Weiss W. J. and Shah S. P. (2000), ”Assessing Damage in Corroded Reinforced Concrete using AE,” J. Engineering Mechanics, 126(3), 273-283.

J. Acoustic Emission, 28 (2010) 215 © 2010 Acoustic Emission Group

ACOUSTIC EMISSION INSPECTION OF RAIL WHEELS

KONSTANTINOS BOLLAS, DIMITRIOS PAPASALOUROS, DIMITRIOS KOUROUSIS and ATHANASIOS ANASTASOPOULOS

Envirocoustics SA, El. Venizelou 7 & Delfon, 14452 Metamorphosis, Athens, Greece Abstract

The ever-increasing demand for safer and faster surface transportation such as railway im-poses many challenges for the inspection of critical components such as train axes and wheels. These components, mostly during usage, are subjected to highly complex dynamic loads and ac-tive defects may result in catastrophic failures, possibly with human casualties. Within the scopes of an R&D project, aiming to develop novel methodologies and techniques for the inspec-tion of wheel sets, extensive acoustic emission (AE) measurements have been performed on various trains and trams. AE sensors were mounted on the rail in order to diagnose wheel prob-lems by monitoring the AE transferred through the rail in real-time and while the vehicles were moving. The purpose of the trials was to investigate the usage of AE for on-line detection of de-fects on wheels such as flats, bearing failures and possibly significant cracks, and to establish optimum setup parameters in this respect. The present paper presents the raw data and evalua-tion results from AE experiments on train and tram wheels, (both healthy ones and wheels hav-ing known defects). During measurements different AE sensors were placed on the side of the rails while the railcars or trams were passing at different speeds. The effect of sensor frequencies and placement were investigated. Multiple AE datasets, i.e., Time Driven Data (TDD), Hit Driven Data (HDD) as well as long (>10 sec.) waveforms were acquired simultaneously. Data analysis involved traditional AE features, source location and digital signal processing of ac-quired waveforms. Initial results presented highlight the different AE behavior for defective and non-defective wheels, and indicate clearly the potential of AE as diagnostic tool. Furthermore, results show that the availability of acquired long, continuous waveforms significantly enhanced analysis capabilities, when combined with advanced AE DSP software and pattern recognition analysis. Keywords: Train wheels inspection, rail axes monitoring, long waveforms Introduction

Inspection of rail wheels poses inherent difficulties due to accessibility issues, when the wheels are mounted on the train, but also due to the complex geometry of the wheels, their shape and their attachments. As a result, current inspection methodologies are mainly based on dis-mounting the wheels and inspecting them off-vehicle, or, in the best case, on an immobile train, either on a periodic or on a need basis (e.g., see [1]), by means of localized NDT (MT, PT, UT, EC, VT etc.). Still, however, failures occur, that occasionally lead to catastrophic accidents. The need to identify wheel flaws at early stages and in a more efficient way, reducing maintenance costs, has steered research efforts towards on-line wheel inspection techniques, such as vibration and AE. The aim of such techniques is to be able to identify flawed train wheels, while the train is in-service (moving) and the “screen out” of bad wheels for further inspection and mainte-nance, in a fast, effective way.

In this respect, AE [2] method has been identified as a very promising tool. Acoustic emis-

sion uses high-frequency, passive piezoelectric transducers, which are mounted on the side of the

216

rails, in order to detect the stress waves generated by the various possible flaws of the wheel, for instance, flats, or other circularity defects, which are impacting on the rail while the wheel is turning, or crack opening, or even faulty bearings from the axes. Since other AE sources are pre-sent, even in a healthy train (engine noise, frictional noise, banging noise, etc.) the technique, at a first stage, usually involves acquiring the AE “signature” of a vehicle with healthy, “reference” wheels, and comparing this to the one of a vehicle with defective wheels. The method offers sig-nificant advantages compared to other passive methods, mainly due to the high sensitivity it of-fers, as well as the fact that the AE transducers operate within high-frequency range (30-1000bkHz), eliminating to a large extent, mechanical noise. Once calibrated and standardized, the method may be employed very effectively, for instance by mounting sensors on all rails of a depot and “listening” to the trains or trams as they exit for their shift. Each wheel is “inspected” as it passes from the point of the rail where the AE sensor is located, thus, within a single day, all wheels in usage may be inspected.

Acoustic emission has already been applied for the detection of rail track faults at laboratory

level [3, 4]. However, only a limited number of field experiments have been carried out [5, 6]. The present research has received funding from the European Community's Seventh Frame-

work Programme (FP7/2007-2013) under grant agreement no 218674 of SAFERAIL project [7]. The SAFERAIL consortium seeks to minimize wheel set failures by developing and successfully implementing a novel on-line system for the inspection of wheels and axles of moving trains, and a combined ultrasonic-electromagnetic system for faster and more reliable inspection of the qual-ity of new and old wheel sets during their production and maintenance. The present work studies the application of AE for the on-line inspection of rail wheels, investigating various hardware parameters such as sensor frequency, test parameters such as type and velocity of vehicles, as well as data acquisition and analysis strategies. Results show the ability of the technique to detect wheel flaws, while the methodology is still under optimization in terms of resources used, and analysis time.

Fig. 1 – 56 tons ALLAN railcar with distances between wheels.

217

Instrumentation

All presented data has been acquired using systems based on the Physical Acoustics Corp. PCI-2 AE boards [8], which have the capability of acquiring AE data with long (>10 s) wave-form streaming directly to hard drive. Four different types of AE sensors have been investigated: Physical Acoustics’ R3i (30 kHz resonant), R15 (150 kHz resonant), R30 (300 kHz resonant) and R50 (500 kHz resonant). Data was acquired by AE-Win real-time data acquisition software [9], while advanced data analysis has been performed using NOESIS, a specialized AE data analysis and pattern recognition software [10].

For the scopes of this work three different data sets of field trials are presented, which were

acquired at three different sites, two involving trains and one involving trams. AE equipment has simultaneously acquired TDD, HDD and waveform streams (WFS). The importance of the avail-ability of each individual data set and its significance in data interpretation are clearly demon-strated. Additionally, advanced and versatile software tools were utilized in order to fully ana-lyze multiple aspects of the acquired data. Case Study - AE Monitoring of Train Wheelsets in EMEF Depot, Oporto, Portugal Experimental Setup and Wave Propagation Study

A 56-tons ALLAN railcar (Fig. 1) was used for the AE measurements and was provided by EMEF. On all experimental setups the sensors were placed on the external rear side of the rail, using silicon grease and supported by magnetic clamp, as shown in Fig. 2a. Pencil-lead breaks (Hsu-Nielsen sources) were performed on the rail track for attenuation and wave velocity estima-tion (Fig. 2b).

(a) (b)

Fig. 2 – a) Sensor placement, b) Hsu-Nielsen source positions for wave propagation study.

Sensors were installed in ways that could lead to location detection of possible wheel defects during train passing along the rail. On the 3&1 setup, three sensors were placed on one rail on a distance equal the half of the wheel perimeter and a fourth sensor was placed on the other rail opposite to sensor No. 1 position (Fig. 3). The same setup was maintained for all sensor types, e.g. 3&1 setup for R15s, the same 3&1 setup for R30s etc. in order to be to able to compare di-rectly the sensitivity differences of each sensor type.

In order to be able to have location evidence of possible wheel defect, different location set-

tings were investigated, using the calculated wave velocities and placing the three (3) sensors at a

218

time within known distances from each other. For the specific setup the distance between sensors was 1.8 m. Pencil-lead breaks (PLB) were performed on distances approximately 0.6 m, 1.2 m, 1.8 m and 2.4 m from sensor No. 1. Figure 4 shows the results of a location check. Y-axis shows the sum of AE events caused by the PLB while X-axis shows the sensor position using the first sensor as reference. It is evident from Fig. 4 that the specific location settings and/or setup are adequate to locate four distinct pencil breaks in a 3.6 m span with high accuracy. Therefore, any possible wheel surface defects are expected to yield similar results.

Fig. 3 – 3 & 1 Sensor setup (3 on one side and 1 on the opposite side).

Fig. 4 - Location check graph showing the real position of the PLB and the position as located by the AE system.

Figure 5 shows the arrival times and amplitudes of a PLB signal performed at 0.6 m from sensor No. 1, as acquired from three sensors located on the same rail track at 1.8-m distance from each other. The corresponding waveform as received for the sensor No. 1 at reference position is shown in Fig. 6, while Fig. 7 shows at sensor No. 2 at 1.8-m distance from sensor No. 1 position and 1.2 m distance from the PLB source. From the sensor No. 3 at 3.6-m distance from sensor No. 1 and 1.8-m distance from the PLB source, waveform in Fig.8 was observed. Note sensor numbers correspond to channel numbers in Fig. 5, which shows 7 dB attenuation over 0.6-m dis-tance from No. 1 to No. 2 and 13 dB over 2.4 m from No. 1 to No. 3.

219

Fig. 5 - Time arrivals and amplitudes of a PLB source at 0.6 m from sensor No. 1 position.

Fig. 6 – Signal amplitude in V vs. time from sensor No. 1.

Fig. 7 – Signal amplitude in V vs time from sensor No. 2.

Fig. 8 - Signal amplitude in V vs time from sensor No. 3.

AE Monitoring of Train Wheelsets

AE acquisition measurements were performed during train passes at the area where the sen-sors were installed. Train passes were made with different speeds from 5 to 40 km/h as reported by the train driver and in different directions (to left or to right). In all cases the train speed was kept constant. Three parameters were selected as most probable either to produce and/or change the AE activity during the test runs. These are the actual speed of the train, the geometry of the railcar and the actual surface condition of the wheels.

220

AE data was analyzed using special pattern recognition software (NOESIS v.5.5). AE fea-tures like signal amplitude, duration, energy, average signal level, etc. of acquired AE signals were compared using the different resulting data sets from:

Same Railcars - Different Railcar Speeds, Different Railcars - Similar Railcar Speeds. Railcar before Defect Generation – Same Railcar after Defect Generation

All comparisons were performed using data from same train direction.

Specifically for the testing of the surface condition parameter, flat surfaces were generated (Fig. 9) on the defect-free wheels by applying the emergency brakes multiple times. The AE af-ter each emergency brake was acquired enabling the comparison of the damage accumulation in different states of the flat surface development. An example of raw waveform streams of 10-s duration after seven emergency brakes is shown in Fig. 10. Here, 3 & 1 setup was used, and the side with three sensors shows slightly lower amplitude. Note increasing signal attenuation in channel 1, 2 and 3. Since the surface of each wheel during emergency braking is accumulating damage in the form of either newly created flat surfaces or enlargement of the existing ones, each wheel can be characterized by a “defect status” directly dependent on the number of emer-gency brakes. The defect status is shown in Fig. 11 as AE events location graphs, representative of AE signals density for 40-km/h railcar travel.

Fig. 9 - Flat defect on a wheel surface created by emergency breaking of the train. Courtesy of EMEF.

The same data analysis was applied on location capabilities of hit-driven AE data (HDD) for

all datasets. In cases where location was applied, more events appeared during defective railcar movement. As can be observed in Fig. 12, at low (5 km/h) and medium (20 km/h) speed the number of located AE events is higher for the defective railcar. In contrast, at higher speed (40 km/h) less AE events can be located.

221

Fig. 10 –Waveform stream of 10-s duration with 3&1 setup using 4 R50 sensors after seven emergency brakes. Amplitude full-scale is 10.2V. Case Study - AE Monitoring of Tram Wheelsets, Antwerp, Belgium Experimental Setup

AE acquisition was performed on four different trams of different types (see an example on Fig. 13a) and defect status (two with defective wheel sets and two with non-defective ones). AE was acquired during 22 tram passes at varying speeds, in both directions. Four different types of AE sensors were used (PAC R3i, R30, R15, R50). Due to the difficulty of accessing the rear part of the rail, no location groups could be setup and no magnetic clamps could be used for placing the sensors. All sensors were attached on the rail using special adhesive grease (see Fig. 13b). Depending on the different resonant frequencies of the sensors, measurements using various low- and high-pass frequency filters were performed.

222

Fig. 11 - AE events location graphs of AE signals density for 40km/h railcar speed with different defects statuses.

Fig. 12 - AE events location graphs of AE signals density for different railcar speeds of 5, 20 and 40 km/h, before any emergency brake (clear wheelsets) and after 7 emergency breaks (defective wheelsets).

223

Fig. 13 – (a) Photo of a PCC vehicle that was used for the AE measurements, (b) sensor position. Various types of analyses have been performed. These include comparison between data of defective and non-defective railcars, repeatability check of data acquired with same conditions (same railcar and railcar speed, same railcar direction, etc.), attempt to separate signals of differ-ent wheels by long-duration waveform analysis, filtering effectiveness check on acquired data. However, the analysis of this particular set of data (tram measurements) focused mainly on long-duration waveform processing with advanced DSP methods.

Prior to experimental measurements, signal attenuation was calculated across the rail. Plac-ing the sensors on a steady point on the rail and using three different AE sources (Hsu-Nielsen, and small/large metallic spherical impactors) across the rail, attenuation measurements were per-formed and graphs were created (Fig. 14).

Fig. 14 - Attenuation graph showing signal amplitude (in dB) vs. distance from sensor, for four different types of sensors (R3i, R15, R30, R50), (a) using a Hsu-Nielsen source (PLB), (b) using a small metallic spherical impactor.

R50 R30

R15 R3i

224

Case Study - AE Monitoring of Tram Wheelsets Analysis of long-duration waveforms either by visual observation or using DSP tools re-

vealed the potential that continuous AE data offers. Observation of the overall signal acquired during passing of a wheel above the sensor position allows the determination of the effective monitoring time, i.e. the period of time during which the train creates detectable AE signals, thus, allowing data collection period determination for future tests (Fig. 15). Focusing on the waveform allows the user to observe peaks in the voltage signal, which are generated when each axis passes right on top of the sensor. Furthermore, the existence of flats may also be demon-strated as voltage peaks, thus, zooming on such, and performing frequency analysis on specific sections, it is possible to observe the frequency contents of specific sources, enhancing signal interpretation. Finally, RMS or ASL processing, which basically offer a “smoother” or averaged representation of the signal, allows the user to conveniently perform time-based calculations, such as observe the time difference of the axes and bogies as they pass from each sensor and al-low macroscopic observation of signal level differences, which may be periodic, indicating tran-sient/banging AE from flats, or continuous, indicating possible bearing problems.

Fig. 15 - AE waveform (top) and RMS graphs (bottom) acquired by a PAC R50 sensor during passes of a railcar (tram) with 3 bogies (6 axes) and a flat on the first axis.

AE acquisition measurements were performed during 22 tram passes at the sensor installation point (see e.g., Fig. 16). Trams were passing with different speeds from 15 to 40 km/h, as re-ported by the train drivers, and different directions (back or front). In all cases the tram speed was kept constant, while passing on the sensor position. Data was analyzed using pattern-recognition software (NOESIS v.5.5). AE features like signal amplitude, duration, energy, aver-age signal level, etc. of acquired AE signals were compared using the different data sets from:

Same Trams - Different Trams Speeds, Different Trams - Similar Trams Speeds. Tram before Defect Generation – Same Tram after Defect Generation

It was expected that tram passes with same characteristics (same tram, same speed, same direc-tion) would result in similar signals and waveforms.

Following the above measurements and preliminary analysis of the acquired data, observa-tions made on the conditions of the experiments produced recommendations below:

225

(1) Location groups have to be setup by placing AE sensors on known distances along the rail. The reporting of vehicle speed has to be done exactly. Signals acquired with similar but not equal speeds may result in inconsistent results. (2) As noted on the last experimentation in Oporto Portugal, the defects have to be created on the wheels surface by controlled damage and their number, size and position on each wheel has to be known on-site. They also have to be characterized as “accepted” or “not accepted” according to international or local (train company) standards. Their position on each wheel has to be known, in order to be able to calculate exactly the position of the contact point between the flat defect and the rail, for location purposes. (3) A trigger, in combination with known train speed, has to be used on the AE system, in order to be able to synchronize the train position with the acquired data.

(a)

(b) Fig. 16 - AE waveforms acquired by (a) R15 sensor and (b) R30 sensor during passing of the same defective tram at 40 km/h. Display duration: 4 s.

Fig. 17 - Sensor setup on the right rail track.

AE Monitoring of Train Wheel sets at VTG Site in Long Marston UK

Previous results have demonstrated the potential of location analysis for feature-based AE data, as well as long-duration AE waveform analysis. It is concluded that geometric wheel

226

anomalies appear in the time domain as specific peaks of the voltage signal. In order to fine-tune the detection capabilities of AE to be used for the real time monitoring a 4&2 (4 sensors on 1 track an 2 on the opposite) setup was used. Two R30a sensors and two R15i sensors were mounted on the same track at 1/3 distance from the tracks ends (Fig. 17). Similarly on the opposite track two R15i were mounted in order to acquire any AE transients that may propagate through various wave paths. The use high-frequency sensors are deemed necessary in a noisy environment. The acquired signals are less prone to active mechanical noise sources (engine, etc) that are usually found at lower frequencies of the spectrum and contaminate or even mask useful AE data. The results presented here are based on raw data acquired with high-density TDD while full analysis is ongoing. Using TDD with a very high sampling rate, a near-continuous acquisition of time-based AE Features can be achieved (ASL, RMS, Absolute Energy, etc.). In order to correlate the AE activity generated with the active defects, two different flat surfaces that spanned 2.5 cm and 5 cm on the wheels circumference were generated with an angle grinder (Fig. 18).

(a) (b)

Fig. 18 - (a) 2.5-cm flat surface on wheel, (b) 5-cm flat surface on wheel.

The train’s configuration was a locomotive and two attached wagons with identical dimen-sions. The first wagon after the locomotive was used as a reference wagon, since its wheel sets had no defects. AE activity was acquired during forward (left-to-right) and backwards (right-to-left) movement. Different data sets were acquired for different speed levels for both directions.

As shown in Fig. 19, both defects are clearly depicted as periodic spikes in the ASL vs. time graphs. Such spikes are completely missing from the reference (healthy) wagon. Additionally, time measurements between the edges of the spikes have shown that these occur at a frequency coinciding with the wheel rotation frequency. Conclusions Work on AE monitoring of train wheelsets on moving trains and trams using AE sensors mounted on the rails has verified the potential of the method in detecting geometric defects, such as flat surfaces on wheel circumference. To achieve optimum results, multi-dimensional analysis has been performed, combining TDD, HDD and long-duration waveform streams acquired si-multaneously. Combined analysis with specialized software is required to reveal the full poten-tial of all available data in the time and frequency domain. Placing more than one AE sensor on the rail track and setting up a location group can show existence of flats on lower railcar speeds, but for exact location of the defect further work is required.

227

Fig. 19 - (a) Reference train movement, (b) 2.5-cm flat defect surface, (c) 5-cm flat defect sur-face. All top graphs: Channel 5 ASL vs. time, bottom: Channel 6 ASL vs. time.

228

Acknowledgment The research leading to these results has received funding from the European Community's Seventh Framework Programme (FP7/2007-2013) under grant agreement no 218674, SAFERAIL project (http://www.saferail.net/). The authors would like to thank all Project Part-ners and gratefully acknowledge the contribution of Vlaamse Vervoersmaatschappij De Lijn, Belgium, VTG Rail, UK and EMEF SA, Portugal for providing access and resources for the tests. References 1. Norme Européenne CEN/TC 256, prEN 15313:2007: Railway applications — In-service wheelset operation requirements and in-service and off-vehicle wheelset maintenance.

2. ASTM Standard E1316, 2005, “Standard Terminology for Nondestructive Examinations”, ASTM International.

3. Bruzelius K., Mba D., “An initial investigation on the potential applicability of Acoustic Emission to rail track fault detection”, NDT&E International 37 (2004) 507–516.

4. Thakkar N.A., Steel J.A., Reuben R.L., “Rail–wheel interaction monitoring using Acoustic Emission: A laboratory study of normal rolling signals with natural rail defects”, Mechanical Systems and Signal Processing, 24 (2010) 256–266.

5. Thakkar N.A., Steel J.A., Reuben R.L., Knabe G., Dixon D., Shanks R.L., “Monitoring of Rail-Wheel Interaction Using Acoustic Emission (AE)”, Advanced Materials Research Vols. 13-14 (2006) 161-168.

6. Private Communication with MISTRAS GROUP SA – EPA, 27 Rue Magellan ZAC des Portes de Sucy, F 94470 Sucy-en-Brie, France.

7. “SAFERAIL - Development of Novel Inspection Systems for Railway Wheelsets”, by Anastassopoulos, A., Barros, P., Bey, F., Blakeley, B., Boynard, C. Braz, R, Courinha, J., Davis, C., Day, N., Dias, D., De Donder, E., Kerkyras, S., Kerkyras, Y., Laeremans, M., Liaptsis, D., Lugg, M., Lugg, P., Nicholson, P.I., Nobre, T., Papaelias, M., Pinto, M., Roberts, C., Vanho-nacker, T., Vermeulen, F., Presented at CM 2010 and MFPT 2010, The Seventh International Conference on Condition Monitoring and Machinery Failure Prevention Technologies, 22-24 June 2010, Ettington Chase, Stratford-upon-Avon, England.

8. PAC’s site: http://www.pacndt.com/index.aspx?go=products&focus=/multichannel/pci2.htm

9. Physical Acoustics Corporation, AEwinTM Software, Installation, Operation and User’s Refer-ence Manual, rev. 2. Princeton Junction, New Jersey, USA.

10. NOESIS Ver. 5.2, Envirocoustics, User’s Manual.

J. Acoustic Emission, 28 (2010) 229 © 2010 Acoustic Emission Group

E07.04 – OVERVIEW OF CURRENT AND DEVELOPING ASTM ACOUS-TIC EMISSION (AE) STANDARDS

MARK F. CARLOS

Chairman, ASTM E07.04 AE Sub-Committee and Physical Acoustics Corporation, 195 Clarksville Rd., Princeton Jct., NJ 08550-5303

Introduction:

The ASTM International, E07.04, Acoustic Emission Subcommittee, is responsible for the formulation, development (writing), standardization and maintenance of test methods, terminol-ogy, practices and guides, related to acoustic emission (AE). Currently, there are about 50 mem-bers in the subcommittee. E07.04 is a Subcommittee under the larger ASTM E07 “Committee on Nondestructive Testing”, a large organization of over 450 worldwide members, who are dedi-cated to developing and writing worldwide NDT consensus standards in Radiography, Digital Radiography, Penetrant Testing, Ultrasonic Testing, Electromagnetic Testing, Leak Testing, NDT Agencies, Emerging NDT Methods and of course, Acoustic Emission. All NDT standard documents developed and maintained by the E07 Committee are published in Volume 03.03 Nondestructive Testing, Annual Book of Standards. Currently there are 24 existing AE related ASTM standards, with three more currently in development.

Four different types of standard documents are developed by the group, each one requiring a specific document format and content in order to qualify. These documents can be broken down into the following standard types.

• Test Method: A definitive procedure, which produces a test result. (e.g., material prop-erty measurement).

• Standard Practice: A definitive set of instructions for performing one or more specific operations that does not produce a test result.

• Guide: A compendium of information or series of options that does not recommend a specific course of action. A tutorial.

• Terminology: A document comprising definitions of terms; explanations of symbols, abbreviations or acronyms.

The goal of this article is to provide an overview of the existing AE standard documents,

their use and importance in industry, an overview of new standard documents in process, and a call for support and participation in the development of new and relevant standard documents related to AE examination. AE Documents Under Jurisdiction in E07.04:

Below is a list of the existing AE standard document numbers and titles that can be found in Volume 03.03 Nondestructive Testing, Annual Book of Standards:

E569, “Standard Practice for Acoustic Emission Monitoring of Structures During Controlled Stimulation”

E650, “Standard Guide for Mounting Piezoelectric Acoustic Emission Sensors”

E749, “Standard Practice for Acoustic Emission Monitoring During Continuous Welding”

230

E750, “Standard Practice for Characterizing Acoustic Emission Instrumentation”

E751, “Standard Practice for Acoustic Emission Monitoring During Resistance Spot-Welding”

E976, “Standard Guide for Determining the Reproducibility of Acoustic Emission Sensors Re-sponse”

E1067, “Standard Practice for AE Examination of Fiberglass Reinforced Plastic Resin (FRP)”

E1106, “Standard Method for Primary Calibration of Acoustic Emission Sensors”

E1118, “Standard Practice for Acoustic Emission Examination of Reinforced Thermosetting Resin Pipe (RTRP) ”

E1139, “Standard Practice for Continuous Monitoring of Acoustic Emission from Metal Pressure Boundaries”

E1211, “Standard Practice for Leak Detection and Location Using Surface-Mounted Acoustic Emission Sensors”

E1419, “Standard Test Method for Examination of Seamless, Gas-Filled, Pressure Vessels Using Acoustic Emission”

E1495, “Standard Guide for Acousto-Ultrasonic Assessment of Composites, Laminates, and Bonded Joints”

E1736, “Standard Practice for Acousto-Ultrasonic Assessment of Filament-Wound Pressure Vessels” E1781, “Standard Practice for Secondary Calibration of Acoustic Emission Sensors”

E1888, “Standard Test Method for Acoustic Emission Examination of Pressurized Containers Made of Fiberglass Reinforced Plastic with Balsa Wood Cores”

E1930, “Standard Test Method for Examination of Liquid Filled Atmospheric and Low Pressure Metal Storage Tanks Using Acoustic Emission”

E1932, “Standard Guide for Acoustic Emission Examination of Small parts”

E2075, “Standard Practice for Verifying the Consistency of AE-Sensor Response Using an Acrylic Rod”

E2076, “Standard Test Method for Examination of Fiberglass Reinforced Plastic Fan Blades Us-ing Acoustic Emission”

E2191, “Standard Test Method for Examination of Gas-Filled Filament-Wound Composite Pres-sure Vessels Using Acoustic Emission”

E2374, “Standard Guide for Acoustic Emission System Performance Verification”

E2478, “Standard Practice for Determining Damage-Based Design Criteria for Fiberglass Rein-forced Plastics (FRP) materials, Using Acoustic Emission”

E2598, “Standard Practice for Acoustic Emission Examination of Cast Iron Yankee and Steam Heated Paper Dryers”

In comparison to the number of standard documents in other test methods in Volume 03.03, Acoustic Emission standards have a respectable showing, indicating the interest in the technol-ogy and the continuing involvement and contributions of the active membership. A comparison table showing the number of standards being maintained by each of the main Test Methods is shown below.

231

Test Method # of Active Standards

Radiography Standards 59

Ultrasonic Standards 29

Acoustic Emission 24

Electromagnetic Standards 20

Liquid Penetrant Standards 14

The 24 AE documents published can be broken out in various ways, the most important of which would be to classify the types of documents from the user perspective as discussed below:

1. Six documents are related to sensor primary and secondary calibration, sensor and system characterization, and verification,

2. Two are tutorial documents, one associated with attaching sensors to the test article, and the other describing how to conduct a typical AE examination.

3. 14 documents are associated with actual AE examinations procedures. 4. Two documents are associated with acousto-ultrasonic applications.

Sensor and system related standards: Of the 6 documents related to sensors and systems, they include the following;

• E1106 describes the primary AE sensor calibration system located at NIST, in Washing-ton DC.

• E1781 describes a secondary AE sensor calibration standard, which parallels the NIST system for sensor developers and users. These two standards are recognized worldwide as the de facto sensor calibration standard.

• For users who need a practical method of verifying the sensor response, we have E976, which also is a worldwide de facto standard and describes various techniques of exciting an AE sensor and measuring its response in all types of situations. The Pencil-Lead Break source is described here.

• E2075 describes a method of measuring consistency of the AE sensor response utilizing an acrylic rod.

• In regards to AE systems, E750 describes various measurements that can and should be made to characterize an AE system (on the bench).

• E2374 describes a method of verifying an AE system before, during and after an AE ex-amination.

Tutorial documents: There are two excellent tutorial documents on the use of AE.

• E650 describes a method for mounting piezoelectric AE sensors for maximum perform-ance for different applications.

• E1932 is a step-by-step guide in preparing for, carrying out, analyzing and reporting on a typical AE examination on smaller parts. Small parts are emphasized in this standard guide in order to minimize complications often associated with the examination of large structures.

232

AE specific examination documents: There are 14 specific application documents, which de-scribe examinations on specific types of structural elements or components. These include the following:

• E569 describes the monitoring of structures, such as pressure vessels and pipelines, which can be stressed by mechanical or thermal means, such as raising the internal pres-sure or temperature.

• E749 describes AE examination procedures during continuous welding. • E751 describes AE examination procedures during resistance spot welding. • E1067 describes AE examination or monitoring of fiberglass-reinforced plastic (FRP)

tanks/vessels under pressure or vacuum to determine structural integrity. • E1118 describes AE examination or monitoring of reinforced thermosetting resin pipe

(RTRP) to determine structural integrity. It is applicable to lined or unlined pipe, fittings, joints and piping systems.

• E1139 describes techniques for continuous AE monitoring from metal pressure bounda-ries in industrial systems during operation. Examples include pressure vessels, piping and other system components, which serve to contain system pressure. Pressure bounda-ries other than metal, such as composites are specifically not covered by this document.

• E1211 describes the method of detecting and locating the steady state source of gas and liquid leaking from pressurized systems.

• E1419 describes AE examination of seamless pressure vessels (tubes) of the type used for distribution or storage of industrial gases. This document is used worldwide and is based on a DOT inspection exemption.

• E1888 describes AE examination techniques of pressurized containers made of fiberglass reinforced plastic (FRP) with balsa wood cores. Containers of this type are commonly used on tank trailers for the transport of hazardous chemicals.

• E1930 describes guidelines for AE examination of new and in-service above-ground storage tanks of the type used for storage of liquids.

• E2076 describes a method for AE examination of fiberglass reinforced plastic (FRP) fan blades of the type used in industrial cooling towers and heat exchangers.

• E2191 describes guidelines for AE examination of filament-wound composite pressure vessels; for example, the type used for fuel tanks in vehicles, which use natural gas.

• E2478 describes how to use AE in the design process of testing of composites samples (specimens) in order to evaluate different designs for damage tolerance.

• E2598 This practice provides guidelines for carrying out acoustic emission (AE) exami-nations of Yankee and steam heated paper dryers (SHPD) of the type to make tissue, pa-per, and paperboard products.

Acousto-ultrasonics standards: Of the 24 existing documents, two relate to acousto-ultrasonics (A-U). Acousto-ultrasonic instruments utilize existing AE instruments but instead of passively listening for generation of stress waves, artificial means are used to pulse or excite the structure, while standard AE techniques are used for detection and processing of the artificial, known sig-nals. Of the two documents,

• E1495 is a general tutorial guide to A-U, in which the rationale and basic technology of the A-U technique is fully discussed. Also, E1495 provides guidelines for NDE of flaws and physical characteristics that influence the mechanical properties and relative strength of composite structures; for example, filament-wound pressure vessels, adhesive bonds and inter-laminar and fiber/matrix bonds in man-made composites and natural composites (such as wood).

233

• E1736 describes an A-U procedure for the assessment of filament-wound pressure ves-sels.

Current AE Work Items in Process Currently there are three standards in progress as discussed below.

• WK658, “Standard Test Method for Acoustic Emission Examination of Seam-Welded High Energy Piping”, provides guidelines for on-line AE monitoring of seam-welded high energy steam piping, of the type found in fossil power plants, for detection of flaws. The AE measurements are used to detect, locate and classify emission sources under normal operational conditions. The primary intent of the examination is to identify the presence and relative severity of high-temperature creep damage to seam welds and is based on an EPRI-funded test method that was developed for the energy industry.

• WK19889: “Standard Practice for AE Examination of Plate-like and Flat Panel Com-posite Structures Used in Aerospace Applications”. This standard practice in develop-ment is based on testing of aerospace composite panels as part of a qualification process. This procedure is being developed in conjunction with the aerospace industry, which re-quires NDT standards for testing and verifying aerospace composites. There will be more documents coming out for the aerospace composites industry as we try to develop inspection standards to assure safety and structural integrity, for this relatively new mate-rial technology in that marketplace.

• WK12759: “Standard Guide for Preparing an AE Examination Plan for Plate-like and Flat Panel Aerospace composite Structures” is a companion document to WK19889, and provides guidance into developing a test plan for various composite materials and geo-metries.

Other documents are in an early formulation stage and we are always interested in receiving

new documents and suggestions. Each AE community member is involved in research and de-veloping new AE test methodology for new applications should consider authoring and submit-ting documents to ASTM for publishing. Summary

ASTM E07.04 AE Sub-Committee currently maintains twenty-four AE-based standard documents, with three additional standards in development. Our standards are used worldwide and some are actually de facto worldwide standards. These days, standard documents and ex-amination procedures are important to the growth and relevance of the AE method (and other NDT techniques, also). E07.04 is a small, close-knit group, working hard to publish meaningful AE standard documents. We welcome new members interested in assisting in the development of new AE standards. We are also interested in new techniques and are open to inputs, suggestions and assistance from all.

Anyone interested in becoming an ASTM - AE Subcommittee member is urged to go to the

ASTM webpage at www.astm.org or to contact the present author, current E07.04 Chairman, at mfcarlos@pacndt.com.

J. Acoustic Emission, 28 (2010) 234 © 2010 Acoustic Emission Group

USE OF AE METHOD FOR DETECTION OF STEEL LAMINATION IN THE INDUSTRIAL PRESSURE VESSEL

V.P. GOMERA1, V.L. SOKOLOV1, V.P. FEDOROV1, A.A. OKHOTNIKOV1

and M.S. SAYKOVA2 1 Kirishinefteorgsintez, 187110, Kirishi, Russia,

2 Central Boilers and Turbine Institute (CKTI), Saint-Petersburg, Russia Abstract

This paper discusses results of two AE tests performed of the pressure vessel within a 5-year period. The AE testing capabilities in detecting the lamination defects are demonstrated. This experience also shows the capabilities of AE technique for condition monitoring of large metallic industrial facility with structural defects evolving with time. This approach allows us to define the main trends in defect development process and to make a decision about what kind of inspec-tion should be performed while facility is under operation or intermittently stopped. Keywords: pressure vessel, lamination, blistering, complex parameter Introduction

The object under observation is a pressure vessel (Fig. 1), where the mixture of hydrocarbons and hydrofluoric acid is separated. Because of the contents and internal operating conditions un-der active stresses, the diffusion of atomic hydrogen occurs in areas of tensile stress concentra-tion. The accumulation of atomic and consequently molecular hydrogen induces the formation of internal cavities with the pressure reaching several hundred atmospheres. This is one of the causes of vessel structural failure. The hydrogen induced cracking (HIC) or blistering usually takes the form of cavities (called blisters when surfaces expand locally). Sometimes blistering takes the form of thin stepped cracks inside a sheet in parallel to the rolling plane. The major fac-tor that determines a tendency of steel to HIC is the presence of atomic hydrogen collectors within the steel matrix. The blistering forms in steels contaminated with sulfide inclusions or in steels that have the banded ferrite-pearlite matrix.

The previous paper [1] presented some results of the first AE testing of the same separator

vessel after seven years of operation. Details of the inner structure and chemical locations are given there. Consideration was also given to characteristics of the steel (A516 Grade 70) used in manufacturing of the vessel including results of its metallographic and chemical analysis. It is the mild steel with banded ferrite-pearlite matrix minimally contaminated with non-metallic in-clusions. Microstructures of the steel, shown in Fig. 2, were taken from the shell of another vessel (the reactor), which was in the process chain immediately before the separator vessel. This reactor was removed and replaced with a new reactor, because of the laminations that occurred at depths of 11.5-12 mm and 6-7 mm in the wall thickness of 18 mm. These laminations were found after 7 years of operation. Both of the vessels were manufactured together from the same steel stock and were working with similar chemicals. However the reactor was working in heavier operating conditions in comparison to the separator (for example, operation pressure of the reactor was twice). Band structures in Fig. 2 generally lead to discontinuous HIC formations.

235

Fig. 1: Vessel for separating the mixture of hydrocarbons and HF acid (V = 243 m3, ID = 4,200

mm, t = 23 mm).

Fig. 2: Metallographic results of vessel material:

(a) specimen No. 1 (x500), (b) specimen No. 2 (x1000).

Fig. 3: Photo of the inside surface of the steel of pipe-manhole with disclosed cavity.

236

Figure 3 shows the part of the inner surface of the pipe-manhole of the removed reactor ves-sel. The small area of surface shows the cavity (it is in red due to dye-penetrant inspection). The lamination was located at 6-mm depth. This lamination was found while planar location AE re-sults were analyzed in the area of the manhole on the vessel head (Fig. 4). Unfortunately, we did not make additional UT-inspection in other parts of the reactor shell after the vessel was re-moved, but we cut several specimens from the shell for steel structure analysis.

Fig. 4: Results of planar location for the part of the reactor shell near the manhole.

Sudakov et al. [2] shows that the principal operational load (internal pressure) usually does

not lead to the development of lamination because the stress intensity factor does not exceed the critical value for lamination of any size. The stress intensity factor reaches its maximum with the definite length of lamination and then with increase of this length it decreases. The internal pres-

237

sure, however, can induce the growth of laminations closer to areas of high stress concentration – around welding seams and holes in the shoulder ring. The probability of growth increases during operation in transient modes, for instance, during startup (or pneumatic tests).

According to the data from non-destructive inspection, continuous lamination detected by UT

inspection actually is the aggregate of closely located relatively small defects [2]. Our practical experience of AE testing of vessels with laminations confirms this [3]. Normally, we detect the indications of lamination at an early stage of the process of large blister buildup in a form of ag-gregate of multiple AE events clustered in the area where blister subsequently appears. Usually these events have relatively low energy characteristics. Afterwards, when a blister is formed (Fig. 5), the number of detected AE events in these areas significantly decreases. For the most part, high-energy events are occasionally detected and localized on blister borders. Apparently they are linked to such rare processes like blister border movement or formation of long cracks at merging of individual blisters. The most serious is a defect in the form of a crack, opening a blis-ter from inside of a vessel (Figs. 6 and 7).

Generally, we detect the presence of blisters with AE during the reduction of vessel pressure.

However, we have also detected clearly AE events from blisters of confirmed presence during rising pressure, not just when the blisters join nearby blisters under stress concentration. For ex-ample, we found an early stage of blisters in a column in 1999 and AE retest in 2010 showed the group of blisters has evolved substantially. This unit was withdrawn from service because of this AE and subsequent UT inspection. We recorded AE waveforms and are in the process of data analysis and will be compared to destructive test data.

In our practice, we have many examples of test vessels with large blisters, located far from each other. Typically, AE does not respond to these defects. Hence, the opportunity to detect the bundle with the AE is strongly dependent on the type of distribution of metallurgical defects in the metal.

Fig. 5: Evolution of lamination to blister for the mild carbon steel with banded ferritic-pearlitic structure having minimal non-metallic inclusions.

238

Fig. 6: The defect opening the blister in the form of crack on the inner side of the vessel.

Fig. 7: Transverse cut of the vessels shell in zone of crack opening the blister.

Results and Discussion

The forecasting of HIC development is considered to be a challenging task. According to the data of ТNK-BP (Fig. 8) [4], the number of failures due to HIC sharply increases with the exten-sion of operating life beyond 10 years.

Fig. 8: Statistics of equipment failure from HIC (data of TNK-BP).

The second AE inspection of the separator vessel was performed after five years from the

first inspection (after 12 years of the vessel operation). The major differences in procedures of these two inspections are the following:

239

1) The number of channels increased from 15 to 27; during the second inspection probes were installed at the same 15 positions they had previously been attached during the first inspec-tion, and 12 additional probes were installed at new positions (Fig. 9);

2) The schedule of vessel pressure increase was faster and nearly linear as shown in Fig. 10.

Figure 9 shows results of planar location for the pressure increase interval P = 0.15 → 0.6 MPa that is imposed on vessel structural parts forming potential areas of stress concentration (welds and holes). The amplitude filtering is used for AE events. It is possible to note the overall high AE activity and existence of several clustering areas of high amplitude AE events.

Fig. 9: Results of planar location of AE sources on shell reamer (inner side view) for the 2nd test (P = 0.15 → 0.6 MPa, V = 3.30 m/ms). ▲ – positions of sensors during both tests, ▼- positions of sensors during the 2nd test only.

Fig. 10: The loading versus time curve during two pressurization tests.

240

Figure 11 shows the result of superposition of two location diagrams and comparison of posi-tion of AE activity areas (without amplitude filtering). Areas with the highest AE activity are pictured as outlines on Fig. 12. It can be seen that the highest AE activity areas during the second inspection are typically located in areas with relatively low density of AE sources during the first inspection. Conversely, high AE activity areas from the first inspection during the second one showed lower number of AE sources. Predominant are the adjacent areas with partial overlap-ping at edges.

Fig. 11: Results of combination of locations graphics for two tests: ▼- AE sources of 1st test (P = 0.13→ 0.5 MPa), ♦- AE sources of 2nd test (P = 0.15→ 0.6 MPa).

These results enable us to suppose that during the 5-year period activity areas from the first

inspection have undergone stress relaxation with disappearance of local stress concentrators around small sulfide inclusions. Aggregation of small defects has taken place and significant size laminations were formed. These laminations have more regular structure than the accumulation of multiple small discontinuities [3]. It defines the response to applied load from the viewpoint of AE testing results. The process of lamination build-up was registered during the second AE in-spection at the new vessel areas.

Figure 12 also shows areas with no activity recorded either during the first or second inspec-

tion. In one instance, these are the areas where lamination processes ended prior to the first in-spection. In the other instance, these are the areas where there are no conditions for development of structural defects under loads active during operation. The order of degradation of different areas of the steel vessel through time is defined by the distribution of products with various cor-rosive activities within the space of vessel and by the location of expanded areas of stress buildup [1].

241

Fig. 12: The contours of zones of the highest AE activity for two tests:

- 1st test (P = 0.13→0.5 MPa), - 2nd test, after 5 year (P = 0.15→0.6 MPa).

Figure 13 presents the result of simulation of interaction of laminations located at various depths within the middle third of sheet. It is defined that these laminations may join together as a result of breakthrough of long cracks between their borders, for instance, during a startup or any other sharp pressure increase.

Fig. 13: The result of lamination’s interaction simulation, located at the different depth in the middle one third of the sheet.

242

The analysis of condition monitoring data for this separator vessel allowed us to make an as-sumption that the development of similar cracks could be registered during the AE inspection. The search for such objects required the application of complex parameter – the combination of several AE signal characteristics (duration, amplitude, and rise time) and their relations with each other taken with weighted coefficients. Preliminary consideration was given to forms of distribution of basic characteristics of AE events that fell into location (Fig. 14). The presence of “heavy tails” in distributions makes possible isolation of AE events that are potentially connected with the specific type of structural alterations. The parameter was deducted with consideration to specific type of defect, i.e. very long cracks.

The AE testing system employed for the inspection is a flexible data processing tool, and the

mathematical expression for complex parameter was completely generated with a help of proces-sors included in the system software. In our opinion, one of the basic principles of operation of this system is in priority of expert decision before formal computer decision-making technolo-gies. Such a feature of the system is specifically useful in abnormal situations of data analysis, when there are not enough regular criteria available for assessment of results. An expert can se-lect necessary functions from a wide range of data processing and presentation means and create arbitrary data analysis routines including his/her own local criteria tuned to any particular appli-cation.

Fig. 14: Distributions for characteristics of AE events used in complex parameter in the 2nd test data analysis. The "heavy tails" in distributions allocated.

Figure 15 shows the diagram similar to the segment given in Fig. 11. This diagram has the

added filtering of AE sources using the K5.2 parameter as defined below. K5.2 = 1.8 log(Duration) + log(A_Lin) + 2 (RiseTime/Duration)

243

Here, A_Lin is linear amplitude. This enables us to visually pinpoint several groups of sources forming bands or lines that cross new and connect old lamination areas. The significant number of individual sources with the high value of K5.2 is located between laminations of various ages.

Fig. 15: The combination of locations graphics for shells fragment:▼- AE sources of 1st test, ♦- AE sources of 2nd test, ■ - AE sources of 2nd test with high value of complex parameter, K5.2.

These findings compare favorably to the preceding AE inspections of vessels with lamina-

tions and provide solution to the lamination problem obtained during simulation study. The feature of this vessel is abnormally large areas of the AE activity. Previously, the AE activity appeared on different stages of the lamination development and we observed only on local sections of the vessel shells. The complex parameter K5.2 turned out to be the helpful character-istic for the localization of position of such critical defects as lamination with linking cracks. This significantly reduced the scope of additional NDT techniques needed for the determination of exact defect geometry.

