safety implications of tofd

HSE Health & Safety

Executive

Safety implications of TOFD for in-manufacture inspections

Prepared by MitsuiBabcock Energy Limited for the Health and Safety Executive 2006

RESEARCH REPORT 433

HSE Health & Safety

Executive

Safety implications of TOFD for in-manufacture inspections

N S Goujon MitsuiBabcock Energy Limited

Porterfield Road Renfrew

Renfrewshire PA4 8DJ

This report describes the results from a project to evaluate the safety implications related to the European project “Effective application of TOFD method for weld inspection at the manufacturing stage of pressure vessels” (acronym: TOFDPROOF).

The TOFDPROOF project was performed by a consortium of European organisations including Mitsui Babcock.

The project aimed at producing a coherent package of EU documents including procedures for applying TOFD, acceptance criteria and recommendations for training and certification.

The work involved ultrasonic inspection of a number of test-specimens containing both synthetic and real service-induced defects (mainly cracks in welds). The project concentrated on detection and sizing trials. Recommendations based both on the results of this project and the views of the partners of the TOFDPROOF project are provided and the safety issues related to the results of the TOFDPROOF project are discussed.

This report and the work it describes were funded by the Health and Safety Executive (HSE). Its contents, including any opinions and/or conclusions expressed, are those of the author alone and do not necessarily reflect HSE policy.

HSE BOOKS

© Crown copyright 2006

First published 2006

All rights reserved. No part of this publication may bereproduced, stored in a retrieval system, or transmitted inany form or by any means (electronic, mechanical,photocopying, recording or otherwise) without the priorwritten permission of the copyright owner.

Applications for reproduction should be made in writing to: Licensing Division, Her Majesty's Stationery Office, St Clements House, 2-16 Colegate, Norwich NR3 1BQ or by e-mail to [email protected]

ii

CONTENTS

DISTRIBUTION LIST.................................................................................................................................. i

AMENDMENT CONTROL.......................................................................................................................... i

CONTENTS .............................................................................................................................................. iii

SUMMARY................................................................................................................................................. v

1. INTRODUCTION .................................................................................................1

2. TOFDPROOF PROJECT ....................................................................................2

2.1 TOFDPROOF CONSORTIUM....................................................................................................... 2

2.2 TOFDPROOF WORK PACKAGES............................................................................................... 2

2.3 WORK RESULTS .......................................................................................................................... 4

2.3.1 Work Package WP1: Trials organisation and justification of representative specimens .......... 42.3.2 Work Package WP2: Blind Trials and Performance Assessment ............................................ 82.3.3 Work Package WP3: Guidelines for training, qualification and certification ........................... 202.3.4 Work Package WP4: Acceptance criteria definition................................................................ 242.3.5 Work Package WP5: Economic analysis ................................................................................ 292.3.6 Work Package WP6: Exploitation and dissemination ............................................................. 332.3.7 Work Package WP7: Data storage, analysis and exchange .................................................. 35

3. SAFETY RELATED ISSUES ............................................................................36

3.1 INTRODUCTION.......................................................................................................................... 36

3.2 ACCEPTANCE CRITERIA .......................................................................................................... 37

4. CONCLUSIONS ................................................................................................39

5. REFERENCES ..................................................................................................41

APPENDICES.......................................................................................................................................... 43

APPENDIX 1: RECOMMENDATIONS FOR APPLYING TOFD (FIELD OF APPLICATION, STRENGTHS, WEAKNESSES) .............................................................................................................. 44

APPENDIX 2 EXAMPLE OF TOFD ERRORS FOR A 15MM THICK SAMPLE ................................... 59

iii

SUMMARY

This final report describes the work carried out on the HSE funded project “Safety Implications of TOFD for In-Manufacture Inspections”.

This report describes the results from a project to evaluate the safety implications related to the European project “Effective application of TOFD method for weld inspection at the manufacturing stage of pressure vessels” (acronym: TOFDPROOF).

The TOFDPROOF project was performed by a consortium of European organisations including Mitsui Babcock.

The project aimed at producing a coherent package of EU documents including procedures for applying TOFD, acceptance criteria and recommendations for training and certification.

The work involved ultrasonic inspection of a number of test-specimens containing both synthetic and real service-induced defects (mainly cracks in welds). The project concentrated on detection and sizing trials.

Recommendations based both on the results of this project and the views of the partners of the TOFDPROOF project are provided and the safety issues related to the results of the TOFDPROOF project are discussed.

v

1. INTRODUCTION This final report describes the work carried out on the HSE funded project “Safety Implications of TOFD for In-Manufacture Inspections”.

Time of Flight Diffraction (TOFD) is an ultrasonic non-destructive testing method which was originally developed as an accurate method for measuring the through-wall extent of defects which had been detected by more conventional methods such as Pulse-Echo (PE) ultrasonics.

TOFD differs from conventional PE examinations in that it relies on tip diffraction from the extremities of the defect for both detection and sizing. With the knowledge of the wave velocity and the spatial relationships of the two probes, the position of the tips of the flaw can be calculated very accurately.

