research article damage detection of composite plates by ...lamb wave based structural health...

Research ArticleDamage Detection of Composite Plates by LambWave Ultrasonic Tomography with a Sparse HexagonalNetwork Using Damage Progression Trends

C J Keulen1 M Yildiz2 and A Suleman1

1 Department of Mechanical Engineering University of Victoria Victoria BC Canada V8W 3P62 Faculty of Engineering and Natural Sciences Sabanci University Orhanli-Tuzla 34956 Istanbul Turkey

Correspondence should be addressed to A Suleman sulemanuvicca

Received 24 July 2013 Accepted 15 November 2013 Published 4 March 2014

Academic Editor Gyuhae Park

Copyright copy 2014 C J Keulen et al This is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

Lamb wave based structural health monitoring shows a lot of potential for damage detection of composite structures Howevercurrently there is no agreement upon optimal network arrangement or detection algorithm The objective of this research is todevelop a sparse network that can be expanded to detect damage over a large area To achieve this a novel technique based ondamage progression history has been developed This technique gives an amplification factor to data along actuator-sensor pathsthat show a steady reduction in transmitted power as induced damage progresses and is implemented with the reconstructionalgorithm for probabilistic inspection of damage (RAPID) technique Two damage metrics are used with the algorithm and acomparison is made to the more commonly used signal difference coefficient (SDC) metric Best case results show that damage isdetected within 12mm The algorithm is also run on a more sparse network with no damage detection therefore indicating thatthe selected arrangement is the most sparse arrangement with this configuration

1 Introduction

To achieve lighter aerospace structures damage is allowedto exist during operation as long as it is within safe pre-determined specifications aircraft structures are designedaccording to a damage tolerant philosophy In more recentyears composite materials are being used to build aerospacestructures because they are lightweight and stiff and haveexcellent fatigue and corrosion resistance The downsideto composites however lies in their damage mechanismsCompositesmay fail or become damaged in a number of waysthat are very different from traditional metallic materialsDefects may arise during manufacture due to voidsporosityply misalignment or inclusion of foreign objects that showno evidence to the naked eye Composites suffer from lowvelocity impacts that can damage the internal structure of alaminate while leaving no visible evidence on the surface

Maintenance and inspection of aircraft is of utmostimportance for safe and efficient operation Aircraft struc-tures operate in harsh conditions sustaining high loads

fatigue cycles and extreme temperature differentials Failureof these structures is not acceptable due to the possibility ofloss of life and assets To ensure aircraft structures are in safeoperational condition costly inspection involving aircraftdowntime and often disassembly of major components isroutinely performed The cost of inspection is about 30of the total cost of acquiring and operating compositestructures [1] Currently damage detection is performedwith techniques referred to as nondestructive testing (NDT)or nondestructive inspection (NDI) to locate and quantifydamage

To reduce operational cost and improve reliability andperformance a common goal of researchers designers andmanufacturers is to develop a real time inspection system thatis permanently installed or embedded within the structureThis technique is commonly referred to as structural healthmonitoring (SMH) [2 3] Such systems typically consist ofa network of transducers that are used to sense physicalparameters that indicate the presence of damage with aninterrogation technique or algorithm

Hindawi Publishing CorporationShock and VibrationVolume 2014 Article ID 949671 8 pageshttpdxdoiorg1011552014949671

2 Shock and Vibration

In 1917 Lamb published his classic analysis and descrip-tion of acoustic waves which included the first consider-ation of Lamb waves [4] In 1961 Worlton [5] proposedthe use of Lamb waves for damage detection and a newNDE potential emerged In 1962 Frederick and Worlont [6]conducted the first experimental study In the beginning ofthe 1990s Hutchins et al [7ndash9] and Nagata et al [10] appliedLamb waves to NDE using medical imaging and seismictomography techniques to bothmetallic and compositemate-rials Their systems utilized pairs of transducers that werepositioned across the material at various locations in orderto obtain a dense collection of pitch-catch signals in orderto reconstruct a tomogram While the techniques producedreasonable results themethods were time consuming requir-ing a lot of repositioning of the transducers and in somecases requiring the specimen to be placed in a water bathPrasad et al [11] implemented an SHM system for compositesusing surface bonded piezoelectric transducers Since thetransducers were permanently bonded in place the systemcould be operated in real time without any repositioningGao et al [12] introduced the reconstruction algorithm forprobabilistic inspection of damage (RAPID) Michaels [13]later investigated the application of tomography algorithmsto sparse networks

Other research efforts were made to use the Lamb wavemode propagation characteristics in order to detect damagespecifically the attenuation and arrival time of the first twowave modes In 1993 Guo and Cawley [14] studied theinteraction of Lamb waves with delaminations in compositematerials both numerically and experimentally Keilers andChang [15] later proposed a built-in damage detection systemusing an array of piezoelectric transducers Giurgiutiu et al[16] discussed the pulse-echo analysis technique and sub-tracting baseline data from damaged data to detect damage

This paper presents the details and results of a study on theimplementation of a sparse piezoelectric transducer networkfor damage detection in composite materials The objectiveof this research is to develop a sparse network that can beexpanded to detect damage over a large area To achieve thisa novel technique based on damage progression history hasbeen developed