Now, the vessel is back in operation. In 2010, two years after the second AE test, two zones

of the vessel shell (areas enclosed in red rectangles; see Fig. 16) were tested with ultrasonic thickens gage during a brief stoppage of operation. The size of the areas was 700 x 920 mm and 660 x 580 mm. A number of areas had high AE activities, but the particular areas were chosen as sample areas because of their accessibility.

The design thickness of the vessel wall is 23 mm and these areas had the thickness of 22 to

22.5 mm when tested in 2008. Thickness scanning (with 2 mm pitch) showed that wall thickness was reduced to 16.7 to 19.6 mm range over the whole scanned areas. More than 98 % of meas-urements were in the range of 18.1-18.2 mm; i.e., smooth UT result. The depth of the potential

244

lamination did not coincide with the usual location of the laminations – in the central part of the sheet. The steel plate typically contains a large number of pearlitic bands (as shown in Fig. 2) located at different depth levels of the plate profile. It can be assumed that any of the bands (and sulfide inclusion stringers also seen in Fig. 2 a) could be collector for hydrogen. In addition, such "smooth" UT results are not typical for the wall thickness measurements of industrial vessels, which had operated for a long time (16 years). However, these results are typical for laminations. This is likely to confirm the hypothesis that the groups of hydrogen-induced blisters are present in the metal shell over a large area. These are located at 5 mm from the inside surface. Another possible cause is erosion-induced thickness loss, but this is unlikely because of location and also the lack of such loss in the first 14 years of operation.

Fig. 16: The positions on the shell of two zones of ultrasonic scanning of wall thickness.

Summary and Conclusions

This paper shows acoustic emission as the optimal inspection method from the time-saving standpoint for obtaining information about the condition of large vessels. After two AE inspections with the total duration of several hours, we obtained the assessment of the quality of steel for almost the whole area of shoulder ring; made correct suggestion about the main type of defects (blistering), compiled the layout of defect distribution along the shoulder ring; localized the most serious defects and also determined the main trends in development of the defect system. In fact, the time of AE testing is actually shorter because it was performed during the mandatory periodical pneumatic tests. The pneumatic test time constitutes a significant part of the whole AE testing procedure.

Specific features of large lamination behavior will probably allow operating this vessel for

additional time. Our recommendations, however, point to optimal decision in replacement of this vessel with a new one considering its current condition and the nature of products circulating

245

inside. The new vessel shall be fitted with permanent AE monitoring system. In an event of peri-odic AE condition monitoring, it is recommended to perform the first test in 3-4 years after the beginning of operation and a follow-up test every 2-3 years. Prior to replacement of the vessel it is necessary to inspect the steel at the boundaries of lamination defined by using AE testing.

Additional results include the data obtained during AE testing that helped reveal technologi-

cal factors initiating the steel degradation process and provided grounds for checking out the ini-tial quality of steel used in manufacturing of the vessel and other process equipment of this facil-ity. The steel microstructure examination has defined that the used steel properties in combina-tion with process conditions can be the source of problems during the operation of some of these facilities. This became the reason for the increase of periodic NDT of all these facilities. References [1] Kabanov B., Gomera V., Sokolov V., Fedorov V., Okhotnikov A., Use of AE method abili-ties for petrochemical equipment inspection, In: Proceedings of the 26th European conference on Acoustic Emission Testing, Berlin, 2004, Germany, Vol. I, pp. 131-138. [2] Sudakov A., Danyushevsky I., Saikova M., Accounting of defects during expert appraisal of industrial safety and operating life of boilers and pipelines: Berg Kollegium, 2 (65), 2010, pp. 20-22. [3] Kabanov B., Gomera V., Sokolov V., Fedorov V., Okhotnikov A., AE Testing of Refinery Structures, In: Proceedings of the 27th European Conference on Acoustic Emission Testing, Car-diff, 2006, Wales, UK, pp. 133-138. [4] Makarenko V., Gorbunov S., Ogorodnikov V., Bakeev R., Shevtsov V., Causes of failure of equipment at gas processing plants of Western Siberia: Mechanical Engineering, #8, 2007, pp. 45-48.

J. Acoustic Emission, 28 (2010) 246 © 2010 Acoustic Emission Group

COMPARISON OF ACOUSTIC EMISSION PRODUCED DURING BENDING OF VARIOUS OXIDE CERAMIC AND SHORT FIBER OXIDE

CERAMIC MATRIX COMPOSITES

S.A. PAPARGYRI-MPENI, D.A. PAPARGYRIS, X. SPILIOTIS and A.D. PAPARGYRIS

Technological Educational Institute, Larissa, General Department of Applied Science, Materials Technology Laboratory, Nea Ktiria, 41110 Larissa, Greece

Abstract

AE can be used in characterization of oxide ceramic and ceramic matrix fiber composites. In oxide systems the AE activity was influenced by chemical composition, sintering temperature and structure development. AE activity increased from kaolin to pottery mixture as a result of the developing a stronger structure with incorporation of a second phase in the matrix, while de-creased in kaolin-fly ash composites due to weakening of the structure and formation of cracks. Kaolin ceramics showed a clear Kaiser effect, while pottery mixture, kaolin-fly ash and brick clay composites were more or less acoustically active during the second loading. Short fiber ce-ramic matrix composites indicated that fiber composites present higher AE activity since, matrix sub-critical events, fracture of the fibers, fiber-debonding or fiber pull-out from the matrix may produce a comparatively large amount of AE events. Increase of sintering temperature produced composites with higher AE activity for similar reasons to that of powder ceramics. Increase of the volume fraction of fibers increased the AE activity. In the examined fiber ceramic systems Kaiser effect was observed in all cases irrespective of the kind of the fibers, a fact that indicates that the sub-critical activity during loading is not a reversible process. AE examination is a valu-able tool giving useful information about the strength, structure and the sintering techniques used during manufacturing of ceramic systems. Keywords: Oxide ceramics, oxide fibers, graphite fibers, short fiber composites 1. Introduction Acoustic emission (AE) is a potential testing method to monitor the integrity of materials, mechanisms of fracture, and other processes. It is nowadays applied in almost all areas of science and technology of materials, e.g. [1, 2]. The monitoring of AE events from ceramic specimens, which are stressed can be divided into two stages. In the first stage the damage (any active dis-continuity) inside the specimen produces an AE event, which propagates through the material and arrives at a transducer. In the second stage, the event is converted into an equivalent elec-tronic signal and finally detected as an AE count. It is generally assumed that in coarse structured poly-phase ceramics where there are mismatch stresses between grains, the minimum stress, at which AE activity starts, coincides with the onset of grain size fracture. During repeated load-ing/unloading of most of ceramic materials, Kaiser effect is observed. This means that during the second and next loadings the AE is zero or close to the background level till the stress reaches the largest previously reached stress level. Above that level the AE activity is increased dramati-cally. In the present work, which is based in a paper presented at the EWGAE 2010 [3], the AE be-havior of a number of oxide ceramics and short fiber composites is compared and discussed.

247

2. Experimental

The chemical composition and other characteristics of the materials compared in the present work are given in Tables 1 and 2.

Table 1. Chemical composition of ceramic materials.

Chemical compound weight %

Remblend Chinaclay

ECC International

Pottery mixture

Brick clay

Fly Ash

SiO2 48.2 59.7 50.6 34.4 Al2O3 37.1 25.5 20.1 19.6 Fe2O3 1.0 1.5 6.7 4.1 TiO2 0.05 1.6 1.2 0.5 CaO 0.07 3.1 3.8 32.2

MgO 0.3 0.2 3.4 1.9 K2O 2.0 1.9 3.2 0.7 Na2O 0.1 0.1 2.1 0.4 L.O.I. 12.1 9.5 9.9 6.0

Mineral phases

83% kaolinite 13% mica 2% feldspar 2% other minerals

39% kaolinite 28.6% lime feldspar 14% quartz 11% potash feldspar 0.8% soda feldspar 6.7% misc. oxides & organic matter

Table 2. Characteristics of fibers.

Characteristics Mullite short fibers Denka B-80

Alpha-alumina short fibers Denka B-95

Delta-alumina short fibers Saffil RF

Graphite PAN short fibers Grafil XAS

Classification temp. oC 1650 1600 > 1000 True density, g/cm3 3.3 3.5 3.3 1.79 Chemmical comp.

Al2O3 SiO2

Al2O3+SiO2 Other

80 20

99.7 0.3

95 5

99.7 0.3

96-97 3-4

graphite

Crystallinity ratio α-Al2O3 mullite

5≥

50≤

30-65 5 ≥

Fiber diameter, µm 2-5 2-5 3.4 (1-10) Strength, MPa 1670 2000 4480

E, GPa 167 300 227 Color white white white black

248

A large number of specimens (75 x 20 x10 mm, length x width x height) was produced for every formulation by slip casting. Mechanical testing was performed using an Instron 1195 ma-chine in compression. The specimens were placed in the three-point bend loading rig with a span of 50 mm. In order to permit sensitive AE monitoring the supports of the bend rig (10-mm steel bars) were covered with neoprene rubber to reduce the possibility of recording noise from the machine and the metal ceramic contact points. The fractured surfaces were examined in a JEOL T-330 scanning electron microscope with resolution of 4.3 nm at 25 kV accelerating voltage. Charging was avoided by application of a gold layer of 200-400 A, under low-pressure argon gas, using a ‘cool’ Edwards S150B Sputter Coater. A Philips X-ray diffractometer with a PW1820/00 vertical goniometer and a PW1710 microprocessor-based control and measuring system were used for XRD analysis. AE analysis was performed using a low noise MRP-01 MARANDY preamplifier with 60-dB gain and processing of the amplified signals was done by a MARANDY/MR 1004 amplitude analysis system, which stores the digital AE data as a function of time together with A-D converted load signals. The system provides information about the peak amplitude of each event (the peak amplitude is given relative to 10 mV). All AE activity was recorded during three-point bending. Schematic representation of the three-point bend test-ing rig and the AE system is given in Fig. 1. Details on specimen preparation and other experi-mental details are given elsewhere [4-6]. 3. Results and Discussion

Selected SEM and AE curves from the examined systems sintered at various temperatures are presented in Figs. 2 to 8.

Fig. 1 Schematic representation of the three-point bending rig and the Marandy AE system [4]. a. Oxide ceramics

Kaolin, a characteristic oxide ceramic, is an acoustically quiet material of low strength in the non-sintered condition. For comparison, specimens of soda-lime glass with similar dimensions were tested in the same rig. The results (Fig. 5) showed that kaolin specimens break with almost no AE. This is probably due to the absence of any connecting phase since it is composed of kao-lin aggregates. Similar behavior was observed in the glass specimens (Fig. 5a). After sintering, they become acoustically active even at very low stresses during flexure testing (Fig. 6a). The acoustic activity increases with increasing sintering temperature probably as a result of structure development during sintering (Fig. 2a); kaolin changes from the weak point-bonded structure to

249

a well-formed glass-mullite structure as sintering temperature is increased from 1000oC to 1300oC. The latter structures are stronger and produce more AE events during testing since more sub-critical activity (micro-cracking) occurs before fracture. The AE activity starts at low stresses, virtually from the very beginning of load application. In most of the examined speci-mens a load application of a few N (2-3% of the ultimate fracture load) was enough for constant AE. The event rate (count/s) is extremely low (less than 5) during loading of specimens sintered at 1000oC and is increased to around 10 at fracture, while it is around 60 in specimens sintered at 1300oC and is increased to around 300 count/s near fracture [4].

(a) (b)

Fig. 2 SEM from (a) Kaolin and (b) Pottery mixture sintered at 1300oC and 1100oC, respectively.

(a) (b) Fig. 3 SEM micrographs from brick-clays sintered at (a) 900oC and (b) 1100oC.

(a) (b) Fig. 4 SEM from (a) Kaolin+8.55 vol.% Grafil, (b) Kaolin+9 vol.% mullite-fiber composites sin-tered at 1000oC.

250

(a) (b)

Fig. 5 (a) AE from glass specimens and (b) from kaolin before sintering.

(a) (b)

Fig. 6 (a) AE and (b) Kaiser effect from kaolin specimens sintered at 1300oC, during 3-point bending.

(a) (b)

Fig. 7 (a) AE cumulative events versus load showing Kaiser effect from a pottery mixture speci-men sintered at 1100oC, (b) AE from a kaolin-fly ash 5 wt.% ceramic sintered at 1000oC

The Kaiser effect (Fig. 6b) was observed in all kaolin specimens during recording of cumula-tive events versus load in repeated loadings and is especially clear in those ceramics with high AE activity. The first loading is sufficient to cause all the subcritical damage in the specimen up to that load. Further damage does not happen until the first load is exceeded. There is absolute lack of any AE activity during the second loading till the same load level, after which AE activ-ity started again till fracture. Although it is not absolutely correct to allocate to each AE event a subcritical event, since the number of AE events collected through the transducer is influenced by many factors, it is reasonable to accept that high AE counts indicates numerous subcritical events.

Pottery mixture specimens generally produced the same behavior as kaolin specimens. Struc-

ture (Fig. 2b) and strength development increases the sub-critical sites for AE activity, which are "activated" during mechanical testing. Pottery mixture specimens sintered at 900oC present the

251

same kind of AE cumulative events, rate, distribution and linear relation between load and de-flection as the kaolin specimens sintered at 1000oC. As the sintering temperature increases from 900 to 1300oC, AE activity becomes more intense in a similar manner to that of kaolin. Pottery mixture specimens show the Kaiser effect (Fig. 7a) in a similar manner to that of kaolin.

SEM examination of brick-clays sintered at various temperatures (900-1100oC) showed that

during sintering a glass structure is gradually developed (Fig. 3) [5]. Brick-clay specimens sin-tered at the same temperatures as pottery mixture generally showed the same behavior and the Kaiser effect, but are more acoustically active (giving higher cumulative events) and have higher event rates. As an example, comparison of AE activity between pottery and brick-clay sintered at 1100oC shows almost four times higher cumulative AE events in the case of brick-clay speci-men. During cooling of the sintered brick-clay, radial tensile stresses are produced forming cracks due to difference in contraction between sand particles and glassy matrix. The application of external stresses produces an increase of the total stress field (enlargement of the process zone) resulting in additional cracking and consequently additional AE events.

AE examination of 5 wt.% and 10 wt.% fly ash-kaolin composites (sintered at 1000oC)

showed low AE activity (Fig. 7b) during mechanical testing. This is lower than that of pure kao-lin, since fly ash is a fine grain constituent (most has a grain size less than 45 µm). When fly ash is added to kaolin, such composites possess a cracked structure, reducing their strength and frac-ture toughness and making the structure weaker and looser. This is why Kaiser effect is less de-fined with large noise background, as has been reported in rocks [7, 8]. In Fig. 7b, the slope of AE activity curve starts to increase when the load reaches the maximum load during the first loading.

b. Fiber oxide ceramic matrix composites SEM micrographs from (a) a kaolin+9 vol.% mullite fiber and (b) kaolin+8.55 vol.% Grafil composites sintered at 1000oC are shown in Fig. 4. Generally, increased sintering temperature produces composites with higher AE activity for similar reasons to that of powder ceramics. In-crease of the volume fraction of fibers in the kaolin matrix also increases the AE activity. In pure kaolin, a low AE event rate increases just before fracture (Fig. 6a), which is a common character-istic of clay ceramic materials. In composite ceramics, while the event rate is low at the begin-ning, later there is a comparatively high AE event rate at loading at about half of the fracture load. The AE event rate eventually drops before fracture. This behavior could be attributed to the presence of fibers, the breakage of which produces a high AE event rate before the final fracture.

SEM examination of oxide-fiber composites (Fig. 4b) shows that the main fracture path is through the matrix and fiber with no indication of fiber pull-out or fiber-debonding. Increase of fiber volume fraction produces a large increase of AE activity. As an example, AE counts under identical conditions from pure kaolin (Fig. 6a) with a kaolin-matrix composite having 9.2% B-95 α-Al2O3 fiber (Fig. 8a) were compared. The comparison shows a five-fold increase, from about 2000 of 8000 events, which means that about 80% of the composite AE activity is due to fibers and only 20% to matrix sub-critical activity. Further measurements showed that increase of α -Al2O3 fiber content to 14% gives an increase to about 26000 AE events (not illustrated), showing that about 90% of the AE activity is due to fiber content. It could be assumed that this increase is a result of numerous sub-critical events due to fiber fracture during the application of load. In almost all the cases, the way that AE events are distributed is similar to that of kaolin specimens. Another general observation is that the event rate in composites under the testing

252

conditions used in this work is about 1000 events/sec, and is almost constant irrespective of sin-tering temperature, fiber content and chemical composition of fibers.

Finally, Kaiser effect was observed in all ceramic composites irrespective of the kind of the fibers (Figs. 8a-9). No AE activity was observed during unloading, a fact that indicates that the sub-critical activity during loading is an irreversible process. Another observation was the differ-ence of AE event rates between the first and second loadings. During the first loading a high event rate (up to 800 counts/s) was recorded in all cases with sudden peaks of high activity, dur-ing the second loading the event rate curve is smooth with a rate of almost nil up to the load of the first loading.

(a)

(b) Fig. 8. Kaiser effect in (a) 9.2 vol.% Denka B-95 α-alumina-fiber composite sintered at 1300oC, (b) 14 vol.% Denka B-80 mullite-fiber composite sintered at 1300oC.

253

Fig. 9. Kaiser effect in kaolin+8.5% Grafil fiber composite sintered at 1300oC.

Fig. 10. SEM micrographs from graphite fiber kaolin composites sintered at 1200oC.

The load-deflection-AE event rate curves of ceramic composites with oxide fiber and graph-

ite fiber show clear differences and similarities, which may be connected to interfaces and frac-ture mechanism of these composites. The oxide fiber composites have a load-deflection curve similar to that of powder ceramics with an almost linear increase of load with deflection up to fracture, since there is a strong interface between matrix and fibers (e.g., Fig. 4b). The fracture proceeds through matrix and fibers. The graphite fiber composites have a curve similar to that of unidirectional composites, where we can detect at first a matrix cracking region, then a region of load transfer to fibers and that of fiber pull-out and finally a fracture region, probably as a result of weak fiber-matrix interface giving fiber pull-out (Figs. 4a and 10).

Comparison of results from Saffil and B-95 Denka α-Al2O3 fiber-composites, sintered at tem-peratures above 1200oC, shows the effect on AE activity, of fibers having the same chemical composition but different fiber diameter. It must be noted that the Saffil fibers before sintering have a δ structure, which is transformed to α-Al2O3 after sintering above 1200oC and the main difference that remains after sintering is the shape and fiber-diameter of the fibers. Comparison of AE from a 9.2 vol.% B-95 α -Al2O3 -kaolin composite with that of 9.7% Saffil fiber-kaolin

254

composite specimen, both sintered at 1300oC, shows about the same AE event rate. It also shows that the B-95 fiber-composites produce a little more cumulative AE events and higher event rate, a fact which justifies the hypothesis that fiber content is directly connected to AE events pro-duced during 3-point bend testing. The difference in fiber diameters (which results in different number of fibers per volume) also affects the AE behavior of the composites. Due to higher range of lengths and diameters, the number of B-95 α-Al2O3 fibers is higher compared to that of Saffil fiber composites with the same fiber content and so the number of AE events is higher.

Comparison of results from B-95 alumina and B-80 mullite-fiber composites showed that there is great difference (probably due to the nature, morphology, diameter distribution etc. of the fibers) in the AE behavior, as shown in Figs. 8a and b. Figure 8a exhibits a high AE activity with about 10000 counts with the Al2O3 fiber-composite sintered at 1300oC. On the other hand, low AE activity (only about 5% of that of B-95 fiber composite) was measured in the B-80-mullite fiber-kaolin composite (Fig. 8b). The comparatively low AE activity of the B-80 mullite composite (which was observed also at other sintering temperatures as well) is probably due to the fact that B-80 Denka-mullite fibers have a high content of non-fibrous materials confirmed by SEM (appearing as large "shots" in the composite structure), which cannot produce as much AE events at the equivalent amount of fibers. The kaolin-graphite fiber composites with Grafil fibers belong to a totally different class of ceramic composites with a different fracture mechanism. SEM examination (Fig. 4a) shows there is a weak fiber-matrix interface and the fracture occurs after matrix cracking and fiber pull-out (Fig. 10), which indicates that any AE activity during loading may be due matrix cracking and fiber pull-out. This fracture mechanism implies a high number of AE events since every matrix fiber pull-out could give at least one AE event. However, it must be noted that the graphite fibers have an almost constant diameter of 6.8 µm, which is about double of that of Saffil fibers. This means that Saffil-kaolin composites will contain about twice as much fibers as the Grafil-kaolin compo-sites for the same volume content and consequently it could be proposed that Grafil-composites with double fiber volume fraction could have about the same AE activity to that of Saffil-kaolin composites with half fiber volume fraction. Comparison of AE activity from a 9.7 vol.% Saffil kaolin composite and a 16.5% Grafil-kaolin composite (which have roughly the same number of fibers), both sintered at 1300oC, shows roughly the same number of events, which means that the above hypothesis could be considered valid. 4. Conclusions AE activity was increased from kaolin to pottery mixture as a result of developing a stronger structure with incorporation of a second phase in the matrix, while decreased in fly ash compo-sites due to weakening of the structure and formation of cracks. In kaolin ceramics, there was a distinct Kaiser effect, while pottery mixture and the kaolin–fly ash and brick clay composites had high background AE activity during the second loading. Fiber-ceramic-matrix composites indi-cated that fiber composites present higher AE activity since the matrix sub-critical events, frac-ture of the fibers, fiber-debonding or fiber pull-out from the matrix may produce a comparatively large amount of AE events. Increase of sintering temperature produced composites with higher AE activity for similar reasons to that of powder ceramics. Increase of the volume fraction of fibers increased the AE activity.

255

Kaiser effect was observed in all the examined systems indicating "memorization" of the ap-plied load due to the fact that the sub-critical activity during loading is an irreversible process. The second loading AE was almost zero or close to the background level. The low AE activity during the second loading found in kaolin-fly ash and B-80 mullite-fiber composites could be attributed to the presence of large fly ash particles, non-fibrous "shots", etc. AE examination under loading can give useful information about the strength, structure, sin-tering techniques used during manufacturing and the "history" of ceramic systems. References 1. S. Momon, M. Moevus, N. Godin, M. R’Mili, P. Reynaud, G. Fantozzi, G. Fayolle, Compos-ites A: Appl. Sci. and Manuf., 41 (7), (2010), 913-918. 2. Tao Fu, Y. Liu, Q. Li, J. Leng, Optics and Lasers in Eng, 47 (10), (2009), 1056-1062 3. A.D. Papargyris et al, Proc. 29th EWGAE, 2010, paper 66, Vienna. 4. A.D. Papargyris and S.A. Papargyri, J. Applied Clay Science, 18, (2001), 191-204. 5. A.D. Papargyris, et al., J. Constr. & Build. Mat., 15, (2001), 361-369. 6. A.D. Papargyris, 4th Euro Ceramics, 12, (1995), 131-138. 7. A. Lavrov, Int. J. of Rock Mech. & Mining Sci., 40 (2), (2003), 151-171. 8. G. G. Zaretskii-Feoktistov and G. N. Tanov, Strength of Materials, 17 (5), (1985) 696-701.

 

J. Acoustic Emission, 28 (2010) 256 © 2010 Acoustic Emission Group  

NEW CHARACTERIZATION METHODS OF AE SENSORS

KANJI ONO 1, HIDEO CHO 2 and TAKUMA MATSUO 2 1 University of California, Los Angeles, Los Angeles, CA 90095 USA;

2 Aoyama Gakuin University, Fuchinobe, Chuo, Sagamihara, Japan Abstract

This paper reports on experiment on characterizing AE sensors commonly available and util-ized. Many AE applications need to detect plate waves, but no calibration scheme for these waves is available. We utilized bar waves with a long metal bar and plate waves using a large aluminum plate. The waves are excited using a transmitter driven by a fast step-down pulse of high voltage. Displacements of the bar are characterized with a laser interferometer. Displace-ment signals lasting over several hundred µs are generated, and the uniformity across the width is adequate. For the plate, ultrasonic transducers are used as reference to obtain power spectra of AE sensors. Frequency from 20 kHz to 1 MHz showed substantial variations in intensity, but the changes are smooth enough for characterization purpose. Using FFT, the power spectra of AE sensors were obtained. Deconvolution procedures were tried, but currently suffer from noise. The bar-wave based approach of sensor evaluation better simulates many AE applications and its further development is warranted. Keywords: Bar waves, plate waves, sensor characterization, laser interferometer 1. Introduction

In any acoustic emission (AE) work, AE sensors play an important role. It is essential that we understand how these are excited and produce outputs. The transfer function describes its input-output relationship. When we set out to determine the transfer function of piezoelectric AE sen-sors using pulse laser input and measuring the displacement response, we anticipated obtaining results in a straightforward fashion. As we reported previously [1, 2], it was anything but simple. Piezoelectric AE sensors have complex behavior and we needed to account for the types of input waves, even among the normal incident waves detected at the epicenter on the opposite face of a plate. With a point-like pulse laser source, spherical waves reach the receiving sensor and its re-sponse depended on the thickness of the plate, which corresponds to the radius of the spherical waves. A broadband ultrasonic transmitter can generate plane (flat wave front) waves within the area facing the transmitter in near-field region. Strictly, these waves should be called quasi-plane waves as diffraction effects produce non-uniform intensity distribution within the near-field re-gion. In this case, most AE sensors with disc-shape piezoelectric elements and piezoelectric discs themselves responded as dictated by the thickness resonance. This offers experimental confirma-tion of the reverse effect of piezoelectric disc vibration, to which a short transient electrical pulse was applied. This result agreed with theory. Sato and Yoshida [3] analyzed the transient behavior using the equivalent circuit method and predicted alternate polarity pulse with period, T = t/v (t = disc thickness and v = wave velocity). This was observed in our experiment. However, no radial resonance was excited in our experiment [1, 2]. Since typical AE sensors are designed using ra-dial resonance mode, the characterization with plane waves has no practical value despite provid-ing substantive physical insight to sensor behavior.

In this work, we first report sensor responses to plate- and bar-wave excitation. This is ex-

pected to lead to useful characterization methods applicable to practical AE work. This is

 

257  

followed by evaluation of sources of out-of-plane displacement using an ultrasonic transmitter. This is used to examine sensor responses to plane wave input, but buffer plate thickness has un-expected effect that needs to be studied further, making the use of transfer function approach even more restrictive than previously reported [1]. Lastly, we report additional findings that raise serious questions on AE sensor calibration methods as currently practiced.

 Fig. 1 Displacement output of FC500 excited by a pulse and its power spectral density.

2. Plate-Wave Experiment

Sensor characterization with plate waves utilized a large aluminum plate, 1230 x 1150 x 12.7 mm (6061-T6 Al). An ultrasonic transducer (AET FC500; 19-mm element diameter, 2.25 MHz nominal resonance) was driven by high-voltage step-down pulses (385 V step maximum, 0.04-0.08 µs rise time). Surface displacement waveform on the face of FC500 was measured using a laser interferometer and is shown in Fig. 1. The initial fast rise and trailing tail over 10 µs resem-bles a one-sided exponential pulse, which has inverse frequency-squared (f –2) power spectral density (PSD). The PSD shown in Fig. 1 follows f –2, illustrated by the slope of –2. The trailing part is due to the radial response of the transmitter and has center frequency of 25 kHs. The peak displacement in Fig. 1 corresponds to 31 nm (100 mV interferometer output = 1 nm).

This transducer was mounted to one edge of the plate at the middle position. After traveling

400 mm, surface motion was detected using another ultrasonic transducer (Aerotech, 5 MHz, 6.4-mm diameter) using 60-dB PAC preamplifier (1220A, 20-1000 kHz filter). The output is shown in Fig. 2a, with FFT power spectrum in Fig. 2b. The spectrum indicates relatively flat (±6 dB) response over 20 kHz to 1.0 MHz. Since the displacement at the sensing position is expected to have f –2-PSD as will be discussed later, the nearly flat PSD response appears to arise from the velocity sensitivity of the receiver. When one differentiates a harmonic displacement signal, D(t) = A sin(ωt), resultant dD(t)/dt contains the proportionality with ω = 2πf and its power density has f 2-term [4]. Therefore, the sensor response becomes flat in frequency.

The waveform may contain reflected waves beyond 250 µs after the initial arrival (at 210 µs

in the figure), but these are buried in noise due to heavily damped nature of this receiver and damping materials (plumber’s putty) placed on the plate surface beyond 450-mm distance. At 400 mm from the transmitter, 25-mm off-center position reduced the intensity 1 dB, while the corresponding reduction was 3 dB at 100 mm. Thus, the larger distance was chosen even though the effect of reflected waves is less at the shorter distance. Without damping, reflected waves were usually visible; flexural waves appear to attenuate only ~10 dB/m at >100 kHz and ~4

 

258  

dB/m at 30-60 kHz. Damping adequately reduced higher frequency components, but was ineffec-tive at lower frequency below 100 kHz.  

 Fig. 2 a) Surface motion detected by Aerotech, 5-MHz transducer using 60-dB PAC pre-amplifier (1220A, 20-1200 kHz filter). b) FFT spectrum in dB scale.  

By placing an AE sensor on the aluminum plate, its response to arriving waves can be deter-mined. Figure 3 shows the power spectral density of four sensors (in dB scale) against frequency. The sensitivity of these sensors is to be compared directly to the reference (Aerotech) sensor. No spectrum subtraction is performed. R15 sensor indicates the peak sensitivity at 160 kHz and sub-sidiary peaks at 250-450 kHz. This is similar to PAC’s calibration curve with the sweep-frequency face-to-face method. FC500 sensor has peaks at 140, 310 and 625 kHz and dips at 225 and 485 kHz. This sensor had a flat response in face-to-face calibration, but the dips appear to result from wave cancellation, as discussed by Beattie [4]. Sensing element was 19-mm diameter so flexural waves can provide this effect. For Pico and WD sensors, the response curves differ markedly from factory calibration, which showed smoother curves. It is possible to subtract the corresponding spectrum of the reference Aerotech sensor. However, the nearly flat PSD-response curve allowed omitting this step. These spectra (Figs. 2-4) were obtained using AEWin software.

Effects of reflected waves were minor after adding damping layer on one surface. However,

at frequency below 100 kHz, these remained clearly visible as shown in Fig. 4 for AET AC30L. This sensor has strong responses at 30-60 kHz. Initial weak waves arrived at 62 µs from the exci-tation pulse, indicating the longitudinal wave velocity of 6.2 mm/µs. This is followed by So Lamb waves at 74 µs and other modes. At 225 µs in Fig. 4, reflected waves started to arrive. Both extensional (225-300 µs portion) and flexural (beyond 300 µs) waves were attenuated only 6-7 dB, but the high frequency components are no longer prominent.

 

259  

It is important to note that radial responses of these AE sensors are excited with the present plate-wave excitation method. When a sensor is excited using plane-wave input [1] or using fast rise-time spot laser excitation, initial sensor response is due to reverberating waves between the front and back faces. In the case of AC30L, radial component at 170 kHz was strong initially, masking the low-frequency sensitivity obtained by mass-loading effect.  

 Fig. 3 FFT results of AE sensors. a) PAC R15. b) AET FC500, c) PAC Pico, and d) PAC WD.  

 Fig. 4 Plate-wave responses of AE sensor: waveform and FFT for AET AC30L.

 

260  

The results shown above indicate that the present plate-wave calibration setup provides real-istic basis for calibrating AE sensors for plate-wave detection applications. We still need to get absolute displacement or velocity responses at the calibration position in order to give the proce-dure the solid basis. One issue is the size of the plate needed. Although the width of the plate may be cut in half, the length is still insufficient to eliminate low-frequency reflection. However, as long as the complete wave train can be characterized, the entire wave can be used for calibra-tion. Even the plate size used here is large for its handling, two to three times longer plates would present a difficulty.

Another aspect of plate wave calibration is the orientation dependence. Some sensors, like

shear sensors (DECI SH225), have high sensitivity direction. In the case of SH225, it responds to waves propagating parallel to the shear sensing direction 7 dB higher than to the normal direc-tion. The spectral response was centered at 225 kHz for the parallel direction, but additional sub-sidiary peaks were present for the normal direction. 3. Bar-Wave Experiment

We next experimented using long aluminum bars as wave propagation media. In order to avoid reflected waves from returning to the calibration position, a longer propagation medium is required. However, it is impractical to use wide and long plates. Hayashi and Tanaka analyzed guided waves in rectangular bars using semi-analytical finite element method [5]. They showed more numerous modes propagating in a flat Al bar (5 x 100 mm), but some modes are similar to plate waves of the same thickness. They also reported that modes giving higher surface dis-placements (on the broad bar face) were those propagating at 3.5 mm/µs or higher and having extensional modes.

Here, we used 6061Al bars with 6.4 x 25.4-mm cross section and 3.66-m long, either straight

or bent into a coil with large radii and a 700-mm long flat section. Shorter bars were also used in trials. One end was polished and AET FC500 was attached to this end. Again, step-pulse excita-tion was applied. Displacement signals were detected using an interferometer as shown in Fig. 5. At 300 mm from transmitter, signal intensity decreased 3 dB at 10 mm off-center. At 640 mm, the difference between the center and near edge positions was 1 dB and this distance was se-lected. Signals lasted well over 600 µ s, indicating the presence of numerous slow-moving modes. The main motions corresponded to initial So and stronger flexural modes. Reflection can be ignored as it returns only after 1 ms or more and attenuated fully with plumber’s putty.

 Fig. 5 Surface displacement waveforms at 300 and 640 mm from transmitter. Edge position is 100 mm off-center.

 

261  

Power spectrum density of the displacement signal shown in Fig. 6 has many peaks and val-leys. The strongest peak appeared at 25 kHz and next strongest was at 330±20 kHz. Generally, the envelope of PSD has a decreasing trend, similar to the inverse frequency squared, found in the input PSD (Fig. 1). From 25-kHz peak to 1 MHz, PSD decreased about 30 dB. (Note here that dB scale for power has the multiplier of 10.) Wavelet transform of displacement signal is shown in Fig. 7. Effects of dispersion are visible, especially in the spreading of signal power over a longer period. Initially, we determined the transfer functions of AE sensors placed on the bar at 600 mm using deconvolution. However, signal-to-noise ratio was less than 10 dB and re-sults were noisy. We need a stronger pulser to increase the amplitude of surface motion.

Fig. 6 Power spectrum of the main part of displacement Fig. 7 Wavelet transform of displace- signal at 600 mm. ment signal at 500 mm.

Fig. 8 Effect of smoothing filter on displacement signals at 600 mm (280-µs segment).

We then examined effects of propagation distance systematically. Because of 10-20 dB dips

observed in the spectra, a smoothing filter was applied, as given below;

fi = [ fi+nn=−2

2

∑ ] / ( n +1)

where fi is the ith component of power spectral density. This reduced large dips, yet preserved most peaks and valleys observed. Effect of smoothing filter on the power spectral density of dis-placement signals at 600 mm can be seen in Fig. 8. All subsequent spectra are treated with this filter. Increasing n value to 3 improved the smoothness slightly, but visibly affected peak height. The smoothed PSD were used in a partial deconvolution scheme (ignoring phase data) to obtain sensor PSD responses by the frequency-domain divisions; i.e., by subtracting logarithmic values of displacement PSD from the corresponding signal PSD values of AE sensor under test.

 

262  

Figure 9 show smoothed power spectra of the main part of displacement signals at 100 mm to 600 mm. Left figure is for the distance of 100 to 300 mm over 0-500 kHz and right figure is for the distance of 400 to 600 mm over 0-1 MHz, respectively. The delay of signal arrival was un-compensated and resulted in some loss of 330-380 kHz components for 600-mm result (signal used was the segment from 110 to 300 µs). Otherwise, the overall trend of inverse frequency-squared is carried over from 100 mm to 600 mm data, although it is obvious that oscillations are larger at shorter distances of 100 and 200 mm. In all distances, peaks exist at 25 kHz and around 300 kHz. As noted elsewhere, the 25-kHz peak comes from the radial resonance of FC500.

Fig. 9 Power spectral density of the main part of displacement signals at 100 to 300 mm (left); Same for the data taken at 400 to 600 mm (right).

Fig. 10 AE sensor output due to displacement signal on the aluminum bar. R15, 600 mm.

As the first step, we placed a PAC R15 sensor at the positions where displacement signals were recorded; i.e., at 100 to 600 mm from the transmitting FC500, driven by step-down pulses. An example waveform taken at 600 mm is shown in Fig. 10, which displays output signal from 100 to 250 µs. Here, R15 output was directly fed to a digital oscilloscope at 1 M-ohm input im-pedance and digitized at 20-ns interval, as was the case for the interferometer output. The wave-form shows So-arrival at 110 µs, followed by Ao-mode at 170 µs. The latter appears about 10% faster than expected in an Al plate of the same thickness, while the So-arrival corresponds to 5.45 mm/µs speed in agreement with the So-mode in a plate. The sensor response was obtained by the frequency-domain division of the sensor power spectral density by displacement power spectral density using the subtraction of respective logarithms at each frequency. Results were converted to dB scale by multiplying 10 as these are power spectra. Figure 11 plots the obtained spectra for an R15 sensor, where six graphs are given representing the data at various distances. The dis-tance values in mm are noted with each figure.

 

263  

Fig. 11 Power spectral density (PSD) representing the sensitivity of an R15 sensor. Vertical scale is in dB relative to the displacement spectra. Horizontal axis: 0 – 1 MHz.

The variation of the sensitivity spectra of the R15 is remarkably small; All six show the main peak frequency at 159 kHz and similar peak height (27±1 dB). A few secondary peaks also ap-pear at same frequencies. The main peak seems to be excited by the So-mode waves at 130-160 µs, while Ao-mode (170-250 µs) contribute to slightly higher frequency (~200 kHz) output. A peak at ~580 kHz seems to exhibits large variation in amplitude and this may be related to a spectral scatter in the displacement spectra. This near independence on the propagation distance on the bar indicates that even though the displacement spectra show large changes as a function of distance, the spectral division method can be utilized to obtain the sensor spectrum.

Comparing Fig. 11 to the plate calibration result in Fig. 3, one notes the similar main peak

position, but secondary peaks have slightly different frequencies and varying relative peak heights. A peak does seem to exist near 600 kHz, as observed in Fig. 11. Since the reference spectrum used in the plate method needs further verification, we have to wait for assessing corre-lation of the two methods.

The use of a shorter propagation distance, like 300 mm, is advantageous as one can use a

shorter overall length. A comparison between spectra taken at 300 and 600 mm is favorable in

 

264  

using the shorter bar. However, the structure of peaks with 600-mm distance appears more natu-ral and is expected to be more representative of the true spectrum. Use of bent bar shape reduces the need to carry a long bar and improves the convenience factor.

The sensor calibration curves from the manufacturers are not useful in our discussion here as

almost all of them are obtained by a face-to-face method. The waves we use are plate or bar waves propagating some distance from a source. Still, the peak resonance frequency of R15 matches what was found in Fig. 11. Since the reference data we utilize is displacement meas-urement, the sensitivity has the basic dimension of sensor output voltage per displacement or V/m. However, both are expressed in power spectral density and results are given non-dimensionally in relative dB values. Here, it is noted that 10-nm displacement is detected as 1.0 V interferometer output. Thus, 0 dB is nominally 1-V sensor output per 10-nm displacement. Further study is needed to verify the absolute calibration scale in our case because the sensor re-sponse must be integrated over the sensing surface, whereas the displacement was taken at a point.

Fig. 12 The power spectral density of six broadband sensors. a) V101; b) V103; c) V112; d) B1080; e) Pinducer; f) WD. See text for details on sensors. PSD in dB vs. frequency (0-1 MHz).

 

265  

Bar-Wave Calibration of AE Sensors Using the method described in the previous section, we determined sensor responses from a

number of sensors from various manufacturers. These include broadband types, such as Panametrics V101, V103 and V112; Digital Wave B1080; Valpey-Fisher Pinducer (VP-1093); PAC WD and S9208. Miniature sensors include: PAC S9220, Pico, HD50 and conventional resonant sensors include: Dunegan S140B, AET AC30L, MAC175L and AC375L; PAC R3 and R6. Results are given in the following figures and appropriate discussion will be provided. Most data was taken at 600 mm propagation distance. When two curves are show, blue curve was taken at 300 mm, while red curve was at 600 mm. We anticipate the 600-mm data is the more reliable spectrum. However, these all show that 300-mm data is quite good approximation for the 600-mm data and may be substituted for field use, for example.

Figure 12 shows the power density spectra of six broadband sensors. First three are well-

damped ultrasonic transducers (UT). Some of them, like V103 has been widely used in concrete and geotechnical applications [6, 7]. B1080 has an integral FET and shows higher sensitivity. Pinducer has perhaps the smallest aperture size (of 1.3 mm diameter) among AE sensors with the highest sensor-element frequency (10 MHz), but requires care to mount it. PAC WD is a popular broadband sensor based on multiple sensor elements.

These six sensors have broadband response, but can hardly be called flat frequency response.

In detecting propagating bar waves, it is inevitable that geometrical cancellation occurs, explain-ing some of the dips in the observed spectra. Figure 12 a) and b) show peaked nature most strongly. These Panametrics UT sensors are difficult to even classify as broadband with a series of distinct frequency peaks. V112 (Fig. 12 c) may be called broadband with the sensitivity range

Fig. 13 The power spectral density of a displacement and three small sensors. a) S9208; b) S9220; c) HD50; d) Pico. PSD in dB vs. frequency (0-1 MHz).

 

266  

of only ±5 dB albeit with many peaks and valleys. B1080 is similar to V112, but sensitivity variation is greater. Among the six, Pinducer (Fig. 12 e) is the smoothest, but it also suffers from non-flatness of the spectrum. WD (Fig. 12 f) was designed to have multiple resonances and gives good sensitivity at the designated frequencies even without integral FET. Figure 13 a) shows the sensitivity of a displacement sensor (S9208). When used in the bar-wave detection, this has two distinct peaks and becomes similar to any common resonant sen-sors. Three other sensors have small sizes. In particular, Pico (Fig. 13 d) has been used often as a broadband detector, even though manufacturer calibration showed a broad peak at ~500 kHz. The present result indicates its response from 150 to 900 kHz is closer to flatness than any other we tested. S9220 does have good high frequency response in the bar-wave condition, but not to the extent shown by the manufacturer calibration.

Fig. 14 The power density spectra of six conventional AE sensors. a) AC30L; b) R3; c) R6; d) S140B; e) MAC175L; f) AC375L. See text for further identification of sensors. PSD in dB vs. frequency (0-1 MHz).

 

267  

Figure 14 shows the spectra of six resonant sensors, commonly used in AE testing. They all show good sensitivity at the peak frequencies. (Except for R3 and R6, these are vintage sensors, all 20-30 years old. Additionally, AC30L was slightly lower as it had been subjected to shock-wave testing.) Again, these show quite different characteristics in comparison to the manufac-turer’s face-to-face calibration curves, as expected. It may be noted that even though some of the sensors have similar element sizes, expected geometrical cancellations do not manifest them-selves. We may need to reexamine this aspect more carefully. When R15, S140B and MAC175 are compared, one finds a large variation in spectral re-sponse in terms of the sharpness of the main peak and the location and height of secondary peaks. All of them have piezoelectric ceramic disks around 12 mm in diameter and 6 mm thick and the differences arise from backing and case design. One may think of them as a group of nearly identical sensors, but each of them provides unique characteristics. The low-frequency sensors (AC30L, R3 and R6) appear to rely on mass-loading effect to ob-tain the low-frequency resonance. In order to monitor low-frequency AE signals, Fig. 14 clearly demonstrates the need of low-pass filtering the natural resonance response of the sensing piezoe-lectric element, as manufacturers often omit the high frequency responses of these sensors. From the results described here, the bar-wave calibration method provides sensor calibration for common measurement needs. It does require an access to a laser interferometer, but once the bent bar setup with a transmitter is characterized it can be taken to field since it is transportable. Additional work is needed for getting averaged displacement history (or velocity) covering an area of typical AE sensor faces. Comparison with the results using a velocity-sensitive interfer-ometer should shed light on the nature of sensor responses. 4. Related Experiments on Sensor Characterization Out-of-plane sources In characterizing AE sensors using normally incident waves, we reported previously a useful setup of using a UT sensor as a transmitter, coupled to a plate [1]. This complements another source using a YAG laser as a point-like source [8]. The latter was also used in NPL study [9] and generates spherical waves at the surface where a sensor-under-test is placed. Our setup with a UT transmitter produces waves closer to planar wavefront due to a larger size of the source. We typically utilize 19-mm diameter FC500 sensor (AET). In the previous study, we used Al plates of two thickness, 25 mm and 100 mm. We added 12.5 mm Al and 11.7-mm and 48-mm thick PMMA plates here and obtained the out-of-plane displacement at the epicenter position. The pulser was the same as before, supplying 280 V step-down pulses (40 ns rise time). Displacement waveforms obtained on the face of FC500 (shown in Fig. 1, but expanded the initial part here) and those after passing various thickness plates are shown in Figs. 15 and 16. The steepest part of the waveforms remains essentially unchanged, but the trailing part varies systematically with increasing thickness of buffer plate. As can be seen in both figures, the dis-placement starts to decrease more rapidly as the thickness of the buffer plate becomes larger. This sharpens the initial broad pulse that lasted over 10 µs into a narrower pulse. With 100-mm Al plate, the unipolar initial pulse has an extrapolated base width of 0.6 µs [1], while the use of 48-mm PMMA plate produced a pulse width of 2 µs at base. At medium thickness, Figs. 15 b, c and 16 b show double-peak shapes, although the second peaks are broader than the first peaks. Higher frequency components are dictated by the initial fast-rising part of the displacement and

 

268  

Fig. 15 Displacement waveforms from FC500 transmitter directly or with Al buffer plate. a) Di-rect, b) with 12.5 mm Al plate, c) with 25 mm Al plate, d) with 100 mm Al plate.