The success in this field, combined with cheaper and more portable TOFD hardware, has resulted in the acceptance within a wide range of industrial sectors (offshore, petrochemical, chemical, defence, conventional power generation). However, it is increasingly being proposed as a rapid search tool, not just for defect sizing (i.e. as an alternative to pulse-echo inspection instead of as a complement to it).

TOFD has been reported as being a technique with high probability of detection for both planar and volumetric defects and of excellent reproducibility and accuracy.

Although the economic advantages of using TOFD for manufacturing inspections are clear (inspection speed is typically 3 times faster than conventional PE techniques) there remain obstacles to its widespread application for general inspections. These include concerns over detection capability, lack of objective guidance on applicability and (particularly for manufacturing inspections) lack of agreed acceptance criteria.

To overcome these obstacles a European consortium submitted a proposal to the European commission for partial funding of a project called “TOFDPROOF”. The TOFDPROOF project aimed to develop procedures for applying TOFD, guidance on applicability, acceptance criteria, and recommendations for training and certification. The intention is that this will allow vessel manufacturers to use TOFD as a standalone method for weld inspection.

This report summarised the work performed on the TOFDPROOF project and covers the evaluation of the implications of the “TOFDPROOF” project for UK manufacturing inspections of safety critical components.

1

2. TOFDPROOF PROJECT

2.1 TOFDPROOF CONSORTIUM

Eight European countries were represented in the TOFDPROOF consortium .

The consortium was a mix of industrial companies and research organisations. Major Non Destructive Testing (NDT) service providers and two inspection and certification notified bodies were represented to enable rapid acceptance and adoption of the TOFD method throughout industry.

An overview of the consortium is given in the Table 1.

2.2 TOFDPROOF WORK PACKAGES

The TOFDPROOF project was divided into the following seven work packages:

• WP1: Trials organisation and justification of representative specimens

• WP2: Blind trials and performance assessing

• WP3: Guidelines for training, qualification and certification

• WP4: Acceptance criteria definition

• WP5: Economic analysis

• WP6: Exploitation and dissemination of results, link with CEN

• WP7: Data storage, analysis and exchange

The objectives and the work performed for each of these work packages are described in the next section.

2

Table 1 Overview of the consortium (extract from TOFDPROOF proposal)

Name Country Main mission / Business activity /Area of activity

Institut de Soudure (IS) (Project Co-ordinator)

France

Non profit company. Research works and expertise in applications of welding and allied techniques and provider of services and training. Member of an accredited notified body for pressure vessels. Certification of NDT operators.

Sonovation Netherlands

Sonovation are a high technology SME company offering sophisticated structural integrity inspection. Provider of NDT services, specifically (mechanised) UT and TOFD. Involved in TOFD inspection since 1988, instrumental in introduction and development of TOFD, involved in trials and validations of TOFD since 1990, initiated and participated in many projects on TOFD, involved in national and international codes and standards development.

TWI Limited UK

A non profit company. Contract research and consultancy in all methods of joining technology and associated technologies. Major worldwide provider of both NDT and welder training and certification. Provider of NDT field services including automated ultrasonics (Pscan, TOFD, phased array), ACFM, Lizard and conventional methods. Many research projects and on-site inspections undertaken with TOFD since 1979.

Mitsui Babcock (MB)

UK

Major provider of power generation equipment. Design, supply and construction of power plant world wide. Major provider of NDT services. On-site NDT inspection, expertise. Expertise in development and qualification of automated ultrasonic inspection techniques.

Staatliche Materialprefungsa nstalt (MPA Stuttgart)

Germany

Major organisation relevant to risk-analysis, reliability and safety of materials, components, processes, structures and systems also component testing and validation of advanced engineering methods as well as intelligent software systems.

Tecnatom S.A. Spain NDE services, NDE in-service inspections, Engineering services, manufacturer of inspection systems

VTT Finland Research and development of NDE, NDE services, materials research, assessment of structural integrity

Instituto de Soldadura e Qualidade (ISQ)

Portugal Non profit company. Institute for welding, materials testing and allied techniques. Major supplier of NDE services.

TUV Suddeutschland Bau und Betrieb GmbH (TUV)

Germany

Major technical control organisation, including power plants, nuclear power plants and plants for process-technology. Member of an accredited notified body for pressure vessels. Focus is put on inspection, testing and assessment of construction of tanks and vessels, pipes, heat exchangers and structural steelwork.

3

2.3 WORK RESULTS

Each work package is discussed in turn in the following paragraphs.

2.3.1 Work Package WP1: Trials organisation and justification of representative specimens

2.3.1.1 Objectives

This task was divided into three sub-tasks, that is, the identification of an appropriate collection of welded specimens for the project, the preparation and agreement on conventional NDT procedures and the organisation of a matrix of tests using various NDT inspections and the manufacture of additional specimens if needed. Each of these sub-tasks is considered in more detail below.

2.3.1.2 Work Performed

Review of Existing Specimens (Task 1.1)

The aim of this task was to establish a population of defects for practical trials. The set of samples to be used in the project was selected from a database compiled by the partners, supplemented by a series of manufactured samples. A review of the original database and the proposals for additional samples was provided. The main outputs are given below.