2 Theory and Implementation

Lamb waves are elastic guided waves that propagate par-allel to the surface in thin structures with free boundariesPlates are the best example however Lamb waves canalso propagate in structures with a shallow curvature Themost advantageous characteristics of these waves are theirsusceptibility to interferences caused by damages or bound-aries (the features of interest) and low amplitude loss Toimplement a Lamb wave based damage detection techniquesome important properties must be determinedWhen Lambwaves propagate they travel in one of two possible ways withrespect to the platersquos mid plane If the motion is symmetricabout the midplane (the peaks and troughs of the waves arein phase) then it is a symmetric mode and if the motionis not symmetric (the peaks and troughs are 180∘ out of

phase) it is an antisymmetric mode An infinite number ofmodes exist each mode is referred to as an 119860 mode or119878 mode if it is antisymmetric or symmetric respectivelywith a subscript indicating its order For example the lowestorderfrequency symmetric mode is referred to as an 119878

0

mode while the second lowest orderfrequency symmetricmode is referred to as an 119860

1mode Each mode exists at

a different frequency depending on the properties of thematerial At lower frequency-thickness values less modesexist It is advantageous to operate in a frequency-thicknessrange where only the 119878

0and119860

0modes existThis is generally

below 15MHzmm

21 Implementation To employ Lamb waves for damagedetection a system to send and receive Lamb waves mustbe developed and a number of parameters must be selectedto tune the system such as the transducer type size andarrangement actuation signal and data acquisition

Piezoelectric transducers are commonly used in SHMsystems They are inherently simple devices consisting ofsimply a piezoelectric material with two conductive surfacesWhen a voltage is applied across the surfaces the materialexpands or contracts (depending on polarity) proportionalto the magnitude of voltage applied Conversely when thematerial is deformed a voltage difference is seen between thesurfaces This property leads to the greatest benefit of piezo-electric transducers their ability to both send and receivesignals Piezoelectric transducers are small and light and canbe bonded to or embedded within a structure with littleeffect Signals are voltage based and therefore easy to generateand acquire with common hardware Lead zirconate titanate(PZT) is the most commonly used material in piezoelectrictransducers They offer excellent performance for both gen-eration and acquisition have excellent mechanical strengthwide frequency responses and low power consumption andcan be obtained at a low cost [17]

When actuating Lamb waves it is desirable to actuate theleast number of modes possible (preferably only the 119878

0and

1198600) so that signal interpretation is simplified As mentioned

earlier this usually occurs below 15MHzmm This range isalso beneficial because there is low dispersion which meansthat if the frequency changes as the wave propagates throughthe material its velocity will remain relatively constanttherefore simplifying signal interpretation It is desirableto send out a single pulse so that the propagation of thewave groups can be analyzed The challenge then lies inproducing an instantaneous pulse that can be controlled andactuated at a desired frequency There is no control over afrequency generated by a simple impulse therefore a shortburst must be emitted If a simple sine wave composed ofa few cycles is emitted then the desired frequency can beactuated however the frequency domain of such a signalshows small secondary peaks in the frequency domain thatare present at other frequencies To eliminate these peaks thesignal can be modulated with a window function to slowlyincrease and decrease the magnitude of the signal [18] Acommonly used signal that provides a good compromisebetween number of cycles and ramp up rate consists of five

Shock and Vibration 3

F

E

D

C B A

L

G H I

J

K

(a)

C

MNO

P

Q

R

S T

D

E

F

G H I

J

K

L

L

B A

(b)

Figure 1 (a) Hexagonal network showing actuator-sensor paths from transducer A (left) and (b) expansion of a single unit cell (right)

sine peaks modulated by a Hann window A signal with anodd number of peaks is used so that there is a clearmaximumpeak that can be used for signal processing (if an even numberof peaks were used then there would be two peaks with thesame maximum amplitude)

22 Proposed Network Technique The technique developedin this research aims to provide a practical modular networkthat is sparse and can be expanded to cover a large area Gen-erally tomography requires a dense network of transducersthat cannot be easily expanded to cover a larger inspectionarea The proposed technique makes use of a hexagonalarrangement which ismodular in the sense that it uses a ldquounitcellrdquo that can be repeated to expand the network to cover alarge inspection area The unit cell consists of 12 transducersin a hexagonal arrangement as shown in Figure 1(a) Thenetwork can be expanded by simply increasing the numberof unit cells as shown in Figure 1(b) A further benefit is thattwo unit cells can share three transducers which means thatanother unit cell only requires nine new transducers

Each transducer can act as both an actuator and a sensorInspection begins by actuating one transducer to send Lambwaves through thematerial and recording the signals with theremaining transducers This is then repeated 11 times suchthat each transducer acts as an actuator once Since there isone actuator and 11 sensors there are 11 actuator-sensor pairsper transducer with direct paths between them as shown inFigure 1(a)

23 Damage Location Algorithm Once data is collected inundamaged and various damaged states a technique mustbe implemented to locate the damage Various techniqueshave been developed each relies on a difference betweenthe damaged and undamaged state Such techniques includedelay-and-sum beam-forming [19] the time-difference-of-arrival method [20] the energy arrival method [21] andthe filtered back-projection method [22] To implement theproposed sparse hex network an algorithm was developedthat incorporates damage progression information This

algorithm can be incorporated with existing algorithms toincrease their accuracy

The algorithm selected for this research was the recon-struction algorithm for probabilistic inspection of damage(RAPID) [12] It was developed for networks based on 8ndash16 transducers and has inherently good signal-to-mean-noise ratios [13] it can accept various input parameters andproduces reasonably accurate results [22] Changes in thetransmitted Lamb wave signal are related to a change inthe material properties (ie damage) between two sensorsThe probability of defect presence at a certain point can bereconstructed from the severity of the signal change and itsrelative position to the actuatorsensor pair [12] The RAPIDalgorithm is based on two assumptions (i) all effects fromevery possible actuatorsensor pair can be expressed as alinear summation across the entire inspection region and (ii)information from a specific actuatorsensor pair contributesto the defect distribution estimation of a subregion is thevicinity of the path between the pair