Fig. 16 Displacement waveforms from FC500 transmitter directly or with PMMA buffer plate. a) Direct, b) with 11.7 mm PMMA plate, c) with 48 mm Al plate, d) same over longer period.

 

269  

produce useful out-of-plane calibration sources at different thickness. Larger thickness reduces the displacement amplitude, especially for Al plates but less so for PMMA, but it acts to separate the radial component of 25 kHz that propagates at shear wave speed. With 48-mm PMMA, the shear waves arrive at 36.5 (=65–28.5) µs while the main out-of-plane displacement step arrives at 18 (=46.5–28.5) µs. By placing a sensor-under-test at the epicenter position, we can get the sensor response. As reported previously [1, 8], the transfer function of a sensor can be determined or the frequency response obtained using the same procedure described in the previous two sections of this paper. Actually, this is partly identical to one of the procedures described in ASTM E-976-10 standard, except E976-10 does not utilize direct optical measurements. By adding the use of modern opti-cal instrumentation, the standard E-976-10 procedures acquire the physical foundation. This is urgently recommended. Buffer plate effects Because the steep initial parts of the displacement pulses observed after passing buffer plates of varying properties and thickness remain basically unchanged, sensor responses at varying thickness should be similar. However, this was not the case for the reason to be discussed in this section. First, we report changes in the responding waveforms produced by sensors as the thick-ness of buffer plates increases.

Fig. 17 Output waveforms from a Pinducer. a) face-to-face to FC500 transmitter, b) after passing 12.5-mm PMMA buffer plate and c) after passing 51-mm PMMA. The first example is given in Fig. 17, which shows the output waveforms from a Pinducer with a 1.3-mm aperture size; a) face-to-face to FC500 transmitter, b) after passing 12.5-mm PMMA buffer plate and c) after passing 51-mm PMMA. The Pinducer was directly connected to a digitizer input (1 M-ohms impedance). The response to direct acoustic coupling to FC500 was a negative-going unipolar pulse, as shown in Fig. 17 a). As the thickness of the buffer plate in-creases, the Pinducer output changes to bipolar, as can be seen in Fig. 17 b) and c). The ampli-tude ratio of the second (positive) to first (negative) peaks was 0.58 or 0.6. It is noted that the shape of the first pulse is similar to the direct pulse even after passing PMMA plates; only the second peak height grew with increasing plate thickness. This behavior was observed with dif-ferent combination of plate material (Al, PMMA and steel) and thickness (up to 154 mm was used). Next two examples are shown in Fig. 18, where V103 sensor was used with Al and steel plates. We used this sensor because Manthei used a capillary break to show its apparent velocity response behavior [7]. In both, the left-most curve is for face-to-face to FC500 transmitter.

 

270  

Figure 18 a) shows the changes due to increasing Al plate thickness (data for 13.1 to 46 mm shown). The amplitude ratio of the second-to-first peaks increased from near zero to 0.34 and 0.93 and, at 154-mm thickness, it was 0.87. Similar trend was found with steel plate (25-mm thickness, Fig. 18 b) and the amplitude ratio was 0.78. The amplitude ratio of the peaks grows linearly at first and tends to saturate at 0.6, 0.8 and 0.9 for PMMA, steel and Al plates. The satu-ration is reached at about 25 mm for steel and Al, and for PMMA at 40~50 mm. Again, the peak shape remains almost identical independent of the thickness. In case of Al plates, the first peak width is slightly reduced when the pulse became bipolar; i.e., after passing 25 mm of Al. In all cases, the bipolar pulse is not the result of differentiation of the initial unipolar pulse because the pulse width did not decrease substantially. A slight decrease of pulse width (20-30%) can be seen in Figs. 17 and 18. This implies that we cannot use the bipolar pulse as an indication of the ve-locity response of a sensor.

(a)

(b) Fig. 18. a) Output waveforms from V103 sensor. Face-to-face to FC500 transmitter, after passing 13.1, 25.1 and 46.2-mm Al buffer plates. b) Same after passing 25.0-mm steel. In order to examine whether the medium has an effect on the observed shape change, we used immersion test method with water as transferring medium. The results for Pinducer are given in Fig. 19. The face-to-face data is similar to Fig. 17 a) with the peak amplitude of –0.4 V. Figures 19 a) and b) correspond to water paths of 9.0 and 20.0 mm. Both show the bipolar nature in contrast to the unipolar nature of direct contact data (Fig. 17 a). Thus, the presence or absence of shear waves is not a factor in the shape change behavior. Immersion test results for larger di-ameter UT sensors with water path of 51 mm are shown in Fig. 20. For V103 (12.7-mm diame-ter), the second peak height now is larger than the first (Fig. 20 a), while the amplitude ratio for FC 500 (19-mm diameter) is almost unity (0.98). Thus, the unipolar to bipolar transition in pulse shape is observed regardless of receiver aperture size. Note that in the immersion tests of three different sensors, the rise time was unchanged as was the case of contact testing.

 

271  

Fig. 19. a) Immersion tests for Pinducer. Water path of 9 mm. b) Same as a), but with 20-mm water path. FC500 transmitter was used.

Fig. 20. a) Immersion test for V103 receiver. Water path of 51 mm. b) Same for FC500 receiver, 51-mm water path. FC500 transmitter was used. It was noted that a similar waveform change results from band-pass filtering of a unipolar pulse, as shown in Fig. 21. Here, a signal from V112 was generated by FC500 transmitter, face-to-face (Fig. 21 a) and was fed to a PAC 2/4/6 preamplifier (20-1200 kHz filter, gain at 20 dB). The output is shown in Fig. 21 b), and has the bipolar nature. Because of the lower input imped-ance of 2/4/6, actual amplification is 3.15x, much less than the nominal value of 10x, and the po-larity was reversed. Obviously, this indicates that cutting off high and low frequency components produces a similar effect on waveforms, adding the trailing half cycle to a unipolar pulse (ampli-tude ratio = 0.46) and increasing the pulse width slightly. Naturally, the similarity of pulse shape change is in form only and based on an unrelated physical phenomenon.

Fig. 21. a) Output from V112 receiver into high impedance input; FC500 transmitter. b) Output in a) passed through PAC 2/4/6 preamplifier at 20 dB gain with 20-1200 kHz filter.

 

272  

Next, we examined the PSD of full- and half-cycle sinusoidal pulses, as shown in Fig. 22. The two pulses compared have 1-V peak amplitude with periods of 2 µs and 1 µs, respectively (Fig. 22 a); 4 µs-portion shown, 10 ns interval, 8192 points). Their FFT are shown in Fig. 22 b). The comparison yields the decreasing low-frequency components of the full-cycle pulse, with similar high-frequency responses. The PSD value decreased 10 dB going from 400 to 25 kHz. Thus, the pulse shape transition reflects the loss of low-frequency components, or from high-pass filtering in electrical circuits. Obviously, this arises from the presence of dc component in the unipolar half-cycle pulse.

Fig. 22 a) Full- and half-cycle sinusoidal pulses, 1 Vp. b) PSD of the two pulses.

It has become apparent that the buffer plates exert similar influence on the frequency spectra of the displacement pulse generated by an UT transmitter. Up to now, the similarity of the initial fast-varying segments was assumed to provide the similarity in PSD. From the above result, such an assumption is untenable. Figure 23 illustrates the PSD of six displacement waveforms, given in Figs. 15 and 16. We used 160 µs segment for this calculation (20 ns interval, 8192 points) and the same spectral smoothing filter was applied as before.

Fig. 23 a) PSD of displacement signals with Al plates, shown in Fig. 15. 160 µs long segment. b) Same for signals with PMMA plates, shown in Fig. 16. PSD for FC500 direct and with 100-mm Al plate shown in both for comparison. PSD results show clearly higher attenuation of the low-frequency components. In the case of Al buffer plates, PSD difference at 1 MHz is 17 dB for 100-mm plate and the difference becomes 28 dB at 200 kHz and 32 dB at 20 kHz, respectively. Similarly, for 48-mm PMMA plate, PSD difference at 1 MHz is 2 dB and the difference becomes 10 dB at 200 kHz and 12 dB at 20 kHz, respectively. Such differences in PSD are adequate to produce the observed changes in received waveforms from unipolar to bipolar pulse shape. Consequently, the unipolar responses of AE sensors, such as Pinducer and V103, can be concluded to be due to their displacement responses. The apparent velocity response of V103, as reported by Manthei [7], arises from the differential

 

273  

attenuation of the low-frequency components in the buffer plate, and it is not from the inherent sensor behavior. The observed differential attenuation of the low-frequency components in buffer plates can-not be attributed to the material absorption because such attenuation is higher at higher fre-quency. This effect is instead due to the directivity of the waves generated by the transmitter. It is well known that a circular piston radiator (or a piezoelectric disk transducer) has the angle of beam spread, θ = sin-1 (λ/D), with λ = wavelength and D = disk diameter [10]. Here, the trans-mitter (FC500) has D = 19 mm. In Al, θ = 22º at 1 MHz, but λ ~ D when the frequency becomes 300 kHz, implying θ = 90º. The equation for the beam spread cannot apply at such large angles, but the beam tends to become omni-directional. The beam spread is less in PMMA and water, but the trend of larger θ is unchanged. In PMMA, wave velocity is 2.67 µs/mm and λ ~ D at 140 kHz. The present finding points to another problem in AE sensor calibration guidelines per ASTM E-976-10. In particular, arbitrary choice of buffer plate thickness must be narrowed and the pre-sumed flatness of transmitter PSD must be replaced by the measured PSD using an independent instrumentation, such as optical interferometers. Radial mode excitation It has been well known that the radial resonance of piezoelectric disks provided the founda-tion of most widely used AE sensors, such as S140B, AC175, R15 and more recently VS150. In Sec. 3, we showed one of them, R15, responds with the peak frequency of 159 kHz with bar-wave excitation. In a previous study [1], we showed that the same sensor does not exhibit this radial resonance until 10+ µs after the initial longitudinal wave arrival (as shown in Fig. 24 c). Initial response of R15 is due to the thickness reverberation, in agreement with theory [1, 3]. In contrast, when this sensor is excited face-to-face with a broadband transmitter (FC500, Fig. 1), it responds immediately from the beginning at the radial resonance frequency (see Fig. 24 a). This excitation of the radial mode comes from the radial vibration of the transmitter, FC500. The ra-dial motion measured on FC500 under pulse excitation is shown in Fig. 25, where part (a) gives details of setup using an interferometer. A small metal block is glued to the face just outside the piezoelectric element and the measuring laser beam was focused at 1-mm height. Figure 25 b) shows the waveform of the radial (or shear) motion, which has 20.4 kHz resonance frequency (the first cycle was at 36 kHz). This radial motion excites the radial resonance of a disk trans-ducer and generates the main peak of sensors such as R15. When R15 is coupled to a Pinducer, the thickness response is detected clearly (see Fig. 24 b), but the 159-kHz component (6.3 µs pe-riod) is difficult to discern. Using S140B, the same face-to-face result was found, showing the radial resonance (Fig. 26 a). When S140B was immersed in a water tank and was separated by 51 mm from FC500 trans-mitter, only the thickness resonance response is clearly visible. Water effectively eliminates the shear motion generated at the transmitter. This observation on radial motion also has practical significance in sensor calibration. It is normal for a sensor manufacturer to provide a calibration curve obtained by the face-to-face cali-bration method as described in ASTM E-976-10. This standard specifies a transmitter of 2.25-5 MHz UT sensor of 1.25-cm or larger diameter with face-to-face arrangement or through a metal buffer plate or block. Such a condition does excite sensor resonances and provides a measure of AE sensor sensitivity. However, the lack of specifics of this procedure contributes to the varia-

 

274  

tions in measuring the sensor response and it is highly doubtful that AE sensor sensitivity can refer to normal incidence acoustic waves or to any physically measurable standards. As shown in Fig. 15, displacement waveforms depend on the buffer plate thickness (also PSD, see Fig. 23).

Fig. 24. a) R15 face-to-face with FC500 transmitter. b) R15 face-to-face with Pinducer. c) R15 excited by FC500 transmitter via 100-mm Al buffer plate [same as Fig. 16 of ref. 1].

Fig. 25. a) Measurement setup of FC500 radial displacement using an interferometer. b) Dis-placement waveform, excited by a step-down pulse. These can be used to convert ASTM E-976-10 procedures into absolute measurement stan-dard, not just relative guidelines. Since E-976-10 does not specify the metal plate thickness, moreover, the arrivals of longitudinal and shear wave components can be separated differently. In Fig. 16 d), for example, the shear components arrive 30 µs after the longitudinal waves for 48-mm PMMA and for the case of 100-mm Al plate, the delay is also about 30 µs (not shown in Fig. 15 d). This gives different responses compared to the face-to-face arrangement, where radial

 

275  

transmitter motion arrives instantaneously. Conversely, the use of face-to-face calibration excites the radial resonance of the sensor-under-test. Thus, the method no longer can refer to the “sound pressure” as has been assumed commonly. Therefore, both of these issues further necessitate the revision of E-976-10 procedures in addition to the problems raised earlier.

Fig. 26. a) S140B face-to-face with FC500 transmitter. b) Immersion test of S140B with 51-mm water path. Reciprocity Breakdown

Reciprocity AE sensor calibration method has been claimed to be “absolute” [11–13] and had recent follow-up studies [14, 15]. However, the fundamental assumption of sensor reciprocity has never been tested or proven experimentally.

We showed previously [1] and here that normal incidence mechanical pulse input to sensors

based on piezoelectric disks results in a series of electrical pulses with the interval equaling the travel time through the thickness. See Fig. 24 c) for an example in the case of R15 sensor or Fig. 26 b) for S140B. This is in accord with theory [3]. Radial resonance builds up only after several oscillations [1]. Figure 11 also shows that the same sensor, R15, is most sensitive at 159-kHz radial resonance when excited by bar waves.

Fig. 27 Displacement response of R15 sensor to a fast-rise step pulse, initial part of the same waveform, and PSD of displacement waveform.

When an electrical pulse is imposed on a disc-based sensor, it responds immediately via ra-dial resonance and thickness resonance effects are secondary. Figure 27 shows the displacement response of R15 sensor to a fast-rise step electrical pulse; the initial seven cycles average 146 kHz and PSD at 150 kHz is 24 dB above the thickness resonance frequency of 350 kHz. These findings clearly show the absence of reciprocity.

The same conclusion is reached when we examined the behavior of PAC WD sensor, which

has three (disc and rings) sensing elements. Figures 28 a) and b) illustrate the displacement

 

276  

response of a WD sensor to electrical pulse of a fast-rise step and its PSD. This was measured by a laser interferometer. The low-frequency (30-kHz) oscillations are the strongest when WD was electrically driven. When a plane-wave mechanical pulse excited the WD sensor at normal inci-dence [1], its electrical output had much higher frequency contents as shown in Fig. 28 c). No reciprocity is observed in this case as well.

Fig. 28 a) Displacement response of a WD sensor to a fast-rise step electrical pulse, b) PSD of waveform in a), c) WD sensor electrical output due to a displacement pulse input. [1]

The results given here demonstrate the invalidity of reciprocity calibration for AE sensors [13] because such methods disregard the nature and condition of incident waves as well as the waves transmitted from the sensors under test. We have previously shown that sensor responses are sensitive to the nature of incident waves, so that current standards of sensor calibration based on the NIST procedure (such as ASTM E-1106-10) also need reevaluation. The methods pro-posed in this work using bar and plate as the propagating media are promising and worth con-ducting further development.

5. Conclusions

We have conducted a series of experiment for AE sensor characterization aided by laser inter-

ferometry and conclude that;

(1) Plate-wave and bar-wave calibration schemes work well in characterizing AE sensor under realistic plate-wave conditions, which simulate practical AE applications. AE sensor behavior was obtained excluding effects of reflected waves in the bar-wave setup, which is the most prom-ising because of its direct traceability to displacement measurements and of its portability.

 

277  

(2) Out-of-plane sources using a UT sensor as a transmitter with or without a buffer plate are characterized. Systematic thickness effect on detected sensor output is noted and is attributed to the frequency-dependent directivity of transmitted ultrasonic beams. Shear motion also plays a significant role in exciting AE sensor resonances of radial mode. (3) Break-down of reciprocity principle is demonstrated for AE sensors, showing that the so-called reciprocity sensor calibration method lacks foundation. (4) Sensor calibration guide, ASTM E976-10, should be revised because of demonstrated vari-ability of PSD as a function of buffer block thickness. Radial motion of transmitter also must be accounted for in the revised procedure. Acknowledgment

One of the authors acknowledges fruitful discussion with Dr. A.G. Beattie.

References [1] Ono, K., Cho, H. and Matsuo, T., J. Acoustic Emission, 26, 72-90 (2008). [2] Ono, K., Cho, H. and Matsuo, T., “Bar-Wave Calibration of Practical AE Sensors”, Proc. EWGAE 29, Vienna, (2010) p. 426. [3] Sato M. and Yoshida, Y., J. Acoustical Soc. Japan, 53 (11), 857-863, (1997). [4] Beattie, A.G., J. Acoustic Emission, 2 (1/2), 95-128 (1983). [5] Hayashi, T., Tanaka, T., Proc. Japan Soc. Mech. Engr., 72 (717), 743-748 (2006). [6] Katsaga, T. and Young, R.P., J. Acoustic Emission, 25, 294-307 (2007). [7] Manthei, G., Bull. Seism. Soc. America, 95 (5), (2005) 1674-1700; “Characterization of Broadband Acoustic Emission Sensors”, Proc. EWGAE 29, Vienna, (2010) p. 435. [8] Ono, K., Cho, H. and Matsuo, T., Proc. EWGAE 28, Krakow (2008), p. 25. [9] Theobald, P., and Pocklington, R., Proc. EWGAE 29, Vienna, (2010), p. 406. [10] Krautkrämer, J. and H., Ultrasonic Testing of Materials, Springer, (1969) p. 57. [11] Hatano, H. and Mori, E. J. Acoustical Soc. America, 59 (2), 344-349 (1976). [12] Hatano, H, Chaya, T., Watanabe, S. and Jinbo, K., IEEE Trans. Ultrasonics, Ferroelectrics, and Frequency Control, 45, 1221-28 (1998). [13] "Absolute calibration of acoustic emission transducers by a reciprocity technique," NDIS 2109, Japanese Society for Nondestructive Inspection, Tokyo, Japan, 1991. [14] Herve, C., Maillard, S., Zhang, F. Jaubert, L., Catty J. and Cherfaoui, M., Proc. EWGAE-28, Krakow 2008, pp. 1-11. [15] Keprt J. and Benes, P., Proc. EWGAE 28, Krakow 2008, pp. 19-24.

1

CUMULATIVE CONTENTS

J. of Acoustic Emission, Volumes 1 - 28, 1982 - 2010 Volume 1 (1982) Number 1 Page 1 An Acoustic Emission Study of the Intergranular Cracking of AISI 4340 Steel A. Nozue and T. Kishi Page 7 Acoustic Emission Behavior of a Low Alloy Steel R. J. Landy and K. Ono Page 21 An Acoustic Measurement of Boiling Instabilities in a Solar Receiver Alan G. Beattie Page 29 A Broadband Acoustic Emission Transducer Mark B. Moffatt, Theodore J. Mapes and Arthur T. Grodotzke Page 35 A Simple Tape Recording System for Acoustic Emission P. G. Bentley and A. Plevin Page 37 Earthquakes as Acoustic Emission - 1980 Izu Peninsula Earthquake ln Particular Kiyoo Mogi Page 45 AE Literature T. F. Drouillard Conferences and Symposia Page 67 The Fifth International Acoustic Emission Symposium K. Ono Page 68 Third Conference on AE/Microseismic Activity in Geologic Structures and Materials H. R. Hardy, Jr. Page 69 The Tenth European Working Group on Acoustic Emission K. Ono Page 72 Future Meetings Number 2 Page 73 Reciprocity and Other Acoustic Emission Transducer Calibration Techniques Roger Hill Page 81 Production Acoustic Emission Testing of Braze Joint T. F. Drouillard and T. G. Glenn Page 87 Acoustic Emission Transducer Calibration by Means of the Seismic Surface Pulse F. R. Breckenridge Page 95 Acoustic Emission Testing of Filament-Wound Pipes under Repeated Loading Leszek Golaski, Maciej Kumosa and Derek Hull Page 103 Source Mechanism and Waveform Analysis of Acoustic Emission in Concrete Masayasu Ohtsu Page 114 Acoustic Emission in Aircraft Structural Integrity and Maintenance Programs J. M. Rodgers Page 121 AE Literature T. F. Drouillard Conferences and Symposia Page 141 The 23rd Meeting of AEWG K. Ono Page 144 The 24th Meeting of AEWG W. F. Hartman and J.W. Whittaker Page 145 AEWG AWARDS K. Ono Page 146 Third National Conference on Acoustic Emission K. Ono Page 147 CARP - Sixth Meeting J. Mitchell Page 148 The XIth EWGAE Meeting B. Audenard BOOK REVIEW

2

Page 149 Elastic Waves in Solids Roger Hill Number 3 Page 151 Testing Fiber Composites with Acouatic Emission Monitoring Marvin A. Hamstad Page 165 On the SPI/CARP Recommended Practice for Acoustic Emission Testing of Fiberglass Tanks and Vessels C. Howard Adams Page 173 Some Details on the NBS Conical Transducer Thomas M. Proctor, Jr. Page 179 Low Temperature Behavior of Liquid Couplants used in Acoustic Emission Experiments J. Baram and J. Avissar Page 183 Acoustic Emission Behavior of Nickel during Tensile Deformation S.-Y. S. Hsu and K. Ono Page 191 Instrumented Impact Testing of Structural Fiber-Reinforced Plastic Sheet Materials and the Simultaneous AE Measurements S. I. Ochiai, K. Q. Lew and J. E. Green Page 193 On the Sensitivity of the Acoustic Barkhausen/Magnetomechanical Acouatic Emission Effect A. E. Lord, Jr. Page 195 AE Literature Thomas F. Drouillard Conferences and Symposia Page 211 1982 ASNT Spring Conference K. Ono Page 211 Fifth Internatl Conf. on NDE in the Nuclear Industry K. Ono Page 211 The Institution of Metallurgists Meeting on AE A. P. G. Rose Page 213 Review of Progress in Quantitative NDE K. Ono Page 214 The 24th Meeting of AEWG K. Ono Page 215 The Sixth International Acoustic Emission Symposium Page 219 Symposium on Structural Faults: Inspection and Repair Page 220 First International Sympoaium on AE from Reinforced Composites Page 220 Sixth Internatl Conf. on NDE in the Nuclear Industry BOOK REVIEW Page 220 AE in Geotechnical Engineering Practice Robert M. Koerner Page 221 Acoustic Emission Clinton Heiple New Products and Services AE Events of Interest / Letter from the Editor Number 4 Page 223 In-Flight Acoustic Emission Monitoring of a Wing Attachment Component S. L. McBride and J. W. Maclachlan Page 229 Effect of Crack PreAence on In-Flight Airframe Noises in a Wing Attachment Component S. L. McBride and J. W. Maclachlan Page 237 Leak Detection Using Acoustic Emission A. A. Pollock and S.-Y. S. Hsu Page 244 Acoustic Emission from Glass/Polyester Composites; Effect of Fibre Orientation F. J. Guild, B. Harris and A. J. Willis Page 251 Changes in Acoustic Emission Peaks in Precipitaion Strengthened Alloya with Heat Treatment C. R. Heiple and S. H. Carpenter Page 263 A Miniature Optical Acoustic Emission Transducer D. C. Emmony, M. W. Godrrey and R. G. White Page 266 Classification of NDE Waveforms with Autoregressive Models Ronald B. Melton Page 271 AE Literature Thomas F. Drouillard

3

Conferences and Symposia Page 294 The l1th Meeting of the European Working Group on Acoustic Emission Roger Hill Page 300 ASTM Subcommittee E 7.04 on Acouatic Emiaaion Alan G. Beattie Page 300 Meetings of Japanese Committee on Acoustic Emission Page 301 The 10th World Conf. on Non-Destructive Testing Page 302 Conference on Periodic Inspection of Pressurized Components Page 302 1983 ASNT Spring Conference Page 302 Symposium; Nondeatructive Methods for Material Property Determination Page 303 First International Symposium on AE from Reinforced Composites Page 303 Review of Progress in Quantitative NDE Page 303 25th Meeting of Acoustic Emission Working Group BOOK REVIEW Page 303 Progress in Acoustic Emission Adrian A. Pollock ANNOUNCEMENTS AE Events of Intereat / Letter from the Editor Page I-1 Index to Volume 1 Volume 2 (1983) Number 1/2 Page 1 Rotating Machinery Diagnosis With Acoustic Emission Techniques Ichiya Sato, Takao Yoneyama, Soji Sasaki, Toshitaka Suzuki, Tomoaki Inoue, Tsuguaki Koga and Takashi Watanabe Page 11 In-Field Experience in Condition Monitoring of Rotating Machinery by Demodulated Resonance Analysis G. Buzzacchi, M. Cartoceti, C. De Michelis and C. Sala Page 19 Acoustic Emission from Environmental Cracking of a High Strength Titanium Alloy S. Yuyama, T. Kishi, Y. Hisamatsu and T. Kakimi Page 29 Detection of Corrosion Fatigue by Acoustic Emission P. Jax and B. Richter Page 39 Effect of Overaging on Acoustic Emission Behaviour of 7075-T651 Aluminum During Crack Growth S.L. McBride and J.W. Maclachlan Page 47 AE Source Identification by Frequency Spectral Analysis for an Aircraft Monitoring Application L. J. Graham and R. K. Elsley Page 57 A User's Perspoctive of Small Computer-Based Acoustic Emission Equipment M.A. Hamstad Page 64 Magnetomechanical Acoustic Emission: A Non-Destructive Characterization Technique of Precipitation Hardened Steels I. Roman, S. Maharshak and G. Amir Page 67 Acoustic Emission Couplants I: The ASTM Survey A.G. Beattie Page 69 Acoustic Emission Couplants II: Coupling Efficiencies of Assorted Materials A. G. Beattie, J. A. Baron, R. S. Algera and C. C. Feng Page 71 AE Analysis During Corrosion, Stress Corrosion Cracking and Corrosion Fatigue Processes S. Yuyama, T. Kishi and Y. Hisamatsu Page 96 Acoustic Emission, Principles and Instrumentation A.G. Beattie Page 129 AE Literature Thomas F. Drouillard Page 143 Conference and Symposia Page 143 The 25th MEETING OF AEWG A.G. Beattie Page 146 AEWG Awards Page i AEWG XXV Page ii Editorial S.L. McBride

4

A Note from the Editor Kanji Ono Number 3 Page 151 Acoustic Emission Source Kinematics Based on the Moving Dislocation Theory Masayasu Ohtsu Page 159 Characterization of Concreto Damages by Acoustic Emission Analysis Marie-Christine Reymond and Andre Raharinaivo Page 169 Effects of Solute Distribution on Acoustic Emission Behavior of Solid Solution Alloys S.-Y. S. Hsu and Kanji Ono Page 179 Acoustic Emission During Fatigue Crack Growth in 7075-T6 Aluminum at 20°C and 120°C S.L. McBride and J.W. Maclachlan Page 187 An Acoustic Pressurementer to Determine In-Situ Soil Properties Arthur E. Lord, Jr., and Robert M. Koerner Page 191 An Acoustic Emission Data Acquisition and Analysis System Using an Apple II Microcomputer P.E. Wilson and S.H. Carpenter Page 195 Acoustic Emissions in Geological Materials Arthur E. Lord, Jr. and Robert M. Koerner Page 221 AE Literature Thomas F. Drouillard Page 239 Conferences and Symposia Page i AE Event of Interest M.A. Hamstad Page ii Editorial Davis M. Egle/ERRATUM Number 4 Page 247 Resonance Analysis of Piezoelectric Transducer Elements M. Ohtsu and K. Ono Page 261 Acoustic Emission Evaluation of Metal-Elastomer Junctions M. Sorel and C. Schepacz Page 267 Partial Discharge Detection in Bushings by an Acoustic Emission Method J. Skubis Page 272 Acoustic Emission as a Measure of Material Damage under Thermal Cycling Ryszard Zuchowski and Leszek Korusiewicz Page 275 Acoustic Emission Sensors C.M. Scala Page 281 Mechanical Properties and Acoustic Emission in Laser Welded HSLA Steel G. Dionoro and R. Teti Page 289 Acoustic Emission Detection of Crack Initiation during Dynamic Fracture Testing of High Strength Materials S. I. Ochiai, M. C. Cheresh and J. E. Green Page 292 AE Literature T.F. Drouillard Page 319 Conferences and Symposia Page 319 12th EWGAE Meeting L.M. Rogers Page 327 Cover Photos Page I-1 Index to Volume 2 Volume 3 (1984) Number 1 Page 1 Acoustic Emission Due to Crack Growth, Crack Face Rubbing and Structural Noise in the CC-130 Hercules Aircraft S. L. Mcbride and J. W. Maclachlan Page 11 Determination of the Source of Acoustic Emission Generated during the Deformation of Magnesium

5

Mark Friesel and Steve H. Carpenter Page 19 Thermal Restoration of Burst Emissions in A533B Steel I. Roman, H. B. Teoh and Kanji Ono Page 27 A Generalized Theory of Acoustic Emission and Green's Functions in a Half Space Masayasu Ohtsu and Kanji Ono Page 41 Acoustic Emission Charachterization of the Mechanical Strength of Sintered SAE-316 Stainless Steel I. Roman, M. Watad and A. Mittelman Page 46 AE Literature Thomas F. Drouillard Page I Editorial S. Vahaviolos Page ii Meeting Calendar Number 2 Page 51 Description of Compound Parameters of Particle-Filled Thermoplastic Materials by Acoustic Emission Techniques Jorg Wolters Page 59 A New Method of Acoustic Emission Transducer Calibration Masayasu Ohtsu and Kanji Ono With Appendix By F.R. Breckenridge and T. Watanabe Page 69 Pattern Recognition Analysis of Magneto-Mechanical Acoustic Emission Signals Masayasu Ohtsu and Kanji Ono Page 81 An Investigation of the Acoustic Emission Generated during The Deformation of Carbon Steel Fabricated by Powder Metallurgy Techniques Yue-Huang Xu, Steve H. Carpenter and Bruce Campbell Page 90 AE Literature Thomas F. Drouillard Page 100 Conferences and Symposia Page 100 AEWG-26 (Abstracts) Page 104 AEWG-27/AEWG AWARDS Page 104 AEWG Chairman's Letter Davis M. Egle Page 105 CARP-8/JCAE/ASNT Methods Committee Page 106 Other Conferences Page 107 Book Review Arthur E. Lord, Jr. Page i AE Events Of Interest Page ii 2nd Internat'l AE Conf/EWGAE-13/7th AE Symp. Page ii Ultrasonics 85/ASME Symposium Number 3 Page 108 Monitoring of Metal Cutting and Grinding Processes by Acoustic Emission Y. Kakino Page 118 Classification of Acoustic Emission Signals from Deformation Mechanisms in Aluminum Alloys D. Robert Hay, Roger W.Y. Chan, Douglas Sharp and Khalid J. Siddiqui Page 130 Effects of Interfacial Segregation on Acoustic Emission Behavior of A533B Steel H.B. Teoh, Kanji Ono, E. Kobayashi and I. Roman Page 144 Magnetomechanical Acoustic Emission of Ferromagnetic Materials at Low Magnetization Levels (Type I Behavior) May Man Kwan, Kanji Ono and M. Shibata Technical Note Page 158 Acoustic Emission during Nickel Electroplating of Copper I. De Iorio, F. Langella and R. Teti Page 164 AE Literature Thomas F. Drouillard Page 172 Conferences and Symposia Page 172 EWGAE-13

6

Page 173 5th Colloquium on AE D. Schumann and W. Morgner Page 174 Book Review Kanji Ono Page 117 Meeting Calendar Page 157 Available Books on AE Page 175 Cover Photograph Number 4 Page 176 Measurement, Detection and Analysis of Longitudinal and Transverse AE Waves Emitted Near A Crack C. Duytsche, P. Fleischmann and D. Rouby Page 182 Three-Dimensional Crack Location by Acoustic Emission C. B. Scruby and G. R. Baldwin Page 190 Magnetomechanical Acoustic Emission of Ferromagnetic Materials At High Magnetization Levels (Type II Behavior) May Man Kwan, Kanji Ono and M. Shibata Technical Note Page 204 A Report on the Pulsed Acoustic Emission Technique Applied to Masonry James D. Leaird Page 212 AE Literature Thomas F. Drouillard Page 224 Conferences and Symposia Page 224 AEWG-27 A. G. Beattie Page 226 EWGAE-13 (Abstracts) Page 233 7th Int'l Ae Symposium Kanji Ono Page 239 ASTM A. G. Beattie/PROC. AE Workshop R. H. Jones et al. Page 242 ASME Symposium D. A. Dornfeld/4th Conf on AE/MA H. R. Hardy, Jr. Page 243 Cover Photograph/Announcements Page 181 Meeting Calendar Page 189 Available Books On AE Page i Ae Events Of Interest Page ii Editorial Kanji Ono Page ii Thomas/Hochwald Award for Dr. Tim Fowler Page I-1 Index to Volume 3 Volume 4 (1985) Number 1 Page 1 Acoustic Emission Monitoring of Flaw Growth in A Graphite-Epoxy Experimental Wing Segment John Rodgers Page 9 Acoustic Emission Measurments Using Point-Contact Transducers C. B. Scruby Page 19 Relationship of Acoustic Emission to Internal Bond Strength of Wood-Based Composite Panel Materials F. C. Beall Page 31 Laboratory Leak Detection in Gas and Liquid Storage Tanks Using Continuous Wave Acoustic Emission A. E. Lord, Jr., R. M. Koerner and R. N. Sands Page 41 AE Literature Thomas F. Drouillard Page 61 Conferences And Symposia Page 61 AEWG-27 (Abstracts) Kanji Ono Page 65 ASTM E-7.04 Alan G. Beattie Page 66 EWGAE-14 (Provisional Programme) Page 67 WCNDT-11/ Other Conferences Page 68 AE in Brazil Page 40 Call for Papers (2nd Int'l Symp AE from RP/8th Int'l AE Symp) Page 69 EWGAE-14 Page 70 2nd Internat'l AE Conf

7

Page i AE Events of Interest Page ii ASME Boiler and Pressure Vessel Code/Activities at AEWG-27 Numbers 2 and 3 Pages S1-S332 Proceedings of the Second International Conference on Acoustic Emission, Oct. 28 - Nov. 1, 1985 S1 Characterizing Fracture Types in Rock/Coal Subjected to Quasi-Static Indentation Using Acoustic Emission Technique, A. Wahab Khair S7 Applications of Statistical Inference to Improve an Evaluation of Rockburst Danger in Underground Coal Mines, Stanislaw Lasocki S11 Field Determination of Prestress (Existing Stress) in Soil and Rock Masses Using Acoustic Emission, A. E. Lord, Jr. and R. M. Koerner S17 Directional Acoustic Emission Activity in Response to Borehole Deformation in Rock Masses, Robert J. Watters and Amir Soltani S19 Acoustic Emission during Dissolution of Salt, H. Reginald Hardy, Jr. S21 Kaiser Experiment in sawcut Rock, J. D. Leaird, J. Dunning and M.E. Miller S26 Acoustic Emission Technique for Solid Propellant Burn Rate Control, V. Lalitha, S. K. Athithan and V. N. Krishnamurthy S30 AE Montioring of Jet Engine Breech Chambers, Nitin Dhond and Davis M. Egle S32 Acoustic Emission Studies for Detection and Monitoring Incipient Cracks in a Simulated Aero Engine Mount under Fatigue, S. C. Pathak and C. R. L. Murthy S35 Post-Test Selective Screening of Acoustic Emission Data - How Helpful Is It?, B.C. Dykes S38 Comparison between Experimentally Detected Surface Motions Due to a Disbonding and Simulated Waveforms, Shigenori Yuyama, Takuichi Imanaka and Masayasu Ohtsu S42 Attenuation and Dispersion in AE Waveforms: a Comparison of Theory and Experiment, R. A. Kline and S. S. Ali S46 The Effects of Transducers on the Decay of a Diffuse Energy Field, H. A. L. Dempsey and Davis M. Egle S50 The Generalized Theory and Source Representations of Acoustic Emission, Masayasu Ohtsu and Kanji Ono S54 Experimental Studies of Diffuse Waves for Source Charcterization, Richard L. Weaver S58 Effect on Flaw Location by the Wave Shape of Acoustic Emission Propogating in a Limited Medium, Yukuan Ma S62 Applications of Quantitative AE Method; Dynamic Fracture, Materials and Transducer Characterization, Wolfgang Sachse and K. Y. Kim S64 A Case for Acoustic Emission Surveilence of Operating Reactors, William F. Hartman S69 Characterization of Acoustic Emission Signals Generated by Water Flow Through Intergranular Stress Corrosion Cracks, T. N. Claytor and D. S. Kupperman S74 On-Line Acoustic Emission Monitoring of Nuclear Reactor Systems - Status and Future, P. H. Hutton S77 Acoustic Leak Detection in Nuclear Power Plants, John W. McElroy

8

S78 On the Application of Acoustic Emission Analysis to Evaluate the Integrity of Protective Oxide Coatings, H. Jonas, D. Stover and R. Hecker S82 Utilization of Acoustic Emission for Detection, Measurement and Location of Partial Discharges, Jerzy Skubis, Jerzy Ranachowski and Boguslaw Gronowski S86 Acoustic Emission Examination of Power Plant Components, G. Tonolini, G. Villa and S. Ghia S90 Acoustic Emission from Aluminum Alloy 6061 Strengthened by Whisker and Particulate Silicon Carbide, James R. Kennedy S94 Acoustic Emission during Phase Transformation in Alloys GCr15 and Fe-32Ni, B. Q. Zhang, C. K. Yao and R. M. Tian S98 Acoustic Emission Studies of Structural Relaxation in a Metallic Glass , G. L. Goswami, P. K. K. Nair, P. Raj and G. P. Tiwari S102 Effects of Secondary Phases on Acoustic Emission in 316 Stainless Steel and a Nimonic Alloy PE-16 during Tensile Deformation and Fracture, Baldev Raj, T. Jayakumar, D. K. Bhattacharya and P. Rodriguez S106 Burst-Type Behavior of Structural Steels, O. Y. Kwon, I. Roman and Kanji Ono S111 Acoustic Emission Behavior of an Advanced Aluminum Alloy, I. Roman, Kanji Ono and C. H. Johnson S116 Acoustic Emission Produced by the Deformation of Uranium, C. R. Heiple and S. S. Christian S119 An Investigation of the Acoustic Emission Generated during the Deformation and Fracture of Molybdenum, J. B. James and S. H. Carpenter S123 Manufacturing Process Monitoring and Analysis using Acoustic Emission, David A. Dornfeld S127 Acoustic Emission of Flexible Disk Magnetic Media Systems, Ming-Kai Tse and Armand F. Lewis S131 Acoustic Emission Monitoring of Drilling, M. W. Hawman S132 Vibro-Acoustic Emission - A Conventional Means of Inspection using AE Technology, J. R. Webster and T. J. Holroyd S134 Weld Penetration Monitoring Using Acoustic Emission, J. Maram and J. Collins S135 Using Acoustic Emission Measurements to Establish the Quality of Bonding of Fine Wires to Microchips, S. H. Carpenter, D. R. Smith and J. H. Armstrong S137 Real-Time Aircraft Structural Monitoring by Acoustic Emission, S. Y. Chuang S138 Aircraft Structure Surveillance in-Flight Using Acoustic Emission, P. H. Hutton S142 In-Flight AE Monitoring, G. G. Martin and I. G. Scott S147 In-Flight Monitoring for Incipient Cracks in An Aero Engine Mount: An Approach Through Pattern Recognition, C. R. L. Murthy, M. A. Majeed, S. C. Pathak and A. K. Rao S151 Acoustic Emission Monitoring of Aircraft Structures, S. L. McBride and J. W. Maclachlan S155 Real-Time Acoustic Emission Monitoring Requirements For Cold Proof Testing of the USAF F-111 Aircraft, J. M. Rodgers S157 Leakage Test by Acoustic Emission Testing (AET) on Flat Bottom Tanks, Peter Tscheliesnig and Heinrich Theiretzbacher

9

S161 Acoustic Emission of Offshore Structures; Attenuation - Noise - Crack Monitoring, Steinar Lφvaas S165 Acoustic Emission Monitoring of a Node in An Off-Shore Platform, A. B. M. Hoff and M. Arrington S166 Acoustic Emission Monitoring of a Bellows and Wye Section on a Fluidized Catalytic Cracking Unit, Thomas Gandy, Martin Peacock and Bobby Wright S170 Acoustic Emission Modulus Determination and Source Location in Unidirectional Fibre Reinforced Polymer Composites, A. M. G. Glennie, T. J. Gulley and J. Summerscales S174 Acoustic Emission Monitoring of Iosipescu Shear Test On Glass Fibre-Epoxy Composites, M. K. Sridhar, Iyer Subramaniam, Chandra Ajay and A. K. Singh S178 Fracture Mechanisms Characterization in Discontinuous Fibre Composites using Acoustic Emission Amplitudes, J-M. Berthelot S182 Testing Stress Corrosion of Glass Reinforced Plastic With Acoustic Emission Monitoring, Leszek Golaski and Andrzej Figiel S186 Analysis of Fatigue Damage in CFR Epoxy Composites by Means of Acoustic Emission: Setting up a Damage Accumulation Theory, M. Wevers, I. Verpoest, E. Aernoudt and P. De Meester S191 Characteristics of Acoustic Emission Generated from GFRP during Tensile Test, Kusuo Yamaguchi, Hirotada Oyaizu, Yasuaki Nagata and Teruo Kishi S195 Acoustic Emission during Load-Holding and Unload-Reload in Fiberglass-Epoxy Composites, M. Shiwa, M. Enoki and T. Kishi S199 Acoustic Emission Monitoring of Composite Damage Occurring under Static and Impact Loading, D. S. Gardiner and L. H. Pearson S203 Fatigue Crack Closure Study, Guozhi Lu S207 Amplitude Distribution Analysis of Acoustic Emission during Fatigue Testing of Steels Used in Offshore Structures, R. Visweswaren, M. Manoharan, G. Jothinathan and O. Prabhakar S211 AE Monitoring of Corrosion Fatigue Growth; Secondary AE Sources, Christian Thaulow S215 Slow Strain Rate Stress Corrosion Cracking of Compact Tension Specimen and Measurement with Acoustic Emission, X. Q. Zhu and J. Z. Xiao S220 Temperature Dependence of Inclusion-Fracture-Related Acoustic Emissions in 7075-T651 Aluminum, S. L. McBride and J. Harvey S224 Asssessment of Fatigue Damage with Acoustic Emission, M. Nabil Bassim S228 Monitoring the Wood Cutting Process with Acoustic Emission, Richard L. Lemaster and David A. Dornfeld S232 Acoustic Emission Characterization of Wood Fiber Hardboard, Henrique L. M. dos Reis S236 Detection of Western Hemlock Wood in Very Early Stages of Decay using Acoustic Emissions, Masami Noguchi and Koichi Nishimoto S240 Application of AE to Mechanical Testing of Wood, Keiichi Sato, Takeshi Okano, Ikuo Asano and Masami Fushitani S244 Effect of Moisture Conditioning on AE from Particleboard, Frank C. Beall