Ideally the specimens had to offer a wide variety of parameters against which to compare time of flight diffraction and conventional NDT methods. Only simple butt welded geometries were considered, such as flat plat and piping. The material was limited to ferritic steel. The sample thicknesses had to be in the range between 6mm and 100mm and both longitudinal and transverse defects were to be included.

The final selection aimed to provide a diverse group of defect types, positions and dimensions spread over the range of material thickness. The group had also to offer a fair balance between plate and pipe specimens.

The database contained more than 190 defects. Weaknesses in the database were identified and additional defect samples were suggested.

Manufacturing of Specimens, Justification of the Collection and Design of the Matrix of Tests (Task 1.2)

The set of samples used in the project were selected from the database compiled by the partners and supplemented by a series of manufactured samples as discussed above. The finalised database is made of 150 samples including 118 samples selected from the original database and 32 additional manufactured samples. The finalised database is given Table 2.

Having established the group of test pieces it was necessary to define the test matrix i.e. the set of tests to be applied to each sample.

4

Tabl

e 2

Sum

mar

y of

the

sam

ple

popu

latio

n

Sa

mpl

e th

ickn

ess

Def

ect

~6m

m

~10m

m

~15m

m

~20m

m

~25m

m

~30m

m

~40m

m

~50m

m

~60m

m

~75m

m

~100

mm

TO

TAL

Type

Lack

of

4 2

11

2 5

7 5

2 4

1 1

44fu

sion

Cra

ck

8 2

2 5

7 1

0 2

5 3

1 36

Poro

sity

0

0 1

1 4

3 1

0 5

0 0

15

Slag

0

3 3

1 5

0 2

1 0

1 1

17

Tran

sver

se

2 0

0 0

2 0

0 3

2 3

0 12

crac

k

Lack

of

0 2

1 0

2 5

0 1

0 1

0 12

pene

trat

ion

Tran

sver

se

0 0

0 0

0 0

0 0

5 0

0 5

LoF

Lack

of I

R

0 0

1 1

3 1

0 0

0 0

0 6

fusi

on

Volu

met

ric

0 0

0 0

0 0

0 0

0 0

2 2

Ove

r 0

1 0

0 0

0 0

0 0

0 0

1pe

netr

atio

n

TOTA

L 14

10

19

10

28

17

8

9 21

9

5 15

0

5

Test Matrix

The test philosophy was to apply conventional pulse echo Ultrasonic Testing (UT), conventional Radiographic Testing (RT) and TOFD to each specimen in accordance with current European standards. To allow comparison, TOFD was applied by at least two different teams and for the common thickness range 15 to 75mm three teams applied TOFD. At least two different teams applied conventional tests to each specimen. For the common thickness range 15 to 50 mm, two teams applied manual UT and at least one of the radiographic methods allowed by European standards was used. Samples less than 10mm thickness were not examined by manual ultrasonic inspection but were examined by two independent X-Ray teams.

The following trials testing organisation was defined:

• 5 teams to perform TOFD testing

• 4 teams to perform UT testing

• 3 teams to perform X-ray testing

• 2 teams to perform γ–ray testing

• 1 team to perform accelerator testing

Test Groups and Tests per Sample

The test method applied to a given test-sample depended on the thickness of the test-sample. The number of times a given method was applied was also dependant on the sample thickness. The number of times a given sample was inspected by a given method is illustrated Table 3.

Table 3 Number of methods/applications applied against sample thickness

Methods / applications

Sample Thickness TOFD Manual UT X Ray Gamma Accelerator

<10 2 0 2 0 0

10 to <15 2 1 1 0 0

15 to <40 3 2 1 1 0

40 to <50 3 2 1 1 1

50 to <75 3 1 0 1 1

75 to 100 2 1 0 1 1

Matrix of Tests

The matrix of tests defined which team was to examine which sample. The number of tests to be applied by a given team was established by determining the total number of tests for a given method and assuring the following:

6

• Conventional methods (including UT and RT) and TOFD were to be applied to each specimen in accordance with current European standards.

• To allow comparisons, TOFD was applied by at least two different teams and for the common thickness (range 15 to 75mm) three teams applied TOFD.

• At least two different teams applied conventional tests to each specimen (three were applied to specimens above 15mm thick).

• The test method applied and the number of times a given method was applied to a given test piece depended on its thickness.

Additionally, it was agreed that the following points had to be addressed in order to preserve the integrity of the work:

• Previous inspection results of the specimens were deliberately withheld.

• A given partner was not allowed to carry out both TOFD and conventional NDT.

• A given partner was only allowed to apply one type of conventional testing method (as far as possible). For practical reasons, the same team would perform gamma-ray inspection and accelerator inspection on the specimens 60mm thick and over.

• The results from the trials were only to be circulated to all the partners after completion of all the trials.

• An acceptable balance of workload between partners had to be achieved.