Equation (1) describes the RAPID algorithm

119875 (119909 119910) =

119873

sum

119896=1

119901119896(119909 119910) =

119873

sum

119896=1

119860119896(120573 minus 119877

120573 minus 1) (1)

where 119875(119909 119910) is the probability of the existence of a defectat position (119909 119910) the Cartesian coordinate of a point inthe inspection area 119860

119896is the damage metric as described

below 120573 is a scaling factor that defines the subregion that theactuatorsensor pair has an effect on (essentially an ellipsewith the actuator and sensor at its foci) and 119877(119909 119910) isdescribed in (2) as

119877 (119909 119910 1199091119896 1199101119896 1199092119896 1199102119896)

=

radic(119909 minus 1199091119896)2

+ (119910 minus 1199101119896)2

+ radic(119909 minus 1199092119896)2

+ (119910 minus 1199102119896)2

radic(1199091119896

minus 1199092119896)2

+ (1199101119896

minus 1199102119896)2

(2)

where (1199091119896 1199101119896) is the Cartesian coordinate of the actuator

and (1199092119896 1199102119896) is the Cartesian coordinate of the sensor In this


(a) (b)

Figure 2 (a) Composite panel with hex network (left) and (b) hex network with induced damage (right)

work a value of 105 was selected for 120573 as it is the commonlyused value [22ndash24]

119860119896is a damage metric that is extracted from the Lamb

wave signals It can be based on various phenomena such asa reduction in transmitted power reduction in magnitude ofwaves or delay in arrival time More recently Moustafa andSalamone [25] used the fractal dimension with a modifiedbox-counting algorithm The fractal dimension is a metricused to compare two curves and is calculated as theHausdorffdimension The most commonly used metric is the signaldifference coefficient (SDC) used by various researchers [22ndash24] This metric will be used as a baseline to compare resultswith the proposed algorithmThe signal difference coefficientbetween two data sets is defined as

SDC = 1 minus1003816100381610038161003816120588119886119887

1003816100381610038161003816 (3)

where

120588119886119887

=1

119878

sum119878

119894=1

(119886119894minus 120583119886) (119887119894minus 120583119887)

radicsum119878

119894=1

(119886119894minus 120583119886)2

sum119878

119894=1

(119887119894minus 120583119887)2

(4)

and 119878 is the total number of samples 119886119894and 119887119894are the initial

data and damaged data respectively at sample 119894 and 120583119886and

120583119887are the arithmetic mean value of the initial data set and

damaged data set respectivelyMost damage detection systems rely solely on detecting

damage as it occurs that is by comparing the change in thedamaged state with the undamaged state In reality howeverdamage often begins as a small flaw that slowly increases insize while in service As the damage grows information canbe collected that can be used to locate the damage beforeit is large enough to be detected by algorithms without thisinformation In this work a novel technique is developedthat incorporates information from the damage progressioninto the RAPID algorithm to increase the effectiveness andenable its use in sparse networks where it may otherwise notbe applicable

To incorporate damage progression information themagnitude of power from the transmitted Lamb waves iscomparedwith that from the previous state (the previous holesize in the case of this study) If the power is less then thereis the possibility that either damage is progressing in that

path or external noise has caused a decrease in the signalA history of the progression is recorded and if the powerconsistently drops across a particular path the probabilityof damage in that location is multiplied by an amplificationfactor This is done in order to differentiate between noiseand damage progression Naturally the initial damage stateswill not accurately show a damage progression however asthe trend continues across more damage states the accuracyincreases and the results become more reliable In this studyan amplification factor of 110 (a 10 increase) was selected

3 Experimental Investigation

To investigate the potential of the proposed network experi-mentswere conducted on a composite panel with a single unitcell of 12 transducers arranged in a hexagonal pattern as inFigure 1(a) Reference data was collected while the materialwas in pristine condition and again with incrementallygreater induced damage in the form of a through hole Thedata was processed using the RAPID algorithm in order todetermine the location of the damage

31 Experimental Setup A 420mm times 420mm panel com-posed of eight plys of 139 gsm unidirectional T700 carbonfiber with West System 105206 epoxy was laminated in a[090plusmn45]

119878orientation to produce a 123mm thick quasi-

isotropic composite A jig was machined to locate and bond12 1mm thick and 756mm diameter PZT transducers tothe panel in a 75mm circumradius hexagonal array Thepanel and network are shown in Figure 2(a) while a closeupof the network showing the damage location is shownin Figure 2(b) Each transducer was assigned a letter forreference as seen in Figures 1 and 2 The coordinates of thetransducers are as follows 119860 (24750 27495) 119861 (2100027495) 119862 (17250 27495) 119863 (15375 24248) 119864 (1350021000) 119865 (15375 17753) 119866 (17250 14505) 119867 (2100014505) 119868 (24750 14505) 119869 (26625 17753) 119870 (2850021000) and 119871 (26625 24248) in mm from the lower leftcorner of the panel Damage was induced at (22875 24185)