10

S247 In-Process Acoustic Emission Monitoring of Laser Welds, J. W. Whittaker, T. M. Mustaleski and K. D. Nicklas S251 Monitoring of Thin Welds by Acoustic Emission, G. L. Goswami and P. R. Roy S255 Defect Detection in Stainless Steel Uranus 45 Tig Welded Joints by Acoustic Emission, Vincenzo Dal Re, B. Birolo and F. Cipri S259 Classification of Acoustic Emission Signals Generatied during Welding, Roger W. Y. Chan, D. Robert Hay, Khalid J. Siddiqui and Douglas R. Sharp S263 Acoustic Emission Behavior of Metal Matrix Composites, C. Johnson, Kanji Ono and D. Chellman S269 Detection of Crack Initiation and Propagation Using Acoustic Emission, W. G. Reuter S270 Fracture Toughness Measurement of a Nicrmov Steel by Acoustic Emission, Vincenzo Dal Re S274 Microcrack Initiation and Acoustic Emission of A533B Steel in Fracture Toughness Tests, Takanori Ohira and Yih-Hsing Pao S278 Quantitative Evaluation of Microcrackings during Fracture Toughness Testing of Al2O3 by AE Source Characterization, S. Wakayama and T. Kishi S282 Three Dimensional Location and Quantitative Evaluation of Cracking Size in Ti Alloy by Acoustic Emission Source Characterization, T. Kishi, H. Ohyama and K. H. Kim S286 The Features and the Mechanism of AE Generation from Fatigue Cracks of SUS304 Piping Components, Kusuo Yamaguchi, Hirotada Oyaizu and Akio Yamashita S290 Acoustic Emission Study in Arctic Sea Ice in a Field Laboratory, N. K. Sinha S294 Acoustic Emission Monitoring of the Main Shaft in the Hydroelectric Power Plant, H. Imaeda, H. Kimura and A. Yasuo S296 Acoustic Emission Applied to Reinforced Concrete Wall, M. C. Reymond and M. Diez S300 Damage Process Characterization in Concrete by Acoustic Emission, J-M. Berthelot and J-L. Robert S304 Detection of Fatigue Cracks in Highway Bridges with Acoustic Emission, David W. Prine and Theodore Hopwood II S307 Laboratory Acoustic Emission Investigation of Full Size ASTM A-588 Bridge Beams, Al Ghorbanpoor and Donald W. Vannoy S311 Acoustic Emission Made Audible by Time Dilation of Digitally-Recorded AE Signals, Nelson N. Hsu and Steven E. Fick S312 Magnetoelastic Resonance Spectroscopy, Wolfgang Stengel S316 Discrimination of Fracture Mechanisms via Pattern Recognition Analysis of AE Signals during Fracture Testing, Kanji Ono and Masayasu Ohtsu S321 Some Design Concepts for an Accurate, High Speed AE Signal Acquisition Module, T. Kevin Bierney S325 Advanced Acoustic Emission Monitoring System by Distributed Processing Waveform Microdata and the System Configuration, Kusuo Yamaguchi, Takashi Hamada, Hatsuo Ichikawa, Hirotada Oyaizu, Teruo Kishi and Hisashi Ishitani

11

S329 Advanced AE Instrumentation Concepts With Real Time Source Identification Through Correlation Plots and Soft Ware Filtering, S. J. Vahaviolos and John M. Carlyle S330 Authors Index Number 4 Page 71 Evaluation of Pattern Recognition Analysis of Acoustic Emission from Stressed Polymers and Composites R. M. Belchamber, D. Betteridge, Y. T. Chow, T. Lilley, M. E. A. Cudby and D. G. M. Wood Page 85 The Use of Acoustic Emission to Measure the Ductility of Hardened Surface Layers J. Roget and D.P. Souquet Page 93 The Detection of Longitudinal Rail Force via Magnetomechanical Acoustic Emission M. Shibata, E. Kobayashi and K. Ono Page 103 Application of Acoustic Emission on Railroad Car Cushioning Devices V. Godinez and W.D. Jolly Page 107 Methods of Calculating Attenuation and Dispersion Effects on Acoustic Emission Signals R. A. Kline and S. S. Ali Page 115 Classification of Acoustic Emission Signals Generated during Welding Roger W.Y. Chan, D. Robert Hay, Victor Caron, Michel Hone and R. Douglas Sharp Page 124 Conferences and Symposia Page 124 The 2nd International Conference on AE/28th AEWG Meeting Page 124 29th AEWG Meeting/AE Primer/The 8th International AE Symposium Page 125 AE Training Course Announced Page S333 Extended Abstracts from The Second International Conference on AE Page I - 1 Index to Volume 4 Volume 5 (1986) Number 1 Page 1 Nondestructive Monitoring of Installed Refractories by Acoustic Emission David A. Bell Page 7 Improving the Reliability of Critical Parts by Acoustic Emission Surveillance during Proof Testing Edward Goliti Page 15 On the Applicability of Amplitude Distribution Analysis to the Fracture Process of Composite Materials Luis Lorenzo and H. Thomas Hahn Page 25 Detection, Measurements and Location of Partial Discharges in High Power Transformers using Acoustic Emission Method Jerzy Skubis, Jerzy Ranachowski and Boguslaw Gronowski Page 31 Is it Time for Acoustic Emission Surveillance of Operating Nuclear Reactors? W. F. Hartman Page 39 Fracture Toughness Measurement of a NiCrMoV Steel by Acoustic Emission V. Dal Re Page 45 Uncommon Cries of Cast Iron Elucidated by Acoustic Emission Analysis Winfred Morgner and Hartmut Heyse

Page 51 The XVth and XIVth EWGAE Meetings Roger Hill Page 53 ASTM E7.04 Meeting A. G. Beattie / French AE Codes / E. German AE Colloquium Page 54 The Second International Conference on AE from Reinforced Plastics Page ii 29th AEWG Meeting / AE Primer Page 50 Meeting Calendar

12

Page i Appreciation K. Ono / European Perspective Roger Hill Page 6 Available Books on AE Page 14 EWGAE Provisional Form Page 59 Montreal Conference Information / Cover Photo Page 60 AE Training Courses Number 2 Page 61 Prediction of Lumber Checking during Drying by Means of Acoustic Emission Technique Shigeru Ogino, Koji Kaino and Masahiko Suzuki Page 67 On the Acousto-Ultrasonic Characterization of Wood Fiber Hardboard Henrique L. M. dos Reis and D. Michael McFarland Page 71 Effect of Moisture Conditioning on Acoustic Emission from Particleboard Frank C. Beall Page 77 Acoustic Emission During the Deformation and Fracture of Molybdenum at Low Temperatures Jay B. James and Steve H. Carpenter Page 85 Acoustic Emission Produced by the Deformation of Uranium C. R. Heiple and S. S. Christiansen Page 95 A Discussion of the Basic Understanding of the Felicity Effect in Fiber Composites M. A. Hamstad Page 103 AE Literature - Concrete Thomas F. Drouillard Page 110 Book Review M. A. Hamstad Page 111 The 29th Meeting of Acoustic Emission Working Group The Second International Symposium on Acoustic Emission from Reinforced Composites XVth Meeting of The European Working Group of Acoustic Emission (EWGAE) The 8th Int'l AE Symposium / The 30th Meeting of AEWG Acousto-Ultrasonics: Theory and Application Page 66 Meeting Calendar Page 94 The Second CARP Symposium Page 102 Cover Photogragh Page i AE Events of Interest Number 3 A SPECIAL ISSUE ON AEWG AND EWGAE MEETINGS - EXTENDED ABSTRACTS Page S1 The 29th Meeting of Acoustic Emission Working Group Page S42 XVth Meeting of The European Working Group of Acoustic Emission (EWGAE) Page 113 The 29th Meeting of Acoustic Emission Working Group (AEWG) XVth Meeting of The European Working Group of Acoustic Emission (EWGAE) The 8th International AE Symposium/ The 30th Meeting of AEWG Acousto-Ultrasonics: Theory and Application Page 114 Program of The 29th Meeting of Acoustic Emission Working Group Page 116 Abstracts of The 29th Meeting of Acoustic Emission Working Group Page 121 Program of XVth Meeting of The EWGAE Page 122 Program of The 8th International AE Symposium Page 123 The 2nd International Symposium on AE from Reinforced Composites M. A. Hamstad Number 4 Page 124 The Generalized Theory and Source Representations of Acoustic Emission Masayasu Ohtsu and Kanji Ono Page 134 More Recent Improvements on the NBS Conical Transducer Thomas M. Proctor, Jr. Page 144 Nondestructive Evaluation of Adhesive Bond Strength Using the Stress Wave Factor Technique Henrique L. M. dos Reis and Harold E. Kautz

13

Page 148 A Note on the Prediction of Fatigue Life of Metal Structures by Use of the Felicity Effect J. W. Whittaker Page 152 Vibro-Acoustic Emission in Plastic Bars A. E. Lord, Jr. Page 156 Acoustic Emission Testing of Glass Fiber Reinforced Plastic Components R. Teti Page 162 Correlation of Acoustic Emission to Microstructural Sources James Mohr and Amiya K. Mukherjee Page 172 Acoustic Emission Characterization of Pinch Welds A. G. Beattie and C. W. Pretzel Page 184 XVth EWGAE Meeting/ 8th Int'l AE Symp. DGM Symp. on AE/ 16th Symp. on NDE/ Prog. QNDE/ AE in Sci. & Tech. 30th AEWG Meeting/ Acousto-Ultrasonics/ XVIth EWGAE Meeting Page 185 9th Int'l AE Symp./ 3rd Int'l AE Conf./ 3rd Int'l Symp. on AE - Composites Page 186 Instruction for AEWG-EWGAE Extended Abstracts Page S69 Abstract; XVth EWGAE Meeting Page 155 Meeting Calendar Page 143 Call for Papers - 30th AEWG Meeting and AEWG Short Course Page 171 Registration Forms for 30th AEWG Meeting and AEWG Short Course Page 185 Call for Papers/ Announcements/ AE Training Courses Page 151 Available Books on AE Page 161 Progress in Acoustic Emission III Page i AE Events of Interest Page ii Acoustic Emission of a Kouros Kanji Ono Page I - 1 Index to Volume 5 Volume 6 (1987) Number 1 Page 1 Fracture-Induced Acoustic Emission during Slow Bend Tests of A533B Steel H.B. Teoh and Kanji Ono Page 13 Real-Time Monitoring of Multi-Pass Welding by Acoustic Emission K. Ishihara and K. Yamada Page 19 Application of Pattern Recognition Concepts to Acoustic Emission Signals Analysis C.R.L. Murthy, B. Dattaguru and A.K. Rao Page 29 Punch Stretching Process Monitoring Using Acoustic Emission Signal Analysis--Part 1: Basic Characteristics Steven Y. Liang and David A. Dornfeld Page 37 Punch Stretching Process Monitoring Using Acoustic Emission Signal Analysis - Part 2: Application of Frequency Domain Deconvolution Steven Y. Liang, David A. Dornfeld and Jackson A. Nickerson Page 43 Modeling Concrete Damage by Acoustic Emission J.M. Berthelot and J.L. Robert Page 61 Pattern Recognition Analysis of Acoustic Emission from Unidirectional Carbon Fiber-Epoxy Composites by using Autoregressive Modeling Masayasu Ohtsu and Kanji Ono Page 73 A New Approach to the Use of Acoustic Emission Peak Amplitude Distribution as a Tool of Characterizing Failure Mechanisms in Composite Materials A. Mittelman and I. Roman Page 84 Waveform Digitizers Kanji Ono Page 79 30th AEWG Meeting/8th Int'l AE Symp. M. Ohtsu Page 80 Int'l Conf. of NDE with AE Technology Page 81 Letter to Editor/An Erratum/ AE Training Courses Page 83 31st AEWG Meeting and AEWG Short Course

14

Page 83 Progress in Acoustic Emission III Number 2 Page 85 Acoustic Emission Wave Characterization: A Numerical Simulation of the Experiments on Cracked and Uncracked Specimens T. Aizawa, T. Kishi and F. Mudry Page 93 Flaw Growth in Alumina Studied by Acoustic Emission M.A. Hamstad, P.M. Thompson and R.D. Young Page 99 Acoustic Emission Characteristics in Concrete and Diagnostic Applications Masayasu Ohtsu Page 109 Measurement of the Maximum Applied Loads to Automobile Components by Acoustic Emission Technique Tatsuhiko Yoshimura and Shigeto Kano Page 115 Effects of Heat Treatment on the Acoustic Emission Generated During the Deformation of 7075-T651 Aluminum Alloy Zu-Ming Zhu and S.H. Carpenter Page 121 An Overview of Acoustic Emission Codes and Standards J.C. Spanner, Sr. Page 125 Acoustic Emission Applied to Pressure Vessels Brian R.A. Wood Page 133 Third Symposium on AE W. Morgner Page 133 Fourth European Conf. on NDT and 16th EWGAE Page 135 AE Codes/Standards Activity within ASME and ASNT J. R. Mitchell Page 136 Prof. S.H. Carpenter, 1987 University Lecturer Number 3 Page 137 Application of Acoustic Emission to the Field of Concrete Engineering Taketo Uomoto Page 145 A Method of Rapidly Estimating the Fatigue Limits by Acoustic Emission Tatsuhiko Yoshimura and Shigeto Kano Page 151 Preliminary Investigation of Acoustic Emission from Wood During Pyrolysis and Combustion Frank C. Beall Page 157 Preliminary Investigation of the Feasibility of Using Acousto-Ultrasonics to Measure Defects in Lumber R.L. Lemaster and D.A. Dornfeld Page 167 Development and Application of Acoustic Emission Methods in the United States - A Status Review P.H. Hutton Page 177 Acoustic Emission Produced by Deformation of Metals and Alloys - A Review: Part I C. R. Heiple and S.H. Carpenter Page 205 30th Meeting of AEWG D.M. Egle and R.A. Kline Page 205 A Brief Note on the Kaiser and Felicity Effects R.A. Kline and D.M. Egle Page 206 The Third Int'l Workshop on Composite Materials / The Sixth (Japanese) Conference on AE Page 166 31st Meeting of AEWG Page 156 Available Books on AE Page 176 Cover Photograph Page 208 AE Training Courses/Announcements Number 4 Page 209 The Application of Acoustic Emission Techniques in High-Temperature Oxidation Studies A. S. Khanna, B. B. Jha and Baldev Raj Page 215 Acoustic Emission Produced by Deformation of Metals and Alloys - A Review: Part II C. R. Heiple and S.H. Carpenter

15

Page 239 Acoustic Emission During Deformation and Crack Initiation of Pipeline Steels M. W. Drew, B. R. A. Wood and R. W. Harris Page 249 Acoustic Emission Waveform Analysis to Identify Fatigue Crack Propagation in a Mirage Aircraft C. M. Scala and R. A. Coyle Page 257 The Detection of Hydrogen-Assisted-Crack Formation in U - 0.8% Ti Alloy Electron-Beam Weldments J. W. Whittaker and M. W. Richey Page 261 The Detection of the Fracture of Autoclaved Aerated Concrete during Autoclave Curing Process by Acoustic Emission Satoshi Teramura, Koichi Tsukiyama and Hideaki Takahashi Page 267 31st Meeting of AEWG - Abstracts Page 273 Meetings on AE and related topics Page 238 9th Int'l AE Symp/World Meeting on AE Page 256 Cover Photograph Page 273 INDEX to Volume 6 Volume 7 (1988) Number 1 Page 1 Evaluation of Fracture Toughness of Autoclaved Lightweight Concrete by Means of Acoustic Emission Technique, Satoshi Teramura, Koichi Tsukiyama and Hideaki Takahashi Page 9 Identification of Crack Propagation Modes in 304 Stainless Steel by Analysis of Their Acoustic Emission Signatures Daniel R. Smith Jr. and Steve H. Carpenter Page 21 A New Sensor for Quantitative Acoustic Emission Measurement Chung Chang and C. T. Sun Page 31 Acoustic Emission Propagation and Source Location in Small, Spherical Composite Test Specimens J. W. Whittaker, W. D. Brosey, O. Burenko and D. A. Waldrop Page 41 A High Fidelity Piezoelectric Tangential Displacement Transducer for Acoustic Emission Thomas M. Proctor, Jr. Page 49 Acoustic Emission from Fatigue Cracks in Chrome-Molybdenum Steel Cylinders P.R. Blackburn Page 57 31st Meeting of AEWG Page 20 Meeting Schedule Page 40A World Meeting on AE Page 48 Instructions to Authors - World Meeting on AE Page 30 Books Available Number 2 Page 59 Acoustic Emission Measurements on PWR Weld Material with Inserted Defects using Advanced Instrumentation P. G. Bentley and M. J. Beesley Page 81 Acoustic Emission Measurements on PWR Weld Material with Inserted Defects C. B. Scruby and K. A. Stacey Page 95 Characterization of Acoustic Emission from Thermally-Cycled Lithium Hydride J. W. Whittaker and D. G. Morris Page 103 Apparatus for Coupling an Acoustic Emission Transducer to a Rotating Circular Saw Richard L. Lemaster and David A. Dornfeld

16

Page 111 Measurement of Density Profiles in Wood Composites Using Acoustic Emission Richard L. Lemaster, Michael F. Gasick and David A. Dornfeld Page 119 Acoustic Emission during Intergranular Stress Corrosion Cracking of Iron M. A. Friesel and R. H. Jones Page 129 XVIIth Meeting of EWGAE - Program Page 129 The 9th International Symposium on AE - Program Page 80 Meeting Schedule Page 94 World Meeting on AE Page 102 Books Available/Training Courses Number 3 Page 135 Acoustic Emission from Porous Films of Aluminum Oxide during the Application of Electrical Voltage J. Sampath Kumar and S.P. Mallikarjun Rao Page 139 Third Domestic Conf. on Subsurface Acoustic Emission (Sendai, Japan) - Program Page 140 Acoustic Emission (UK) - Abstracts Page 141 Seminar on Acoustic Emission (India) - Abstracts Extended Abstracts of Presentations at AEWG(I) Seminar on ACOUSTIC EMISSION Page S1 Acoustic Emission during Tensile Deformation and Fracture in Austenitic Stainless Steels Baldev Raj and T. Jayakumar Page S13 Detection of Breakaway Oxidation of Zircaloy-2 by Acoustic Emission Technique B. K. Gaur, A. K. Sinha, B. K. Shah, P. G. Kulkarni and R. Vijayaraghavan Page S18 Magnetomechanical Acoustic Emission Behaviour of Some Structural Steels S. G. Savanur and C. R. L. Murthy Page S29 A Study of Acoustic Emission Activity in Granites during Stress Cycling Experiments M.V.M.S. Rao Page S35 Acoustic Emission Studies on Adhesive Potted Inserts of Honeycomb Sandwich Panels T.S. Sriranga and R. Samuel Page S40 Location of Weld Defects by Acoustic Emission G.L. Goswami and P.R. Roy Page S43 Thermally Induced Relaxation Behaviour in a Metallic Glass G. L. Goswami, S. K. Jha and G. P. Tiwari Page S47 Meeting Schedule/Training Courses Page S48 Announcements Number 4 Page 145 Acoustic Emission During Quasi-Static Loading/Hold /Unloading in Notched Reinforced Fiber Composite Materials S. V. Hoa and L. Li Page 161 The Acoustic Emission Generated during the Plastic Deformation of High Purity Zinc S. H. Carpenter and Chung-Mei Chen Page 167 Evaluation of Concrete Structure Deterioration via AE Observation of Core Tests Masayasu Ohtsu, Tatsuro Sakimoto, Yutaka Kawai and Syuro Yuji Page 173 Diagnosis of Rotating Slides in Rotary Compressors using Acoustic Emission Technique Ichiya Sato, Takao Yoneyama, Kouichi Sato, Toshiyuki Tanaka and Hiroaki Hata Page 179 AE-Monitoring Systems for the Detection of Single-Point and Multipoint Cutting Tool Failures Thomas Blum, Ippei Suzuki and Ichiro Inasaki

17

Page 185 Direct Measurement of Water Hammer Pressure by AE Source Wave Analysis Mikio Takemoto and Yasuhisa Hayashi Page 193 Acoustic Emission from an Industrial Applications Viewpoint Trevor J. Holroyd Page 201 Downhole AE Measurement Technique and Its Application to Geothermal Fields Hiroaki Niitsuma Page 211 Acoustic Emission Source Location in Fiber Reinforced Plastic Composites D. J. Buttle and C. B. Scruby Page 225 The Acoustic Emission Source Mechanism for Fatigue Crack Propagation in 7075 Aluminum S. L. McBride, P. Bowman and K. I. McRae Page 231 Conferences and Symposia Page 231 Third International Symposium on AE from Composite Materials/EWGAE Meeting Page 192 Meeting Schedule Page 200 Books Available Page 210 Progress in AE IV, Proc. The 9th International AE Symposium Page 224 The World Meeting on AE/ASNT Recognition/Codes and Standards Committee Page I-1 Index to Volume 7 Volume 8 (1989) Numbers 1 and 2 Pages S1-S338 Proceedings of the World Meeting on Acoustic Emission, March 20-23, 1989 S1 "Acoustic Emission Technology using Multi-parameter Analysis of Waveform and the Applications to Fracture Modes and Growth Recognition in Composites", K. Yamaguchi, H. Oyaizu, J. Johkaji and Y. Kobayashi S4 "Acoustic Emission Detection of Crack Presence and Crack Advance During Flight", S. L. McBride, M. D. Pollard, J. D. MacPhail, P. S. Bowman and D. T. Peters S8 "Time-frequency domain (3-D) Analysis of AE Signals using Simple Instrumentation Techniques", S. V. Subba Rao, K. V. Srincivasan and M. Annamalai S12 "Improving Acoustic Emission Crack/Leak Detection in Pressurized Piping by Pattern Recognition Techniques", R. W. Y. Chan, D. R. Hay, J. R. Hay and H. B. Patel S16 "An Efficient Unsupervised Pattern Recognition Procedure for Acoustic Emission Signal Analysis", M. A. Majeed and C. R. L. Murthy S20 "Solving AE Problems by a Neural Network", I. Grabec and W. Sachse S24 "The General Problems of AE Sensors", Y. Higo and H. Inaba S28 "Acoustic Emission Transducer Modelling Using System Identification Techniques", S. Kallara, P. K. Rajan and J. R. Houghton S32 "Design of 3-Dimensional AE/MS Transducer Arrays", M. Ge and H. R. Hardy S38 "Simultaneous Velocity Tomography and Source Location of Synthetic Acoustic Emission Data", S. C. Maxwell, R. P. Young, and D. A. Hutchins S42 "Use of Mechanical Waveguides and Acoustic Antennae in Geotechnical AE/MS Studies", H. R. Hardy, Jr., F. Taioli and M. E. Hager S49 "Linear Location of AE Simulated Sources on Steel Pipelines with Waveguides", B. Q. Zhang

18

S53 "AE Application and Recent Results on Nuclear Components", P. Jax and V. Streicher S57 "Acoustic Emission from Steel Structures", A. Nielsen S62 "AE Role in the Diagnosis and Prognosis of Defects in Industrial Plant Steel Components", F. Tonolini S66 "Structural Integrity Evaluation Using Acoustic Emission Techniques", B. R. A. Wood and R. W. Harris S70 "On-line Application of Acoustic Emission Analysis", W. Morgner S71 "AE - For when you absolutely cannot afford a Failure of a Critical Pressure Boundary", H. L. Dunegan S75 "Periodic Inspection of Compressed Gas Cylinders and Transport Vessels by Using Acoustic Emission Testing", H. Barthelemy S79 "Locating Fatigue Cracks by Acoustic Emission Testing", P. M. Horrigan, J. F. Finn, F. R. Tuler and J. H. Smith S84 "A New Acoustic Emission Measurement System and its Application to the Local Monitoring of a Crack in a Pressure Vessel", B. Tirbonod and L. Hanacek S88 "An Approach for the Integrity Assessment of M250 Maraging Steel Pressurized Systems", T. Chelladurai, R. Krishnamurthy and A. R. Acharya S93 "Detectability of Defects in Reactor Pressure Components by Location and Interpretation of AE-Sources", C. Sklarczyk and E. Waschkies, S97 "Application of Acoustic Emission Technique during In-service Pressure Vessel Inspection", S. Liu, G. Shen, Y. Wan and Q. Duan S101 "Acoustic Emission Leak Monitoring in Pressurized Piping", H. B. Patel and A. W. Cook S103 "Application of Acoustic Emission Technique in the High-temperature Oxidation Studies - A Review", A. S. Khanna S105 "Acoustic Emission during Phase Transformation in Cr12MoV Steel and Fe-33Ni-4Ti-10Co Alloy", B. Q. Zhang, C. K. Yao and C. G. Jiao S109 "Effect of Pre-exposure to Water on the Acoustic Emission Behavior of 2091-T3 Al-Li Alloy", F. M. Zeides and I. Roman S114 "Acoustic Emission from Weld-seam of 16MnR Steel during Stress Corrosion", B. Q. Zhang and J. Q. Sun S118 "The Influence of Crack Front Geometry on Acoustic Emission During Fatigue of Al2024-T351", F. A. Veer and J. Zuidema S122 "Acoustic Emission Monitoring of Incipient Crack Propagation and its Growth Rate in EN-24 Steel under Fatigue", S. C. Pathak and C. R. L. Murthy S126 "Acoustic Emission during Tensile Deformation and Fracture in Austenitic Alloys", B. Raj and T. Jayakumar S131 "Relationship between Acoustic Emission and Flaw Size in Si3N4 Ceramics", Y. Mori, M. Nishino, K.-I. Aoki, T. Kishi and K. Kitadate S135 "A Comparison of the Acoustic Emission Generated from the Fracture and Decohesion of Graphite Nodules with Theoretical Predictions", S. H. Carpenter and Z. Zhu

19

S140 "Influence of Grain Size on Frequency Spectra of AE Signal Generated during Tensile Deformation in an AISI Type 316 Stainless Steel", B. Raj, P. Kalyanasundaram, T. Jayakumar, P. Barat and P. Rodriguez S145 "Evaluation of Fatigue Crack Growth Rate of Carburized Gear by Acoustic Emission Technique", Y. Obata, H. Kobayashi, K. Aoki, T. Yamaguchi and K. Shibata S149 "Acoustic Emission Investigations in Poland", J. Ranachowski, Poland "Comparative Studies on Acoustic Emission Generated During Lüder's Deformation in Mild Steel and Portevin-Le Chatelier Effect in Austenitic Stainless Steel", B. Raj, T. Jayakumar, P. Kalyanasundaram, P. Barat, B. B. Jha, D. K. Bhattacharya, P. Rodriguez S154 "Development and Future Aspects in AE Source Characterization", M. Enoki and T. Kishi S158 "AE Source Modelling-Comparison of Inversion and Forward Techniques", D. J. Buttle and C. B. Scruby S162 "Source Inversion of Acoustic Emission for the Determination of Crack Kinematics and Kinetics", M. Ohtsu S166 "Acoustic Emission Analysis and Ultrasonic Velocity Imaging in the Study of Rock Failure", S. D. Falls, T. Chow, R. P. Young and D. A. Hutchins S170 "Thin-Film Acoustics: Line and Point Sources Generation and the Testing of thin Films", K. Y. Kim and W. Sachse S175 "Acousto-Ultrasonics: An Update", A. Vary S179 "Theoretical Basis of Acousto-Ultrasonics", M. T. Kiernan and J. C. Duke S184 "Fracture of Boron Particles in 2219 Aluminum as a Known Acoustic Emission Source", C. R. Heiple, S. H. Carpenter and S. S. Christiansen S188 "Surface Analysis by Tribo-acoustic Emission", M.-K. Tse and P.-Y. Gu S192 "Acoustic Emission Measurements of Rubbing Surfaces", S. L. McBride, R. J. Boness, M. Sobczyk and M. R. Viner S197 "Vibro-Acoustic Emission Rubbing Sources", J. R. Webster S201 "Acoustic Characterisation of Small Particle Impact", D. J. Buttle and C. B. Scruby S205 "Estimation of Impact Force between Rough Surfaces by Means of Acoustic Emission", I. Kukman and I. Grabec S209 "Acoustic Emission Monitoring of Magnetic Hard Disks during Accelerated Wear Testing", J. C. Briggs, M. M. Besen and M.-K. Tse S213 "Applications of Acoustic Emission Techniques for Diagnosis of Large Rotating Machinery and Mass Products", I. Sato, T. Yoneyama, K. Sato, T. Tanaka, M. Yanagibashi and K. Takigawa S217 "Quality Inspection of Rolling Element Bearings using Acoustic Emission Technique", V. Bansal, A. Prakash, V. A. Eshwar and B. C. Gupta S219 "Stress Wave Sensing - Affordable AE for Industry", T. J. Holroyd S223 "Cavitation Monitoring of Hydroturbines with True-RMS Acoustic Emission Measurement", O. Derakhshan, J. R. Houghton, R. K. Jones and P. A. March S227 "Monitoring of the Cutting Process by Means of AE Sensor", D. A. Dornfeld

20

S231 "Tool Monitoring by Acoustic Emission", J. Roget, P. Souquet, M. Deschamps and N. Gsib S236 "MONPAC - An Acoustic Emission based System for Evaluating the Structural Integrity of Metal Vessels", T. J. Fowler, J. A. Blessing and T. L. Swanson S238 "ICI's Perspective on Acoustic Emission Monitoring", S. Hewerdine S239 "MONPAC - Condition Monitoring for Static Plant - Case Histories", P. T. Cole S240 "Acoustic Emission Monitoring of a Large Pressure Vessel during a Pneumatic Re-Qualification Test", M. Peacock S241 "Summary of Experiences with MONPAC Testing by MQS/Dunegan Testing Group", R. K. Miller S242 "Microseismics and Geotechnical Applications", M. Ohtsu S246 "Acoustic Emission Phenomena in Geological Media", R. A. Kline, A. S. Khan, Y. Xiang, J. Shlyapobersky and F. Irani S250 "Acoustic Emission/Microseismic Activity at very low Strain Levels", B. H. Armstrong S254 "Focal Mechanism Determinations for a Sequence of Mining-Induced Seismic Events Recorded at a Hard Rock Mine", T. I. Urbancic, S. Talebi and R. P. Young, S258 "Effects of Porosity on Acoustic Emission Signatures", R. J. Watters and D. M. Chuck S262 "Acoustic Emission Monitoring and Analysis Procedures Utilized during Deformation Studies on Geologic Materials", X. Sun, H. R. Hardy, Jr. and M. V. M. S. Rao S266 "PDP 11/34 Based Microseismic Monitoring System for Kolar Gold Fields, India", G. Jayachandran Nair S268 "Fracture Mechanism Studies of Carbon/PMR-15 Composites by Acoustic Emission", J. S. Jeng, K. Ono and J. M. Yang S272 "Assessment of Fatigue Damage in Carbon Fibre Reinforced Epoxy Laminates with the Energy Discrimating Acoustic Emission Method", M. Wevers, I. Verpoest, P. De Meester and E. Aernoudt, S276 "NDE Procedure for Predicting the Fatigue Life of Composite Structural Members", M. J. Sundaresan, E. G. HennekeII and A. Gavens S280 "Detection of Impact Damage in Composite Structures by Use of Thermally-activated Acoustic Emission", J. W. Whittaker and W. D. Brosey S284 "Characterization of Failure Modes in Glass Fibre Reinforced Plastic Composites", M. N. Raghavendra Rao and C. R. L. Murthy S288 "Spectrum Analysis of Acoustic Emission Signals from Carbon-Glass Hybrid Composites", M. R. Madhava and H. N. Sudheendra S292 "Analysis of AE Signals in Time and Frequency Domains Coupled to Pattern Recognition to Identify Fracture Mechanisms in CFRP", A. Maslouhi and C. Roy S297 "Some New Results in the Damage Identification in Kevlar-Epoxy Composites", D. S. Rajan, N. N. Kishore and B. D. Agarwal S301 "Monitoring Initiation and Growth of Matrix Splitting in a Uni-directional Graphite/Epoxy Composite", S. Ghaffari and J. Awerbuch S306 "Correlation of Internal Bond Strength of Particleboard with Acousto-Ultrasonics", A. T. Green

21

S311 "Acoustic Emission during Contact Drying of Southern Pine Veneer", F. C. Beall S314 "Nondestructive Evaluation of Adhesively Bonded Joints using Acousto-Ultrasonics", S. Tanary, A. Fahr and Y. Haddad S317 "AE Research on Adhesion in Composite Materials", H. G. Moslé S318 "Acoustic Emission Testing of Flexed Concrete Beams Reinforced with Bonded Surface Plates", D. P. Henkel and J. D. Wood S322 "Acoustic Emission Investigation into some Concrete Construction Problems", J. D. Leaird and M. A. Taylor S326 "Correlation of AE and Fracture Intensity during Impact Indentation of Coal Block", S. J. Jung and A. W. Khair S330 "Acoustic Emission Monitoring of Pressure Vessel Preventing Catastrophic Failure", S. F. Botten S334 Limiting State Prediction from AE Signals for Large-size Structures under Static, Cyclic and Thermocyclic Loads, V.A. Strizhalo and V.A. Strelchenko S337 Authors Index Number 3 Page 1 The MONPAC System Timothy J. Fowler, James A. Blessing, Peter J. Conlisk and Terry L. Swanson Page 11 Acoustic Emission Monitoring of a Large Pressure Vessel during a Pneumatic Re-qualification Test M. J. Peacock Page 21 ICI's Perspective on Acoustic Emission Monitoring S. Hewerdine Page 25 A Summary of Experiences with MONPAC™ Testing by the MQS/Dunegan Testing Group R.K. Miller, R.G. Tobin, D.J. Gross and D.T. Tran Page 31 MONPAC - Condition Monitoring for Static Plant - Case Histories Phillip T. Cole Page 35 Detection of an Impulse Force in Head-Disk Media Contact using Small Piezoelectric Transducer Kenji Mochizuki, Isamu Sato and Takefumi Hayashi Page 41 Investigations of Sensor Placement for Monitoring Acoustic Emission in Machining David V. Hutton and Qing Huan Yu Page 47 Use of Plane-Strain Compression for the Diagnosis of Acoustic Emission Source during Plastic Deformation F. Zeides and I. Roman Page 51 Transverse Cracking and Longitudinal Splitting in Graphite/Epoxy Tensile Coupons as Determined by Acoustic Emission Steven M. Ziola and Michael R. Gorman Page 8 Cover Photographs Page 9 Meeting Calendar -- 33rd AEWG Meeting, 10th IAES, Kumamoto Workshops Page 61 Third International Symposium on AE from Composite Materials (AECM-3) Page 61 EWGAE Meeting/Future Meetings/Errata Page 34 The Status of EWGAE Panel on Standards J. Roget /Codes and Standards Committee Page 20, 30, 40 Abstracts of Third International Symposium on AE from Composite Materials (AECM-3) Page 50, 50, 62 Selected Abstracts of AECM-3 (continued)

22

Number 4 Page 65 A Review of International Research Relative to the Geotechnical Field Application of Acoustic Emission/ Microseismic Techniques H. Reginald Hardy, Jr. Page 93 A Review of Acoustic Emission in Civil Engineering with Emphasis on Concrete Masayasu Ohtsu Page 99 Application of Acoustic Emission for the Evaluation of Microseismic Source Location Techniques M. Kat and F. P. Hassani Page 107 Acoustic Emission Characteristics of Unstable Slopes A. Chichibu, K. Jo, M. Nakamura, T. Goto and M. Kamata Page 113 Experimental Studies on the Effect of Stress History on Acoustic Emission Activity -- A Possibility for Estimation of Rock Stress Sumio Yoshikawa and Kiyoo Mogi Page 125 Acoustic Emission from Phase Transformations in Au - 47.5 at. % Cd C. R. Heiple and S. H. Carpenter Page 135 The Effect of Same-side and Through-Thickness Transmission Modes on Signal Propagation in Wood Richard L. Lemaster and Stephen L. Quarles Page 92 Meeting Schedule Page 143 Abstracts of Third International Symposium on AE from Composite Materials Page 200 EWGAE Meeting/AEWG -33 Books Available Page 210 Progress in Acoustic Emission IV, Proc. The 9th International AE Symposium Page I-1 Index to Volume 8 Volume 9 (1990) Number 1 Page 1 Origin of Acoustic Emission Produced during Deformation of Beryllium C. R. Heiple and S. H. Carpenter Page 9 Optimal Waveform Feature Selection Using a Pseudo-Similarity Method K. J. Siddiqui, Y.-H. Liu, D. R. Hay and C. Y. Suen Page 17 The Effect of Same-Side and Through-Thickness Transmission Modes on Signal Propagation in Wood Richard L. Lemaster and Stephen L. Quarles Page 25 Defect Detection in Rolling Element Bearings by Acoustic Emission Method N. Tandon and B. C. Nakra Page 29 Monitoring of a Pressure Vessel (ZB2) by means of Acoustic Emission H.-A. Crostack and P. Böhm Page 37 A Procedure for Acceptance Testing of FRP Balsa Wood Core Pressure Vessels P. Ouellette, S.V. Hoa and L. Li AE Literature Page 45 A Comprehensive Guide to the Literature on Acoustic Emission from Composites, Supplement II Thomas F. Drouillard Conferences and Symposia Page 69 Abstracts of 33rd Acoustic Emission Work Group Meeting Page 72 ASNT Spring Conference Page 73 1st International Conf. on AE in Manufacturing, AECM-4

23

Page 8 Available Books on AE Page 31 Cover Photograph Page ii Meeting Calendar Number 2 Composites Page 75 Felicity Ratio Behavior of Pneumatically and Hydraulically Loaded Spherical Composite Test Specimens J. W. Whittaker, W. D. Brosey and M. A. Hamstad Page 84 Correlation of Felicity Ratio and Strength Behavior of Impact-Damaged Spherical Composite Test Specimens J. W. Whittaker, W. D. Brosey and M. A. Hamstad Page 91 Delayed Acoustic Emission: A Rheological Approach N. Rochat, R. Fougeres and P. Fleischmann Page 97 Acoustic Emission Analysis of the Accumulation of Cracks in CFRP Cross-ply Laminates under Tensile Loading J.-P. Favre and J.-C. Laizet Page 103 Analysis of Acoustic Emission Events from Single Fiber Pullout Experiments W. Mielke, A. Hampe, O. Hoyer and K. Schumacher Page 109 Digital Signal Analysis of Acoustic Emission from Carbon Fiber/ Epoxy Composites Kanji Ono and Kenji Kawamoto Page 117 Acoustic Emission: A Micro-Investigation Technique for Interface Mechanisms in Fiber Composites D. Rouby Page 123 Acoustic Emission Characterization of the Deformation and Fracture of an SiC-Reinforced, Aluminum Matrix Composite Oh-Yang Kwon and Kanji Ono Page 131 Burst Prediction by Acoustic Emission in Filament-Wound Pressure Vessels Michael R. Gorman Page 140 Critical AE Problems for the Researcher Adrian A. Pollock Page 142 Quality Inspection of Rolling Element Bearing using Acoustic Emission Technique Vibha Bansal, B.C. Gupta, Arun Prakash and V.A. Eshwar Conferences and Symposia Page 147 XIX Meeting of EWGAE, 10th International Acoustic Emission Symposium Page 152 1st Symp. Evaluation of Adv. Materials by AE Page 153 Intl Joint Meeting at Kumamoto Page 154 Korean Working Group on Acoustic Emission (KWGAE), AE Training Courses Page 122 Meeting Schedule Page 130 34th AEWG Meeting Page 154 Cover Photograph Number 3 Wood Frank C. Beall, Topical Editor Page 155 Anecdotal History of Acoustic Emission from Wood Thomas F. Drouillard Page 177 An Experiment on the Progression of Fracture (A Preliminary Report) Fuyuhiko Kishinouye (Translated by Kanji Ono) Page 181 Acoustic Emission from Drought-Stressed Red Pine (Pinus resinosa) Robert A. Haack and Richard W. Blank Page 189 The Effect of Moisture Content and Ring Angle on the Propagation of Acoustic Signals in Wood Stephen L. Quarles

24

Page 197 Nondestructive Evaluation of Adhesive Bond Strength of Finger Joints in Structural Lumber Using the Acousto-Ultrasonic Approach Henrique L. M. dos Reis, Frank C. Beall, Michael J. Chica and Dick W. Caster Page 203 Determining the Abrasiveness to Tools of Wood-Based Composites with Acoustic Emission Richard L. Lemaster Page 209 Lumber Stress Grading utilizing the Acoustic Emission Technique Keiichi Sato, Hajime Takeuchi, Katsuya Yamaguchi, Naoto Ando and Masami Fushitani Page 215 AE Literature - Wood Thomas F. Drouillard and Frank C. Beall Conferences and Symposia 196, 214, 223 10th International Acoustic Emission Symposium (abstracts) Page 226 Intl Joint Meeting at Kumamoto/ KWGAE/ 34th AEWG Meeting, AE Training Courses Page 188 Meeting Schedule Page 180 Fuyuhiko Kishinouye (biography) / Cover Photograph Number 4 Page 227 Detection of Irradiation Effects on Reactor Vessel Steels by Magneto-Acoustic Emission Oh-Yang Kwon and Kanji Ono Page 237 New Algorithm for Acoustic Emission Source Location in Cylindrical Structures Dong-Jin Yoon, Young H. Kim and Oh-Yang Kwon Page 243 Characterization of Fatigue of Aluminum Alloys by Acoustic Emission, Part I - Identification of Source Mechanism D. J. Buttle and C. B. Scruby Page 255 Characterization of Fatigue of Aluminum Alloys by Acoustic Emission, Part II - Discrimination Between Primary and Other Emissions D. J. Buttle and C. B. Scruby Page 271 Acoustic Emission Source Location Using Simplex Optimization M. P. Collins and R. M. Belchamber Page 277 Acoustic Emission during Lumber Drying S. Ogino, K. Kaino and M. Suzuki Page 283 AE Source Orientation by Plate Wave Analysis Michael R. Gorman and William H. Prosser Page 289 Conferences and Symposia 10th International AE Symp. Page 270 Cover Photograph AE Inspection of Monorail Trains Page 276A Meeting Calendar I-i Index to Volume 9 Volume 10 (1991/92) Number 1/2 Page i Current Research and Future Trend of AE Applications to Civil Engineering and Geological Technology Masayasu Ohtsu, Topical Editor S1-S12 Variety of Acoustic Emission Waveforms Produced by Discrete Crack Growth in Rock Steven D. Glaser and Priscilla P. Nelson S13-S17 Estimation of Maximum Stress in Old Railway Riveted I-Girder Bridges using Acoustic Emission Signals Hisanori Otsuka, Hiroshi Hikosaka, Hiroyuki Miyatake and Syouzou Nakamura S18-S21 Another Look at Booming Sand Marcel F. Leach and Gottfried A. Rubin

25

S22-S28 Expected Acoustic Emission from Around a Shaft in Intact Rock B.J.S. Wilkins and G.L. Rigby S29-S34 Acoustic Emission Monitoring during Microseismic Activity Caused by Mine Subsidence Brian R. A. Wood and Robert W. Harris S35-S41 Acoustic Emission/Microseismic Activity Monitoring of Salt Crystallization for Stone Conservation M. Montoto, R.M. Esbert, L.M. Suárez del Río, V.G. Ruiz de Argandoña and C.M. Grossi S42-S48 Acoustic Emission Monitoring during In-situ Heater Test of Granite T. Ishida, K. Kitano, N. Kinoshita and N. Wakabayashi S49-S54 Using Acoustic Emission Testing in Seepage Investigations Andrew R. Blystra S55-S58 Acoustic Emission of Penetration Experiments to Judge Soil Condition S. Naemura, M. Tanaka, S. Nishikawa, M. Nakamura, K. Jo and T. Kishishita S59-S62 A Laboratory Investigation of AE from Coal R. W. Harris, B. R. A. Wood and T. Flynn S63-S76 Determination of the Initial Stresses on Rock Mass using Acoustic Emission Method K. Michihiro, K. Hata, H. Yoshioka and T. Fujiwara S77-S89 U.S. Bureau of Mines Research on the Kaiser Effect for Determining Stress in Rock Michael J. Friedel and Richard E. Thill S90-S96 Evaluation of Joint Properties of Anti-washout Underwater Concrete by Acoustic Emission Measurement Kazuya Miyano, Tatsuo Kita, Yuji Murakami and Takako Inaba S97-S103 Automated Determination of First P-Wave Arrival and Acoustic Emission Source Location E. Landis, C. Ouyang and S. P. Shah S104-S109 Application of Acoustic Emission Techniques in the Evaluation of Frost Damage in Mortar Hisatoshi Shimada and Koji Sakai Number 3/4 Pages 1-11 Acoustic Emission of the 45HNMFA Structural Steel during Low-Cycle Fatigue J. Siedlaczek, S. Pilecki and F. Dusek Pages 13-17 Parameter Estimation in Acoustic Emission Signals C. E. D'Attellis, L. V. Perez, D. Rubio and J. E. Ruzzante Pages 19-23 Acoustic Emission Technique at Proof Tests of Nuclear Pressure Vessels in Hungary Peter Pellionisz and János Geréb Pages 25-29 Effects of Wave Velocity Change on Magnetomechanical AE in Sintered Iron Noboru Shinke and Yoshitugu Ohigashi Pages 31-33 Origin of Acoustic Emission in Naturally Aged Aluminum-Lithium Alloys F. Zeides and I. Roman Pages 35-41 Analysis of Artificial Acoustic Emission Waveforms Using a Neural Network Hironobu Yuki and Kyoji Homma Pages 43-48 Source Force Waveforms: The Use of a Calibrated Transducer in Obtaining an Accurate Waveform of a Source Thomas M. Proctor, Jr. and Franklin R. Breckenridge Pages 49-60 Acoustic Emission Monitoring of a Fatigue Test of an F/A-18 Bulkhead C. M. Scala, J. F. McCardle and S. J. Bowles Pages 61-65 Maximum Curvature Method: A Technique to Estimate Kaiser-Effect Load from Acoustic Emission Data