Design and agreement on ionising radiation and UT procedures (Task 1.3)

A UT manual procedure was written in accordance with EN 1714 [1] and EN 1713 [2]; as well as a series of ionising radiation procedures written in accordance with EN 1435 class B [3]. These procedures were reviewed by a working group including most of the partners and were all approved.

2.3.1.3 Comments

It has to be stressed that the samples used for the study are only of simple geometry (plate and pipe) and that the defects they contained are essentially manufactured defects.

Some of the samples contained additional defects which were not included in the study by choice or because they were not intended defects. In some instances, these extra defects made the study more difficult when these defects interfered with the selected defects. Moreover, one of the main difficulties encountered later on, related to the matrix of the test samples, was the quality of information provided on the defects including defects type, position and size. Wrong information but mainly incomplete information made the data study difficult.

7

2.3.2 Work Package WP2: Blind Trials and Performance Assessment

2.3.2.1 Objectives

The Objectives were to carry out blind trials in order to assess the performance, the influence of scan set-up, the reliability and reproducibility of the TOFD technique:

• To compare TOFD performance (both in term of acceptance criteria and best practice) with UT and RT, as applied according to the European standards defined by CEN/TC 121 "Welding";

• To define the field of application of TOFD, highlighting weaknesses and strengths and the possible need to use TOFD in combination with other NDT techniques;

• To optimise the methodology of application in order to ensure reproducible inspections with different pieces of equipment and inspectors;

• To verify how TOFD allows for detection of transverse defects;

• To perform sectioning or advanced NDT techniques when discrepancies were observed between TOFD and conventional inspection.

The work package was divided into sub-packages, that is:

a) Design of TOFD procedure and reference blocks;

b) Experimental work: TOFD and conventional NDT procedures application;

c) Technical analysis;

d) Optimisation of TOFD procedure;

e) Recommendations.

The work carried out under each sub-work packages is discuss below.


a) Design and agreement on set of TOFD procedures and manufacturing of test-blocks

TOFD procedure

A TOFD procedure was produced by IS and reviewed by the partners.

The choice of set-ups to be applied (including: probe centre separation, probe frequency, etc,…) in relation to the sample thickness was determined using the results of beam coverage modelling.

200 cases in terms of thickness, probe diameter, probe frequency, refraction angle were studied. A table of recommended values was produced, see Table 4.

8

Table 4 Recommended TOFD Set-up

Thickness t (mm)

Number of TOFD set-ups

Depth-range (mm)

Centre frequency

(MHz)

Beam-angle (°)

Element diameter

(mm)

Beam intersection

depth

6-10 1 0-t 15 70 2-3 2/3 of t

>10-15 1 0-t 15-10 70 2-3 2/3 of t

>15-35 1 0-t 10-5 70-60 2-6 2/3 of t

>35-50 1 0-t 5-3.5 70-60 3-6 2/3 of t

0-t/2 5-3.5 70-60 3-6 1/3 of t >50-100 2

t/2-t 5-3.5 60-45 6-12 5/6 of t for 60° or t for 45º

The classification of relevant indications was specified by analysing the following features:

• Disturbance of the lateral wave

• Disturbance of the back wall echo

• Pattern between lateral wave and back wall echo

• Signal phase pattern between lateral wave and back wall echo

• Mode converted signal after the first back wall reflection.

The indications were classified as follows:

• Surface breaking discontinuities

• Embedded discontinuities (point like, elongated without measurable height, elongated with measurable height)

• Unclassified.

Reference block

For the application of this procedure, a reference block had to be used. A report describing the reference blocks was written by IS and reviewed by the other partners involved in the task. Two new designs were proposed for the reference blocks, one with side drilled holes and one with notches. Note that the reflectors machined in the reference blocks were side drilled holes which were not connected to the scanning surface (as suggested by ENV 583-6:2000 [4]).

9

b) Experimental work: Round Robin Trials (RRT)

The RRT were carried out using the NDT procedures developed during the project.

Conventional inspections

The data were compiled by the partners and stored into a data base. The data base was available through the TOFDPROOF web site. The data was only available to the TOFDPROOF partners.

TOFD inspections

The TOFD inspections were carried out by five TOFD inspection teams (IS, SONOVATION, TWI, VTT, ISQ). The team who supplied the specimens were not involved in the examination of their own specimens.

Following the round robin trials and the reporting of the results, the data were collated and a review focusing on the cause of discrepancies of the results was carried out by MBEL and TWI. TOFD images and the inspection settings used for the round robin were requested to enable the identification of discrepancy of the results.

Selected results from the round robin trial were examined and the source of the discrepancies between the intended defects and the inspection results was identified. This was carried out in order to highlight the strengths and weaknesses of the TOFD technique. The lack of information on some of the defects limited the analysis.

The results were provided in a discrepancies analysis report. The report includes a table summarising the discrepancies between the TOFD results for the defects for which sufficient information was provided.

The report discussion is divided into six groups including: general comments, effect of the set-up, surface defects, transverse defects, sizing errors and acceptance criteria. For each group, comments and recommendations are provided. Recommendations on when complementary NDT techniques should be used are also provided.