A National Instruments NI PXI-5421 arbitrary waveformgenerator was used to generate a signal while a NI PXI-5105digitizeroscilloscope was used to acquire the signal Both


4 6 8 10 12 14

times10minus5

minus025

minus02

minus015

minus01

minus005

0

005

01

015

02

Volta

ge (V

)

Seconds (s)

(a)

935922591

times10minus5

005

01

Volta

ge (V

)

Seconds (s)

(b)

Figure 3 (a) Recieved Lamb wave signals at all seven damage states along path 119860-119866 (left) and (b) closeup of 1198600

wave (right)

units were installed in a NI PXI-1033 chassis and controlledwith a custom National Instruments LabView program Amodel A-303 amplifier from AA Labs Ltd was used toamplify the generated signal

32 Signal Generation and Acquisition An actuation signalconsisting of a sine wave with five peaks modulated by aHann window was employed An amplitude of plusmn8V wasproduced by the waveform generator and amplified by afactor of 20 before it was sent to the actuator resulting ina maximum amplitude of plusmn160V The actuation frequencywas selected based on a number of criteria In order to onlyactuate the 119878

0and119860

0modes the frequency-thickness product

was kept below 15MHzmm A frequency scan from 100 kHzto 400 kHz was performed to determine the frequency thattransmitted the greatest amplitude It was found that thefrequency range of 260ndash270 kHz transmitted the greatestamplitude therefore 265 kHz was used for the experiments

With these properties selected the waveform generatorwas programed to output the actuation signal at 100MHzData was acquired at 60MHz for 30 times 103 samples (500 120583s)An ASCII text file was written after each run and saved fordata processing

33 Experimental Procedure Once the experimental appara-tus was set up as described above a number of experimentswere performed under various damage conditions Eachexperiment consists of actuating one transducer and sensingthe other eleven then actuating the adjacent transducer andsensing with the remaining eleven This process is repeateduntil all 12 transducers have actuated the system once At thispoint the damage is increased and the process is repeated

Initially the experiment was conducted on the panel inpristine condition before damage was inflicted in the formof a hole drilled through the panel at a location such thatit intersected the paths between PZT pairs 119860-119866 and 119862-119870 asshown in Figure 2(b) The initial diameter of the hole was159mmwhichwas increased to 238mm 318mm 400mm476mm and finally 635mm

4 Discussion

Themain objective of this work is to develop amodular sparsenetwork To achieve this an algorithm that uses damageprogression information is introduced With this algorithmtwo damage metrics were implemented The first is basedon the SDC in order to compare the proposed algorithmto a conventional metric (SDC) The second damage metricis based on the transmitted power of the signal from anFFT analysis For both metrics a comparison between theresults with and without the damage progression algorithmis made to assess the effectiveness A third case was alsoconsidered that neglected information from every secondsensor essentially creating a more sparse network of sixtransducers however this did not produce any reasonableresults therefore implying that a minimum of 12 transducersare required in this situation

Figure 3(a) presents an example of the received 1198780and

1198600waveforms actuated by transducer 119860 and received by

transducer 119866 All damage states are plotted together on thesame chart with their peaks marked with an asterisk (lowast)Figure 3(b) shows a closeup of the 119860

0peak a consistent

decrease in amplitude due to damage is clearly seen Signalslike these from all paths are processed to determine if adecreasing trend exists

To benchmark the results the SDC was calculated foreach actuator-sensor path and implemented in the RAPIDalgorithm A contour plot indicating the probability of dam-age is presented in Figure 4(a) The damage location is indi-cated with the yellow cross and circle With this informationno damage could be detected This is reasonable accordingto Michaels [13] who reported that the technique was nothighly effective on large sparse arrays The aforementioneddamage progression information was also implemented withthe SDC metric The results are shown in Figure 4(b) Whilethe algorithm was not able to detect the exact location of thedamage it did locate a region close to the general area of thedamage therefore showing a marked increase in accuracy byusing damage progression information

In the second implementation the transmitted power wasused as the damage metric An FFT analysis was performed


260

240

220

200

180

160

280260240220200180160140

Position (mm)

Posit

ion

(mm

)

(a)

260

240

220

200

180

160

280260240220200180160140

Position (mm)

Posit

ion

(mm

)

(b)

Figure 4 (a) SDC results with no damage progression factor (left) and (b) SDC results with damage progression factor of 110 (right) lowastdamagelocation indicated by yellow cross

260

240

220

200

180

160

280260240220200180160140

Position (mm)

Posit

ion

(mm

)

(a)

260

240

220

200

180

160

280260240220200180160140

Position (mm)

Posit

ion

(mm

)

(b)

Figure 5 (a) Power amplitude results with no damage progression factor (left) and (b) power amplitude results with damage progressionfactor of 110 (right) lowastdamage location indicated by yellow cross

on data from every actuator-sensor path to extract themagnitude of transmitted power The RAPID algorithm wasimplemented and the results are shown in Figure 5(a) At thispoint the algorithm has located three possible damage loca-tions compared to zero locations with the SDC metric Thealgorithm was implemented with the damage progressioninformation as shown in Figure 5(b) With this informationthe algorithm has located a small region of damage roughlythe same size within a 12mm radius While these resultsare not highly accurate they do demonstrate that the use ofinformation from the damage progression does increase theprobability of damage detection and allows a large sparsenetwork to be used