26

M. Momayez, F. P. Hassani and H. R. Hardy, Jr. Pages 67-70 Acoustic Emission Measurements on Bridges H. Hick, H. Willer, E. Winter and F. Simacek Pages 71-82 Acoustic Emission Monitoring of the Sensitivity of Chemicals to Impact Timothy G. Crowther, Adrian P. Wade and Nancy Brown Pages 83-89 Study of Acoustic Emission Generation in Sliding Motion Musa K. Jouaneh, Richard Lemaster and Frank C. Beall Pages 91-95 The Role of Acoustic Monitoring as a Diagnostic Tools in Nuclear Reactors Shahla Keyvan and Ron King Pages 97-101 Acoustic Emission Produced by Sliding Friction and its Relationship to AE from Machining S. H. Carpenter, C. R. Heiple, D. L. Armentrout, F. M. Kustas and J. S. Schwartzberg Pages 103-106 Comments on the Origin of Acoustic Emission in Fatigue Testing of Aluminum Alloys C. R. Heiple, S. H. Carpenter and D. L. Armentrout Pages 107-111 Acoustic Emission of Wood during Swelling in Water Stefan Poliszko, Waldemar Molinski and Jan Raczkowski Pages 113-116 Characterization of the ASL Parameter J. W. Whittaker Pages 117-121 Acoustic Emission from Bubbles in a Water Column, Mark A. Friesel and Jack F. Dawson Pages 122-124 In Memoriam, Dr. Raymond W. B. Stephens (1902-1990) Conferences and Symposia 12, 18, 24, 30, 34 Progress in Acoustic Emission VI, Proceedings of The 11th International AE Symposium 42, 66, 90, 96, 102 36th Meeting of Acoustic Emission Working Group Page I-i Index to Volume 10

27

INDEX to Journal of Acoustic Emission, 1993-2002 Volume 11, 1993 Number 1 Pages 1-4 Acoustic Emission Inspection of Defects under Coating in Nozzles of Vessels - Fatigue Study on a Coated Steel Sample E. Verbrugghe and M. Cherfaoui Pages 5-10 Acoustic Emisslon Generated from Silicon Particles during Deformation of Al-Si Alloys Min Wu and Steve H. Carpenter Pages 11-18 Frequency Analysis of Acoustic Emission Signals in Concrete J.-M. Berthelot, M. Ben Souda and J. L. Robert Pages 19-20 Acoustic Emission of Booming Sand Analyzed in the Laboratory Marcel F. Leach and Gottfried A. Rubin Pages 21-26 Comparison of AE Source Location Methods in Paper Sheets under Tension T. Fuketa, S. Okumura, M. Noguchi and T. Yamauchi Pages 27-32 Performance of a Noncontact Magnetostrictive AE Sensor on a Steel Rod H. Kwun, J. J. Hanley and C. M. Teller Pages 33-41 Acoustic Emission Technology for Smart Structures M. A. Hamstad and G. P. Sendeckyj Pages 43-51 Influence of MC-Type Carbides on Acoustic Emission Generated during Tensile Deformation in a Nimonic Alloy PE16 T. Jayakumar, Baldev Raj, D. K. Bhattacharya, P. Rodriguez and O. Prabhakar

Conferences and Symposia Pages 42, 52 Second International Conference on Acousto-Ultrasonics, Review of Progress in Quantitative Nondestructive Evaluation, Future Meetings Pages 53-59 AE Literature T. F. Drouillard Pages 60 Available Books on AE Number 2 Pages 61-63 Broadband Acoustic Emission Sensor with a Conical Active Element in Practice Miroslav Koberna Pages 65-70 Acoustic Emission Signal Trends during High Cycle Fatigue of FRP/Balsa Wood Core Vessels P. Ouellette and S.V. Hoa Pages 71-78 Analysis of the Acoustic Emission Generated by the Failure of Oxide Scales and Brittle Lacquer Layers M. M. Nagl, Y. S. Chin and W. T. Evans Pages 79-84 Solution of a Simple Inverse Source Characterization Problem using Associative Recall Kornelija Zgonc, Igor Grabec and Wolfgang Sachse Pages 85-94 Acoustic Emission during Fatigue of a Nickel Base Superalloy Daining Fang and Avraham Berkovits Pages I - XXXI Cumulative Index, J. of Acoustic Emission, Volumes 1 - 10, 1982 - 92 Pages I - VII Author Index Pages VII-XXXI Contents, Volumes 1 - 10 Page 64 Future Meetings on AE Page XXXI Cover Photograph, Available Books on AE

28

Number 3 Page C1-C24 Guidance for Development of AE Applications on Composites CARP Aerospace/Advanced Composites Subcommittee Page 95-99 Finite Element Simulation of Acoustic Emission due to Fiber Failure in a Single Fiber Composite S. De Bondt, L. Froyen, L. Delaey and A. Deruyttere Page 101-106 Acoustic Emission from Aluminum during Hydrostatic Extrusion L. Hanumantha Sastry and S. P. Mallikarjun Rao Page 107-115 Nondestructive Evaluation of Damage in Steel-Belted Radial Tires using Acousto-Ultrasonics Henrique L. M. dos Reis and Kris A. Warmann Page 117-128 A Study of Fracture Dynamics in a Model Composite by Acoustic Emission Signal Processing Hiroaki Suzuki, Mikio Takemoto and Kanji Ono Page 116 Conferences and Symposia Page 129 The 37th Meeting of Acoustic Emission Working Group Page 100 Available Books on AE Page 115 Cover Photograph Page 130 Early Days of AEWG Allen T. Green Page 130 Internet Address List, Erratum Number 4 Fifth National Conference on Subsurface and Civil Engineering Acoustic Emission, Japan Page i Preface Hiroaki Niitsuma, Topical Co-Editor Page i Research Activities on Acoustic Emission in Civil Engineering in Japan Masayasu Ohtsu, Topical Co-Editor Page ii Papers Presented at the Conference Page S1-S18 Analysis of Acoustic Emission from Hydraulically Induced Tensile Fracture of Rock Hiroaki Niitsuma, Koji Nagano and Koji Hisamatsu Page S19-S26 Acoustic Emission Activities during the Injection of High Pressure Water into Coal Measures M. Seto and K. Katsuyama Page S27-S36 The Variation of Hypocenter Distribution of AE Events in Coal under Triaxial Compression Masahiro Seto, Osamu Nishizawa and Kunihisa Katsuyama Page S37-S46 Assessment of Concrete Deterioration using Plastic Analysis and Acoustic Emission Technique Ahmed M. Farahat and Masayasu Ohtsu Page S47-S56 Principal Components Analysis of AE Waveform Parameters for Investigating an Instability of Geotechnical Structures Akiyoshi Chichibu, Tadashi Kikuchi and Takahiro Kishishita Page S57-S63 Observation of Mixed-Mode Fracture Mechanism by SiGMA-2D Mitsuhiro Shigeishi and Masayasu Ohtsu Page S65-S73 Field Application of Acoustic Emission for the Diagnosis of Structural Deterioration of Concrete K. Matsuyama, T. Fujiwara, A. Ishibashi and M. Ohtsu Page S75-S88 An Evaluation of Subsurface Fracture Extension Using AE Measurement in Hydraulic Fracturing of a Geothermal Well Masayuki Tateno, Mineyuki Hanano and Qiang Wei Page S89-S98 Assessment of Concrete Deterioration by Acoustic Emission Rate Analysis Masayasu Ohtsu, Kunihiro Yuno and Yoshiki Inoue Page S56 Cover Photograph Page S64 Papers Presented at the Conference (continued) Page S74 Available Books on AE Page S98 AECM-5

29

Volume 12, 1994 Number 1/2 Acousto-Ultrasonics Pages i-ii Foreword Frank C. Beall and Alex Vary, Topical Co-Editors Pages 1-14 Nondestructive Evaluation of Adhesively Bonded Joints by Acousto- Ultrasonic Technique and Acoustic Emission H. Nayeb-Hashemi and J. N. Rossettos Pages 15-21 Acousto-Ultrasonic Nondestructive Evaluation of Porosity in Polymer- Composite Structures of Complex Geometry Henrique L. M. dos Reis Pages 23-26 Wave Mechanics in Acousto-Ultrasonic Nondestructive Evaluation Joseph L. Rose, John J. Ditri and Aleksander Pilarski Pages 27-38 Laser-based Techniques to Resolve Mode Propagation of Lamb Waves in Plates R. Daniel Costley, Jr., Yves H. Berthelot and Laurence J. Jacobs Pages 39-44 Adhesive Bond Evaluation Using Acousto-Ultrasonics and Pattern Recognition Analysis A. Fahr, Y. Youssef and S. Tanary Pages 45-54 Acousto-Ultrasonic Signal Classification to Evaluate High Temperature Degradation in Composites A. Maslouhi, H. Saadaoui, S. Béland and C. Roy Pages 55-64 The Use of Acousto-Ultrasonics to Detect Biodeterioration in Utility Poles Frank C. Beall, Jacek M. Biernacki and Richard L. Lemaster Pages 65-70 Determination of Plate Wave Velocities and Diffuse Field Decay Rates with Broadband Acousto-Ultrasonic Signals Harold E. Kautz AE Literature Pages 71-78 Acousto-Ultrasonic Reflections Thomas F. Drouillard and Alex Vary Pages 79-103 Acousto-Ultrasonics Thomas F. Drouillard and Alex Vary Conferences and Symposia Page 22 37th Meeting of AEWG Pages 104-106 12th International AE Symposium Page 106 Cover Photograph Number 3/4 Pages 107-110 A Double Exponential Model for AE Signals M. A. Majeed and C. R. L. Murthy Pages 111-115 Acoustic Emission Response of Centre Cracked M250 Maraging Steel Welded Specimens T. Chelladurai, A. S. Sankaranarayanan and K. K. Purushothaman Pages 117-126 Low Strain Level Acoustic Emission due to Seismic Waves and Tidal/Thermoelastic Strains Observed at the San Francisco Presidio Baxter H. Armstrong, Carlos M. Valdes-Gonzalez, Malcolm J. S. Johnston and James D. Leaird Pages 127-140 Fracture Analysis of Mullite Ceramics using Acoustic Emission Technique Yoshiaki Yamade, Yoshiaki Kawaguchi, Nobuo Takeda and Teruo Kishi Pages 141-148 Acoustic Emission from AISI 4340 Steel as a Function of Strength Steve H. Carpenter and Christian Pfleiderer Pages 149-155 Acoustic Emission in Laser Bending of Steel Sheets H. Frackiewicz, J. Królikowski, S. Pilecki, A. M. Leksowskij, B. L. Baskin and E. W. Khokhlova Pages 157-170 On the Far-field Structure of Waves Generated by a Pencil Lead Break on a Thin Plate John Gary and Marvin A. Hamstad Pages 171-176 Improving the Coupling Reproducibility of Piezoelectric Transducers D. Geisse AE Literature Pages 177-198 Trends of Recent Acoustic Emission Literature Kanji Ono

30

12th International Acoustic Emission Symposium Pages S1-S6 Acoustic Emission Monitoring for Shield Tunneling Akiyoshi Chichibu, Akimasa Waku, Teruyuki Waki and Hiroshi Yoshino Pages S7-S11 Acoustic Emission of Coal Induced by Gas and Water Flow, Gas Sorption or Stress Z. J. Majewska, S.A. Majewski, H. Marcak, W. J. Mos cicki, S. Tomecka-Suchon and J. Zie tek Pages S12-S17 Acoustic Emission Study of Anisotropic Stress Memory in Rock Subjected to Cyclic Polyaxial Loading C. E. Stuart, P. G. Meredith and S. A. F. Murrell Pages S18-S23 Characterization of Acoustic Emission Signals During Phase Transformations in a TiNiFe Shape Memory Alloy Kazuki Takashima and Minoru Nishida Pages S24-S28 Simulation of AE Generation Behavior during Fracture of Alumina Ceramics Byung-Nam Kim, Hidehumi Naito and Shuichi Wakayama Conferences and Symposia Page 116 Future Meetings of AE Page 156 38th Meeting of AEWG Pages 199-200 5th International Symposium on Acoustic Emission from Composite Materials Page 200 On the Cover Pages I-i, I-ii Index to Volume 12 Volume 13, 1995 Numbers 1/2 Pages 1-10 Early Detection of Damages in Journal Bearings by Acoustic Emission Monitoring Dong-Jin Yoon, Oh-Yang Kwon, Min-Hwa Chung and Kyung-Woong Kim Pages 11-22 Clustering Methodology for the Evaluation of Acoustic Emission from Composites A. A. Anastassopoulos and T. P. Philippidis Pages 23-29 Investigation of AE Signals Emitted from an SiOx Layer Deposited on a PET Film Masa-aki Yanaka, Noritaka Nakaso, Yusuke Tsukahara and Nelson N. Hsu Pages 31-41 On Characterization and Location of Acoustic Emission Sources in Real Size Composite Structures -A Waveform Study M. A. Hamstad and K. S. Downs Conferences and Symposia: 12th International Acoustic Emission Symposium Pages S01-07 A Waveform Investigation of the Acoustic Emission Generated during the Deformation and Cracking of 7075 Aluminum Steve H. Carpenter and Michael R. Gorman Pages S08-13 Acoustic Emission and Damage Evolution in an SiC Fiber Reinforced Ti Alloy Composite K. Takashima, H. Tonda and P. Bowen Pages S14-20 Cracking Process Evaluation in Reinforced Concrete by Moment Tensor Analysis of Acoustic Emission Shigenori Yuyama, Takahisa Okamoto, Mitsuhiro Shigeishi and Masayasu Ohtsu Pages S21-28 The Interaction between Pore Fluid Pressure Changes and Crack Damage Evolution in Rocks And Subsurface Rock Structures Modeled from Acoustic Emission Data Peter Sammonds, Philip Meredith, Javier Gomez and Ian Main Pages S29-34 Acoustic Emission Analysis of TiAl Intermetallics Manabu Enoki and Teruo Kishi Pages S35-41 Recent Applications of Acoustic Emission Testing for Plant Equipment Masashi Amaya Pages S42-46 Effects of Soil Acidity on Acoustic Emission Properties of Sugi (Cryptomeria Japonica) Seedling Keiichi Sato, Atsushi Uchiyama, Takeshi Izuta, Makoto Miwa, Naoaki Watanabe, Takafumi Kubo and Masami Fushitani Pages S47-53 Acoustic Emission of Bending Fatigue Process of Spur Gear Teeth Kouitsu Miyachika, Satoshi Oda and Takao Koide

31

Pages S54-59 Characterization of Thermal Cracking by Acoustic Emission Time Series Analysis Koji Nagano, Katsuhiro Sugawara, Ken-Ichi Itakura and Kazuhiko Sato Page 30 Future Meetings on Acoustic Emission Pages 42-44 Historic Files (AEWG Meetings, AEWG Awards), T.F. Drouillard / Call for papers Page S60 Available Books on Acoustic Emission Page S07 Cover Photograph Numbers 3/4

Composites Pages 45-55 Correlation of Acoustic Emission Felicity Ratios and Hold-Based Rate Moments with Burst Strengths of Spherical Graphite/Epoxy Pressure Vessels, Karyn S. Downs and Marvin A. Hamstad Pages 56-66 Correlation of Regions of Acoustic Emission Activity with Burst Locations for Spherical Graphite/Epoxy Pressure Vessels, Karyn S. Downs and Marvin A. Hamstad Pages 67-77 A Study of Acoustic Emission-Rate Behavior in Glass Fiber-Reinforced Plastics, A.J. Brunner, R. Nordstrom and P. Flüeler Pages 79-86 The Deterioration of Foamglas® under Compression Studied with the Acoustic Emission Technique, M. Wevers, D. Tsamtsakis, P. De Meester, E. Uria and H. Strauven Pages 87-96 Localization of Acoustic Emission in the Fracture of Fiber Composites, Vladimir Krivobodrov Pages 97-100 Investigation of Damage Development in Paper Using Acoustic Emission Monotoring, Per A Gradin and Staffan Nyström Conferences and Symposia: 12th International AE Symposium, Sapporo, Japan Pages S61-S67 Acoustic Emission Behavior during Plastic Deformation of 8090 Al-Li Alloy, Ki-Jung Hong, Hee-Don Jeong and Chong Soo Lee Pages S68-S74 Development Of Thermal Shock and Fatigue Tests of Ceramic Coatings for Gas Turbine Blades by AE Technique, C.Y. Jian, Tatsuya Shimizu, Toshiyuki Hashida, Hideaki Takahashi and Masahiro Saito Pages S75-S82 Fractals on Acoustic Emission during Hydraulic Fracturing Ken-Ichi Itakura, Kazuhiko Sato, Koji Nagano and Yasufumi Kusano Pages S83-S88 Acoustic Emission during Tensile Loading of Low Velocity Impact-Damaged CFRP Laminates, Oh-Yang Kwon, Joon-Hyun Lee and Dong-Jin Yoon Pages S89-S94 AE Characterization of Compressive Residual Strength of Impact-Damaged CFRP Laminates, Isamu Ohsawa, Isao Kimpara, Kazuro Kageyama, Toshio Suzuki and Akihiko Yamashita Pages S95-S102 Acoustic Emission Analysis on Interfacial Fracture of Laminated Fabric Polymer Matrix Composites, Toshiyuki Uenoya Page 78 Future Meetings on Acoustic Emission Page 100 Cover Photograph and AECM-5 Page S74 Professor Hideaki Takahashi (1940-1995) Pages I-i Index to Journal of Acoustic Emission, Volume 13 Volume 14, 1996 Number 1 Pages 1-34 A History of Acoustic Emission Thomas F. Drouillard Pages 35-50 The Fracture Dynamics in a Dissipative Glass-Fiber/ Epoxy Model Composite with AE

32

Source Simulation Analysis Hiroaki Suzuki, Mikio Takemoto and Kanji Ono Conferences and Symposia Pages 51-52 22nd European Conference on AE Page 52 Cover Photograph Number 2 Pages 53-59 Modeling of Stress-Strain Response of Unidirectional and Cross-Ply SiC/CAS-II Ceramic Composites by Acousto-Ultrasonic Parameters Anil Tiwari, Edmund G. Henneke II and Alex Vary Pages 61-68 Neural Network Approach to Acoustic Emission Source Location Vasisht Venkatesh and J.R. Houghton Pages 69-84 Wavelet Transform of Acoustic Emission Signals Hiroaki Suzuki, Tetsuo Kinjo, Yasuhisa Hayashi, Mikio Takemoto and Kanji Ono with Appendix by Yasuhisa Hayashi Pages 85-95 Acoustic Emission in a Nextel 440 Fiber Reinforced 6061 Al Composite T. Pocheco, H. Nayeb-Hashemi and H. M. Sallam Pages 97-102 Pattern Recognition of Acoustic Signatures Using ART2-A Neural Network Shahla Keyvan and Jyothi Nagaraj Pages 103-114 Far-field Acoustic Emission Waves by Three-Dimensional Finite Element Modeling of Pencil-Lead Breaks on a Thick Plate M. A. Hamstad, J. Gary and A. O'Gallagher Pages 115-118 Acoustic Emission Testing of Bolted Connections under Tensile Stress V. Hänel and W. Thelen Pages 119-126 A Method to Determine the Sensor Transfer Function and its Deconvolution from Acoustic Emission Signals Bernhard Allemann, Ludwig Gauckler, Wolfgang Hundt and F. Rehsteiner Conferences and Symposia Page 60 39th Meeting of the Acoustic Emission Working Group and Primer Page 96 40th Meeting of the Acoustic Emission Working Group and Primer Page 127-128 13th International AE Symposium (IAES-13) Number 3/4 Proceedings of International Workshop at Schloss Ringberg Pages i-iv Materials Research with Advanced Acoustic Emission Techniques Alexander Wanner and Michael R. Gorman, Topical Co-Editors Page v Friedrich Förster and Erich Scheil, Two Pioneers of Acoustic Emission Alexander Wanner Pages vi-viii Acoustic Investigation of Martensite Needle Formation by Fritz Förster and Erich Scheil, Translated by Peter G. Thwaite and Alexander Wanner Pages S1-S11 Advanced AE Techniques in Composite Materials Research William H. Prosser Pages S12-S18 Digital Signal Processing of Modal Acoustic Emission Signals Steve Ziola Pages S19-S46 Wave Theory of Acoustic Emission in Composite Laminates Dawei Guo, Ajit Mal and Kanji Ono Pages S47-S60 Fiber Fragmentation and Acoustic Emission Alexander Wanner, Thomas Bidlingmaier and Steffen Ritter Pages S61-S73 Wave Propagation Effects Relative to AE Source Distinction of Wideband AE Signals from a Composite Pressure Vessel Karyn S. Downs and Marvin A. Hamstad Pages S74-S87 Relative Moment Tensor Inversion Applied to Concrete Fracture Tests C. U. Grosse, B. Weiler and H. W. Reinhardt Pages S88-S101 Brittle Fracture as an Analog to Earthquakes: Can Acoustic Emission Be Used to Develop a

33

Viable Prediction Strategy? David A. Lockner Pages S102-S105 Abstracts of Talks Presented Conferences and Symposia Page S106 6th International Symp. on Acoustic Emission from Reinforced Composites Page S107 14th International Acoustic Emission Symposium and 5th Acoustic Emission World Meeting Pages S108-S109 40th Meeting of the Acoustic Emission Working Group and Primer Page S110 Available Books on AE Pages I-i - I-ii Index to Vol. 14 Volume 15, 1997 Page 1-18 Wideband and Narrowband Acoustic Emission Waveforms from Extraneous Sources during Fatigue of Steel Samples M.A. Hamstad and J.D. McColskey Page 19-32 Fracture-Mode Classification Using Wavelet-Transformed AE Signals from a Composite Tetsuo Kinjo, Hiroaki Suzuki, Naoya Saito, Mikio Takemoto and Kanji Ono Page 33-42 Nondestructive Evaluation of Fiberglass-Reinforced Plastic Subjected to Localized Heat Damage Using Acoustic Emission H. Nayeb-Hashemi, P. Kisnomo and N. Saniei Page 43-52 An Investigation of Lüders Band Deformation and the Associated Acoustic Emission in Al - 4.5% Mg Alloys D. L. Armentrout and S. H. Carpenter Page 53-61 Modal Analysis of Acoustic Emission Signals H.L. Dunegan Page 63-68 An Acoustic Emission Tester for Aircraft Halon-1301 Fire-Extinguisher Bottles Alan G. Beattie Page 69-78 AE Detection of Cracking in Pipe Socket Welds Bryan C. Morgan Page 79-87 Feature Extraction of Metal Impact Acoustic Signals For Pattern Classification by Neural Networks Shahla Keyvan and Rodney G. Pickard Conferences and Symposia

Fourth Far East Conference On NDT (FENDT '97) October 8-11, 1997, Cheju-Do, Korea, sponsored by Korean Society of Nondestructive Testing.

Page S1-S10 Source Location in Highly Dispersive Media by Wavelet Transform. of AE Signals Oh-Yang Kwon and Young-Chan Joo Page S11-S18 Acoustic Emission Monitoring of the Fatigue Crack Activity in Steel Bridge Members Dong-Jin Yoon, Seung-Seok Lee, Philip Park, Sang-Hyo Kim, Sang-Ho Lee and Young-Jin Park Page S19-S30 Acousto-Ultrasonic Evaluation of Adhesively Bonded CFRP-Aluminum Joints Seung-Hwan Lee and Oh-Yang Kwon Page S31-S39 Estimation of Initial Damage in Concrete By Acoustic Emission Masayasu Ohtsu, Yuichi Tomoda and Taisaku Fujioka Page S40-S49 Application of AE to Evaluate Deterioration of Port and Harbor Structures Kimitoshi Matsuyama, Akichika Ishibashi, Tetsuro Fujiwara, Yasuhiro Kanemoto, Shiro Ohta, Shigenori Hamada and Masayasu Ohtsu Page S50-S59 Acoustic Emission Diagnosis of Concrete-Piles Damaged By Earthquakes Tomoki Shiotani, Norio Sakaino, Masayasu Ohtsu and Mitsuhiro Shigeishi Page S60-S69 Observation of Damage Process in RC Beams under Cyclic Bending by Acoustic Emission

34

Mitsuhiro Shigeishi, Masayasu Ohtsu, Nobuyuki Tsuji and Daisuke Yasuoka Page S70-S79 Spectral Response and Acoustic Emission of Reinforced Concrete Members under Fatigue Bending Yasunori Sakata and Masayasu Ohtsu Page S80-S88 Acoustic Emission Behavior during Tensile Deformation of Welded Steel Joints J. H. Huh, K. A. Lee and C. S. Lee Page S90-S94 Acoustic Emission Signal Analysis in C/C Composites Ja-Ho Koo, Byung-Nam Kim, Manabu Enoki and Teruo Kishi

Congreso Regional de Ensayos No Destructivos y Estructurales Oct. 27-30, 1997, Mendoza, Argentina, sponsored by Comisión Nacional de Energía Atómica and Universidad Tecnologica Nacional.

Page S95-S102 Recent Development in Acoustic Emission Kanji Ono Page S103-S110 40th AEWG Meeting, Program and Abstracts Page S111-S124 14th International Acoustic Emission Symposium & 5th Acoustic Emission World Meeting, Abstracts of Oral Briefs Page S125-S126 Program of the 6th AECM Symposium Page 62 Meeting Calendar Page 68 Cover Photograph Page I-1 Index to Volume 15 Volume 16, 1998 Number 1-4 Page S1 Improvements Of Grinding/Dressing Monitoring Using Acoustic Emission Jason W.P. Dong Page S10 An Investigation Of Brittle Failure In Composite Materials Used For High Voltage Insulators D. Armentrout, T. Ely, S. Carpenter, and M. Kumosa Page S19 AE in Tooth Surface Failure Process of Spur Gears Hirofumi Sentoku Page S25 Long-Term Continuous Monitoring Of Structural Integrity Of Steel Storage Tanks Hiroyasu Nakasa and Hiroaki Sasaki Page S35 Using Of Non-Stationary Thermal Fields and Thermal Stresses As A Method For Evaluating The Danger Of Damage Development In Chemical and Refinery Equipment Boris Muravin, Luidmila Lezvinsky, Gregory Muravin Page S45 Acoustic Emission and Electric Potential Changes of Rock Sample under Cyclic Loading Y. Mori, K. Sato, Y. Obata and K. Mogi Page S53 Correlations Of AE Signatures To Mechanical and Petrologic Properties Of Four Types Of Rocks A. Wahab Khair Page S65 Damage Mechanics and Fracture Mechanics Of Concrete By SiGMA M. Ohtsu, M. Shigeishi and M. C. Mumwam Page S75 Acoustic Emission Applications To An Arch Dam Under Construction S. Yuyama, T. Okamoto, O. Minemura, N. Sakata, K. Murayama Page S85 Acoustic Emission Measurements During Hydraulic Fracturing Tests In A Salt Mine Using A

Special Borehole Probe Gerd Manthei, Jürgen Eisenblätter, Peter Kamlot and Stefan Heusermann Page S95 Evaluation Of Progressive Slope-Failure By Acoustic Emission Tomoki Shiotani and Masayasu Ohtsu Page S105 Fractal Description Of Acoustic Emission Produced In Systems: Coal-Gas and Coal-Water Zofia Majewska and Zofia Mortimer

35

Page S115 Characterization Of The Lamb Waves Produced By Local Impact Fracture In Thin Brittle Plates Yoshihiro Mizutani, Hideo Cho , Mikio Takemoto and Kanji Ono Page S125 Acoustic Emission and Magnetic Flux Leakage Associated With Magnetisation Of Cracked and Uncracked Ferromagnetic Materials R. Hill, A-A. R. Choudhury and L. Morgan Page S134 Wavelet Transform Of Magnetomechanical Acoustic Emission Under Elastic Tensile Stress With Displacement Sensor Masanori Takuma, Noboru Shinke, and Kanji Ono Page S142 Effect Of Boron Addition On Acoustic Emission Behavior During Tensile Deformation Of Ni3Al Intermetallic Compound Single Crystals K. Yoshida, Y. Iwata, H. Takagi and K. Sakamaki Page S150 Acoustic Emission Characteristics During Deformation Of Polyether Ether Ketone (PEEK) M. Gakumazawa, M. Akiyama, C. Ishiyama, J. Hu, K. Takashima, Y. Higo and C. Nojiri Page S158 Principals Of Statistical and Spectral Analysis of Acoustic Emission and Their Application To Plastic Deformation of Metallic Glasses A. Yu. Vinogradov Page S170 Microfracture Process In Ceramics Under Thermal Shock Fracture Characterized By Acoustic Emission Shuichi Wakayama Page S178 AE Study Of Stress Corrosion Cracking Mechanism Of Stainless Foil Using Quantitative Lamb Wave Analysis and Video Images Mikio Takemoto, Okiharu Tamura and Hiroaki Suzuki Page S186 A Study Of The Acoustic Emission From Musical Sand and Silica Gel Marcel F. Leach, Douglas E. Goldsack, Cindi Kilkenny and Chantal Filion Page S196 Effects Of Humidity On Acoustic Emission Characteristics During Environmental Stress Cracking In Polymethyl Methacrilate (PMMA) Chiemi Ishiyama, Takumi Sakuma, Yasuyuki Bokoi, Masayuki Shimojo and Yakichi Higo Page S204 Optimizing AE Location Accuracy: A Measurement Approach Richard Nordstrom Page S212 Source Location in Plates by Using Wavelet Transform of AE Signals Oh-Yang Kwon and Young-Chan Joo Page S222 Waveform Analysis Of Acoustic Emission Signals Ajit Mal and Dawei Guo and Marvin Hamstad Page S233 Selection Of Acoustic Emissions and Classification Of Damage Mechanisms In Fiber Composite Materials Torsten Krietsch and Jurgen Bohse Page S243 Grey Correlation Analysis Method Of Acoustic Emission Signals For Pressure Vessels Gongtian Shen, Qingru Duan and Bangxian Li Page S251 On Wideband Acoustic Emission Displacement Signals As a Function Of Source Rise- Time and Plate Thickness M. A. Hamstad, J. Gary and A. O'Gallagher Page S261 2-D AE Source Localization On The Material With Unknown Propagation Velocity Of AE Wave Kyung-Young Jhang, Weon-Heum Lee, Dal-Jung Kim Page S269 Three Dimensional Acoustic Emission Signal Analysis In C/C Composites With Anisotropic Structure Ja-Ho Koo, Manabu Enoki and Teruo Kishi and Byung-Nam Kim Page S277 Calibration Of Low-Frequency Acoustic Emission Transducers H. Reginald Hardy, Jr. and Euiseok Oh (The Pennsylvania State University) Page S289 Advanced AE Signal Classification For Studying The Progression Of Fracture Modes In Loaded UD-GFRP Naoya Saito, Hiroaki Suzuki, Mikio Takemoto and Kanji Ono Page S299 Fatigue Monitoring of Heat Exposed Carbon Fiber/Epoxy By Means Acoustic Emission and Acousto-Ultrasonic A. Maslouhi and V.L. Tahiri Page S309 Thermal Shock Evaluation Of Functionally Graded Ceramic/Metal Composites By AE Jae-Kyoo Lim and Jun-Hee Song Page S317 Effect Of Surface Modification Of SiC Fiber On Acoustic Emission Behaviors and Interface Strength Of SiCf/Al Composite Zuming Zhu, Yanfeng Guo, Nanling Shi Page S324 Characterization Of Fracture Process In Short-Fiber-Reinforced Plastics By Acoustic Emission K. Takahashi and N. S. Choi Page S333 Effects of Foam Thermal Insulation and Previous Thermal Exposure on the Acoustic

36

Emission Recorded From Graphite/Epoxy Pressure Vessels With and Without Impact Damage K. S. Downs and M. A. Hamstad Page S343 Interpretation Of Fracture Toughness In Unidirectional Glass-Fiber/Polypropylene Composites By Acoustic Emission Analysis Of Damage Mechanisms Jurgen Bohse and Torsten Krietsch Page i-iii Preface Page iii Conference Events Page iv Theme Discussion Page v In Memoriam Page vi-viii Contents Page viii Cover Photograph Page I-1 Authors index Page I-2 Selected e-mail address of authors Volume 17, 1999 Numbers 1/2 Page i Announcement: New Format for Journal of Acoustic Emission Page ii Founding members of GLEA, Grupo Latinoamericano de Emisión Acústica (Cover photograph); Color plate for Figure 8 on page 9. Page 1-13 Classification of Acoustic Emissions in Metallic Glasses A. Vinogradov Page 15-21 Acoustic Emission Characteristics of Soil and Sand in Response to Simulated Root

Growth C. Divaker Durairaj, L. Okushima and S. Sase Page 23-27 Detection of Defects in Gears by Acoustic Emission Measurements

N. Tandon and S. Mata Page 29-36 Acoustic Emission Signals in Thin Plates Produced by Impact Damage William H. Prosser, Michael R. Gorman and Donald H. Humes Page 37-47 Reflections of AE Waves in Finite Plates: Finite Element Modeling and Experimental Measurements W. H. Prosser, M. A. Hamstad, J. Gary and A. O’Gallagher Page 49-59 Classification of Acoustic Emission Signatures Using a Self-organization Neural Network Tinghu Yan, Karen Holford, Damian Carter and John Brandon Page 61-67 Discussion of the Log-Normal Distribution of Amplitude in Acoustic Emission Signals

M. I. López Pumarega, R. Piotrkowski and J. E. Ruzzante Page 69-81 Unsupervised Pattern Recognition Techniques for the Prediction of Composite Failure

T. P. Philippidis, V. N. Nikolaidis and J. G. Kolaxis Page 83-93 Real-Time Tool Condition Monitoring in Cold Heading Machine Processes Using an Acoustic Approach Henrique L.M. dos Reis, David B. Cook and Aaron C. Voegele Conferences and Symposia Page 14 Meeting Calendar Page 22-95 42nd Meeting of The Acoustic Emission Working Group Page 96 Available Books on Acoustic Emission/Short Courses Numbers 3/4 Page 97 Modeling of Buried Monopole and Dipole Sources of Acoustic Emission with a Finite Element Technique M. A. Hamstad, A. O'Gallagher and J. Gary Page 111 Numerical Assessment of the Quality of AE Source Locations Gang Qi and Jose Pujol Page 121 Structural Integrity and Remnant Life Evaluation Using Acoustic Emission Techniques

37

Brian R. A. Wood, Robert W. Harris and Elizabeth L. Porter Page S1 Selected Papers from International Conference Acoustic Emission ’99 Page S2 Report on International Conference Acoustic Emission ’99 Pavel Mazal and Václav Svoboda Page S7 Identification of fundamental forms of partial discharges based on the results of frequency analysis of their acoustic emission T. Boczar Page S13 Technical possibilities of the non-contact acoustic emission method at testing hollow articles integrity G. Budenkov, O. Nedzvetskaya, E. Bulatova Page S20 Method of AE and possibilities of corrosion degradation detection M. Cerny, P. Mazal, V. Suba Page S29 Application of acoustic emission in metal physics and materials science F. Chmelík, P. Lukác Page S37 Waveform analysis of acoustic emission during pressurization of glass-fiber composite pipes L. Golaski, P. Gebski, I. Baran, Kanji Ono Page S45 Acoustic emission monitoring during solidification processes F. Havlícek, J. Crha Page S51 Radiation of acoustic emission waves during stress corrosion cracking of the metal A. Kotolomov, G. Budenkov, O. Nedzvetskaya Page S57 NDE of phase transformations in Cu based shape memory alloys by ultrasonic techniques M. Landa, M. Chlada, Z. Prevorovsky Page S65 VVER steam generators and acoustic emission O. Matal, J. Zaloudek, T. Simo Page S70 Application of acoustic emission technique on fatigue testing machine Rumul P. Mazal, J. Richter Page S78 Acoustic emission and state of fatigue of ferroelectric Pb(ZrxTi1-x)O3 ceramics J. Nuffer, D. Lupascu, J. Rödel Page S83 Application of AE method at pressure tests of boiler header V. Svoboda, J. Petrasek, A. Proust Page S92 Acoustic emissions of vessels with partially penetrated longitudinal seams F. Rauscher Page S100 Electromagnetic emission from polycrystalline solids J. Sikula, B. Koktavy, I. Kosiková, J. Pavelka, T. Lokajícek Page S108 The testing of LPG vessels with acoustic emission examination P. Tscheliesnig, J. Liöka Conferences and Symposia Page S116 24th EWGAE Meeting (EWGAE 2000)/ The 43rd Meeting of Acoustic Emission Working Group/ 15th International AE Symposium (IAES-15)/ Short courses Page S6 Next EWGAE Meeting / Cover Photograph Page I-i Index to Volume 17 Volume 18, 2000 Selected papers from “EWGAE 2000, 24th European Conference on Acoustic Emission Testing”, published by CETIM, Senlis, France Page 1 Studies of the non-linear dynamics of acoustic emission generated in rocks Z. Majewska and Z. Mortimer Page 8 Acoustic emission as result of tensile and shearing processes in stable and unstable fracturing of rocks J. Pininska Page 15 Relation between acoustic emission signals sequences induced by thermal loading and the structure of sedimentary rocks B. Zogala, and R. Dubiel Page 21 Acoustic emission/Acousto-Ultrasonic data fusion for damage evaluation in concrete

38

A. Tsimogiannis, B. Georgali, and A. Anastassopulos (Envirocoustics S.A. - Greece) Page 29 Concrete Crossbeam Diagnostic by acoustic emission method Z. Weber, P. Svadbik, M. Korenska and L. Pazdera Page 34 Waveform based analysis techniques for the reliable acoustic emission testing of composite structures M. Surgeon, C. Buelens, M. Wevers, and P. De Meester Page 41 Optical fibres for in situ monitoring the damage development in composites and the relation with acoustic emission measurements M. Wevers, L. Rippert, and S. Van Huffel Page 51 Lamb-wave source location of impact on anisotropic plates H. Yamada, Y Mizutani, H Nishino, M. Takemoto and K. Ono Page 61 Characterisation of the damage and fracture mechanisms in Ti3SiC2 using acoustic emission P. Finkel, R.K. Miller, M.A. Friesel, R.D. Finlayson, P.T. Cole, M.W. Barsoum and T. El-Raghy Page 68 Evaluation of martensitic transformation dynamics of Cu-Al-Ni shape memory alloy single crystals by acoustic emission method K. Yoshida, S. Kihara, and K. Sakamaki Page 75 The identification of basic fatigue parameters on electroresonance pulsator with help of acoustic emission technology P. Mazal, and J. Petras Page 81 The phase of contact damage and its description by help of acoustic emission J. Dvoracek, J. Pazdera, and L. Petras (Brno University of Technology - Czech Republic) Page 87 Acoustic emission monitoring of delayed hydride cracking in Zirconium A. Barron and C. Rowland Page 96 Examination of plate valve behaviour in a small reciprocating compressor using acoustic emission J.D. Gill, R.D. Douglas, Y.S. Neo, R.L. Reuben and J.A. Steel Page 102 A new method of acoustic emission source location in Pipes using cylindrical guided waves H. Nishino, F. Uchida, S. Takashina , M. Takemoto and K. Ono Page 111 Acoustic emission method for pressure vessel diagnostics at a refinery B.S. Kabanov, V.P. Gomera, V.L. Sokolov, and A.A. Okhotnikov Page 118 Optimisation of acquisition parameters for acoustic emission measurements on small pressure vessels F. Rauscher Page 125 Monitoring of weld's defects evolution submit to static and dynamic loading thanks to the acoustic emission method C. Hervé, R. Pensec, and A. Laksimi, Page 131 Acoustic emission due to cyclic pressurisation of vessels with partially penetrated longitudinal seams M. Bayray Page 138 Inspection of LPG vessels with AE examination P. Tscheliesnig, and G. Schauritsch Page 144 Inspection of pressure vessels used in refrigeration and air conditioning systems A. Skraber, F. Zhang, M. Cherfaoui, and L. Legin Page 150 The new Russian standards in the field of acoustic emission V.I. Ivanov, and L.E. Vlasov Page 155 Using acoustic emission to monitor metal dusting F. Ferrer, E. Andres, J. Goudiakas, and C. Brun (Elf Atochem - France) Page 161 Use of acoustic emission to detect localised corrosion philosophy of industrial use, illustrated with real examples A. Proust, and J. C. Lenain Page 167 Inspection of flat bottomed storage tanks by acoustical methods. Classification of corrosion related signals P. Tscheliesnig, G. Lackner, M. Gori, H. Vallen and B. Herrmann Page 174 Screening of tank bottom corrosion with a single point AE detector: AE-Simple P.J. Van De Loo, and D.A. Kronemeijer Page 180 Case histories from ten years of testing storage tank floors using acoustic emission S.N. Gautrey, P.T. Cole, and H.J. Schoorlemmer, Page 189 Acoustic emission detection of damage in reinforced concrete conduit H.W. Shen, S. Iyer, M.A. Friesel, F. Mostert, R.D. Finlayson, R.K. Miller, M.F. Carlos

39

and S. Vahaviolos, Page 196 Comparison of artificial acoustic emission sources as calibration sources for tool wear monitoring in single-point machining A. Prateepasen, Y.H.J. Au and B.E. Jones Page 205 Development of an equipment to monitoring and control the quality of resistance welding (CRAFT Project) J. Catty, Page 211 A study of small HSDI Diesel engine fuel injection equipment faults using acoustic emission J.D. Gill, R.L. Reuben, J.A. Steel, M.W. Scaife and J. Asquith Page 217 Unsupervised pattern recognition of acoustic emission from full scale testing of a wind turbine blade D. Kouroussis, A. Anastassopoulos, P. Vionis and V. Kolovos Page 224 Acoustic emission proof testing of insulated aerial man lift devices A. Anastassopoulos, A. Tsimogianis and D. Kourousis Page 232 BOXMAP - Non-invasive detection of cracks in steel box girders J.R. Watson, K.M. Holford, A.W. Davies and P.T. Cole, Page 239 Monitoring failure mechanisms in CFRP orthopaedic implants during fatigue testing A. Taylor, S. Gross, C. Rowland and P. Gregson Page 248 Continuous monitoring of rock failure by a remote AE system T. Shiotani, S. Yuyama, M. Carlos and S. Vahaviolos Page 258 New AE signal conditioner for industrial use H. Vallen, J. Forker and J. von Stebut, Page 265 New software tools for the AE-practitioner H. Vallen and J. Vallen, Page 272 Improved source location methods for pressure vessels V. Godinez, S. Vahaviolos, R.D. Finlayson, R.K. Miller, and M.F. Carlos Page 279 Neural network localization of noisy AE events in dispersive media M. Blahacek, Z. Prevorovsky, and J. Krofta Page 286 Dynamics and damage assessment in impacted cross-ply CFRP plate utilizing the wavaform simulation of Lamb wave acoustic emission Y. Mizutani, H. Nishino, M. Takemoto and K. Ono Page 293 Acoustic emission detection during stress corrosion cracking at elevated pressure and temperature R. Van Nieuwenhove, and R.W. Bosch Page 299 Detection of pitting corrosion of aluminiurn alloys by acoustic emission technique H. Idrissi, J. Derenne, and H. Mazille Page 307 Reliability of acoustic emission technique to assess corrosion of reinforced concrete H. Idrissi, and A. Limam AEWG43 Presentation Page S1 Theoretical Treatment of AE in Massive Solid….M. Ohtsu Page S7 Diagnosis of Concrete Structures by AE….M. Ohtsu EWGAE 2000 (EWGAE.pdf) Page i – v SOMMAIRE -- Content of EWGAE 2000 Proceedings