The results confirm some of the TOFD limitations and highlight the need for an appropriate procedure, a skilled data analyst and for realistic acceptance criteria.

c) Technical analysis of the results

A technical analysis of results (performance, reproducibility) was carried out.

A template was defined for reporting the results and each partners completed their template either offline (e.g. Excel) or on-line on the web site. In the course of data gathering, more than 600 documents, including single test reports have been gathered and processed.

Definition of the parameters to store for each NDT Technique was done for each specific procedure. MPA’s “ALIAS” software or optionally Access or Excel software using links to images and files were used for data storage, processing and analysis.

The existing (MPA ALIAS system) tool was used as an initial platform for results analysis. The database contains a list of specimens with basic information as well as the information about intended and detected defects.

10

Based on the literature overview carried out and on the results of similar projects, it was decided to perform the hit-miss data analysis using either a log-logistic function or using the Weibull three parameter cumulative distribution.

Due to the fact that the Weibull distribution has already been used for structure assessment and development of the acceptance criteria in other similar projects, it was favoured by the project partners against the log-logistic or log-normal approaches.

The assumed probability of detection (POD) curves for TOFD are based on a 3parameter Weibull equation to describe the mean POD as a function of defect height,

POD (a) = 1 – exp { - [( a - γ ) / α ]β }

Where: α: Scale parameter, calculated using selected mean POD values (for given shape and threshold parameters)

β: Shape parameter γ: Threshold parameter, below which size no defects are detected

(POD=0)

It should be noted that a further comparison with accepted approaches in other NDT areas (RT with lognormal cumulative distribution approach, for example) will be needed in order to obtain better benchmarking results. Furthermore, one of the results of the interaction between the project partners and other interested groups in this area showed that they might be objections to the shape of the curves obtained in the project. These objections were based on the comparison of the curves proposed with those obtained in other projects using similar approaches for other NDT techniques. However, for the purpose of data analysis and development of acceptance criteria, the applied data analysis methods was considered to give more than reasonable and feasible results, comparable with similar results obtained in other related projects.

For the hit-miss data analysis (approach by which the inspection results can be recorded only in terms of whether or not a flaw was found), a comparison of expected results against obtained results has been performed.

The data fitting process has been performed using, in addition to pure analytical/statistical data analysis, some expert judgement and taking into account the results and experience of other projects, namely:

1. It has been assumed that the gamma parameter has the physical meaning of the non-detection limit, that is, under the size of gamma (in millimetres) the system is either not capable or it is not possible to detect the defect.

2. It is assumed that the curves have similar shapes (similarity principle), therefore, an average value of beta parameter has been fixed as 0.65.

3. A lower bound curve can be constructed based on confidence level of 95% and shape similarity.

The achieved mean POD curves are shown in Figure 1 to Figure 6, whereas Figure 7 shows the whole family of curves.

11

Figure 1 POD curve for plate thickness up to 6 mm

Figure 2 POD curve for plate thickness 6 to 15 mm

Figure 3 POD curve for plate thickness 15 to 25.4 mm

12

Figure 4 POD curve for plate thickness 25.4 to 40 mm

Figure 5 POD curve for plate thickness 40 to 60 mm

Figure 6 POD curve for plate thickness over 60 mm

13

14

Figure 7 Family of POD curves for defect height detection

Figure 8 Comparison of TOFD, RT and MUT, defect height, for a 25.4mm thick plate

The Figure 8 shows an example of overall NDT methods comparison for the methods considered in the project (TOFD, RT, MUT) for a 25.4mm thick plate.

For sizing performance, one of the greatest drawbacks in the analysis process was the lack of information on the true size of the defects which could have been provided through sectioning (destructive confirmation of defects). The information may have explained some gaps and discrepancies in the results obtained.

However, using the available data, and combining it with other available data provided by Sonovation, the data analysis for sizing was performed; the summary of the results is given in Table 5.

Table 5 Performances comparison

Height Mean Error

∆h (mm)

Height Standard Deviation

Length Mean Error

∆l (mm)

Length Standard Deviation

TOFD 1.0* 1.9 4.5 6.5

RT NA NA 1.5 18

MUT 0.4 4 1.5 25 * Mean value on the full range of samples 6mm < t <100mm. Note that the mean value obtained on another population of defects provided a ∆h of 0.3mm for TOFD after sectioning.

Other comparison performances for POD and false call rate (FCR) were carried out and are summarised in Table 6. Table 6 also includes performances obtained during the Dutch acceptance criteria project by the Dutch Quality Surveillance and Non-Destructive Testing Society (KINT).

Table 6 Performances comparison – POD & FCR

TOFDPROOF project KINT project

POD FCR POD FCR

TOFD 70-90% <10% 82.4% 11.1%

RT 60-70% NA 60.1%* 10.8%*

MUT 55-65% NA 52.3% 22.7% * Using γ-ray only.

False Call Probability analysis was not performed for RT and MUT due to the lack of relevant data and no possibility to objectively verify the results by sectioning the samples.

d) Optimisation of TOFD procedures and application

The TOFD procedure developed for the trials was updated to reflect the remarks and conclusions from the discrepancy analysis report and the comments from the partners involved in the TOFD examinations. A final report of the procedure was produced. The main items modified are listed below:

• When using the recommended TOFD set-ups presented in Table 4, the maximum recommended probe frequency should be considered first. Lower frequencies may be used if the required sensitivity setting cannot be achieved with higher frequencies.