Many similar research efforts focus on aluminum ratherthan composite making it difficult to compare results Theresults from the experiments performed here are an improve-ment over the work on an aluminum panel by Hay et al[22] as a larger spaced array is implemented with fewertransducer with results comparable to Liu et alrsquos [26] whowere able to detect a 30mm times 30mm delamination to within

102mm using an array of four transducer pairs arrangedin a square with sides of 225mm in length In general thistechnique allows for a larger spaced arraywith a lower densityof transducers

In real world structures the effects of temperature fluc-tuation and material anisotropy would have an effect onthe signals used in this algorithm and could possibly createinaccurate results In the case of anisotropy this wouldbe known a priori as the laminate schedule and materialproperties of the structure are known It could be takeninto account by normalizing the signals to compare allsensor-actuator pairs if needed Since this technique relieson signals that travel in a straight line between sensor-actuator pair that are compared to the same signal from aprevious damage state the effect of anisotropy would nothave an influence because it remains constant (aside fromthe effects of damage) Temperature however can changedrastically during service As the temperature of a carbonfiberepoxy laminate increases the modulus of elasticitydecreases and therefore the magnitude of the signal that it


transmits decreases To address this a temperature sensorcould be used to measure the temperature of the laminate inthe region beingmonitored andwith the effect of temperatureon the signal transmission known the effects of temperaturecould be accounted for in the algorithm

5 Conclusion

A novel sparse network that can be expanded to cover anysize area has been proposed along with a novel technique forincorporating information on the progression of damage toimprove the accuracy of existing algorithms

The network consists of a unit cell of 12 piezoelectrictransducers arranged in a hexagonal pattern as shown inFigure 1(a) The network is easily expanded by adding ninemore transducers to the existing 12 as shown in Figure 1(b)(two unit cells) The hexagonal arrangement allows forefficient stacking of unit cells that would not be otherwisepossible with a circular array

Damage progression informationwas implemented in theform of a 10 increase in signal magnitude if the actuator-sensor path showed a consistent decrease in power acrosseach damage state (ie if the power was consistently lessthan the previous state) The results presented in Figure 4show that incorporating the damage progression informationincreases the chance of detection with SDC as a damagemetric When the power of the transmitted wave is used asa damage metric some potential damage locations appear asshown in Figure 5(a) When the damage progression infor-mation is incorporated the accuracy is increased locatinga potential damage site that is within 12mm of the actualdamage as shown in Figure 5(b) While the results do notdirectly pinpoint the damage location they do show thatan improvement over the existing RAPID algorithm can bemade by incorporating damage progression informationwiththe use of a larger area sparse network on a carbon fibercomposite material

Conflict of Interests

The authors declare that there is no conflict of interestsregarding the publication of this paper

References

[1] Y Bar-Cohen A Mal S-S Lih and Z Chang ldquoCompositematerials stiff- ness determination and defects characterizationusing enhanced leaky Lamb wave dispersion data acquisitionmethodrdquo in Proceedings of the SPIE Annual InternationalSymposium on NDE of Aging Aircraft Airports and AerospaceHardware 1999

[2] K Worden C R Farrar G Manson and G Park ldquoThe funda-mental axioms of structural health monitoringrdquo Proceedings ofthe Royal Society A vol 463 no 2082 pp 1639ndash1664 2007

[3] K Worden ldquoAn introduction to structural health monitoringrdquoThe Philosophical Transactions of the Royal Society A vol 365no 1851 pp 303ndash315 2007

[4] H Lamb ldquoOn waves in an elastic platerdquo Proceedings of the RoyalSociety A vol 93 pp 114ndash128 1917

[5] D Worlton ldquoExperimental confirmation of lamb waves atmegacycle frequenciesrdquo Journal of Applied Physics vol 32 no6 pp 967ndash971 1961

[6] C Frederick and D Worlont ldquoUltrasonic thickness measure-ments with Lamb wavesrdquo Journal of Nondestructive Testing vol20 pp 51ndash55 1962

[7] D Jansen and D Hutchins ldquoLamb wave tomographyrdquo inProceedings of the IEEE 1990 Ultrasonics Symposium pp 1017ndash1020 December 1990

[8] D Hutchins D Jansen and C Edwards ldquoLamb-wave tomogra-phy using non-contact transductionrdquo Ultrasonics vol 31 no 2pp 97ndash103 1993

[9] D Jansen D Hutchins and J Mottram ldquoLamb wave tomog-raphy of advanced composite laminates containing damagerdquoUltrasonics vol 32 no 2 pp 83ndash90 1994

[10] Y Nagata J Huang J Achenback and S Krishnaswamy ldquoLambwave tomography using laser-based ultrasonicsrdquo in Review ofProgress in Quantitative Nondestrcutive Evaluation vol 14 pp561ndash568 1995

[11] S Prasad K Balasubramaniam and C Krishnamurthy ldquoStruc-tural health monitoring of composite structures using Lambwave tomographyrdquo Smart Materials and Structures vol 13 no5 pp N77ndashN79 2004

[12] H Gao Y Shi and J L Rose ldquoGuided wave tomography onan aircraft wing with leave in place sensorsrdquo in Proceedings ofthe Review of Progress in Quantitative Nondestructive EvaluationConference (AIP rsquo05) pp 1788ndash1794 2005