Page vi AVANT-PROPOS Mohammed CHERFAOUI, Christel RIGAULT Page vii Présentation – EWGAE

Authors Index (18Auindx.pdf) Contents of Volume 18, 2000 (18Conts.pdf)

Page I-1 – I-2 e-mail Addresses of Authors (e-mail.pdf) Page I-3 – I-6 EWGAE 2000 Participant List (Senlis.pdf)

40

VOLUME 19, 2001 ACOUSTIC EMISSION EXAMINATION OF MODE I, MODE II AND MIXED-MODE I/II INTERLAMINAR FRACTURE OF UNIDIRECTIONAL FIBER-REINFORCED POLYMERS Jürgen Bohse and Jihua Chen 1

FRACTURE DYNAMICS IN NOTCHED PMMA PLATES BY LAMB WAVE ACOUSTIC EMISSION ANALYSIS Kenji Nagashima, Hideo Nishino, Mikio Takemoto and Kanji Ono 11

MONITORING OF MICRO-CRACKING DURING HEATING OF HYDROGEN DOPED GERMANIUM AND SILICON SINGLE CRYSTALS BY ACOUSTIC EMISSION METHOD K. Yoshida, Y. Wang, K. Horikawa, K. Sakamaki and K. Kajiyama 22 DAMAGE DETECTION IN A FIBER-REINFORCED CYLINDER (FISHING ROD) BY GUIDED WAVE ACOUSTIC EMISSION ANALYSIS Yoshie Hayashi, Yoshihiro Mizutani, Hideo Nishino, Mikio Takemoto and Kanji Ono 35

ACOUSTIC EMISSION CHARACTERIZATION AND NUMERICAL SIMULATION OF INTERNAL DAMAGE PROGRESSION IN CFRP MULTI-DIRECTIONAL SYMMETRIC LAMINATES Isamu Ohsawa, Isao Kimpara, Kazuro Kageyama, Satoshi Abe, and Kazuo Hiekata 45 SCC MONITORING OF ZIRCONIUM IN BOILING NITRIC ACID BY ACOUSTIC EMISSION METHOD Chiaki Kato and Kiyoshi Kiuchi 53

ACOUSTIC EMISSION MONITORING OF CHLORIDE STRESS CORROSION CRACKING OF AUSTENITIC STAINLESS STEEL Shinya Fujimoto, Mikio Takemoto and Kanji Ono 63 CYLINDER WAVE ANALYSIS FOR AE SOURCE LOCATION AND FRACTURE DYNAMICS OF STRESS CORROSION CRACKING OF BRASS TUBE Fukutoshi Uchida, Hideo Nishino, Mikio Takemoto and Kanji Ono 75 DETECTION OF PRE-MARTENSITIC TRANSFORMATION PHENOMENA IN AUSTENITIC STAINLESS STEELS USING AN ACOUSTIC EMISSION TECHNIQUE T. Inamura, S. Nagano, M. Shimojo, K. Takashima and Y. Higo 85 ACOUSTIC EMISSION ANALYSIS OF CARBIDE CRACKING IN TOOL STEELS Kenzo Fukaura and Kanji Ono 91 SOURCE PARAMETERS OF ACOUSTIC EMISSION EVENTS IN SALT ROCK Gerd Manthei, Jürgen Eisenblätter, Thomas Spies and Gernot Eilers 100

GEOMETRICAL COMPLEXITY OF ROCK INCLUSION AND ITS INFLUENCE ON ACOUSTIC EMISSION ACTIVITY Ken-Ichi Itakura, Atsushi Takashima, Tatsuma Ohnishi and Kazuhiko Sato 109 APPLICATION OF AE IMPROVED b-VALUE TO QUANTITATIVE EVALUATION OF FRACTURE PROCESS IN CONCRETE MATERIALS T. Shiotani, S. Yuyama, Z. W. Li and M. Ohtsu 118 EVALUATION OF FRACTURE PROCESS IN CONCRETE JOINT BY ACOUSTIC EMISSION T. Kamada, M. Asano, S. Lim, M. Kunieda and K. Rokugo 134

41

DAMAGE DIAGNOSIS OF CONCRETE-PILES BY MACHINERY-INDUCED ACOUSTIC EMISSION T. Shiotani, S. Miwa, Y. Ichimura and M. Ohtsu 142 ACOUSTIC EMISSION MONITORING OF CLOSELY SPACED EXCAVATIONS IN AN UNDERGROUND REPOSITORY Thomas Spies and Jürgen Eisenblätter 153 ACOUSTIC EMISSION MONITORING OF THE JAS 39 GRIPEN COMBAT AIRCRAFT Dan Lindahl and Markku Knuuttila 162 DETECTION AND LOCATION OF CRACKS AND LEAKS IN BURIED PIPELINES USING ACOUSTIC EMISSION S.J. Vahaviolos, R.K. Miller, D.J. Watts, V.V. Shemyakin, and S.A. Strizkov 172 RECOMMENDED PRACTICE FOR IN SITU MONITORING OF CONCRETE STRUCTURES BY ACOUSTIC EMISSION Masayasu Ohtsu and Shigenori Yuyama 184 ACOUSTIC EMISSION FROM ACTIVE CORROSION UNDER THE INSULATION OF A SULPHUR TANK Phillip T. Cole and Stephen N. Gautrey 191 A NEW SYSTEM FOR MACHINERY DIAGNOSIS USING AE AND VIBRATION SIGNALS Atsushi Korenaga, Shigeo Shimizu, Takeo Yoshioka, Hidehiro Inaba, Hidemichi Komura and Koji Yamamoto 196

DEVELOPMENT OF ABNORMALITY DETECTION TECHNOLOGY FOR ELECTRIC GENERATION STEAM TURBINES Akihiro Sato, Eisaku Nakashima, Masami Koike, Morihiko Maeda, Toshikatsu Yoshiara and Shigeto Nishimoto 202 ACOUSTIC EMISSION SIGNAL CLASSIFICATION IN CONDITION MONITORING USING THE KOLMOGOROV-SMIRNOV STATISTIC L. D. Hall, D. Mba and R.H. Bannister 209

CHARACTERIZATION BY AE TECHNIQUE OF EMISSIVE PHENOMENA DURING STRESS CORROSION CRACKING OF STAINLESS STEELS A. Proust, H. Mazille, P. Fleischmann and R. Rothea 229 INVESTIGATION OF ACOUSTIC EMISSION WAVEFORMS ON A PRESSURE VESSEL Mulu Bayray 241 EFFECTS OF LATERAL PLATE DIMENSIONS ON ACOUSTIC EMISSION SIGNALS FROM DIPOLE SOURCES M. A. Hamstad, A. O'gallagher and J. Gary 258 A WAVELET-BASED AMPLITUDE THRESHOLDING TECHNIQUE FOR AE DATA COMPRESSION Gang Qi and Eng T. Ng 275 ACOUSTIC EMISSION MONITORING OF FATIGUE OF GLASS-FIBER WOUND PIPES UNDER BIAXIAL LOADING Pawel Gebski, Leszek Golaski and Kanji Ono, 285 Contents of Volume 19 i Authors Index iii e-Mail Addresses of Selected Authors v AE Literature (CARP/TEXAS DOT; AGU-Vallen Wavelet; JAE Indices) vi Meeting Calendar ix

42

VOLUME 20, 2002 Contents 20-001 DAMAGE ESTIMATION OF CONCRETE BY AE RATE PROCESS ANALYSIS

Masayasu Ohtsu, Makoto Ichinose and Hiroshi Watanabe 1 20-016 EXTRACTION OF HISTORIES OF DAMAGE MICRO-MECHANISMS IN

UNIDIRECTIONAL COMPOSITES BY TRANSIENT-PARAMETRIC ANALYSIS Jie Qian and Yuris Dzenis 16

20-025 ACOUSTIC EMISSION SOURCES BY ATOMISTIC SIMULATIONS M. Landa, J. Cerv, A. Machová and Z. Rosecky 25 20-039 A WAVELET TRANSFORM APPLIED TO ACOUSTIC EMISSION SIGNALS: PART 1: SOURCE IDENTIFICATION M. A. Hamstad, A. O’Gallagher and J. Gary 39 20-062 A WAVELET TRANSFORM APPLIED TO ACOUSTIC EMISSION SIGNALS: PART 2: SOURCE LOCATION M. A. Hamstad, A. O’Gallagher and J. Gary 62 20-083 DIAGNOSTICS OF REINFORCED CONCRETE BRIDGES

BY ACOUSTIC EMISSION Leszek Golaski, Pawel Gebski and Kanji Ono 83 20-099 ANALYSIS AND IDENTIFICATION OF ACOUSTIC EMISSION FROM DAMAGE AND

INTERNAL FRETTING IN ADVANCED COMPOSITES UNDER FATIGUE Yuris Dzenis and Jie Qian 99

20-108 ACOUSTIC EMISSION FROM MAGNESIUM-BASED ALLOYS AND METAL MATRIX COMPOSITES Frantisek Chmelík, Florian Moll, Jens Kiehn, Kristian Mathis, Pavel Lukác, Karl-Ulrich Kainer and Terence G. Langdon 108 20-129 TRAINING AND CERTIFICATION ON THE FIELD OF ACOUSTIC EMISSION

TESTING (AT) IN ACCORDANCE WITH THE EUROPEAN STANDARDISATION (EN 473) P. Tscheliesnig 129

20-134 THRESHOLD COUNTING IN WAVELET DOMAIN Milan Chlada and Zdenek Prevorovsky 134 20-145 DAMAGE DIAGNOSIS TECHNIQUE FOR BRICK STRUCTURES USING ACOUSTIC

EMISSION Takuo Shinomiya, Yasuhiro Nakanishi, Hiroyuki Morishima and Tomoki Shiotani 145

20-153 EVALUATION OF DRYING SHRINKAGE MICROCRACKING IN CEMENTITIOUS MATERIALS USING ACOUSTIC EMISSION

Tomoki Shiotani, Jan Bisschop and J. G. M. Van Mier 153

43

20-163 TRANSFORMATION PROCESSES IN SHAPE MEMORY ALLOYS BASED ON MONITORING ACOUSTIC EMISSION ACTIVITY Michal Landa, Václav Novák, Michal Blahácek and Petr Sittner 163

20-172 CHARACTERIZATION OF STRESS CORROSION CRACKING OF CuZn-ALLOYS BY ACOUSTIC EMISSION TESTING U.-D. Hünicke, M. Schulz, R. Budzier and J. Eberlein 172

20-179 ACOUSTIC EMISSION TESTING ON FLAT-BOTTOMED STORAGE TANKS: HOW TO CONDENSE ACQUIRED DATA TO A RELIABLE STATEMENT REGARDING FLOOR CONDITION

G. Lackner and P. Tscheliesnig 179 20-188 ACOUSTIC EMISSION MEASUREMENTS ON SHELL STRUCTURES WITH

DIRECTLY ATTACHED PIEZO-CERAMIC Franz Rauscher and Mulu Bayray 188 20-194 AE MONITORING OF CRYOGENIC PROPELLANT TANK

Yoshihiro Mizutani, Takayuki Shimoda, Jianmei He, Yoshiki Morino and Souichi Mizutani 194

20-206 NON-DESTRUCTIVE TESTING FOR CORROSION MONITORING IN CHEMICAL PLANTS M. Winkelmans and M. Wevers 206

20-218 AE TECHNOLOGY AS A KEY ELEMENT OF THE OPERATION SAFETY SYSTEM AT REFINERY

B. S. Kabanov, V. P. Gomera, V. L. Sokolov, A. A. Okhotnikov and V. P. Fedorov 218 20-229 STRUCTURAL INTEGRITY EVALUATION OF WIND TURBINE BLADES USING

PATTERN RECOGNITION ANALYSIS ON ACOUSTIC EMISSION DATA A. A. Anastassopoulos, D. A. Kouroussis, V. N. Nikolaidis, A. Proust, A. G. Dutton, M. J.

Blanch, L. E. Jones, P. Vionis, D. J. Lekou, D. R. V. Van Delft, P. A. Joosse, T. P. Philippidis, T. Kossivas and G. Fernando 229

20-238 LOW ALLOY STEEL METAL DUSTING: DETAILED ANALYSIS BY MEANS OF ACOUSTIC EMISSION P. J. Van De Loo, A. Wolfert, R. Schelling, H. J. Schoorlemmer and T. M. Kooistra 238 20-248 RECONSTRUCTION METHOD OF DYNAMIC FRACTURE PROCESS INSIDE THE

MATERIAL WITH THE AID OF ACOUSTIC EMISSION Yasuhiko Mori, Yoshihiko Obata and Takateru Umeda 248

20-257 NON-DESTRUCTIVE EVALUATION OF BRAZED JOINTS BY MEANS OF ACOUSTIC

EMISSION H. Traxler, W. Arnold, W. Knabl and P. Rödhammer 257 20-265 DAMAGE MODE IDENTIFICATION AND ANALYSIS OF COATED GAS TURBINE

MATERIALS USING A NON-DESTRUCTIVE EVALUATION TECHNIQUE Y. Vougiouklakis, P. Hähner, V. Kostopoulos and S. Peteves 265

20-274 ACOUSTIC EMISSION DURING STRUCTURE CHANGES IN SEMI-CRYSTALLINE

POLYMERS J. Krofta, Z. Prevorovsky, M. Blahácek and M. Raab 274

44

20-285 ACOUSTIC EMISSION CHARACTERISTICS OF SURFACE FRICTION IN BIO-MEDICAL APPLICATION D. Prevorovsky, Z. Prevorovsky, J. Asserin, and D. Varchon 285

20-292 ACOUSTIC EMISSION DURING MONOTONIC AND CYCLIC DEFORMATION OF A BRITTLE LIMESTONE A. Lavrov, M. Wevers and A. Vervoort 292

20-300 LEAK DETECTION BY ACOUSTIC EMISSION USING SUBSPACE METHODS Amani Raad, Fan Zhang and Ménad Sidahmed 300 Contents20.pdf Contents of Volume 20 (2002) I-1 - I-3 AuIndex20.pdf Authors Index of Volume 20 I-4 AusNotes.pdf Policy/Author’s Notes/Authors e-mail/Subscription Information I-5 - I-7 Ewgae02.pdf EWGAE 2002, 25-th European Acoustic Emission Conference 10 p. APPENDICES (Available on CD-ROM only) Prosser.ppt Presentation at AEWG45 by W. Prosser (NASA Langley) PAC Folder Presentation at AEWG45 by T. Tamutas (Physical Acoustics Corp.) Vallen Folder Presentation at AEWG45 by H. Vallen (Vallen Systeme) AGU-Vallen Wavelet Transform Freeware and Introduction by M. Hamstad

45

Volume 21, 2003 21-001 Identifying Acoustic Emission Sources In Aging Bridge Steel Takao Kobayashi And Donald A. Shockey 1 21-014 Analysis Of Source Location Algorithms, Part I: Overview And Non-Iterative Methods Maochen Ge 14

21-029 Analysis Of Source Location Algorithms, Part Ii: Iterative Methods

Maochen Ge 29 21-052 Wavelet Transform Signal Processing To Distinguish Different Acoustic Emission Sources K. S. Downs, M. A. Hamstad And A. O’Gallagher 52

21-070 Practical Aspects Of Acoustic Emission Source Location By A Wavelet Transform M. A. Hamstad, K. S. Downs And A. O’Gallagher 70 21-A01 Appendices: Practical Aspects Of Acoustic Emission Source Location By A Wavelet Transform M. A. Hamstad, K. S. Downs And A. O’Gallagher A1 21-095 Acoustic Emission Monitoring Of A High Pressure Test Of A Steel Reactor Containment Vessel Model A. G. Beattie 95 21-112 Micro-Cracking And Breakdown Of Kaiser Effect In Ultra High Strength Steels Hideo Cho, Kenzo Fukaura And Kanji Ono 112

21-120 Acoustic Emission From The Fracture Of Atmospheric Rust M. Takemoto, T. Sogabe, K. Matsuura And K. Ono 120

21-131 Acoustic Property Of Cvd-Diamond Film And Acoustic Emission Analysis For Integrity Evaluation R. Ikeda, Y. Hayashi And M. Takemoto 131 21-142 Evaluation Of Coated Film By Laser-Based Ae-Ut Technique M. Enoki And T. Kusu 142 21-149 Acoustic Emission From Micro-Fracture Processes Of Bio-Ceramics In Simulated Body Environment Shuichi Wakayama, Teppei Kawakami, Satoshi Kobayashi, Mamoru Aizawa And Akira Nozue 149 21-157 Corrosion Monitoring In Reinforced Concrete By Acoustic Emission Masayasu Ohtsu And Yuichi Tomoda 157 21-166 Evaluation Of Bond Behavior Of Reinforcing Bars In Concrete Structures By Acoustic Emission K. Iwaki, O. Makishima, H. Tanaka, T. Shiotani And K. Ozawa 166 21-176 Development Of A Novel Optical Fiber Sensor For Ae Detection In Composites Isamu Ohsawa, Kazuro Kageyama, Yukiya Tsuchida And Makoto Kanai 176 21-187 Acoustic Emission Evaluation Of Corrosion Damages In Buried Pipes Of Refinery S. Yuyama And T. Nishida 187

46

21-197 New Concept Of Ae Standard: Jis Z 2342-2002 “Method For Acoustic Emission Testing Of Pressure Vessels During Pressure Tests And Classification Of Test Results” Y. Mori, M. Shiwa, M. Nakano And K. Iwai 197

21-206 Acoustic Emission Caused By Environmental Embrittlement Of An Al-Mg-Si Alloy Keitaro Horikawa, Kenichi Yoshida, A. Ohmori And Kiyoshi Sakamaki 206

21-213 Quantitative Study Of Acoustic Emission Due To Leaks From Water Tanks K. Morofuji, N. Tsui, M. Yamada, A. Maie, S. Yuyama And Z. W. Li 213

21-223 Effect Of Pinhole Shape With Divergent Exit On Ae Characteristics During Gas Leak K. Yoshida, Y. Akematsu, K. Sakamaki And K. Horikawa 223 21-230 Operation Monitoring Of Roll Cover By Acoustic Emission Juha Miettinen And Pekka Salmenperä 230 Contents21.Pdf Contents Of Volume 21 (2003) I-1 - I-3 Auindex21.Pdf Authors Index Of Volume 21 I-4 Ausnotes.Pdf Policy/Author’s Notes/Meeting Calendar/Subscription Information I-5 - I-7 Iaes16.Pdf Iaes16, 16-th International Acoustic Emission Symposium Appendices (Available On Cd-Rom Only) AGU-Vallen Wavelet Transform Freeware And Introduction By M. Hamstad Volume 22, 2004 22-S01 The Kaiser-Effect And Its Scientific Background Hans Maria Tensi S1

22-001 Modal-Based Identification Of Acoustic Emission Sources In The Presence Of Electronic Noise M. A. Hamstad And A. O’gallagher 1

22-A01* Appendix A: Details Of The Application Of The Source Identification Scheme M. A. Hamstad And A. O’gallagher A1 22-022 Experience With Acoustic Emission Monitoring Of New Vessels During Initial Proof Test Phillip Cole And Stephen Gautrey 22 22-030 Quantitative Damage Estimation Of Concrete Core Based On Ae Rate Process Analysis Masayasu Ohtsu And Tetsuya Suzuki 30 22-039 Damage Assessment In Deteriorated Railway Sub-Structures Using Ae Technique Tomoki Shiotani, Yasuhiro Nakanishi, Xiu Luo And Hiroshi Haya 39

22-049 Defect Detection By Acoustic Emission Examination Of Metallic Pressure Vessels Franz Rauscher 49 22-059 Evaluation Of Acoustic Emission Signals During Monitoring Of Thick-Wall Vessels Operating At

Elevated Temperatures. Athanasios Anastasopoulos And Apostolos Tsimogiannis 59 22-071 Acoustic Emission Technique And Potential Difference Method For Detecting The Different

Stages Of Crack Propagation In Carbon And Stainless Steels C. Ennaceur, A. Laksimi, C. Hervé, M. Mediouni And M. Cherfaoui 71

47

22-077 Detection Of Incipient Cavitation And Best Efficiency Point In A 2.2mw Centrifugal Pump Using Acoustic Emission

L. Alfayez And D. Mba 77 22-083 Investigation Of Fracture Processes Using Moment Tensor Inversion Technique

F. Finck, C. U. Grosse And H.-W. Reinhardt 83 22-091 AE Kaiser Effect And Electromagnetic Emission In The Deformation Of Rock Sample

Yasuhiko Mori, Yoshihiko Obata, Jan Pavelka, Josef Sikula And Thomas Lokajicek 91

22-102 Acoustic Emissions From Transpiring Plants– New Results And Conclusions Ralf Laschimke, Maria Burger And Hartmut Vallen 102 22-110 Acoustic Detection Of Cavitation Events In Water Conducting Elements Of Norway Spruce

Sapwood Sabine Rosner 110 22-119 AE Monitoring From Cvd-Diamond Film Subjected To Micro-Indentation And Pulse Laser

Spallation R. Ikeda, H. Cho, M. Takemoto And Kanji Ono 119

22-127 Composites From Piezoelectric Fibers As Sensors And Emitters For Acoustic Applications Andreas J. Brunner, Michel Barbezat, Peter Flüeler And Christian Huber 127

22-138 Processing Of Ae Signals In Dispersive Media Michal Blahacek, Zdenek Prevorovsky, And Michal Landa 138

22-147 Basic Principles Of Acoustic Emission Tomography Frank Schubert 147

22-159 Acoustic Emission Behavior Of Martensitic Transformation During Deformation Of Cu-Al-Ni Shape-Memory Alloy Single Crystals Kenichi Yoshida, Kotaro Hanabusa And Takuo Nagamachi 159

22-166 Acoustic Emission Monitoring Of Concrete Hinge Joint Models

K. M. Holford, R. Pullin And R. J. Lark 166 22-173 Characterization Of Acoustic Emission Sources In A Rock Salt Specimen Under Triaxial Load

Gerd Manthei 173 22-190 Testing Of Diamond-Like Carbon Coatings Under Slip-Rolling Friction Monitored By Acoustic

Emission Manuel Löhr 190

22-201 Field Testing Of Flat Bottomed Storage Tanks With Acoustic Emission – A Review On The Gained Experience

Gerald Lackner And Peter Tscheliesnig 201 22-208 Acoustic Emission Examination Of Polymer-Matrix Composites

Jürgen Bohse 208 22-224 Acoustic Emission From Rust In Stress Corrosion Cracking

Hideo Cho And Mikio Takemoto 224

22-236 Acoustic Emission Measurement System For The Orthopedic Diagnostics Of The Human Femur And Knee Joint

R.P. Franke, P. Dörner, H.-J. Schwalbe And B. Ziegler 236

48

22-243 Rods And Tubes As Ae Waveguides Kanji Ono And Hideo Cho 243

22-253 Acoustic Emission For On-Line Monitoring Of Damage In Various Application Fields Martine Wevers, Gert Van Dijck, Wendy Desadeleer, Mark Winkelmans And Koen Van Den Abeele 253

22-264 The Effect Of Waveguide Material And Shape On Acoustic Emission Transmission

Characteristics - Part 1: Traditional Features Joanna Sikorska And Jie Pan 264 22-274 The Effect Of Waveguide Material And Shape On Ae Transmission Characteristics - Part 2:

Frequency And Joint-Time-Frequency Characteristics Joanna Sikorska And Jie Pan 274

Contents22.Pdf Contents Of Volume 22 (2004) I-1 - I-4 Auindex22.Pdf Authors Index Of Volume 22 I-5 Ausnotes.Pdf Policy/Author’s Notes/Meeting Calendar/Subscription Information I-6 - I-8 Berlin2004.Pdf Activities At 2004 Ewgae Meeting; I-9 Inmemorium.Pdf Dick Blackburn (T.F. Drouillard) I-10 Tensi.Pdf Professor H.M. Tensi I-12

Ewgae Folder* 2004 And 2006 Ewgae Meetings, 2004 Program, 2006 Local Information

Jae Index Folder* Cumulative Indices Of J. Of Acoustic Emission, 1982 - 2004

* Indicates The Availability In CD-Rom Only.

Volume 23, 2005 23-001 Effects Of Noise On Lamb-Mode Acoustic-Emission Arrival Times Determined By Wavelet Transform M. A. Hamstad And A. O’gallagher 1

23-025 Acoustic Emission Technique For Detecting Damage And Mechanisms Of Fracture In A Knitted Fabric Reinforced Composite Carlos R. Rios, Steve L. Ogin, Constantina Lekakou And K. H. Leong 25 23-037 Quantitative Damage Estimation Of Concrete Core Based On Ae Rate Process Analysis William Prosser, Eric Madaras, George Studor, And Michael Gorman 37 23-047 Moment Tensors Of In-Plane Waves Analyzed By Sigma-2D

Masayasu Ohtsu, Kentaro Ohno And Marvin A. Hamstad 47 23-064 Development Of An Optical Micro AE Sensor With An Automatic Tuning System Hiroshi Asanuma, Hironobu Ohishi, And Hiroaki Niitsuma 64 23-072 Development Of Stabilized And High Sensitive Optical

Fiber Acoustic Emission System And Its Application Hideo Cho, Ryouhei Arai And Mikio Takemoto 72 23-081 High Precision Geophone Calibration Masahiro Kamata 81

23-091 Development Of Heat-Resistant Optical Fiber Ae Sensor

49

Pornthep Chivavibul, Hiroyuki Fukutomi, Shin Takahashi And Yuichi Machijima 91

23-096 Damage Detection System For Structures With

Smart Ae Sensors Takahito Yanase And Sei Ikegaya 96

23-102 Hierarchical Fracture Process In Brittle Rocks By Means Of High-Speed Monitoring Of Ae Hypocenter

Xinglin Lei, Osamu Nishizawa, Andre Moura And Takashi Satoh 102

23-113 Measurement Of Hydraulically Activated Subsurface Fracture System In Geothermal Reservoir

By Using Acoustic Emission Multiplet-Clustering Analysis Hirokazu Moriya, Hiroaki Niitsuma And Roy Baria 113

23-119 A Modeling Method On Fractal Distribution Of Cracks In Rocks Using Ae Monitoring

Yoshinori Watanabe, Ken-Ichi Itakura, Kazuhiko Sato, Yoshiaki Fujii, Rao Balusu, Hua Guo And Xun Luo 119

23-129 Interpretation Of Reservoir Creation Process At Cooper Basin, Australia By Acoustic Emission

Yusuke Kumano, Hirokazu Moriya, Hiroshi Asanuma, Nobukazu Soma, Hideshi Kaieda, Kazuhiko Tezuka, Doone Wyborn And Hiroaki Niitsuma 129

23-136 Micromechanics Of Corrosion Cracking In Concrete By Ae-Sigma

Farid A. K. M. Uddin And Masayasu Ohtsu 136 23-142 Evaluation Of Parameter Dependencies Of Ae Accompanying Sliding Along A Rough Simulated Fracture

Katsumi Nemoto, Hirokazu Moriya And Hiroaki Niitsuma 142 23-150 Ae Characterization Of Thermal Shock Crack Growth Behavior In Alumina Ceramics By Disc-

On-Rod Test Huichi Wakayama, Satoshi Kobayashi And Toshiya Wada 150

23-156 Fatigue Damage Progression In Plastics During Cyclic Ball Indentation

Akio Yonezu, Takayasu Hirakawa, Takeshi Ogawa

And Mikio Takemoto 156 23-164 Evaluation Of Fatigue Damage For Frm With Ae Method

Masanori Takuma And Noboru Shinke 164 23-173 Acoustic Emission Behavior Of Failure Processes Of

Glass-Fiber Laminates Under Complex State Of Loading Jerzy Schmidt, Ireneusz Baran And Kanji Ono 173

23-181 Rolling Contact Fatigue Damage Of WC-Co Cermet Sprayed Coating And Its Ae Analysis

Junichi Uchida, Takeshi Ogawa, Mikio Takemoto Yoshifumi Kobayashi And Yoshio Harada 181 23-189 Boron Effects On Ae Event Rate Peaks During Tensile Deformation Of Ni3al Intermetallic Compound

K. Yoshida, Y. Masui, T. Nagamachi And H. Nishino 189

50

23-196 Ae And Electrochemical Noise Analysis For Fracture Study Of Hard Surface Film Akio Yonezu, Hideo Cho, Takeshi Ogawa And Mikio Takemoto 196 23-206 The Origin Of Continuous Emissions

Kanji Ono, Hideo Cho And M. Takuma 206

23-215 Precursor Of Hydroigen Induced Glass Lining Chipping By Ae Monitoring

Kohei Murakami And Mikio Takemoto 215

23-224 Real-Time Executing Source Location System Applicable To Anisotropic Thin Structures

Yu Kurokawa, Yoshihiro Mizutani And Masami Mayuzumi 224 23-233 Investigation On Ae Signal/Noise Processing In Corrosion Damage Evaluation Of Tank Bottom Zhengwang Li , Shigenori Yuyama, Minoru Yamada, Kazuyoshi Sekine, Shigeo Kitsukawa, Hiroaki Maruyama And Shigeo Konno 233

23-243 Examination of AE Wave Propagation Routes In A Small Model Tank Hideyuki Nakamura, Takahiro Arakawa, Minoru Yamada 243 23-249 Integrity Evaluation Of Glass-Fiber Reinforced Plastic Vessels By Lamb Wave Ae Analysis

Takashi Futatsugi 249

23-260 Evaluation Of Reinforcement In Damaged Railway

Concrete Piers By Means Of Acoustic Emission Tomoki Shiotani, Yasuhiro Nakanishi, Keisuke Iwaki,

Xiu Luo Hiroshi Haya 260

23-272 Water-Leak Evaluation Of Existing Pipeline By Acoustic Emission Tetsuya Suzuki, Yukifumi Ikeda, Yuichi Tomoda And Masayasu Ohtsu 272

23-277 Acoustic Emission For Fatigue Damage Detection Of Stainless Steel Bellows Koji Kagayama, Akio Yonezu, Hideo Cho, Takeshi Ogawa, And Mikio Takemoto 277

23-285 Plastic Region Bolt Tightening Controlled By Acoustic Emission Monitoring

Tadashi Onishi, Yoshihiro Mizutani And Masami Mayuzumi 285

23-292 Acoustic Emission Behaviors Of Recovery For Mg Alloy At Room Temperature

Y. P. Li And M. Enoki 292

23-299 An Acoustic Emission Test System For Airline Steel

Oxygen Cylinders: System Design And Test Program Alan G. Beattie 299

23-310 Development Of In-Situ Monitoring System For

Sintering Of Ceramics Using Laser Ae Technique S. Nishinoiri And M. Enoki 310

51

23-318 Pattern Recognition Techniques For Acoustic Emission Based Condition Assessment Of Unfired Pressure Vessels Athanasios Anastasopoulos 318

23-331 The Acoustic Emission Halon 1301 Fire Extinguisher Bottle Tester: Results Of Tests On 649

Bottles Alan G. Beattie And D. D. Thornton 331

Contents23.Pdf Contents Of Volume 23 (2005) I-1 - I-4 Auindex23.Pdf Authors Index Of Volume 23 I-5 Aueaddress.Pdf Authors E-Mail Addresses I-6 Ausnotes.Pdf Policy/Author’s Notes/Meeting Calendar/ Subscription InformationI-7 - I-9

Jae Index Folder* Cumulative Indices Of J. Of Acoustic Emission, 1982 - 2005

* Indicates The Availability In CD-Rom Only.

Volume 24, 2006

24-001 A Variable Velocity Approach To Locate Fatigue-Induced Microcracks Occurred In Structures With Multiple Material Layers Jihui Li And Gang Qi 1

24-012 Lamb-Wave Acoustic Emission For Condition Monitoring Of Tank Bottom Plates Mikio Takemoto, Hideo Cho And Hiroaki Suzuki 12

24-022 Wavelet Transform Analysis Of Experimental Ae Waveforms On Steel Pressure Vessel Mulu Bayray And Franz Rauscher 22 24-044 Acoustic Emission Pattern Recognition Analysis Applied To The Over-Strained Pipes In A

Polyethylene Reactor Ireneusz Baran, Marek Nowak And Kanji Ono 44 24-052 Acoustic Emission Evaluation Systems Of Tool Life Forshearing Of Piano And Stainless Steel Wires

Masanori Takuma, Noboru Shinke, Takako Nishiura And Kensuke Akamatu 52

24-067 Optical Fiber System For Ae Monitoring Of High Temperature Damage Of Stainless Steel Tubing Tomoharu Hayano, Takuma Matsuo, Hideo Cho And Mikio Takemoto 67

24-076 Development Of Measurement System Using Optical Fiber Ae Sensors For Actual Piping Satoshi Nishinoiri, Pornthep Chivavibul, Hiroyuki Fukutomi And Takashi Ogata 76

24-084 Utilization Of Cascade Optical Fiber Ae System For Source Location Of Lamb Waves Through A

Cross-Ply Cfrp Plate Takuma Matsuo, Hideo Cho And Mikio Takemoto 84

24-097 Elastic Waves From Fast Heavy-Ion Irradiation On Solids

Tadashi Kambara, Yasuyuki Kanai, Takao M. Kojima, Yoichi Nakai, Akira Yoneda, Yasunori Yamazaki And Kensuke Kageyama

97 24-104 Acoustic Emission Rate Behavior Of Laminated Wood Specimens Under Tensile Loading

Andreas J. Brunner, Martin T. Howald And Peter Niemz 104 24-111 Ae Measurements For Superconducting Devices

52

Kazuaki Arai, Katsuyuki Kaiho, Hiroshi Yamaguchi, Hirofumi Yamasaki, Akira Ninomiya, Takeshi Ishigohka, Katsutoshi Takano, Hideo Nakajima And Kiyoshi Okuno 111

24-119 Acoustic Emission Of Sensitized 304 Stainless Steel With Simultaneous Hydrogen Charging S. H. Carpenter, Kanji Ono And D. Armentrout 119

24-127 Ae And Corrosion Potential Fluctuation (Cpf) For Environmental Assisted Fracture

Koji Kagayama, Takeshi Ogawa, Akio Yonezu, Hideo Cho And Mikio Takemoto 127

24-139 Damage Evaluation By Frequency Analysis Of Continuous Recorded Ae Waveform

Kaita Ito And Manabu Enoki 139 24-145 Frequency Filtering Algorithms Of Plate Wave Ae For Source Location

Yu Kurokawa, Yoshihiro Mizutani And Masami Mayuzumi 145 24-153 Evaluation Of Two Types Of Martensitic Transformation In Cu-Al-Ni Shape Memory Alloy Single

Crystal By Acoustic Emission Waveform Analysis Takeshi Yasuda, Daiki Tani, Hideo Nishino And Kenichi Yoshida 153 24-161 Fatigue Fracture Dynamics Of High Strength Steel Studied By Acoustic Emission Technique Akio Yonezu, Takeshi Ogawa And Mikio Takemoto 161

24-173 Quantitative Detection Of Microcracks In Bioceramics By Acoustic Emission Source

Characterization Shuichi Wakayama, Takehiko Jibiki And Junji Ikeda 173 24-179 Determination Of Wave Attenuation In Rock Salt In The Frequency Range 1 - 100 Khz Using

Located Acoustic Emission Events Gerd Manthei, Jürgen Eisenblätter And Thomas Spies 179 24-187 Acoustic Emission Behavior Of Prestressed Concrete Girder During Proof Loading Leszek Gołaski, Grzegorz Swit, Małgorzata Kalicka And Kanji Ono 187 24-196 Multiplet Analysis For Estimation Of Structures Inside An Ae Cloud Associated With A

Compression Test Of A Salt Rock Specimen Hirokazu Moriya, Gerd Manthei, Hiroaki Niitsuma And Jürgen Eisenblätter 196 24-205 Damage Diagnosis Of Railway Concrete Structures By Means Of One-Dimensional Ae Sources Tomoki Shiotani, Xiu Luo And Hiroshi Haya 205 24-215 Charactaristics Of Damage And Fracture Process Of Solid Oxide Fuel Cells Under Simulated

Operating Conditions By Using Ae Method Kazuhisa Sato, Toshiyuki Hashida, Hiroo Yugami, Keiji Yashiro, Tatsuya Kawada And Junichiro Mizusaki 215 24-222 Acoustic Emission Detection Of Damage Evolution In Short-Fiber Composites Jerzy Schmidt, Ireneusz Baran, Marek Nowak And Kanji Ono 222 24-228 Ae Monitoring Of Microdamage During Proof Test Of Bioceramics For Artificial Joints Shuichi Wakayama, Chikako Ikeda And Junji Ikeda 228 24-234 Small Diameter Waveguide For Wideband Acoustic Emission M. A. Hamstad 234

53

Contents24.pdf Contents of Volume 24 (2006)I-1, -2, -3 AUindex24.pdf Authors Index of Volume 24 I-4, -5 AusNotes.pdf Policy/Author’s Notes/Meeting Calendar/Subscription Information I-6 - I-8 JAE Index Folder* Cumulative Indices of J. of Acoustic Emission, 1982 - 2006

* indicates the availability in CD-ROM only.

Volume 25, 2007 25-001 STRUCTURAL INTEGRITY EVALUATION USING ACOUSTIC EMISSION KANJI ONO 25-021 ACOUSTIC EMISSION TECHNIQUES STANDARDIZED FOR CONCRETE STRUCTURES MASAYASU OHTSU, TOSHIRO ISODA and YUICHI TOMODA 25-033 ACOUSTIC EMISSION MONITORING OF REINFORCED CONCRETE FRAME DURING SEISMIC LOADING A. ANASTASOPOULOS, S. BOUSIAS and T. TOUTOUNTZAKIS 25-042 ACOUSTIC EMISSION LEAK TESTING OF PIPES FOR PRESSURIZED GAS USING ACTIVE FIBER COMPOSITE ELEMENTS AS SENSORS ANDREAS J. BRUNNER and MICHEL BARBEZAT 25-051 ACOUSTIC EMISSION TECHNIQUE APPLIED FOR MONITORING AND INSPECTION OF CEMENTITIOUS STRUCTURES ENCAPSULATING ALUMINIUM L. M. SPASOVA, M. I. OJOVAN and C. R. SCALES 25-069 EVALUATION OF REPAIR EFFECT FOR DETERIORATED CONCRETE PIERS OF INTAKE DAM USING AE ACTIVITY TOMOKI SHIOTANI and DIMITRIOS G. AGGELIS 25-080 ACOUSTIC EMISSION MONITORING OF FLEXURALLY LOADED ARAMID/EPOXY COMPOSITES BY EMBEDDED PVDF SENSORS CLAUDIO CANEVA, IGOR MARIA DE ROSA and FABRIZIO SARASINI 25-092 ACOUSTIC EMISSION SIGNALS GENERATED BY MONOPOLE (PENCIL-

LEAD BREAK) VERSUS DIPOLE SOURCES: FINITE ELEMENT MODELING AND EXPERIMENTS

M. A. HAMSTAD 25-107 HIGH-TEMPERATURE ACOUSTIC EMISSION SENSING USING ALUMINUM NITRIDE SENSOR HIROAKI NOMA, TATSUO TABARU, MORITO AKIYAMA, NORIKO MIYOSHI, TOMOHARU HAYANO and HIDEO CHO 25-115 DAMPING, NOISE, AND IN-PLANE RESPONSE OF MEMS ACOUSTIC EMISSION SENSORS

AMELIA P. WRIGHT, WEI WU, IRVING J. OPPENHEIM and DAVID W. GREVE

25-124 IMMERSION-TYPE QUADRIDIRECTIONAL OPTICAL FIBER AE SENSOR FOR LIQUID-BORNE AE TAKUMA MATSUO, HIDEO CHO, TAKESHI OGAWA and MIKIO TAKEMOTO

54

25-132 A SIMPLE METHOD TO COMPARE THE SENSITIVITY OF DIFFERENT AE SENSORS FOR TANK FLOOR TESTING HARTMUT VALLEN, JOCHEN VALLEN and JENS FORKER 25-140 DAMAGE IN CARBON FIBRE COMPOSITES: THE DISCRIMINATION OF ACOUSTIC EMISSION SIGNALS USING FREQUENCY MARK EATON, KAREN HOLFORD, CAROL FEATHERSTON and RHYS PULLIN 25-149 CHARACTERISTICS OFACOUSTIC EMISSIONS FROM DEHYDRATING WOOD RELATED TO SHRINKAGE PROCESSES SABINE ROSNER 25-157 CHARACTERIZATION OF TITANIUM HYDRIDES USING A HYBRID TECHNIQUE OF AE AND FEM DURING INDENTATION TEST

YOSHIHIRO TANIYAMA, HIDEO CHO, MIKIO TAKEMOTO and GEN NAKAYAMA

25-166 ANALYSIS OF THE HYDROGEN DEGRADATION OF LOW-ALLOY STEEL BY ACOUSTIC EMISSION KRYSTIAN PARADOWSKI, WOJCIECH SPYCHALSKI, KRYSTYNA LUBLINSKA and KRZYSZTOF J. KURZYDLOWSKI 25-172 HYDROGEN RELATED BRITTLE CRACKING OF METASTABLE TYPE- 304 STAINLESS STEEL HIDEO CHO and MIKIO TAKEMOTO 25-179 ANALYSIS OF ACOUSTIC EMISSION FROM IMPACT AND FRACTURE

OF CFRP LAMINATES KANJI ONO, YOSHIHIRO MIZUTANI AND MIKIO TAKEMOTO 25-187 NEURAL NETWORK BURST PRESSURE PREDICTION IN COMPOSITE OVERWRAPPED PRESSURE VESSELS ERIC v. K. HILL, SETH-ANDREW T. DION, JUSTIN O. KARL, NICHOLAS S. SPIVEY and JAMES L. WALKER II 25-194 ACOUSTIC EMISSION SOURCE LOCATION IN A THICK STEEL PLATE BY LAMB MODES M. A. HAMSTAD 25-215 NOVEL ACOUSTIC EMISSION SOURCE LOCATION RHYS PULLIN, MATTHEW BAXTER, MARK EATON, KAREN HOLFORD and SAM EVANS 25-224 ACOUSTIC EMISSION SOURCE LOCATION ON AN ARBITRARY SURFACE BY GEODESIC CURVE EVOLUTION G. PRASANNA, M. R. BHAT and C. R. L. MURTHY 25-231 PROBABILITY OF DETECTION FOR ACOUSTIC EMISSION ADRIAN POLLOCK 25-238 PLASTIC-REGION TIGHTENING OF BOLTS CONTROLLED BY ACOUSTIC EMISSION METHOD YOSHIHIRO MIZUTANI, TADASHI ONISHI and MASAMI MAYUZUMI 25-247 REAL-TIME DENOISING OF AE SIGNALS BY SHORT TIME FOURIER TRANSFORM AND WAVELET TRANSFORM

55

KAITA ITO and MANABU ENOKI 25-252 MONITORING THE EVOLUTION OF INDIVIDUAL AE SOURCES IN CYCLICALLY LOADED FRP COMPOSITES RUNAR UNNTHORSSON, THOMAS P. RUNARSSON AND MAGNUS T. JONSSON 25-260 ON USING AE-HIT PATTERNS FOR MONITORING CYCLICALLY LOADED CFRP RUNAR UNNTHORSSON, THOMAS P. RUNARSSON and MAGNUS T. JONSSON 25-267 AE MONITORING OF SOIL CORROSION OF BURIED PIPE HIDEO CHO and MIKIO TAKEMOTO 25-276 THIRTY YEARS EXPERIENCE OF INDUSTRIAL APPLICATIONS OF ACOUSTIC EMISSION TESTING AT TÜV AUSTRIA PETER TSCHELIESNIG

25-286 ACOUSTIC EMISSION TESTING OF SEAM-WELDED HIGH ENERGY PIPING SYSTEMS IN FOSSIL POWER PLANTS

JOHN M. RODGERS 25-294 ACOUSTIC EMISSION AND X-RAY TOMOGRAPHY IMAGING OF SHEAR FRACTURE FORMATION IN CONCRETE TATYANA KATSAGA and R. PAUL YOUNG 25-308 GLOBAL MONITORING OF CONCRETE BRIDGE USING ACOUSTIC EMISSION T. SHIOTANI, D. G. AGGELIS and O. MAKISHIMA 25-316 DEMAND ON FLEXURAL TENSION STEEL REINFORCEMENT ANCHORAGE ZONES IN FULL-SCALE BRIDGE BENT CAPS QUANTIFIED BY MEANS OF ACOUSTIC EMISSION THOMAS SCHUMACHER, CHRISTOPHER HIGGINS, STEVEN GLASER and CHRISTIAN GROSSE 25-324 DAMAGE EVALUATION OF POST-TENSIONED CONCRETE VIADUCT BY AE DURING PROOF LOADING EDOARDO PROVERBIO, GIUSEPPE CAMPANELLA and VINCENZO VENTURI 25-331 EARLY FAULT DETECTION AT GEAR UNITS BY ACOUSTIC EMISSION AND WAVELET ANALYSIS CHRISTIAN SCHEER, WILFRIED REIMCHE and FRIEDRICH-WILHELM BACH 25-341 APPLICATION OF ACOUSTIC EMISSION IN MONITORING OF FAILURE IN SLIDE BEARINGS IRENEUSZ BARAN, MAREK NOWAK and WOJCIECH DARSKI 25-348 MAPPING OF WHEEL FLANGE RUBBING ON RAIL USING AE: LABORATORY TEST N. A. THAKKAR, R. L. REUBEN and J. A. STEEL 25-355 DAMAGE ASSESSMENT OF GEARBOX OPERATING IN HIGH NOISY ENVIRONMENT USING WAVEFORM STREAMING APPROACH DIDEM OZEVIN, JASON DONG, VALERY GODINEZ and MARK CARLOS

25-364 CLUSTERING ANALYSIS OF AE IN ROCK N. IVERSON, C-S. KAO and J.F. LABUZ

56

25-373 IMPACT IMAGING METHOD TO MAP DAMAGE IN CONCRETE BRIDGE DECK SLABS

STEPHEN D. BUTT, VIDYADHAR LIMAYE and JOHN P. NEWHOOK

Contents25 Contents of Volume 25 (2007)I-1 – I-4 AUindex25 Authors Index of Volume 25 I-5 – I-7 AusNotes Policy/Author’s Notes/Meeting Calendar/Subscription Information I-8 – I-10 In Memoriam Professor Reginald Hardy, Jr. I-11 EWGAE27 Note on expanded contributions from Cardiff Conference. I-12 JAE Index Folder* Cumulative Indices of J. of Acoustic Emission, 1982 - 2007

* indicates the availability in CD-ROM only.