• The techniques to measure length and height were more clearly defined. This part of the procedure has also been added in the proposed acceptance criteria and in the recommendation for applying TOFD (see Appendix 1).

• When the specific pattern (replication of mode converted signals) shown in Figure 10 is observed one can conclude that there is evidence of transverse defects.

15

• The end of time window shall be at least 1 µs after the 1st mode converted signal.

Specific p

Transverse defects

attern

Figure 10 TOFD image of a non-parallel scan of a sample containing 3 transverse defects.

e) Recommendations for applying TOFD

Taking account of the discrepancies identified during the analysis of the round robin trials TOFD data, recommendations for applying TOFD and on when a complementary NDT technique could be required, were provided.

A report on recommendations, report No 2-31-D-2004-02-1 [5] given in Appendix 1 (the report can also be downloaded from the TOFDPROOF web site: http://www.mpa-lifetech.de/TOFD) was written by MB and TWI and includes comments from the other TOFDPROOF partners.

The recommendations were transmitted to CEN/TC 121, CEN/TC 54, CEN/TC 138 and EPERC.

The recommendations (complete with observations) were provided for seven categories, including: procedure, identification of reference marks, set-up, classification of indications, evaluation of indications, personnel qualification and acceptance criteria. For each category, comments and recommendations were provided. A summary is provided hereafter.

16

• A specific procedure shall be written in accordance with the guidelines given in ENV 583-6 [4] and DD CEN/TS 14751 [6] for each individual type of inspection.

• The identification and inscription of reference marks (including datum on the component and reference point on the inspection probes array) is critical to allow repeatability of the inspection and results comparison.

• Where a defect extends before and after a datum, an extra scan covering the entire defect area (without a break) should be carried out in order to provide a more accurate value of the defect length.

• Where a large number of point-like indications have been detected which creates a cluster of indications, the inspection should be supported by another NDT technique.

• It became clearer after the review of the analysis that high frequency small crystal diameter probes (15MHz, 3mm) should be recommended for the inspection of thin samples (up to 15mm) especially when the weld surfaces are as-welded, an alternative choice (e.g., 10MHz) may not be appropriate.

• The presence of the lateral wave and the back wall echo restrict the inspection zone and small defects in these zones can be missed. High frequency (15MHz) probes provided a better means of detection and evaluation of defects which are surface breaking or just below the surface.

• A separate root scan should be considered.

• Inspection from both surfaces is recommended when access permits.

• Inspection for transverse indications can be limited especially when the weld cap is still present. The normal TOFD configuration is not optimised for transverse defect inspection. However, the presence of some mode converted echoes may suggest that transverse defects could be present.

• Transverse indications were in general not reported or wrongly reported as point- like or as longitudinal defects. When transverse defects are expected or/and when indications on the TOFD image suggest the presence of such defects (especially if the weld is as-welded), additional technique(s) should be used.

• The indications shall be classified into categories clearly defined in the inspection procedure.

• For interpretation of the images, initial analysis has to be carried out on unprocessed data. Straightening and removal (for lateral wave and back wall echo) tools can be use for subsequent analysis e.g. confirmation of presence/absence of surface defects.

• The technique used to determine the indications dimensions should be clearly defined.

In addition, guidelines on how to determine defect length, depth and height were provided.

17

The results confirm some of the TOFD limitations and highlight the need for an appropriate procedure, a skilled data analyst and for realistic acceptance criteria.

2.3.2.3 Comments

General

Most of the specimens used during the round robin trials contained additional defects which were not included in the matrix of defects. The lack of information / wrong information provided by the owner of the sample used for the round robin trials as well as the presence of additional unwanted defects restricted the study.

Technical analysis of the results

The technical analysis of the results provided was not as detailed as expected. This is probably partly due to the comments above; however, the results from the conventional inspections were not presented in any detail. In order to analyse the large quantity of data, MPA used their ALIAS software. The results provided are limited to Table 5 and Table 6 and the example presented Figure 8.

It is not clear how the information from the round robin was treated. Having carried out a preliminary analysis of the TOFD data it is not clear how much of the results from the round robin were used for the construction of the POD curves. The preliminary analysis carried out highlights the difficulty of identifying the defects selected for the study, mainly due to: the inconsistency of the referencing of the TOFD scans, the presence of unwanted defects and the lack of information on the defects. The conclusions are therefore unlikely to be based on the full population of defects. Moreover, the PODs (Figures 1 to 8) relate to defects as small as 0.5mm, it is not clear how those dimensions are known.

Note that the dimensions used for the analysis are based on manufacturing values. No sectioning was performed on any of the specimens even when discrepancies between the TOFD results (and/or NDT techniques) were observed.