[13] J Michaels ldquoEffectiveness of in situ damage localization meth-ods using sparse ultrasonic sensor arraysrdquo in Proceedings of theSPIE March 2008

[14] N Guo and P Cawley ldquoThe interaction of Lamb waves withdelaminations in composite laminatesrdquo Journal of the AcousticalSociety of America vol 94 no 4 pp 2240ndash2246 1993

[15] C H Keilers Jr and F K Chang ldquoIdentifying delamina-tion in composite beams using built-in piezoelectrics partI-experiments and analysisrdquo Journal of Intelligent MaterialSystems and Structures vol 6 no 5 pp 649ndash663 1995

[16] V Giurgiutiu A Zagrai and J J Bao ldquoPiezoelectric waferembedded active sensors for aging aircraft structural healthmonitoringrdquo Structural Health Monitoring vol 1 no 1 pp 41ndash61 2002

[17] Z Su L Ye and Y Lu ldquoGuided Lamb waves for identificationof damage in composite structures a reviewrdquo Journal of Soundand Vibration vol 295 no 3ndash5 pp 753ndash780 2006

[18] B Rocha C Silva and A Suleman ldquoStructural health mon-itoring system using piezoelectric networks with tuned lambwavesrdquo Shock and Vibration vol 17 no 4-5 pp 677ndash695 2010

[19] CWang J Rose and F Chang ldquoA synthetic time-reversal imag-ing method for structural health monitoringrdquo Smart Materialsand Structures vol 13 no 2 pp 415ndash423 2004

[20] JMichaels A Croxford and PWilcox ldquoImaging algorithms forlocating damage via in situ ultrasonic sensorsrdquo in Proceedings ofthe 3rd IEEE Sensors Applications Symposium (SAS rsquo08) pp 63ndash67 February 2008

[21] J Michaels and T Michaels ldquoDamage localization in inhomo-geneous plates using a sparse array of ultrasonic transducersrdquoin AIP Conference Proceedings of the Review of Progress inQuantitative Nondestructive Evaluation vol 26 pp 846ndash853August 2006

[22] T Hay R Royer H Gao X Zhao and J L Rose ldquoA compar-ison of embedded sensor Lamb wave ultrasonic tomography


approaches for material loss detectionrdquo Smart Materials andStructures vol 15 no 4 pp 946ndash951 2006

[23] J V Velsor H Gao and J Rose ldquoGuided-wave tomographicimaging of defects in pipe using a probabilistic reconstructionalgorithmrdquo Insight vol 49 no 9 pp 532ndash537 2007

[24] F Yan R Royer Jr and J Rose ldquoUltrasonic guided waveimaging techniques in structural health monitoringrdquo Journal ofIntelligentMaterial Systems and Structures vol 21 no 3 pp 377ndash384 2010

[25] A Moustafa and S Salamone ldquoFractal dimension-based Lambwave tomography algorithm for damage detection in plate-like structuresrdquo Journal of Intelligent Material Systems andStructures vol 23 no 11 pp 1269ndash1276 2012

[26] Y Liu M Fard A Chattopadhyay and D Doyle ldquoDam-age assessment of CFRP composites using a time-frequencyapproachrdquo Journal of IntelligentMaterial Systems and Structuresvol 23 no 4 pp 397ndash413 2012

International Journal of

AerospaceEngineeringHindawi Publishing Corporationhttpwwwhindawicom Volume 2014

RoboticsJournal of

Hindawi Publishing Corporationhttpwwwhindawicom Volume 2014


Active and Passive Electronic Components

Control Scienceand Engineering

Journal of



RotatingMachinery


Hindawi Publishing Corporation httpwwwhindawicom

Journal ofEngineeringVolume 2014

Submit your manuscripts athttpwwwhindawicom

VLSI Design



Shock and Vibration


Civil EngineeringAdvances in

Acoustics and VibrationAdvances in



Electrical and Computer Engineering

Journal of

Advances inOptoElectronics


Volume 2014

The Scientific World JournalHindawi Publishing Corporation httpwwwhindawicom Volume 2014

SensorsJournal of


Modelling amp Simulation in EngineeringHindawi Publishing Corporation httpwwwhindawicom Volume 2014


Chemical EngineeringInternational Journal of Antennas and

Propagation




Navigation and Observation



DistributedSensor Networks



In 1917 Lamb published his classic analysis and descrip-tion of acoustic waves which included the first consider-ation of Lamb waves [4] In 1961 Worlton [5] proposedthe use of Lamb waves for damage detection and a newNDE potential emerged In 1962 Frederick and Worlont [6]conducted the first experimental study In the beginning ofthe 1990s Hutchins et al [7ndash9] and Nagata et al [10] appliedLamb waves to NDE using medical imaging and seismictomography techniques to bothmetallic and compositemate-rials Their systems utilized pairs of transducers that werepositioned across the material at various locations in orderto obtain a dense collection of pitch-catch signals in orderto reconstruct a tomogram While the techniques producedreasonable results themethods were time consuming requir-ing a lot of repositioning of the transducers and in somecases requiring the specimen to be placed in a water bathPrasad et al [11] implemented an SHM system for compositesusing surface bonded piezoelectric transducers Since thetransducers were permanently bonded in place the systemcould be operated in real time without any repositioningGao et al [12] introduced the reconstruction algorithm forprobabilistic inspection of damage (RAPID) Michaels [13]later investigated the application of tomography algorithmsto sparse networks