Cover photograph is from 25-294 by TATYANA KATSAGA and R. PAUL YOUNG. Volume 26 (2008) 26-001 Acoustic Emission Investigation of Coating Fracture and Delamination in Hybrid Carbon Fiber Reinforced Plastic Structures Markus G. R. Sause, Daniel Schultheiß

and Siegfried Horn 1-13

26-014 Acoustic Emissions Related to the Dehydration Stress Behavior of Green Norway Spruce Wood Sabine Rosner, Bo Karlsson, Johannes Konnerth and Christian Hansmann 14-22 26-023 Investigation of the Z-Direction Strength Properties of Paper by Use of Acoustic Emission Monitoring S. Norgren, P. A. Gradin, S. Nyström and M. Gullikson 23-31 26-032 Assessment of Stress Corrosion Cracking in Prestressing Strands Using AE Technique Marianne Perrin, Laurent Gaillet, Christian Tessier and Hassane Idrissi 32-39 26-040 Comparison of Wavelet Transform and Choi-Williams Distribution to Determine Group Velocities for Different Acoustic Emission Sensors M. A. Hamstad 40-59 26-060 A Comparison of AE Sensor Calibration Methods Jiri Keprt and Petr Benes 60-71

26-072 Experimental Transfer Functions of Acoustic Emission Sensors Kanji Ono, Hideo Cho and Takuma Matsuo 72-90 26-091 Couplants and Their Influence on AE Sensor Sensitivity Pete Theobald, Bajram Zeqiri and Janine Avison 91-97

26-098 Laboratory Experiments for Assessing the Detectability of Specific Defects by Acoustic Emission Testing Franz Rauscher 98-108 26-109 Integrity Evaluation of COPVs by Means of Acoustic Emission Testing Yoshihiro Mizutani, Kouki Saiga, Hideyuki Nakamura, Nobuhito Takizawa, Takahiro Arakawa and Akira Todoroki 109-119

26-120 Structural Integrity Evaluation of CNG Composite Cylinders by Acoustic Emission Monitoring Olivier Skawinski, Patrice Hulot, Christophe Binétruy and Christian Rasche 120-131 26-132 Automated Method for Statistical Processing of AE Testing Data

57

V. A. Barat and A. L. Alyakritskiy 132-141 26-142 Termites Detection via Spectral Kurtosis and Wavelet De-Noising of Acoustic Emission Signals Juan-José G. De La Rosa, Antolino Gallego, Rosa Piotrkowski, Enrique Castro and Antonio Moreno-Muñoz 142-151 26-152 Acousto-Ultrasonic Signal Analysis for Damage Detection in GFRP Adhesive Joints Andreas J. Brunner and Giovanni P. Terrasi 152-159 26-160 Bending Fracture Behavior of 3D-Woven SiC/SiC Composites with Transpiration Cooling Structure Characterized by AE Wavelet Analysis Toshimitsu Hayashi and Shuichi Wakayama 160-171 26-172 Acoustic Emission Monitoring of Bridge Structures in the Field and Laboratory Rhys Pullin, Karen M. Holford, Robert J. Lark and Mark J. Eaton 172-181 26-182 Arrival Time Detection in Thin Multilayer Plates on the Basis of Akaike Information Criterion Petr Sedlak, Yuichiro Hirose, Manabu Enoki and Josef Sikula 182-188 26-189 Some Possibilities of AE Signal Treatment at Contact Damage Tests of Materials and Bearings Pavel Mazal, Filip Hort, Martin Drab and Tomas Slunecko 189-198 26-199 Laser Cutting and Acoustic Emission Signals Tomaž Kek and Janez Grum 199-207 26-208 Online Monitoring of Hot Die Forging Processes Using Acoustic Emission (Part I) Islam El-Galy and Bernd-Arno Behrens 208-219

26-220 Natural Fiber Composites Monitored by Acoustic Emission Igor Maria De Rosa, Carlo Santulli and Fabrizio Sarasini 220-228 26-229 Acoustic Emission Feature for Early Failure Warning of CFRP Composites Subjected to Cyclic Fatigue Runar Unnthorsson, Thomas P. Runarsson and Magnus T. Jonsson 229-239 26-240 Identification of Damage Initiation and Development in Textile Composite Materials Using Acoustic

Emission D.S. Ivanov, S.V. Lomov, I. Verpoest and M. Wevers 240-246 26-247 Damage Identification in Corroded Galvanized and Duplex Coatings Using Wavelet Power and Entropy Rosa Piotrkowski, Antolino Gallego and Enrique Castro 247-261 26-262 AE Entropy for the Condition Monitoring of CFRP Subjected to Cyclic Fatigue Runar Unnthorsson, Thomas P. Runarsson and Magnus T. Jonsson 262-269 26-270 Experimental Simulation and Dynamic Behavior of the AE due to Martensitic Transformation Using Shear Wave Transmission Sensor Takeshi Yasuda, Shinya Kondo, Hideo Nishino and Kenichi Yoshida 270-278 26-279 Implementation of Acoustic Emission Method to the Conventional NDT Structure in Oil Refinery V.P. Gomera, V.L. Sokolov and V.P. Fedorov 279-289 26-290 An Experimental Analysis of Frequency Emission and Noise Diagnosis of Commercial Aircraft on Approach S. Khardi 290-310

26-311 New Developments of Software for A-line Family AE Systems Sergey Elizarov, Аnton Bukatin, Мikhail Rostovtsev and Denis Terentyev 311-317

26-317 Application of Acoustic Emission in Optimizing the Design of New Generation castings of

58

High-Voltage Electric Devices Jan Płowiec, Wojciech L. Spychalski, Huber Matysiak and Jakub Michalski 318-325 Contents26 Contents of Volume 26 (2008) I-1 – I-3 AUindex26 Authors Index of Volume 26 I-4 AusNotes Policy/Author’s Notes/Meeting Calendar/DVD/Subscription Information I-5 – I-7 JAE Index Folder* Cumulative Indices of J. of Acoustic Emission, 1982 - 2008 PhD Thesis of Runar Unnthorsson, University of Iceland* * indicates the availability in CD-ROM only. Cover illustration is from 26-160 by Toshimitsu Hayashi and Shuichi Wakayama. This figure shows the WT diagram, AE signal, projected WT-frequency curve and FFT spectrum. While WT identifies the first major frequency at 0.56 MHz and the second major frequency at 0.32 MHz, FFT shows the 1st and 2nd major frequencies of ~0.3 MHz as these frequency components have long duration. This clearly shows that FFT analysis failed to detect the highest characteristic frequency at 0.56 MHz. In using FFT, it is essential to be cognizant of this shortcoming. Volume 27, 2009 27-001 MONITORING THE CIVIL INFRASTRUCTURE WITH ACOUSTIC EMISSION: BRIDGE CASE STUDIES D. ROBERT HAY, JOSE A. CAVACO and VASILE MUSTAFA 1-10 27-011 ACOUSTIC EMISSION TESTING OF A DIFFICULT-TO-REACH STEEL BRIDGE DETAIL DAVID E. KOSNIK 11-17 27-018 ACOUSTIC EMISSION AS A MONITORING METHOD IN PRESTRESSED CONCRETE BRIDGES HEALTH CONDITION EVALUATION MAŁGORZATA KALICKA 18-26

27-027 ACOUSTIC EMISSION LEAK DETECTION OF LIQUID FILLED BURIED PIPELINE ATHANASIOS ANASTASOPOULOS, DIMITRIOS KOUROUSIS 27-39 and KONSTANTINOS BOLLAS 27-040 ACOUSTIC EMISSION MONITORING AND FATIGUE LIFE PREDICTION IN AXIALLY LOADED NOTCHED STEEL SPECIMENS FADY F. BARSOUM, JAMIL SULEMAN, ANDREJ KORCAK and ERIC V. K. HILL 40-63 27-064 AE ANALYSIS ON BLADE CUTTING PRESSURE ADJUSTMENT IN DYNAMIC CUTTING OF PAPERBOARD DARULIHSAN A. HAMID, SHIGERU NAGASAWA, YASUSHI FUKUZAWA, YUUKI KOMIYAMA and AKIRA HINE 64-76

59

27-077 DAMAGE ONSET AND GROWTH IN CARBON-CARBON COMPOSITE MONITORED BY ACOUSTIC EMISSION TECHNIQUE ARIE BUSSIBA, ROMANA PIAT, MOSHE KUPIEC, RAMI CARMI, IGAL ALON and THOMAS BÖHLKE 77-88 27-089 FUNDAMENTAL STUDY ON INTEGRITY EVALUATION METHOD FOR COPVS BY MEANS OF ACOUSTIC EMISSION TESTING YOSHIHIRO MIZUTANI, SOTA SUGIMOTO, RYOSUKE MATSUZAKI and AKIRA TODOROKI 89-97 27-098 ACOUSTIC EMISSION FROM IMPACTS OF RIGID BODIES TATIANA B. PETERSEN 98-113 27-114 SOME OBSERVATIONS ON RAYLEIGH WAVES AND ACOUSTIC EMISSION IN THICK STEEL PLATES M. A. HAMSTAD 114-136

27-137 FRACTURE BEHAVIOR IN BONE CHARACTERIZED BY AE WAVELET ANALYSIS SHUICHI WAKAYAMA, KEISUKE MOGI and TETSUYA SUEMUNE 137-143 27-144 ABOUT PLASTIC INSTABILITIES IN IRON AND POWER SPECTRUM OF ACOUSTIC EMISSION ALEXEY LAZAREV and ALEXEI VINOGRADOV 144-156 27-157 ACOUSTIC AND ELECTROMAGNETIC EMISSION FROM CRACK CREATED IN ROCK SAMPLE UNDER DEFORMATION YASUHIKO MORI, YOSHIHIKO OBATA and JOSEF SIKULA 157-166

27-167 IDENTIFICATION OF AE MULTIPLETS IN THE TIME AND FREQUENCY DOMAINS HIROSHI ASANUMA, YUSUKE KUMANO, HIROAKI NIITSUMA, DOONE WYBORN and ULRICH SCANZ 167-175 27-176 CRACK GROWTH MONITORING WITH HIERARCHICAL CLUSTERING OF AE N. F. INCE, CHU-SHU KAO, M. KAVEH, A. TEWFIK and J. F. LABUZ 176-185 27-186 ACOUSTIC EMISSION FOR CHARACTERIZING BEHAVIOR OF COMPOSITE CONCRETE ELEMENTS UNDER FLEXURE SHOHEI MOMOKI, HWAKIAN CHAI, DIMITRIOS G. AGGELIS, AKINOBU HIRAMA and TOMOKI SHIOTANI 186-193 27-194 DISTINCT ELEMENT ANALYSIS FOR ROCK FAILURE CONSIDERING AE EVENTS GENERATED BY THE SLIP AT CRACK SURFACES HIROYUKI SHIMIZU, SUMIHIKO MURATA and TSUYOSHI ISHIDA 194-211

60

27-212 ELECTROMAGNETIC METHOD OF ELASTIC WAVE EXCITATION FOR CALIBRATION OF ACOUSTIC EMISSION SENSORS AND APPARATUS SERGEY LAZAREV, ALEXANDER MOZGOVOI, ALEXEI VINOGRADOV, ALEXEY LAZAREV and ANDREY SHVEDOV 212-223

27-224 MONITORING OF PIPE CLOGGING BY MUSSELS UTILIZING AN OPTICAL FIBER AE SYSTEM TAKUMA MATSUO, YUTA MIZUNO and HIDEO CHO 224-232 27-233 CORROSION DETECTION BY FIBER OPTIC AE SENSOR YUICHI MACHIJIMA, MASAHIRO AZEMOTO, TOYOKAZU TADA and HISAKAZU MORI 233-240 27-241 EFFECT OF SHOT PEENING ON THE DELAYED FRACTURE USING THE ALMEN STRIP AND AE TECHNIQUE MIKIO TAKEMOTO, MOTOAKI NAKAMURA, SEIJI MASANO and SHUICHI UENO 241-253 27-254 CONTRIBUTION OF ACOUSTIC EMISSION TO EVALUATE CABLE STRESS CORROSION CRACKING IN SIMULATED CONCRETE PORE SOLUTION S. RAMADAN, L. GAILLET, C. TESSIER and H. IDRISSI 254-262 27-263 FLEXURAL FAILURE BEHAVIOR OF RC BEAMS WITH REBAR CORROSION AND DAMAGE EVALUATION BY ACOUSTIC EMMISSION NOBUHIRO OKUDE, MINORU KUNIEDA, TOMOKI SHIOTANI and HIKARU NAKAMURA 263-271

27-272 ACOUSTIC EMISSION METHOD FOR SOLVING PROBLEMS IN DOUBLE-BOTTOM STORAGE TANKS MAREK NOWAK, IRENEUSZ BARAN, JERZY SCHMIDT and KANJI ONO 272-280 27-281 STUDY OF IDENTIFICATION AND REMOVAL METHOD FOR DROP NOISE IN AE MEASUREMENT OF TANKS HIDEYUKI NAKAMURA, TAKAHIRO ARAKAWA, HIRAKU KAWASAKI, KAZUYOSHI SEKINE and NAOYA KASAI 281-290 27-291 A GENERIC TECHNIQUE FOR ACOUSTIC EMISSION SOURCE LOCATION JONATHAN J. SCHOLEY, PAUL D. WILCOX, MICHAEL R. WISNOM, MIKE I. FRISWELL, MARTYN PAVIER and MOHAMMAD R ALIHA 291-298

27-299 ACOUSTIC EMISSION TESTING – DEFINING A NEW STANDARD OF ACOUSTIC EMISSION TESTING FOR PRESSURE VESSELS Part 1: Quantitative and comparative performance analysis of zonal location and triangulation methods JOHANN CATTY 299-313

Contents27 Contents of Volume 27 (2009) I-1 – I-3 AUindex27 Authors Index of Volume 27 I-4 AusNotes Policy/Author’s Notes/Meeting Calendar/DVD/Subscription Information I-5 – I-7

 

61

Volume 28, 2010 28-001 ESTIMATION OF CRUSTAL STRUCTURE IN HORONOBE AREA, HOKKAIDO, JAPAN, USING MULTIPLET-CLUSTERING ANALYSIS HIROKAZU MORIYA, KOICHI ASAMORI, ITARU KITAMURA, HIKARU HOTTA, HIDEFUMI OHARA and TADAFUMI NIIZATO 1-10 28-011 ACOUSTIC EMISSION TESTING - DEFINING A NEW STANDARD OF ACOUSTIC EMISSION TESTING FOR PRESSURE VESSELS Part 2: Performance analysis of different configurations of real case testing and recommendations for developing a new guide for the application of acoustic emission JOHANN CATTY 11-31 28-032 ACOUSTIC EMISSION MONITORING OF STEEL-FIBER REINFORCED CONCRETE BEAMS UNDER BENDING DIMITRIOS G. AGGELIS, DIMITRA SOULIOTI, NEKTARIA M. BARKOULA, ALKIVIADIS S. PAIPETIS, THEODORE E. MATIKAS and TOMOKI SHIOTANI 32-40 28-041 ON LAMB MODES AS A FUNCTION OF ACOUSTIC EMISSION SOURCE RISE TIME M. A. HAMSTAD 41-58 28-059 WAVEFORM ANALYSIS OF ACOUSTIC EMISSION MONITORING OF TENSILE TESTS ON WELDED WOOD-JOINTS ANDREAS J. BRUNNER, THOMAS TANNERT and TILL VALLÉE 59-67 28-068 USE OF ACOUSTO-ULTRASONIC TECHNIQUESTO DETERMINE PROPERTIES OF REMANUFACTURED PARTICLEBOARDS MADE SOLELY FROM RECYCLED PARTICLES SUMIRE KAWAMOTO 68-75

28-076 ACOUSTIC EMISSION ACTIVITY OF SPRUCE SAPWOOD BECOMES WEAKER AFTER EACH DEHYDRATION-REWETTING CYCLE SABINE ROSNER and SUMIRE KAWAMOTO 76-84

 

62

28-085 ACOUSTIC EMISSION SOURCE LOCATION IN PLATE-LIKE STRUCTURES USING A CLOSELY ARRANGED TRIANGULAR SENSOR ARRAY DIRK ALJETS , ALEX CHONG , STEVE WILCOX and KAREN HOLFORD 85-98 28-099 NEURAL NETWORK AE SOURCE LOCATION APART FROM STRUCTURE SIZE AND MATERIAL MILAN CHLADA, ZDENEK PREVOROVSKY and MICHAL BLAHACEK 99-108 28-109 INTELLIGENT AE SIGNAL FILTERING METHODS VERA BARAT, YRIJ BORODIN and ALEXEY KUZMIN 109-119 28-120 DISCRIMINATION OF ACOUSTIC EMISSION HITS FROM DYNAMIC TESTS OF A REINFORCED CONCRETE SLAB ENRIQUE CASTRO, ROSA PIOTRKOWSKI, ANTOLINO GALLEGO and AMADEO BENAVENT CLIMENT 120-128

28-129 USE OF CLUSTER ANALYSIS OF ACOUSTIC EMISSION SIGNALS IN EVALUATING DAMAGE SEVERITY IN CONCRETE STRUCTURES L. CALABRESE, G. CAMPANELLA and E. PROVERBIO 129-141 28-142 SIMULATION OF LAMB WAVE EXCITATION FOR DIFFERENT ELASTIC PROPERTIES AND ACOUSTIC EMISSION SOURCE GEOMETRIES MARKUS G. R. SAUSE and SIEGFRIED HORN 142-154

28-155 ACOUSTIC EMISSION EVENT IDENTIFICATION WITH SIMILAR TRANSFER FUNCTIONS FRANZ RAUSCHER 155-162 28-163 ANALYSIS OF FRACTURE RESISTANCE OF TOOL STEELS BY MEANS OF ACOUSTIC EMISSION EVA MARTINEZ-GONZALEZ, INGRID PICAS, DANIEL CASELLAS and JORDI ROMEU

163-169

28-170 COMPARISON OF ACOUSTIC EMISSION SIGNAL AND X-RAY DIFFRACTION AT INITIAL STAGES OF FATIGUE DAMAGE FRANTISEK VLASIC, PAVEL MAZAL and FILIP HORT 170-178 28-179 AE SIGNALS DURING LASER CUTTING OF DIFFERENT STEEL SHEET THICKNESSES TOMAŽ KEK and JANEZ GRUM 179-187 28-188 ACOUSTIC EMISSION ANALYSIS AND THERMO-HYGRO-MECHANICAL MODEL FOR CONCRETE EXPOSED TO FIRE CHRISTIAN GROSSE, JOŠKO OŽBOLT, RONALD RICHTER and GORAN PERIŠKIĆ 188-203

 

63

28-204 AE-SiGMA ANALYSIS IN BRAZILIAN TEST AND ACCELERATED CORROSION TEST OF CONCRETE MASAYASU OHTSU and YUMA KAWASAKI 204-214 28-215 ACOUSTIC EMISSION INSPECTION OF RAIL WHEELS KONSTANTINOS BOLLAS, DIMITRIOS PAPASALOUROS, DIMITRIOS KOUROUSIS and ATHANASIOS ANASTASOPOULOS 215-228 28-229 E07.04 – OVERVIEW OF CURRENT AND DEVELOPING ASTM ACOUSTIC EMISSION (AE) STANDARDS MARK F. CARLOS 229-­‐233 28-234 USE OF AE METHOD FOR DETECTION OF STEEL LAMINATION IN THE INDUSTRIAL PRESSURE VESSEL V.P. GOMERA, V.L. SOKOLOV, V.P. FEDOROV, A.A. OKHOTNIKOV and M.S. SAYKOVA 234-245 28-246 COMPARISON OF ACOUSTIC EMISSION PRODUCED DURING BENDING OF VARIOUS OXIDE CERAMIC AND SHORT FIBER OXIDE CERAMIC MATRIX COMPOSITES S.A. PAPARGYRI-MPENI, D.A. PAPARGYRIS, X. SPILIOTIS and A.D. PAPARGYRIS

246-275

28-256 NEW CHARACTERIZATION METHODS OF AE SENSORS KANJI ONO, HIDEO CHO and TAKUMA MATSUO 256-277 Contents28 Contents of Volume 28 (2010) I-1 – I-3 AUindex28 Authors Index of Volume 28 I-4 AusNotes Policy/Author’s Notes/Meeting Calendar/DVD/ I-5 – I-7 Subscription Information I-7 IAES20 JCAE Kishinoue Awards I-8 AE Literature AELit28 Book on AE by Markus Sause 12th AE conference proceeding in China, 2009: Gongtian Shen I-9 – I-11 Cover illustration See 28-142 by Sause and Horn for details. JAE Index Folder* Cumulative Indices of J. of Acoustic Emission, 1982 – 2010 Contents1-28 Contents Volumes 1-28 Authors Index1-28 Authors Index Volumes 1-28

* indicates the availability in CD-ROM only. Indices are also available for download from www.aewg.org.

1

Authors Index, Volumes 1-28 (1982-2010) Satoshi Abe 19-045 A. R. Acharya 8-S88 C. Howard Adams 1-165 E. Aernoudt 4-S186 B. D. Agarwal 8-S297 Dimitrios G. Aggelis, 25-069, 25-308, 27-186, 28-032 Mamoru Aizawa 21-149 T. Aizawa 6-85 Chandra Ajay 4-S174 Kensuke Akamatu 24-052 Y. Akematsu 21-223 M. Akiyama 16-S150 Morito Akiyama, 25-107 L. Alfayez 22-077 R.S. Algera 2-69 S. S. Ali 4-S42, 4-107 Mohammad R Aliha 27-291 Dirk ALJETS 28-085 Bernhard Allemann 14-119 A.F. Almeida 15-S107, 15-S108 Igal Alon 27-077 A. L. Alyakritskiy 26-132 Masashi Amaya, 13-S35 J. F. R. Ambler 5-S16 G. Amir 2-64 A. A. Anastassopoulos, 13-011, 18-021, 18-217, 18-224, 20-229, 22-059, 23-318, 25-033, 27-027, 28-215 Naoto Ando 9-209 E. Andres, 18-155 M. Annamalai 8-S8 K.-I. Aoki 8-S131 K. Aoki 8-S145 Kazuaki Arai 24-111 Ryouhei Arai 23-072 Takahiro Arakawa 23-243, 26-109, 27-281 D. Armentrout 10-97, 10-103, 15-43, 16-S10, 24-119 Baxter H. Armstrong, 8-S250, 12-117 J. H. Armstrong 4-S135 M. Arrington 4-S165 Koichi Asamori 28-001 Ikuo Asano 4-S240 M. Asano, 19-134 Hiroshi Asanuma 23-064, 23-129, 27-167 J. Asquith, 18-211 S. K. Athithan 4-S26

2

Y.H.J. Au, 18-196 B. Audenard 1-148 J. Aviassar 1-179 Janine Avison 26-091 J. Awerbuch 8-S301 Masahiro Azemoto, 27-233 Friedrich-Wilhelm Bach, 25-331 Baldev Raj 4-S102, 6-209, 7-S1, 8-S126, 8-S140, 8-S149, 11-43 G. R. Baldwin 3-182 Rao Balusu 23-119 R.H. Bannister 19-209 V. Bansal 8-S217, 9-142 J. Baram 1-179 Ireneusz Baran, 17-S37, 23-173, 24-044, 24-222, 25-341, 27-272 P. Barat 8-S140, 8-S149 V. A. Barat 26-132, 28-109 Michel Barbezat, 22-127, 25-042 Roy Baria 23-113

Nektaria M. Barkoula 28-032 J.A. Baron 2-69 A. Barron, 18-87 Fady F. Barsoum, 27-040 M.W. Barsoum, 18-61 H. Barthelemy 8-S75 B. L. Baskin, 12-149 M. Nabil Bassim 4-S224 Matthew Baxter, 25-215 M. Bayray, 18-131, 19-241, 20-188 Mulu Bayray 24-022 Frank C. Beall, 4-S244, 4-19, 5-71, 6-151, 8-S311, 9-197, 9-215, 10-83, 12-i (1/2), 12-55 Alan G. Beattie 1-21, 1-300, 2-67, 2-69, 2-95, 2-143, 3-224, 3-239, 4-65, 5-53, 5-172, 15-63, 15-S111, 21-095, 23-299, 23-331 M. J. Beesley 7-59 Bernd-Arno Behrens 26-208 S. Béland, 12-45 R. M. Belchamber 4-71, 9-271 David A. Bell 5-1 S. H. Benabdallah 15-S117 Amadeo Benavent Climent 28-120 Petr Benes 26-060

P. G. Bentley 1-35, 7-59 Avraham Berkovits 11-85 C. C. Berndt 15-S117 J. M. Berthelot, 4-S178, 4-S300, 6-43, 11-11 Yves H. Berthelot, 12-27 M. M. Besen 8-S209

3

D. Betteridge 4-71 M. R. Bhat, 25-224 D. K. Bhattacharya, 4-S102, 8-S149, 11-43 Frank S. Biancaniello 5-S69 Thomas Bidlingmaier 14-S47 Jacek M. Biernacki, 12-55 T. Kevin Bierney 4-S321 Christophe Binétruy 26-120 B. Birolo 4-S255 Jan Bisschop 20-153 Michal Blahácek 18-279, 20-163, 20-274, 22-138, 28-099 P.R. Blackburn 7-49 M. J. Blanch 20-229 Richard W. Blank 9-181 J. A. Blessing 8-S236, 8-1 Thomas Blum 7-179 Andrew R. Blystra 10-S49 T. Boczar 17-S7 Thomas Böhlke 27-077 P. Böhm 9-29 Jurgen Bohse 15-S108, 16-S233, 16-S343, 19-001, 22-208 Yasuyuki Bokoi 16-S196 Konstantinos Bollas 27-027, 28-215 R. J. Boness 8-S192, 15-S117 Yrij Borodin 28-109 R.W. Bosch, 18-293 S. F. Botten 8-S330 S. Bousias, 25-033 P. Bowen, 13-S08 S. J. Bowles 10-49 P. Bowman 7-225, 8-S4 Marcelle Brachet 2-159 John Brandon 17-49 Franklin R. Breckenridge 1-87, 3-59, 10-43 J. C. Briggs 8-S209 W. D. Brosey 7-31, 8-S280, 9-75, 9-84 Nancy Brown 10-71 C. Brun, 18-155 Andreas J. Brunner, 15-S108, 22-127, 24-104, 25-042, 28-059 G. Budenkov 17-S13, 17-S51 R. Budzier 20-172 C. Buelens, 18-34 Аnton Bukatin 26-311 E. Bulatova 17-S13 O. Burenko 7-31 Maria Burger 22-102 Arie Bussiba, 27-077 Stephen D. Butt, 25-373 D. J. Buttle 7-211, 8-S158, 8-S201, 9-243, 9-255

4

G. Buzzacchi 2-11 L. Calabrese 28-129 CARP Aerospace/Advanced Composites Subcommittee 11-C1 Giuseppe Campanella, 25-324, 28-129 Bruce Campbell 3-81 Claudio Caneva, 25-080 Nicholas J. Carino 5-S24 Mark Carlos 15-S104, 15-S107, 15-S109, 18-189, 18-248, 18-272, 25-355, 28-229 John M. Carlyle 4-S329 Rami Carmi, 27-077 Victor Caron 4-S259, 4-115 Steve H. Carpenter, 1-251, 2-191, 3-11, 3-81, 4-S119, 4-S135, 5-77, 6-115, 6-136, 6-177, 6-215, 7-9, 7-161, 8-S135, 8- S184, 8-125, 9-1, 10-97, 10-103, 11-5, 12-141, 13-S01, 15-43, 16-S10, 24-119 Damian Carter 17-49 M. Cartoceti 2-11 Daniel Casellas 28-163 Dick W. Caster 9-197 Enrique Castro 26-142, 26-247, 28-120 Johann Catty 18-205, 27-299, 28-011 Jose A. Cavaco 27-001 M. Cerny 17-S20 J. Cerv 20-025 Hwakian Chai, 27-186 Roger W.Y. Chan 3-118, 4-S259, 4-115, 8-S12 C. Chang 4-S62 Chung Chang 7-21 C. Chapelier 5-S52 T. Chelladurai 8-S88, 12-111 D. Chellman 4-S263 Chung-Mei Chen 7-161 Jihua Chen 19-001

M.C.Cheresh 2-289 M. Cherfaoui 5-S66, 22-071, 11-1, 18-144 Michael J. Chica 9-197 A. Chichibu 8-107 Chang-Sheng Chien, 15-S118 Y. S. Chin 11-71 Pornthep Chivavibul 23-091, 24-076 Robert Chivers 10-123 Akiyoshi Chichibu, 11-S47, 12-S1 Milan Chlada 17-S57, 20-134, 28-099 Frantisek Chmelík 17-S29, 20-108 Hideo Cho 16-S115, 21-112, 22-119, 22-224, 22-243, 23-072, 23-196, 23-206, 23-277, 24-012, 24-067, 24-084, 24-127, 24-161, 25-107, 25-124, 25-157, 25-172, 25-267, 26-072, 27-224,

5

28-256 N. S. Choi 16-S324 Alex Chong 28-085 A-A. R. Choudhury 16-S125 T. Chow 8-S166 Y. T. Chow 4-71 Min-Hwa Chung, 13-1 S. Y. Chuang 4-S137 S. S. Christiansen 4-S116, 5-85, 8-S184 F. Cipri 4-S255 M.A. Clark 15-S107 T. N. Claytor 4-S69 Roger B. Clough 5-S69 D. M. Chuck 8-S258 P. Cole 15-S109 Phillip T. Cole 8-S239, 8-31, 18-61, 18-180, 18-232, 19-191, 22-022 C. E. Coleman 5-S16 J. Collins 4-S134 M. P. Collins 9-271 Peter J. Conlisk 8-1 A. W. Cook 8-S101 David B. Cook 17-83 R. Daniel Costley, Jr., 12-27 R. A. Coyle 6-249 J. Crha 17-S45 H.-A. Crostack 9-29 Timothy G. Crowther 10-71 M. E. A. Cudby 4-71 C. E. D'Attellis 10-13 V. Dal Re 4-S255, 4-S270, 5-39 Wojciech Darski, 25-341 B. Dattaguru 6-19 A.W. Davies, 18-232 Jack F. Dawson 10-117 V. G. Ruiz de Argandoña 10-S35 Igor Maria De Rosa, 25-080, 26-220 Juan-José G. De La Rosa 26-142 S. De Bondt, 11-95 I. De Iorio 3-158 P. De Meester 4-S186, 8-S272, 13-79, 15-S105, 18-34 C. De Michelis 2-11 L. M. Suárez del Río 10-S35 L. Delaey 11-95 H. A. L. Dempsey 4-S46 O. Derakhshan 8-S223 J. Derenne, 18-299 A. Deruyttere 11-95 Wendy Desadeleer, 22-253

6

M. Diez 4-S296 Seth-Andrew T. Dion, 25-187 John J. Ditri, 12-23 Nitin Dhond 4-S30 G. Dionoro 2-281 Jason W.P. Dong 16-S1 Jason Dong, 25-355 P. Dörner, 22-236 David A. Dornfeld 3-242, 4-S123, 4-S228, 7-103, 7-111, 8-S227, 6-29, 6-37, 6-157 Henrique L. M. dos Reis, see Reis R.D. Douglas, 18-96 Karyn S. Downs, 13-31, 13-45, 13-56, 14-S61, 15-S109, 16-iv, 16-S333, 21-052, 21-070, 21-A01 Martin Drab 26-189 M. W. Drew 6-239 T.F. Drouillard, 1-45, 1-81, 1-121, 1-195, 1-271, 2-129, 2-221, 2-292, 3-46, 3-90, 3-164, 3-212, 4-41, 5-103, 9-45, 9-155, 9-215, 11-53, 12-71, 12-79, 13-42, 14-1, 16-v Q. Duan 8-S97 Qingru Duan 16-S243 R. Dubiel, 18-15 J. C. Duke 8-S179 H.L. Dunegan 8-S71, 15-53, 15-S106, 16-v J. Dunning 4-S22 C. Divaker Durairaj 17-15 F. Dusek 10-1 A. G. Dutton 20-229 C. Duytsche 3-176 J. Dvoracek, 18-81 B.C. Dykes 4-S35 Yuris A. Dzenis 15-S112, 20-016, 20-099 Mark Eaton, 25-140, 25-215, 26-172 J. Eberlein 20-172 Davis M. Egle 2-ii (3), 3-104, 4-S30, 4-S46, 6-205 Gernot Eilers 19-100 J. Eisenblätter 15-S119, 16-S85, 19-100, 19-153, 24-179, 24-196 Islam El-Galy 26-208 T. El-Raghy, 18-61 Sergey Elizarov 26-311 R.K. Elsley 2-47 T. Ely 16-S10 D. C. Emmony 1-263 C. Ennaceur, 22-071 Manabu Enoki, 4-S195, 8-S154, 13-S29, 15-S90, 15-S120, 16-S269, 21-142, 23-292, 23-310, 24-139, 25-247, 26-182 R. M. Esbert 10-S35 V. A. Eshwar 8-S217, 9-142

7

Sam Evans, 25-215 W. T. Evans 11-71 A. Fahr, 8-S314, 12-39, 15-S122 S. D. Falls 8-S166 Daining Fang 11-85 Ahmed M. Farahat 11-S37 J.-P. Favre 9-97 Carol Featherston, 25-140 V. P. Fedorov 20-218, 26-279, 28-234 G. Fernando 20-229 F. Ferrer, 18-155 Steven E. Fick 4-S311 Andrzej Figiel 4-S182 Chantal Filion 16-S186 F. Finck, 22-083 P. Finkel, 18-61 R.D. Finlayson, 18-61, 18-189, 18-272 J. F. Finn 8-S79 P. Fleischmann 3-176, 5-S42, 9-91, 19-229 Peter Flüeler 22-127 T. Flynn 10-S59 Jens Forker, 18-258, 25-132 D. S. Forsyth 15-S122 R. Fougeres 5-S42, 9-91 Timothy J. Fowler 8-S236, 8-1 H. Frackiewicz, 12-149 R.P. Franke, 22-236 Michael J. Friedel 10-S77 M.A. Friesel, 3-11, 3-239, 7-119, 10-117, 18-61, 18-189 Mike I. Friswell, 27-291 L. Froyen, 11-95 Yoshiaki Fujii 23-119 Taisaku Fujioka 15-S31 Tetsuro Fujiwara 10-S63, 11-S65, 15-S40, 15-S113 Kenzo Fukaura 19-091, 21-112 T. Fuketa, 11-21 Hiroyuki Fukutomi 23-091, 24-076 Yasushi Fukuzawa, 27-064 Roy D. Fultineer, Jr. 15-S103 Masami Fushitani, 4-S240, 9-209, 13-S42 Takashi Futatsugi 23-249 Laurent Gaillet 26-032, 27-254 M. Gakumazawa 16-S150 Antolino Gallego 26-142, 26-247, 28-120 Thomas Gandy 4-S166 D. S. Gardiner 4-S199

8

John Gary, 12-157, 14-103, 15-S115, 16-S251, 17-37, 17-97, 19-258, 20-039, 20-062 Michael F. Gasick 7-111 Ludwig Gauckler 14-119 B. K. Gaur 7-S13 Stephen N. Gautrey 18-180, 19-191, 22-022 A. Gavens 8-S277 Maochen Ge 15-S105, 8-S32, 21-014, 21-029 P. Gebski 17-S37, 19-285, 20-083 D. Geisse, 12-171 B. Georgali, 18-21 János Geréb 10-19 S. Ghaffari 8-S301 S. Ghia 4-S86 Al Ghorbanpoor 4-S307 J.D. Gill, 18-96, 18-211 Steven D. Glaser 10-S1, 25-316 T. G. Glenn 1-81 A. M. G. Glennie 4-S170 M. W. Godfrey 1-263, Valery Godinez, 4-103, 18-272, 25-355 L. Golaski 1-95, 4-S182, 17-S37, 19-285, 20-083, 24-187 Douglas E. Goldsack 16-S186 Edward Goliti 5-7 V.P. Gomera, 18-111, 20-218, 26-279, 28-234 Javier Gomez, 13-S21 Carlos M. Valdes-Gonzalez, 12-117 M. Gori, 18-167 Michael R. Gorman 8-51, 9-131, 9-283, 13-S01, 14-i (3/4), 17-29, 23-037 G.L. Goswami 4-S98, 4-S251, 7-S40, 7-S43 J. Goudiakas, 18-155 Per A Gradin, 13-97, 26-023 Igor Grabec 8-S20, 8-S205, 11-79 L. J. Graham 2-47 A. T. Green 4-124, 8-S306 J. E. Green 1-191, 2-289 P. Gregson, 18-239 David W. Greve, 25-115 Arthur T. Grodotzke 1-29 Boguslaw Gronowski 4-S82, 5-25 D.J. Gross 8-25 S. Gross, 18-239 Christian U. Grosse 14-S74, 22-083, 25-316, 28-188 C. M. Grossi 10-S35 Janez Grum 26-199, 28-179 N. Gsib 5-S60 P.-Y. Gu 8-S188 F. J. Guild 1-244 T. J. Gulley 4-S170

9

M. Gullikson 26-023 Dawei Guo 14-S19, 16-S222 Hua Guo 23-119 Yanfeng Guo 16-S317 B.C. Gupta 8-S217, 9-142 Robert A. Haack 9-181 Y. Haddad 8-S314 M. E. Hager 8-S42 H. Thomas Hahn 5-15 P. Hähner 20-265 Shigenori Hamada 15-S40 Takashi Hamada 4-S325 Darulihsan A. Hamid, 27-064 A. Hampe 9-103 M. A. Hamstad 2-57, 2-i(3), 5-95, 5-110, 5-123, 6-93, 9-75, 9-84, 11-33, 12-157, 13-31, 13-45, 13-56, 14-103, 14-S61, 15-1, 15-S108, 15-S109, 15-S115, 16-S222, 16-S251, 16-S333, 17-37, 17-97, 19-258, 20-039, 20-062, 21-052, 21-070, 21-A01, 22-001, 22-A01, 23-001, 23-047, 24-234, 25-092, 25-194, 26-040, 27-114, 28-041 L. Hanacek 8-S84 Kotaro Hanabusa 22-159 Mineyuki Hanano 11-S75 L. D. Hall, 19-209 V. Hänel 14-115 J. J. Hanley 11-27 Christian Hansmann 26-014 Yoshio Harada 23-181

H.R. Hardy, Jr. 3-242, 4-S19, 8-S32, 8-S42, 8-S262, 8-65, 10-61, 15-S105, 15-S118, 16-S277 Robert W. Harris 6-239, 8-S14, 8-S66, 10-S29, 10-S59, 15-S113, 17-121 W. F. Hartman 1-144, 4-S64, 5-31 J. Harvey 4-S220 H. Nayeb-Hashemi, 12-1 Toshiyuki Hashida, 13-S68, 24-215 F. P. Hassani 8-99, 10-61 Hiroaki Hata 7-173, 10-S63 Hajime Hatano 15-S115 D. Robert Hay 3-118, 4-S259, 4-115, 8-S12, 9-9, 27-001 J. R. Hay 8-S12 Hiroshi Haya 22-039, 23-260, 24-205

Tomoharu Hayano 24-067, 25-107 F. Havlícek 17-S45 M. W. Hawman 4-S131 Takefumi Hayashi 8-35 Toshimitsu Hayashi 26-160 Yoshie Hayashi, 19-035 Y. Hayashi 21-131

10

Yasuhisa Hayashi 7-185, 14-69, 15-S108 Jianmei He 20-194 R. Hecker 4-S78 C. R. Heiple 1-221, 1-251, 4-S116, 5-85, 6-177, 6-215, 8-S184, 8-125, 9-1, 10-97, 10-103 D. P. Henkel, 8-S318 Edmund G. Henneke II 8-S277, 14-53 B. Herrmann, 18-167 C. Hervé, 18-125, 22-071 Stefan Heusermann 16-S85 S. Hewerdine 8-S238, 8-21 Hartmut Heyse 5-45 H. Hick 10-67 Kazuo Hiekata 19-045 Christopher Higgins, 25-316 Y. Higo 8-S24, 16-S150, 16-S196, 19-085 Hiroshi Hikosaka 10-S13 Eric v. K. Hill, 25-187, 27-040 Roger Hill 1-73, 1-149, 1-294, 5-51, 5-i(1), 10-124, 16-S125 Akira Hine 27-064 Takayasu Hirakawa 23-156 Akinobu Hirama 27-186 Yuichiro Hirose 26-182 Y. Hisamatsu 2-19, 2-71 Koji Hisamatsu 11-S1 S.V. Hoa 7-145, 9-37, 11-65 A. B. M. Hoff 4-S165 Karen Holford 17-49, 18-232, 22-166, 25-140, 25-215, 26-172, 28-085 T. J. Holroyd 4-S132, 7-193, 8-S219 Kyoji Homma 10-35 Michel Hone 4-S259, 4-115 K1-Jung Hong, 13-S61 Theodore Hopwood II 4-S304 K. Horikawa, 19-022 Keitaro Horikawa 21-206, 21-223 Siegfried Horn 26-001, 28-142 P. M. Horrigan 8-S79 Filip Hort 26-189, 28-170 Hikaru Hotta 28-001 J.R. Houghton 8-S28, 8-S223, 14-61 Martin T. Howald 24-104 O. Hoyer 9-103 Nelson N. Hsu, 4-S311, 5-S24, 5-S28, 5-S29, 13-23 S.-Y. S. Hsu 1-183, 1-237, 2-169 J. Hu 16-S150 Christian Huber 22-127 J. H. Huh 15-S80 Derek Hull 1-95 Patrice Hulot 26-120

11

Donald H. Humes 17-29 Wolfgang Hundt 14-119 U.-D. Hünicke 20-172 David A. Hutchins 5-S29, 5-S34, 8-S38, 8-S166 David V. Hutton 8-41 H. Hutton 3-239, 4-S74, 4-S138, 6-167 Hatsuo Ichikawa 4-S325 Y. Ichimura 19-142 Makoto Ichinose 20-001 Hassane Idrissi, 18-299, 18-307, 26-032, 27-254 Chikako Ikeda 24-228 Junji Ikeda 24-173, 24-228 R. Ikeda 21-131, 22-119 Yukifumi Ikeda 23-272 Sei Ikegaya 23-096 H. Imaeda 4-S294 Takuichi Imanaka 4-S38 Hidehiro Inaba, 8-S24, 19-196 Takako Inaba 10-S90 T. Inamura, 19-085 Ichiro Inasaki 7-179 N. F. Ince, 27-176 Tomoaki Inoue 2-1 Yoshiki Inoue 11-S89 Akichika Ishibashi 11-S65, 15-S40, 15-S113 T. Ishida 10-S42 Tsuyoshi Ishida 27-194 Takeshi Ishigohka 24-111 K. Ishihara 6-13 Hisashi Ishitani 4-S325 C. Ishiyama 16-S150, 16-S196 Toshiro Isoda, 25-021 Ken-Ichi Itakura, 13-S54, 13-S75, 19-109, 23-119 Kaita Ito 24-139, 25-247 D.S. Ivanov 26-240 V.I. Ivanov, 18-144 N. Iverson, 25-364 K. Iwai 21-197 K. Iwaki 21-166 Keisuke Iwaki 23-260 Y. Iwata 16-S142 S. Iyer, 18-189 Takeshi Izuta, 13-S42 Laurence J. Jacobs, 12-27 Jay B. James 4-S119, 5-77 P. Jax 2-29, 8-S53 T. Jayakumar 4-S102, 7-S1, 8-S126, 8-S140, 8-S149, 11-43