Recommendations

The recommendations provided in recommendations for applying TOFD report [5] were based on the preliminary analysis of the TOFD data. The report provides important information. However, an important recommendation was omitted in the final TOFDPROOF report, that is, a separate root scan should be considered. It is well recognised that one of the main limitations of TOFD is the capability of detecting small defects at the root area especially when the root bead is present. It is therefore important to concentrate on this area, by using specific TOFD scans but more appropriately to use additional NDT techniques.

TOFD procedure

The Optimisation of the TOFD procedures was carried out using the comments from the TOFDPROOF partners. Although the procedure is specific to the TOFDPROOF round robin, it can be used as an example for other inspections.

The procedure was supplied to the Technical Committee CEN/TC 121 “Welding” and the information was used to update the technique specification CEN/TC 121/SC 5/WG 2 N 146 document “Welding – Use of time-of-flight diffraction technique (TOFD) for testing of welds” and the draft European standard DD CEN/TS

18

14751:2004 document “Welding – Use of time-of-flight diffraction technique (TOFD) for examination of welds” [6]. These documents were approved after the completion of the TOFDPROOF project.

The TOFD procedure used for the TOFDPROOF round robin trial has a lot of similarity with the DD CEN/TS 14751:2004 document (being developed at similar time). The scope of this project did not include the review of the CEN standard however, comparisons and comments on ‘the preparation for examination’ section(s) of the documents are presented hereafter.

The main differences are that the CEN standard extends its scope to in-service inspections and to thicker specimens (note that the standard does not provide an upper thickness limit but provides recommended TOFD set-ups for thicknesses up to 300mm).

The type of calibration blocks recommended under both the TOFDPROOF procedure and the DD CEN/TS 14751:2004, does not include one of the two type of blocks proposed under ENV 583-6:2000. The European pre-standard DD ENV 583-6 proposed a choice of blocks containing two types of diffractors:

• Machined notches, open to the scanning surface of the reference block; or

• Side drilled holes with a diameter of at least twice the wavelength of the nominal frequency of the probes utilised in the inspection. The holes were to be cut to the scanning surface in order to block the direct reflection from the top of the hole.

It is not clear why the block containing the second type of diffractor is not included into the recommendations. Especially because the manufacture of this type of diffractor is easier than the 60° notches and allows greater reliability in calibration than the reference block containing side-drilled holes without being cut to the surface.

Both documents state that the sensitivity setting shall be performed in accordance with ENV 583-6:2000 which specifies that the settings in the region of the timebase after the arrival of the lateral wave should be set to approximately 5% of the amplitude scale. However, both documents favour the setting with reference to the lateral wave. The sensitivity should be set such that the amplitude of the lateral wave is between 40% and 80% full screen height (FSH). Then, in the case where the use of the lateral wave is not appropriate the sensitivity shall be set such that the amplitude of the backwall signal is between 18dB and 30dB above FSH. If neither lateral wave nor backwall signal is appropriate, the sensitivity should be set such that the material grain noise is between 5% and 10% FSH.

The variation between the two documents are:

• Under the TOFDPROOF procedure the sensitivity check shall be performed on a reference block, under the CEN standard the sensitivity shall be set on the test object.

• Under the TOFDPROOF procedure an additional note is provided when the sensitivity is set with reference to the lateral wave, that is, if the noise level exceeds 10% FSH the lateral wave amplitude shall be decreased accordingly.

Both documents cover simple geometry in plates, pipes and vessels where both the weld and the parent material are low alloy carbon steel. However, it is likely that the

19

standard will also be used as a reference for other material. Moreover, it seems more appropriate to use the noise level as a preferred setting and if the material grain noise is not appropriate (lateral wave or backwall signal not appropriate e.g., saturation of the lateral wave) use the lateral wave or the backwall signal (as per ENV 583-6:2000). Too high amplitude of the lateral wave or the backwall echo may increase difficulties in identifying small defects at these areas.

As the statement “a sensitivity check shall be performed at least every four hours and after completion of the examination” covers both pre-service and in-service inspections for the CEN standard, It should be noted that if the inspection is applied to radiation environments the four hours may not be always practical from an ALARP point of view (e.g. during the inspection of a large vessel).

Furthermore, the sensitivity and range corrections checking proposed by both documents, suggest that all examinations shall be repeated since the last valid check if any deviation on the sensitivity greater than 6dB is observed. It has to be noted that local deviations are frequently observed and local deviations of more than 6dB should not be considered as a source of rejection of acquired data. The sensitivity variation should mainly be used to verify the proper working condition of the equipment.

Due to the similarity between the TOFD procedure used under the TOFDPROOF project and the DD CEN/TS 14751:2004 document, The TOFDPROOF round robin trials could have been a good opportunity to test the capability of the new EN technical specification. Unfortunately, there were evidences that the partners involved in the round robin trials did not always follow the TOFD procedure.

2.3.3 Work Package WP3: Guidelines for training, qualification and certification

2.3.3.1 Objectives

This work package was to define a framework for operators qualification and certification and assessing the influence of the objectiveness of the inspector interpreting the results.


A. Interactive guideline

Interactive training guidelines for interpreting TOFD images were produced. They were designed as a computer software programme. It contains a database of more than 25 typical TOFD images. TOFD images used for the software were provided by the partners.