Other research efforts were made to use the Lamb wavemode propagation characteristics in order to detect damagespecifically the attenuation and arrival time of the first twowave modes In 1993 Guo and Cawley [14] studied theinteraction of Lamb waves with delaminations in compositematerials both numerically and experimentally Keilers andChang [15] later proposed a built-in damage detection systemusing an array of piezoelectric transducers Giurgiutiu et al[16] discussed the pulse-echo analysis technique and sub-tracting baseline data from damaged data to detect damage

This paper presents the details and results of a study on theimplementation of a sparse piezoelectric transducer networkfor damage detection in composite materials The objectiveof this research is to develop a sparse network that can beexpanded to detect damage over a large area To achieve thisa novel technique based on damage progression history hasbeen developed

2 Theory and Implementation

Lamb waves are elastic guided waves that propagate par-allel to the surface in thin structures with free boundariesPlates are the best example however Lamb waves canalso propagate in structures with a shallow curvature Themost advantageous characteristics of these waves are theirsusceptibility to interferences caused by damages or bound-aries (the features of interest) and low amplitude loss Toimplement a Lamb wave based damage detection techniquesome important properties must be determinedWhen Lambwaves propagate they travel in one of two possible ways withrespect to the platersquos mid plane If the motion is symmetricabout the midplane (the peaks and troughs of the waves arein phase) then it is a symmetric mode and if the motionis not symmetric (the peaks and troughs are 180∘ out of

phase) it is an antisymmetric mode An infinite number ofmodes exist each mode is referred to as an 119860 mode or119878 mode if it is antisymmetric or symmetric respectivelywith a subscript indicating its order For example the lowestorderfrequency symmetric mode is referred to as an 119878

0

mode while the second lowest orderfrequency symmetricmode is referred to as an 119860

1mode Each mode exists at

a different frequency depending on the properties of thematerial At lower frequency-thickness values less modesexist It is advantageous to operate in a frequency-thicknessrange where only the 119878

0and119860

0modes existThis is generally

below 15MHzmm

21 Implementation To employ Lamb waves for damagedetection a system to send and receive Lamb waves mustbe developed and a number of parameters must be selectedto tune the system such as the transducer type size andarrangement actuation signal and data acquisition

Piezoelectric transducers are commonly used in SHMsystems They are inherently simple devices consisting ofsimply a piezoelectric material with two conductive surfacesWhen a voltage is applied across the surfaces the materialexpands or contracts (depending on polarity) proportionalto the magnitude of voltage applied Conversely when thematerial is deformed a voltage difference is seen between thesurfaces This property leads to the greatest benefit of piezo-electric transducers their ability to both send and receivesignals Piezoelectric transducers are small and light and canbe bonded to or embedded within a structure with littleeffect Signals are voltage based and therefore easy to generateand acquire with common hardware Lead zirconate titanate(PZT) is the most commonly used material in piezoelectrictransducers They offer excellent performance for both gen-eration and acquisition have excellent mechanical strengthwide frequency responses and low power consumption andcan be obtained at a low cost [17]

When actuating Lamb waves it is desirable to actuate theleast number of modes possible (preferably only the 119878

0and

1198600) so that signal interpretation is simplified As mentioned

earlier this usually occurs below 15MHzmm This range isalso beneficial because there is low dispersion which meansthat if the frequency changes as the wave propagates throughthe material its velocity will remain relatively constanttherefore simplifying signal interpretation It is desirableto send out a single pulse so that the propagation of thewave groups can be analyzed The challenge then lies inproducing an instantaneous pulse that can be controlled andactuated at a desired frequency There is no control over afrequency generated by a simple impulse therefore a shortburst must be emitted If a simple sine wave composed ofa few cycles is emitted then the desired frequency can beactuated however the frequency domain of such a signalshows small secondary peaks in the frequency domain thatare present at other frequencies To eliminate these peaks thesignal can be modulated with a window function to slowlyincrease and decrease the magnitude of the signal [18] Acommonly used signal that provides a good compromisebetween number of cycles and ramp up rate consists of five


F

E

D

C B A

L

G H I

J

K

(a)

C

MNO

P

Q

R

S T

D

E

F

G H I

J

K

L

L

B A

(b)









119875 (119909 119910) =

119873

sum

119896=1

119901119896(119909 119910) =

119873

sum

119896=1

119860119896(120573 minus 119877

120573 minus 1) (1)




119877 (119909 119910 1199091119896 1199101119896 1199092119896 1199102119896)

=

radic(119909 minus 1199091119896)2

+ (119910 minus 1199101119896)2

+ radic(119909 minus 1199092119896)2

+ (119910 minus 1199102119896)2

radic(1199091119896

minus 1199092119896)2

+ (1199101119896

minus 1199102119896)2

(2)




(a) (b)





SDC = 1 minus1003816100381610038161003816120588119886119887

1003816100381610038161003816 (3)

where

120588119886119887

=1

119878

sum119878

119894=1

(119886119894minus 120583119886) (119887119894minus 120583119887)

radicsum119878

119894=1

(119886119894minus 120583119886)2

sum119878

119894=1

(119887119894minus 120583119887)2

(4)















4 6 8 10 12 14

times10minus5

minus025

minus02

minus015

minus01

minus005

0

005

01

015

02

Volta

ge (V

)

Seconds (s)