12

J. S. Jeng 8-S268 Hee-Don Jeong, 13-S61 B. B. Jha 6-209, 8-S149 S. K. Jha 7-S43 Kyung-Young Jhang 16-S261 C.Y. Jian, 13-S68 C. G. Jiao 8-S105 Takehiko Jibiki 24-173 K. Jo 8-107, 10-S55 J. Johkaji 8-S1 C. H. Johnson 4-S111, 4-S263 Malcolm J. S. Johnston, 12-117 W.D. Jolly 4-103 H. Jonas 4-S78 L. E. Jones 20-229 R. H. Jones 3-239 R. K. Jones 7-119, 8-S223 Magnus T. Jonsson, 25-252, 25-260, 26-229, 26-262 Young-Chan Joo 15-S1, 16-S212 P. A. Joosse, 20-229 G. Jothinathan 4-S207 Musa K. Jouaneh 10-83 S. J. Jung, 8-S326 B.S. Kabanov, 18-111, 20-218 Koji Kagayama 23-277, 24-127 Kazuro Kageyama, 13-S89, 19-045, 21-176 Kensuke Kageyama 24-097 Hideshi Kaieda 23-129 Katsuyuki Kaiho 24-111 Karl-Ulrich Kainer 20-108 Koji Kaino 5-61, 9-277 K. Kajiyama 19-022 T. Kakimi 2-19 Y. Kakino 3-108 Małgorzata Kalicka 24-187, 27-018 S. Kallara 8-S28 P. Kalyanasundaram 8-S140, 8-S149 T. Kamada, 19-134 M. Kamata 8-107 Masahiro Kamata 23-081 Tadashi Kambara 24-097 Peter Kamlot 16-S85 Makoto Kanai 21-176 Yasuyuki Kanai 24-097 Yasuhiro Kanemoto 15-S40 Elijah Kannatey-Asibu, Jr. 15-S118 Shigeto Kano 6-109, 6-145 C-S. Kao, 25-364

13

Chu-Shu Kao, 27-176 Justin O. Karl, 25-187 Bo Karlsson 26-014 Naoya Kasai 27-281 M. Kat 8-99 Chiaki Kato 19-053 Tatyana Katsaga, 25-294 K. Katsuyama 11-S19 Kunihisa Katsuyama 11-S27 Harold E. Kautz, 5-144, 12-65 M. Kaveh, 27-176 Tatsuya Kawada 24-215 Yoshiaki Kawaguchi, 12-127 Yutaka Kawai, 7-167 Teppei Kawakami 21-149 Kenji Kawamoto 9-109 Sumire Kawamoto 28-068, 28-076 Hiraku Kawasaki, 27-281 Yuma Kawasaki 28-204 Tomaž Kek 26-199, 28-179 James R. Kennedy 4-S90 Jiri Keprt 26-060 Shahla Keyvan 10-91, 14-97, 15-79 A. Wahab Khair 4-S1, 8-S326, 15-S105, 16-S53 A. S. Khan, 8-S246 A. S. Khanna 6-209, 8-S103 S. Khardi 26-290 E. W. Khokhlova, 12-149 Jens Kiehn 20-108 M. T. Kiernan 8-S176 S. Kihara, 18-68 Tadashi Kikuchi 11-S47 Cindi Kilkenny 16-S186 Byoung-Geuk Kim 15-S120 Byung-Nam Kim 12-S24, 15-S90, 16-S269 Dal-Jung Kim 16-S261 K. H. Kim 4-S282 K. Y. Kim 4-S62, 8-S170 Kyung-Woong Kim, 13-1 Sang-Hyo Kim 15-S11 Isao Kimpara, 13-S89, 19-045 H. Kimura 4-S294 Ron King 10-91 Tetsuo Kinjo 15-19, 14-69 N. Kinoshita 10-S42 Teruo Kishi 1-1, 2-19, 2-71, 4-S191, 4-S195, 4-S278, 4-S282, 4-S325, 6-85, 8-S131, 8-S154, 12-127, 13-S29, 15-S90, 15-S120, 16-S269 Fuyuhiko Kishinouye 9-177, 9-180

14

Takahiro Kishishita 10-S55, 11-S47 N. N. Kishore 8-S297 P. Kisnomo 15-33 Tatsuo Kita 10-S90 K. Kitadate 8-S131 Itaru Kitamura 28-001 K. Kitano 10-S42 Shigeo Kitsukawa 23-233 Kiyoshi Kiuchi 19-053 R. A. Kline 4-S42, 4-107, 6-205, 8-S246 W. Knabl 20-257 Markku Knuuttila 19-162 E. Kobayashi 3-130, 4-93 H. Kobayashi 8-S145 Satoshi Kobayashi 21-149, 23-150 Takao Kobayashi 21-001 Yoshifumi Kobayashi 23-181 Y. Kobayashi 8-S1 Miroslav Koberna 11-61 R. M. Koerner 1-220, 2-187, 2-195, 4-S11, 4-31 Tsuguaki Koga 2-1 Takao Koide, 13-S47 Masami Koike, 19-202 Takao M. Kojima 24-097 B. Koktavy, 17-S100 J. G. Kolaxis 17-69 V. Kolovos, 18-217 Yuuki Komiyama 27-064 Hidemichi Komura 19-196 Shinya Kondo 26-270 Johannes Konnerth 26-014 Shigeo Konno 23-233

Ja-Ho Koo 15-S90, 16-S269 T. M. Kooistra 20-238 Atsushi Korenaga, 19-196 Andrej Korcak 27-040 M. Korenska, 18-29 Leszek Korusiewicz 2-272 I. Kosiková 17-S100 David E. Kosnik, 27-011 T. Kossivas 20-229 V. Kostopoulos 20-265 A. Kotolomov 17-S51 D. A. Kouroussis, 18-217, 18-224, 20-229 Dimitrios Kourousis 27-027, 28-215 Torsten Krietsch 16-S233, 16-S343 R. Krishnamurthy 4-S26, 8-S88 Vladimir Krivobodrov, 13-87 J. Krofta, 18-279, 20-274

15

J. Królikowski, 12-149 D.A. Kronemeijer, 18-174 Joseph Krynicki 15-S116 Takafumi Kubo, 13-S42 I. Kukman 8-S205 P. G. Kulkarni, 7-S13 Yusuke Kumano 23-129, 27-167 J. Sampath Kumar 7-135 M. Kumosa 1-95, 16-S10 M. Kunieda 19-134 Minoru Kunieda, 27-263 Moshe Kupiec, 27-077 D. S. Kupperman 4-S69 Yu Kurokawa 23-224, 24-145 Krzysztof J. Kurzydlowski 25-166 Yasufumi Kusano, 13-S75 F. M. Kustas 10-97 T. Kusu 21-142 Alexey Kuzmin 28-109 May Man Kwan 3-144, 3-190 Oh-Yang Kwon 4-S106, 9-123, 9-227, 9-237, 13-1, 13-S83, 15-S1, 15-S19, 16-S212 H. Kwun, 11-27 J.F. Labuz, 25-364, 27-176 G. Lackner, 18-167, 20-179, 22-201 J.-C. Laizet 9-97 A. Laksimi, 18-125, 22-071 V. Lalitha 4-S26 Michal Landa 17-S57, 20-025, 20-163, 22-138 E. Landis 10-S97, 15-S104 R. J. Landy 1-7 Terence G. Langdon 20-108 F. Langella 3-158 Stanislaw Lasocki 4-S7 Robert J. Lark 22-166, 26-172 Ralf Laschimke, 22-102 A. Lavrov 20-292 Alexey Lazarev 27-144, 27-212 Sergey Lazarev, 27-212 Marcel F. Leach 10-S18, 11-19, 16-S186 James D. Leaird, 3-204, 4-S22, 8-S322, 12-117 R. D. Leblanc 15-S122 Chong Soo Lee, 13-S61, 15-S80 Joon-Hyun Lee, 13-S83 K. A. Lee 15-S80 P. Y. Lee 15-S117 Sang-Ho Lee 15-S11 Seung-Hwan Lee 15-S19

16

Sekyung Lee 15-S120 Seung-Seok Lee 15-S11 Weon-Heum Lee 16-S261 L. Legin 18-144 Xinglin Lei 23-102 Jack Leifer 4-S127 Constantina Lekakou 23-025 D. J. Lekou 20-229 A. M. Leksowskij, 12-149 A. Lemascon 5-S66 Richard L. Lemaster, 4-S228, 6-157, 7-103, 7-111, 8-135, 9-17, 9-203, 10-83, 12-55 J. C. Lenain, 18-161 K. H. Leong 23-025

Armand F. Lewis 4-S127 Luidmila Lezvinsky 16-S35 Steinar Lfvaas 4-S161 Bang xian Li 16-S243 Jihui Li 24-001 L. Li 7-145, 9-37 Y. P. Li 23-292 Zhengwang (Z. W.) Li 21-213, 23-233 Z. W. Li 19-118 Steven Y. Liang 6-29, 6-37 T. Lilley 4-71 S. Lim, 19-134 Jae-Kyoo Lim 16-S309 R. Lima 15-S117 A. Limam, 18-307 Vidyadhar Limaye, 25-373 C. K. Lin 15-S117 Dan Lindahl 19-162 J. Liöka 17-S108 S. Liu 8-S97 Shifeng Liu 15-S124 T. Liu 15-S117 X1-qiang Liu 15-S123 Y.-H. Liu 9-9 David A. Lockner 14-S88 Manuel Löhr, 22-190 T. Lokajícek 17-S100 Thomas Lokajicek 22-091

S.V. Lomov 26-240 M. I. López Pumarega 17-61 Arthur E. Lord, Jr. 2-187, 2-195, 3-107, 4-S11, 4-31, 5-152 Luis Lorenzo 5-15 E. Lowenhar 15-S109 M.G. Lozev 15-S104 Guozhi Lu 4-S203

17

Krystyna Lublinska, 25-166 P. Lukác 17-S29, 20-108 K. Lundgren 5-S29 Xiu Luo 22-039, 23-260, 24-205 Xun Luo 23-119

D. Lupascu 17-S78 Yukuan Ma 4-S58 J. D. MacPhail 8-S4 Yuichi Machijima 23-091, 27-233 A. Machová 20-025 J. W. Maclachlan 1-223, 1-229, 2-39, 2-179, 3-1, 4-S151 Eric Madaras 23-037 M. R. Madhava 8-S288 Morihiko Maeda, 19-202 S. Maharshak 2-64 A. Maie 21-213 Ian Main, 13-S21 M. A. Majeed, 4-S147, 8-S16, 12-107 Z. J. Majewska, 12-S7 Zofia Majewska 16-S105, 18-1 S.A. Majewski, 12-S7 Arup Maji 15-S116 O. Makishima 21-166, 25-308 Ajit Mal 14-S19, 16-S222 M. Manoharan 4-S207 Ll. Mañosa 5-S49 Gerd Manthei 15-S119, 16-S85, 19-100, 22-173, 24-179, 24-196 Theodore J. Mapes 1-29 J. Maram 4-S134 H. Marcak, 12-S7 P. A. March 8-S223 K. Marsh 15-S104 G. G. Martin 4-S142 Eva Martinez-Gonzalez 28-163 Hiroaki Maruyama 23-233 Seiji Masano 27-241 A. Maslouhi, 8-S292, 12-45, 15-S122, 16-S299 Y. Masui 23-189 S. Mata 17-23 O. Matal 17-S65 Kristian Mathis 20-108 Theodore E. Matikas 28-032 Takuma Matsuo 24-067, 24-084, 25-124, 26-072, 27-224, 28-256 K. Matsuura 21-120 Kimitoshi Matsuyama 11-S65, 15-S40 Ryosuke Matsuzaki 27-089 Huber Matysiak 26-317 S. C. Maxwell 8-S38

18

Masami Mayuzumi 23-224, 23-285, 24-145, 25-238 P. Mazal 17-S2, 17-S20, 17-S70, 18-75, 26-189, 28-170 H. Mazille, 18-299, 19-229 D. Mba 19-209, 22-077 Stuart L. McBride 1-223, 1-229, 2-39, 2-179, 3-1, 4-S151, 4-S220, 7-225, 8-S4, 8-S192 J. F. McCardle 10-49 J.D. McColskey 15-1, 15-S108, 15-S111, 15-S119 John W. McElroy 4-S77 D. Michael McFarland 5-67 K. I. McRae 7-225 M. Mediouni 22-071 Ronald B. Melton 1-266 P. G. Meredith, 12-S12 Philip Meredith, 13-S21 Jakub Michalski 26-317 K. Michihiro 10-S63 W. Mielke 9-103 Juha Miettinen 21-230 Hamish D. S. Miller 5-S1 M. E. Miller 4-S22 R. (K.) Miller 8-S241, 8-25, 15-S104, 15-S107, 15-S108, 18-61, 18-189, 18-272 O. Minemura 16-S75 J. R. Mitchell 6-135 J. Mitchell 15-S109 A. Mittelman 3-41, 6-73 S. Miwa, 19-142 Makoto Miwa, 13-S42 Kouitsu Miyachika, 13-S47 Kazuya Miyano 10-S90 Hiroyuki Miyatake 10-S13 Noriko Miyoshi, 25-107 Yuta Mizuno 27-224 Junichiro Mizusaki 24-215 Souichi Mizutani 20-194

Yoshihiro Mizutani 16-S115, 18-51, 18-286, 19-035, 20-194, 23-224, 23-285, 24-145, 25-179, 25-238, 26-109, 27-089 Kenji Mochizuki 8-35 Mark B. Moffatt 1-29 K. Mogi 1-37, 8-113, 16-S45 Keisuke Mogi 27-137 James Mohr 5-162 Waldemar Molinski 10-107 Florian Moll 20-108 M. Momayez 10-61 Shohei Momoki, 27-186 Keiichi Monma 15-S113 M. Montoto 10-S35

19

Antonio Moreno-Muñoz 26-142 Bryan C. Morgan 15-69, 15-S103 L. Morgan 16-S125 Winfred Morgner 3-172, 5-45, 6-133, 8-S70 Hisakazu Mori 27-233 Y. Mori 8-S131, 16-S45, 20-248, 21-197, 22-091, 27-157 Yoshiki Morino 20-194 Hiroyuki Morishima 20-145 Hirokazu Moriya 23-113, 23-129, 23-142, 24-196, 28-001 D. G. Morris 7-95 K. Morofuji 21-213 Zofia Mortimer 16-S105, 18-1 W. J. Mos cicki, 12-S7 H. G. Moslé 8-S317 F. Mostert 18-189 Andre Moura 23-102 Alexander Mozgovoi, 27-212 F. Mudry 6-85 Amiya K. Mukherjee 5-162 M. C. Mumwam 16-S65 Kohei Murakami 23-215 Yuji Murakami 10-S90 Sumihiko Murata 27-194 K. Murayama 16-S75 Boris Muravin 16-S35 Gregory Muravin 16-S35 S. A. F. Murrell, 12-S12 C. R. L. Murthy, 4-S30, 4-S147, 6-19, 7-S18, 8-S16, 8-S122, 8-S284, 12-107, 25-224 Vasile Mustafa 27-001 T. M. Mustaleski 4-S247 S. Naemura 10-S55 T. Nagamachi 23-189 Takuo Nagamachi 22-159 Koji Nagano, 11-S1, 13-S54, 13-S75 S. Nagano, 19-085 Jyothi Nagaraj 14-97 Shigeru Nagasawa, 27-064 Kenji Nagashima, 19-011 Yasuaki Nagata 4-S191 M. M. Nagl, 11-71 G. Jayachandran Nair 8-S266 P. K. K. Nair 4-S98 Hidehumi Naito, 12-S24 Yoichi Nakai 24-097 Hideo Nakajima 24-111 Hideyuki Nakamura 23-243, 26-109, 27-281 M. Nakamura 10-S55, 8-107

20

Motoaki Nakamura, 27-241 Syouzou Nakamura 10-S13 Yasuhiro Nakanishi 20-145, 22-039, 23-260 Hiroyasu Nakasa 16-S25 Eisaku Nakashima, 19-202 M. Nakano 21-197 Noritaka Nakaso, 13-23 Gen Nakayama, 25-157

B. C. Nakra 9-25 H. Nayeb-Hashemi, 12-1, 14-85, 15-33 O. Nedzvetskaya 17-S13, 17-S51 Priscilla P. Nelson 10-S1 Katsumi Nemoto 23-142 Y.S. Neo, 18-96 K. P. Nerz 5-S56 John P. Newhook 25-373 Eng T. Ng 19-275 K. D. Nicklas 4-S247 Jackson A. Nickerson 6-37 A. Nielsen 8-S57 Peter Niemz 24-104 Hiroaki Niitsuma 7-201, 11-i(4), 11-S1, 23-064, 23-113, 23-129, 23-142, 24-196, 27-167 Tadafumi Niizato 28-001 Akira Ninomiya 24-111 T. Nishida 21-187 S. Nishikawa 10-S55 Koichi Nishimoto 4-S236 Shigeto Nishimoto 19-202 Hideo Nishino 23-189, 24-153, 26-270 M. Nishino 8-S131 Satoshi Nishinoiri 23-310, 24-076 V. N. Nikolaidis, 17-69, 20-229 Minoru Nishida, 12-S18 H. Nishino, 18-51, 18-102, 18-286, 19-011, 19-035, 19-075 Takako Nishiura 24-052 Osamu Nishizawa 11-S27, 23-102 Masami Noguchi 4-S236 M. Noguchi 11-21 C. Nojiri 16-S150 Hiroaki Noma, 25-107 Richard Nordstrom 15-S108, 16-S204 S. Norgren 26-023 Václav Novák 20-163 Marek Nowak 24-044, 24-222, 25-341, 27-272 A. Nozue 1-1 Akira Nozue 21-149 J. Nuffer 17-S78 Staffan Nyström, 13-97, 26-023

21

Y. Obata 8-S145, 16-S45, 20-248, 22-091, 27-157 S. I. Ochiai 2-289 Satoshi Oda, 13-S47 A. O’Gallagher 14-103, 15-S115, 16-S251, 17-37, 17-97, 19-258, 20-039, 20-062, 21-052, 21-070, 21-A01, 22-001, 22-A01, 23-001 Takashi Ogata 24-076 Takeshi Ogawa 23-156, 23-196, 23-181, 23-277, 24-127, 24-161, 25-124 Steve L. Ogin 23-025 S. Ogino 5-61, 9-277 Eu Seok Oh 16-S277 Hidefumi Ohara 28-001 Yoshitugu Ohigashi 10-25 Takanori Ohira 4-S274 Hironobu Ohishi 23-064 A. Ohmori 21-206 Tatsuma Ohnishi 19-109 Tadashi Onishi, 25-238 Kentaro Ohno 23-047

Isamu Ohsawa, 13-S89, 19-045 Shiro Ohta 15-S40 M. Ohtsu 1-103, 2-151, 2-247, 3-27, 3-59, 3-69, 4-S38, 4-S50, 4-S316, 5-124, 6-43, 6-79, 6-99, 7-167, 8-S162, 8-S242, 8-93, 10-i (1/2), 11-i(4), 11-S37, 11-S57, 11-S65, 11-S89, 13-S14, 15-S31, 15-S40, 15-S50, 15-S60, 15-S70 16-S65, 16-S95, 18-S1, 18-S7, 19-118, 19-142, 19-184, 20-001, 21-157, 22-030, 23-047, 23-136, 23-272, 25-021, 28-204 H. Ohyama 4-S282 M. I. Ojovan, 25-051 Takahisa Okamoto, 13-S14 T. Okamoto 16-S75 Takeshi Okano 4-S240 A.A. Okhotnikov, 18-111, 20-218, 28-234 Nobuhiro Okude, 27-263 S. Okumura 11-21 L. Okushima 17-15 Kanji Ono, 1-7, 1-67, 1-69, 1-141, 1-145, 1-146, 1-183, 1-211, 1-213, 1-214, 2-169, 2-247, 2-ii(1/2), 3-19, 3-27, 3-59, 3-69, 3-130, 3-144, 3-174, 3-190, 3-233, 3-ii(4), 4-S50, 4-S106, 4-S111, 4-S263, 4-S316, 4-61, 4-93, 5-124, 5-i(1), 5-ii(4), 6-1, 6-43, 6-84, 8-S268, 9-109, 9-123, 9-177, 9-227, 9-270, 11-117, 12-177, 14-35, 14-69, 14-S19, 15-19, 15-S95, 15-S105, 15-S108,

16-S115, 16-S134, 16-S289, 17-S37, 18-51, 18-102, 18-286, 19-011, 19-035, 19-063, 19-075, 19-091, 19-285, 20-083,

21-112, 21-120, 22-119, 22-243, 23-173, 23-206, 24-044, 24-119, 24-187, 24-222, 25-001, 25-179, 26-072, 27-272, 28-256

22

M. Onoe 5-i(4) Irving J. Oppenheim, 25-115 Hisanori Otsuka 10-S13 P. Ouellette 9-37, 11-65 C. Ouyang 10-S97 H. Oyaizu 8-S1 K. Ozawa 21-166 Joško Ožbolt 28-188 Didem Ozevin, 25-355

Alkiviadis S. Paipetis 28-032 Jie Pan 22-264, 22-274 Yih-Hsing Pao 4-S274 Dimitrios Papasalouros 28-215 S.A. Papargyri-Mpeni 28-246 A.D. Papargyris 28-246 D.A. Papargyris 28-246 Krystian Paradowski, 25-166 Philip Park 15-S11 Young-Jin Park 15-S11 H. B. Patel 8-S12, 8-S101 S. C. Pathak 4-S32, 4-S147, 8-S122 J. Pavelka 17-S100, 22-091 Martyn Pavier 27-291 J. Pazdera, 18-81 L. Pazdera, 18-29 Martin Peacock 4-S166, 8-S240, 8-11 L. H. Pearson 4-S199 Peter Pellionisz 10-19 R. Pensec, 18-125 L. V. Perez 10-13 Goran Periškić 28-188 Marianne Perrin 26-032 D. T. Peters, 8-S4 Tatiana B. Petersen 27-098 S. Peteves 20-265

J. Petras, 18-75 L. Petras, 18-81 J. Petrasek 17-S83 Christian Pfleiderer, 12-141 T. P. Philippidis, 13-11, 17-69, 20-229 Romana Piat, 27-077 Ingrid Picas 28-163 Rodney G. Pickard 15-79 C. Picornell 5-S49 Aleksander Pilarski, 12-23 S. Pilecki, 10-1, 12-149 J. Pininska, 18-8 Rosa Piotrkowski 17-61, 26-142, 26-247, 28-120

23

A. Plevin 1-35 Jan Płowiec 26-317 T. Pocheco 14-85 Stefan Poliszko 10-107 M. D. Pollard, 8-S4 Adrian A. Pollock 1-237. 1-303, 9-140, 10-122, 15-S107, 15-S108, 25-231 Elizabeth L. Porter 17-121 O. Prabhakar 4-S207 Arun Prakash 9-142, 8-S217 A. Prateepasen, 18-196 C. W. Pretzel 5-172 D. Prevorovsky 20-285 Zdenek Prevorovsky 17-S57, 18-279, 20-134, 20-274, 20-285, 22-138, 28-099 David W. Prine 4-S304, 15-S106 Thomas M. Proctor, Jr. 1-173, 5-134, 7-41, 10-43 William H. Prosser 9-283, 14-S1, 15-S106, 15-S115, 17-29, 17-37, 23-037 A. Proust 17-S83, 18-161, 19-229, 20-229 E. Proverbio 28-129 Jose Pujol 17-111 R. Pullin 22-166, 26-172 M. I. López Pumarega 17-61 K. K. Purushothaman, 12-111 Gang Qi 17-111, 19-275, 24-001 Jie Qian 20-016, 20-099 Stephen L. Quarles 8-134, 9-17, 9-189 M. Raab 20-274

Amani Raad 20-300 Jan Raczkowski 10-107 Andre Raharinaivo 2-159 Baldev Raj, 4-S102, 6-209, 7-S1, 8-S126, 8-S140, 8-S149, 11-43 P. Raj 4-S98 D. S. Rajan 8-S297 P. K. Rajan 8-S28 S. Ramadan, 27-254 Jerzy Ranachowski 4-S82, 5-25 A.K. Rao 4-S147, 6-19 M. V. M. S. Rao 7-S29, 8-S262 S. P. Mallikarjun Rao 7-135, 11-101 M. N. Raghavendra Rao 8-S284 Christian Rasche 26-120 Franz Rauscher 17-S92, 18-118, 20-188, 22-049, 24-022, 26-098, 28-155 Henrique L. M. dos Reis, 4-S232, 5-67, 5-144, 9-197, 11-107, 12-15, 17-83 F. Rehsteiner 14-119 Wilfried Reimche, 25-331 H. W. Reinhardt 14-S74, 22-083 R.L. Reuben, 18-96, 18-211, 25-348

24

W. G. Reuter 4-S269 L.E. Rewerts 15-S107 Marie-Christine Reymond 2-159, 4-S296, 5-S63 M. W. Richey 6-257 J. Richter 17-S70 Ronald Richter 28-188 B. Richtor 2-29 G. L. Rigby 10-S22 Carlos R. Rios 23-025 L. Rippert, 18-41 Steffen Ritter 14-S47 J. L. Robert 4-S300, 6-43, 11-11 R.A. Roberts 15-S107 N. Rochat 9-91 J. Rödel 17-S78 John M. Rodgers 1-114, 4-S155, 4-1, 25-286 P. Rödhammer 20-257 P. Rodriguez 4-S102, 8-S140, 8-S149, 11-43 L. M. Rogers 2-319 J. Roget 4-85, 5-S60, 5-S66, 8-S231, 8-34 K. Rokugo 19-134 I. Roman 2-64, 3-19, 3-41, 3-130, 4-S106, 4-S111, 6-73, 8- S109, 8-47, 10-31 Jordi Romeu 28-163 Igor Maria De Rosa, 25-080 A. P. G. Rose 1-213 Joseph L. Rose, 12-23 Z. Rosecky 20-025 Sabine Rosner 22-110, 25-149, 26-014, 28-076 J. N. Rossettos, 12-1 Мikhail Rostovtsev 26-311 R. Rothea 19-229 D. Rouby 3-176, 9-117 C. Rowland, 18-87, 18-239 C. Roy 8-S292, 12-45 P. R. Roy 7-S40, 4-S251 Gottfried A. Rubin 10-S18, 11-19 D. Rubio 10-13 Thomas P. Runarsson, 25-252, 25-260, 26-229, 26-262 J. E. Ruzzante 10-13, 17-61 H. Saadaoui, 12-45 Wolfgang Sachse 4-S62, 8-S20, 8-S170, 11-79 S. Sagat 5-S16 Kouki Saiga 26-109 Masahiro Saito, 13-S68 Naoya Saito 15-19, 16-S289 Koji Sakai 10-S104 Norio Sakaino 15-S50

25

K. Sakamaki 16-S134, 18-68 Kiyoshi Sakamaki 21-206, 21-223

N. Sakata 16-S75 Yasunori Sakata 15-S70 Tatsuro Sakimoto 7-167 Takumi Sakuma 16-S196 C. Sala 2-11 H. M. Sallam 14-85 Pekka Salmenperä 21-230 Peter Sammonds, 13-S21 A. Sampath 5-S12 R. Samuel 7-S35 R. N. Sands 4-31 A. S. Sankaranarayanan, 12-111 N. Saniei 15-33 Mary Sansalone 5-S24 Carlo Santulli 26-220 Fabrizio Sarasini, 25-080 Hiroaki Sasaki 16-S25 Soji Sasaki 2-1 S. Sase 17-15 L. Hanumantha Sastry 11-101 Akihiro Sato, 19-202 Ichiya Sato 2-1, 7-173, 8-S213 Isamu Sato 8-35 Kazuhiko Sato, 13-S54, 13-S75, 19-109, 23-119 Kazuhisa Sato 24-215 K. Sato 16-S45 Keiichi Sato, 4-S240, 8-S213, 9-209, 13-S42 Kouichi Sato 7-173 Takashi Satoh 23-102 Markus G. R. Sause 26-001, 28-142 S. G. Savanur 7-S18 M.S. Saykova 28-234 M.W. Scaife, 18-211 C. M. Scala 2-261, 6-249, 10-49 C. R. Scales, 25-051 Ulrich Scanz 27-167 G. Schauritsch, 18-138 Christian Scheer, 25-331 R. Schelling 20-238 C. Schepacz 2-267 Jerzy Schmidt 23-173, 24-222, 27-272 Jonathan J. Scholey, 27-291 H.J. Schoorlemmer, 18-180, 20-238 Frank Schubert 22-147 M. Schulz 20-172 Daniel Schultheiß

26-001

K. Schumacher 9-103

26

Thomas Schumacher, 25-316 D. Schumann 3-172 H.-J. Schwalbe 22-236 J. S. Schwartzberg 10-97 Ian G. Scott 4-S142 C. B. Scruby 3-182, 4-9, 7-81, 7-211, 8-S158, 8-S201, 9-243, 9-255 Petr Sedlak 26-182 C. Seguf 5-S49 Kazuyoshi Sekine 23-233, 27-281 G. P. Sendeckyj 11-33 Hirofumi Sentoku 16-S19 U. Senturk 15-S117 M. Seto 11-S19 Masahiro Seto, 11-S27 B. K. Shah 7-S13 S. P. Shah 10-S97 R. Douglas Sharp 3-118, 4-S259, 4-115 V.V. Shemyakin, 19-172 G. Shen, 8-S97 Gongtian Shen 16-S243 H.W. Shen 15-S105, 18-189 Ping Shen 15-S123 Nanling Shi 16-S317 K. Shibata 8-S145 M. Shibata 3-144, 3-190, 4-93, Mitsuhiro Shigeishi, 11-S57, 13-S14, 15-S50, 15-S60, 16-S65 Hisatoshi Shimada 10-S104 Hiroyuki Shimizu, 27-194 Shigeo Shimizu, 19-196 Tatsuya Shimizu, 13-S68 Takayuki Shimoda 20-194 J. Sikula 17-S100 T. Simo 17-S65 Masayuki Shimojo 16-S196, 19-085 Noboru Shinke 10-25, 16-S134 , 23-164, 24-052 Takuo Shinomiya 20-145 Tomoki Shiotani 15-S50, 16-S95, 18-248, 19-118, 19-142, 20-145, 20-153. 21-166, 22-039, 23-260, 24-205, 25-069, 25-308, 27-186, 27-263, 28-032 M. Shiwa 4-S195, 21-197, Donald A. Shockey 21-001 Andrey Shvedov 27-212

Ménad Sidahmed 20-300 Khalid J. Siddiqui 3-118, 9-9 J. Siedlaczek 10-1 Joanna Sikorska 22-264, 22-274 Josef Sikula 22-091, 26-182, 27-157 F. Simacek 10-67 A. K. Singh 4-S174

27

A. K. Sinha 7-S13 N. K. Sinha 4-S290 Petr Sittner 20-163 Olivier Skawinski 26-120 C. Sklarczyk 8-S93 A. Skraber, 18-144 Jerzy Skubis 2-261, 4-S82, 5-25 A. Slimani 5-S42 Daniel R. Smith Jr. 4-S135, 7-9 J. H. Smith 8-S79 M. Sobczyk 8-S192 T. Sogabe 21-120 V.L. Sokolov, 18-111, 20-218, 26-279, 28-234 Amir Soltani 4-S17 Nobukazu Soma 23-129 Jun-Hee Song 16-S309 M. Sorel 2-261 M. Ben Souda 11-11 Dimitra Soulioti 28-032 P. Souquet 4-85, 5-S60, 8-S231 J.C. Spanner, Sr. 6-121 L. M. Spasova, 25-051 Th. Spies 15-S119, 19-100, 19-153, 24-179 X. Spiliotis 28-246 Nicholas S. Spivey, 25-187 Wojciech Spychalski, 25-166, 26-317 M. K. Sridhar 4-S174 K. V. Srincivasan, 8-S8 T.S. Sriranga 7-S35 K. A. Stacey 7-81 J.A. Steel, 18-96, 18-211, 25-348 Wolfgang Stengel 4-S312 D. Stöver 4-S78 H. Strauven, 13-79 V. Streicher 8-S53 V. A. Strelchenko 8-S334 V. A. Strizhalo 8-S334 S.A. Strizkov 19-172 C. E. Stuart, 12-S12 George Studor 23-037 V. Suba 17-S20 S. V. Subba Rao 8-S8 Iyer Subramaniam 4-S174 H. N. Sudheendra 8-S288 Tetsuya Suemune 27-137 C. Y. Suen 9-9 Katsuhiro Sugawara, 13-S54 Sota Sugimoto, 27-089 Jamil Suleman, 27-040

28

J. Summerscales 4-S170 C. T. Sun 7-21 X. Sun 8-S262 J. Q. Sun 8-S114 M. J. Sundaresan 8-S277 M. Surgeon 15-S105, 18-34 Hiroaki Suzuki 11-117, 14-35, 14-69, 15-19, 15-S108, 16-S178, 16-S289, 24-012 Ippei Suzuki 7-179 M. Suzuki 9-277 Masahiko Suzuki 5-61 Tetsuya Suzuki 22-030, 23-272 Toshio Suzuki, 13-S89 Toshitaka Suzuki 2-1 P. Svadbik, 18-29 V. Svoboda 17-S2, 17-S83 Terry L. Swanson 8-1, 8-S236 Grzegorz Swit 24-187 Tatsuo Tabaru, 25-107 Toyokazu Tada 27-233 A.N. Tafuri 15-S104 V.L. Tahiri 16-S299 F. Taioli, 8-S42 H. Takagi 16-S142 Hideaki Takahashi, 6-261, 7-1, 13-S68 K. Takahashi 16-S324 Shin Takahashi 23-091 Katsutoshi Takano 24-111 Atsushi Takashima, 19-109 Kazuki Takashima, 12-S18, 13-S08, 16-S150, 19-085 S. Takashina, 18-102 Nobuo Takeda, 12-127 Mikio Takemoto 7-185, 11-117, 14-35, 14-69, 15-19, 15-S108, 16-S115, 16-S178, 16-S289, 18-51, 18-102, 18-286, 19-011, 19-035, 19-063, 19-075, 21-120, 21-131, 22-119, 21-120, 21-131, 22-119, 22-224, 23-072, 23-156, 23-196, 23-181, 23-215, 23-277, 24-012, 24-067, 24-084, 24-127, 24-161, 25-124, 25-157, 25-172, 25-179, 25-267 27-241 Hajime Takeuchi 9-209 K. Takigawa 8-S213 Masanori Takuma 16-S134, 23-164, 23-206, 24-052 S. Talebi 8-S254 Okiharu Tamura 16-S178 H. Tanaka 21-166 M. Tanaka 10-S55 T. Tanaka 8-S213 Toshiyuki Tanaka 7-173

29

S. Tanary, 8-S314, 12-39 N. Tandon 9-25, 17-23 Daiki Tani 24-153 Yoshihiro Taniyama, 25-157 Thomas Tannert 28-059 O.B. Tarutin 15-S120 Masayuki Tateno, 11-S75 A. Taylor, 18-239 M. A. Taylor 8-S322 C. M. Teller 11-27 Hans Maria Tensi 22-S01 H.B. Teoh 3-19, 3-130, 6-1 Satoshi Teramura 6-261, 7-1 Giovanni P. Terrasi 26-152 Christian Tessier 26-032, 27-254 R. Teti 3-158, 5-156 Lawrence W. Teufel 15-S124 A. Tewfik 27-176 Kazuhiko Tezuka 23-129 N. A. Thakkar, 25-348 Christian Thaulow 4-S211 Heinrich Theiretzbacher 4-S157 W. Thelen 14-115 Pete Theobald 26-091 Richard E. Thill 10-S77 P.M. Thompson 6-93 D. D. Thornton 23-331 Peter G. Thwaite 14-vi (3/4) R. M. Tian 4-S94 B. Tirbonod 8-S84 Anil Tiwari 14-53 G. P. Tiwari 4-S102, 7-S43 R.G. Tobin 8-25 Akira Todoroki 26-109, 27-089 S. Tomecka-Suchon , 12-S7 Yuichi Tomoda 15-S31, 21-157, 23-272, 25-021 H. Tonda, 13-S08 F. Tonolini 8-S62 G. Tonolini 4-S86 V. Torra 5-S49 T. Toutountzakis, 25-033 D.T. Tran 8-25 H. Traxler 20-257 D. Tsamtsakis, 13-79 P. Tscheliesnig 4-S157, 17-S108, 18-138, 18-167, 20-129, 20-179, 22-201, 25-276 Ming-Kai Tse 4-S127, 8-S188, 8-S209 A. Tsimogiannis, 18-21, 18-224, 22-059 Yukiya Tsuchida 21-176

30

N. Tsui 21-213 Nobuyuki Tsuji 15-S60 Yusuke Tsukahara, 13-23 Koichi Tsukiyama 6-261, 7-1 F. R. Tuler 8-S79 F. Uchida, 18-102, 19-075 Junichi Uchida 23-181 Atsushi Uchiyama, 13-S42 Farid A. K. M. Uddin 23-136 Shuichi Ueno 27-241 Toshiyuki Uenoya, 13-S95, 15-S112 Takateru Umeda 20-248

Runar Unnthorsson, 25-252, 25-260, 26-229, 26-262 Taketo Uomoto 6-137 T. I. Urbancic 8-S254 E. Uria, 13-79 S.J. Vahaviolos 3-i(1), 4-S329, 15-S109, 18-189, 18-248, 18-272, 19-172 S. Vajpayee 5-S12 Carlos M. Valdes-Gonzalez, 12-117 Till Vallée 28-059 H. Vallen, 18-167, 18-258, 18-265, 22-102, 25-132 J. Vallen, 18-265, 25-132 P.J. Van De Loo, 18-174, 20-238 D. R. V. Van Delft 20-229 Koen Van Den Abeele 22-253

Gert Van Dijck, 22-253 S. Van Huffel, 18-41 J. G. M. Van Mier 20-153

R. Van Nieuwenhove, 18-293 Donald W. Vannoy 4-S307 D. Varchon 20-285 Alex Vary, 8-S175, 12-i (1/2), 12-71, 12-79, 14-53 F. A. Veer 8-S118 Vasisht Venkatesh 14-61 Vincenzo Venturi, 25-324 E. Verbrugghe 11-1 I. Verpoest 4-S186, 8-S272, 26-240 A. Vervoort 20-292 R. Vijayaraghavan, 7-S13 G. Villa 4-S86 M. R. Viner 8-S192 A. Vinogradov 17-1, 27-144, 27-212 A. Yu. Vinogradov 16-S158 P. Vionis 18-217, 20-229 R. Visweswaren 4-S207 Frantisek Vlasic 28-170 L.E. Vlasov, 18-150

31

Aaron C. Voegele 17-83 J. von Stebut, 18-258 Y. Vougiouklakis 20-265 Toshiya Wada 23-150 Adrian P. Wade 10-71 Haydn N.G. Wadley 5-S69 N. Wakabayashi 10-S42 Shuichi Wakayama, 4-S278, 12-S24, 16-S170, 21-149, 23-150, 24-173, 24-228, 26-160, 27-137 Teruyuki Waki, 12-S1 Akimasa Waku, 12-S1 D. A. Waldrop 7-31 James L. Walker II, 25-187 Y. Wan, 8-S97 Y. Wang, 19-022 Alexander Wanner 14-i, 14-v, 14-vi (3/4), 14-S47 Kris A. Warmann 11-107 E. Waschkies, 8-S93 G. Washer 15-S104 M. Watad 3-41 Hiroshi Watanabe 20-001 Naoaki Watanabe, 13-S42 T. Watanabe 3-59 Takashi Watanabe 2-1 Yoshinori Watanabe 23-119 J.R. Watson, 18-232 Robert J. Watters 4-S17, 8-S258 D.J. Watts 15-S104, 19-172 Richard L. Weaver 4-S54, 5-S40 Z. Weber, 18-29 J. R. Webster 4-S132, 8-S197 Qiang Wei 11-S75 B. Weiler 14-S74 M. Wevers, 4-S186, 8-S272, 13-79, 15-S105, 18-34, 18-41, 20-206, 20-292, 22-253, 26-240 R. G. White 1-263 J. W. Whittaker 1-147, 4-S247, 5-148, 6-257, 7-31, 7-95, 8-S280, 9-75, 9-84, 10-113 Paul D. Wilcox, 27-291 Steve Wilcox 28-085 B. J. S. Wilkins 10-S22 H. Willer 10-67 A. J. Willis 1-244 P. E. Wilson 2-191 Leo Windecker Icae Banquet M. Winkelmans 20-206, 22-253 Michael R. Wisnom, 27-291 E. Winter 10-67

32

S. M. Wolf 3-239 A. Wolfert 20-238 Jörg Wolters 3-51 Brian R. A. Wood 6-125, 6-239, 8-S14, 8-S66, 10-S29, 10-S59, 15-S113, 17-121 J. D. Wood 8-S318 D. G. M. Wood 4-71 G. Wormser 5-S52 Amelia P. Wright, 25-115 Bobby Wright 4-S166 Min Wu 11-5 Wei Wu, 25-115 Doone Wyborn 23-129, 27-167 Y. Xiang, 8-S246 J. Z. Xiao 4-S215 Yue-Huang Xu 3-81 H. Yamada, 18-51 K. Yamada 6-13 M. Yamada 21-213 Minoru Yamada 23-233, 23-243 Yoshiaki Yamade, 12-127 Hiroshi Yamaguchi 24-111 Katsuya Yamaguchi 9-209 Kusuo Yamaguchi 4-S191, 4-S286, 4-S325, 8-S1 T. Yamaguchi 8-S145 Koji Yamamoto 19-196 Hirofumi Yamasaki 24-111 Akihiko Yamashita, 13-S89 Akio Yamashita 4-S286 T. Yamauchi 11-21 Yasunori Yamazaki 24-097 Tinghu Yan 17-49 W. Yan 15-S109 M. Yanagibashi 8-S213 Masa-aki Yanaka, 13-23 Takahito Yanase 23-096 J. M. Yang 8-S268 Xuanhui Yang 15-S123 C. K. Yao 4-S94, 8-S105 Takeshi Yasuda 24-153, 26-270 A. Yasuo 4-S294 Daisuke Yasuoka 15-S60 J.J. Yezzi, Jr. 15-S104 Akira Yoneda 24-097 Takao Yoneyama 2-1, 7-173, 8-S213 Akio Yonezu 23-156, 23-196, 23-277, 24-127, 24-161 Dong-Jin Yoon, 9-237, 13-S83, 15-S11

33

Kenichi Yoshida 16-S142, 18-68, 19-022, 21-206, 21-223, 22-159, 23-189, 24-153, 26-270 Toshikatsu Yoshiara 19-202 Sumio Yoshikawa 8-113 Tatsuhiko Yoshimura 6-109, 6-145 Hiroshi Yoshino, 12-S1 H. Yoshioka 10-S63 Takeo Yoshioka, 19-196 R.D. Young 6-93 R. P. Young 8-S38, 8-S166, 8-S254 R. Paul Young 5-S29, 5-S34, 25-294 Y. Youssef, 12-39 Qing Huan Yu 8-41 Hiroo Yugami 24-215 Syuro Yuji, 7-167 Hironobu Yuki 10-35 Kunihiro Yuno 11-S89 Shigenori Yuyama, 2-19, 2-71, 4-S38, 13-S14, 15-S107, 16-S75, 18-248, 19-118, 19-184, 21-187, 21-213, 23-233 J. Zaloudek 17-S65 F. Zeides 8-S109, 8-47, 10-31 Bajram Zeqiri 26-091 Kornelija Zgonc, 11-79 B. Q. Zhang 4-S94, 8-S49, 8-S105, 8-S114 Fan Zhang 18-144, 20-300 Zhizhen Zheng 15-S123 X. Q. Zhu 4-S215 Z. Zhu 8-S135 Zu-Ming Zhu 6-115 Zuming Zhu 16-S317 B. Ziegler 22-236

J. Zietek, 12-S7 Steve Ziola 8-51, 14-S12 B. Zogala, 18-15 Daihua Zou 5-S1 Ryszard Zuchowski 2-272 J. Zuidema 8-S118