(a)

935922591

times10minus5

005

01

Volta

ge (V

)

Seconds (s)

(b)


wave (right)



0and119860






4 Discussion





0peak a consistent





260

240

220

200

180

160

280260240220200180160140

Position (mm)

Posit

ion

(mm

)

(a)

260

240

220

200

180

160

280260240220200180160140

Position (mm)

Posit

ion

(mm

)

(b)


260

240

220

200

180

160

280260240220200180160140

Position (mm)

Posit

ion

(mm

)

(a)

260

240

220

200

180

160

280260240220200180160140

Position (mm)

Posit

ion

(mm

)

(b)








5 Conclusion






References































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










F

E

D

C B A

L

G H I

J

K

(a)

C

MNO

P

Q

R

S T

D

E

F

G H I

J

K

L

L

B A

(b)









119875 (119909 119910) =

119873

sum

119896=1

119901119896(119909 119910) =

119873

sum

119896=1

119860119896(120573 minus 119877

120573 minus 1) (1)




119877 (119909 119910 1199091119896 1199101119896 1199092119896 1199102119896)

=

radic(119909 minus 1199091119896)2

+ (119910 minus 1199101119896)2

+ radic(119909 minus 1199092119896)2

+ (119910 minus 1199102119896)2

radic(1199091119896

minus 1199092119896)2

+ (1199101119896

minus 1199102119896)2

(2)




(a) (b)





SDC = 1 minus1003816100381610038161003816120588119886119887

1003816100381610038161003816 (3)

where

120588119886119887

=1

119878

sum119878

119894=1

(119886119894minus 120583119886) (119887119894minus 120583119887)

radicsum119878

119894=1

(119886119894minus 120583119886)2

sum119878

119894=1

(119887119894minus 120583119887)2

(4)















4 6 8 10 12 14

times10minus5

minus025

minus02

minus015

minus01

minus005

0

005

01

015

02

Volta

ge (V

)

Seconds (s)

(a)

935922591

times10minus5

005

01

Volta

ge (V

)

Seconds (s)

(b)


wave (right)



0and119860






4 Discussion





0peak a consistent





260

240

220

200

180

160

280260240220200180160140

Position (mm)

Posit

ion

(mm

)

(a)

260

240

220

200

180

160

280260240220200180160140

Position (mm)

Posit

ion

(mm

)

(b)


260

240

220

200

180

160

280260240220200180160140

Position (mm)

Posit

ion

(mm

)

(a)

260

240

220

200

180

160

280260240220200180160140

Position (mm)

Posit

ion

(mm

)

(b)








5 Conclusion






References































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










(a) (b)





SDC = 1 minus1003816100381610038161003816120588119886119887

1003816100381610038161003816 (3)

where

120588119886119887

=1

119878

sum119878

119894=1

(119886119894minus 120583119886) (119887119894minus 120583119887)

radicsum119878

119894=1

(119886119894minus 120583119886)2

sum119878

119894=1

(119887119894minus 120583119887)2

(4)















4 6 8 10 12 14

times10minus5

minus025

minus02

minus015

minus01

minus005

0

005

01

015

02

Volta

ge (V

)

Seconds (s)

(a)

935922591

times10minus5

005

01

Volta

ge (V

)

Seconds (s)

(b)


wave (right)



0and119860






4 Discussion





0peak a consistent





260

240

220

200

180

160

280260240220200180160140

Position (mm)

Posit

ion

(mm

)

(a)

260

240

220

200

180

160

280260240220200180160140

Position (mm)

Posit

ion

(mm

)

(b)


260

240

220

200

180

160

280260240220200180160140

Position (mm)

Posit

ion

(mm

)

(a)

260

240

220

200

180

160

280260240220200180160140

Position (mm)

Posit

ion

(mm

)

(b)








5 Conclusion






References































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










4 6 8 10 12 14

times10minus5

minus025

minus02

minus015

minus01

minus005

0

005

01

015

02

Volta

ge (V

)

Seconds (s)

(a)

935922591

times10minus5

005

01

Volta

ge (V

)

Seconds (s)

(b)


wave (right)



0and119860






4 Discussion





0peak a consistent





260

240

220

200

180

160

280260240220200180160140

Position (mm)

Posit

ion

(mm

)

(a)

260

240

220

200

180

160

280260240220200180160140

Position (mm)

Posit

ion

(mm

)

(b)


260

240

220

200

180

160

280260240220200180160140

Position (mm)

Posit

ion

(mm

)

(a)

260

240

220

200

180

160

280260240220200180160140

Position (mm)

Posit

ion

(mm

)

(b)








5 Conclusion






References































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation










260

240

220

200

180

160

280260240220200180160140

Position (mm)

Posit

ion

(mm

)

(a)

260

240

220

200

180

160

280260240220200180160140

Position (mm)

Posit

ion

(mm

)

(b)


260

240

220

200

180

160

280260240220200180160140

Position (mm)

Posit

ion

(mm

)

(a)

260

240

220

200

180

160

280260240220200180160140

Position (mm)

Posit

ion

(mm

)

(b)








5 Conclusion






References































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation











5 Conclusion






References































RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation

















RoboticsJournal of





Journal of



RotatingMachinery





VLSI Design



Shock and Vibration







Journal of



Volume 2014


SensorsJournal of





Propagation









research article damage detection of composite plates by ...lamb wave based structural health...

Documents