The guidelines are divided into three parts, providing:

• Basic information on TOFD

• TOFD image assessment methodology

• TOFD indication classification and sizing methodology

20

The last part doesn’t take into account the optimising sizing techniques recommended by the “optimised TOFD procedure” presented in Work package 2; and the individual A-scans of a given TOFD image were not made available.

As a part of the training, the inspector learns to assess the quality of the TOFD image. Unsatisfactory TOFD images are provided with explanation on the cause of the problems. This section also provides a way of identifying defects on a TOFD image and to determine their type and dimensions (length and height).

The interactive training software can be used on line free of charge on the web site: http://www.mpa-lifetech.de/TOFD.

B. Objectivity assessment

As well as the training part, the software also contains a ‘testing’ part that has been used to assess the objectivity of the inspector’s interpretation of TOFD images.

Inspectors from the TOFD partners have taken part in the objectivity assessment, and the results of this exercise have been analysed and reported.

The discrepancies seen in the results, during the objectivity assessment, have been fed back to the guidelines designer to enable the interactive training guidelines to be improved and updated.

The conclusions from the survey were:

• The greater degree of confusion was observed in the defects type E (elongated without measurable height) and D (point like).

• For most of the defects it was observed that there was a tendency to undersize defect length and over size defect height.

• In order to improve the operator capability for defect characterisation and sizing, it is recommended to increase the number of defects available for the training.

C. Recommendations for TOFD training and certification

Two documents containing recommendations for TOFD training and certification have been produced (reports [7] & [8] respectively). These have taken account of existing certification schemes in the UK (administered by PCN) and The Netherlands (administered by SKO).

The documents contain:

• details of the recommended training syllabus for TOFD levels 1, 2 and 3,

• the necessary experience and pre-qualifications required, and

• a recommended format and marking scheme for the theoretical and practical parts of the examinations.

The recommendations have been transmitted to one of the technical committee for standardisation, CEN/TC 138 (NDT). They can be downloaded from the TOFDPROOF website (http://www.mpa-lifetech.de/TOFD).

21

One of the pre-requirement for the certification described in the document is that it can only be available to holders of current, valid ultrasonic weld testing certification. The certification is initially valid for a period of 5 years. The candidates are required to demonstrate that they meet the minimum supplementary training and certification requirements presented in Table 7 before they will be allowed to take TOFD examinations.

Table 7 Minimum training and experience requirements

Minimum Training Requirements

Level 1 3 Days (24 Hours)

Level 2 5 Days (40 Hours)*

Level 2 (direct) 8 Days (64 Hours)

Minimum Experience Requirements

Level 1 3 Month

Level 2 6 Months

* Contains additional modules covering interpretation of TOFD images and instruction (procedure) writing.

2.3.3.3 Comments

The number of defects provided in the ‘testing’ study is limited. They were no opposite surface discontinuities in the catalogue of defects included. From experience these defects may be difficult to identify/characterised and are often wrongly sized in both depth and length.

The tool is very limited as the A-scans are not available and the curser available may not be adequate for an accurate sizing performance. Moreover, information on the defects used for the survey is based on intended defect type and dimensions, with no confirmation from sectioning.

Although the training tool in the TOFDPROOF website is useful, it is very basic and only offers a restrictive number of defects which are relatively easy to interpret. However, it still provides a good introduction to TOFD images and data analysis; besides, the web training tool was never designed to replace more sophisticated software available with the TOFD equipment.

Training in TOFD is critical as training and experience in TOFD are essential for an appropriate application of the TOFD technique.

The recommendations for TOFD training and certification proposed under the TOFDPROOF project are similar to the BINDT ‘specific requirements for the certification of personnel engaged in ultrasonic time of flight diffraction testing of linear butt welds in ferritic steel’ which was issued in January 2002 [9]. The variations between the PCN/GEN document and the TOFDPROOF document could be due to the later issue of the TOFDPROOF document.

22

The main differences are:

• Durations for training and experience as summarised in Table 8 below. The TOFDPROOF figures presented were modified numerous times throughout the project. The new requirement figures seem more practical.

Table 8 Comparison of durations for training and experience

TOFDPROOF requirements PCN/GEN requirements

Minimum Training Requirements

Level 1 24 Hours 40 Hours

Level 2 40 Hours 40 Hours

Level 2 (direct) 64 Hours 80 Hours

Minimum Experience Requirements

Level 1 3 Months 3 Months

Level 2 6 Months 9 Months

Level 2 (direct) 9 Months 12 Months

• The syllabus presented by the TOFDPROOF project is generally less demanding for a level 1.

• Two items are allocated to the level 3 syllabus with TOFDPROOF and to the level 1 syllabus for the PCN/GEN, that is:

o TOFD on complex geometry

o The use of Synthetic Aperture Focussing Technique (SAFT)

These two items should certainly not be included into the level 1 syllabus. It is believed that for a level 1 only simple analysis and geometry should be covered. It can however be argued that these two items could have been included into the level 2 syllabus.

Additional requirements to those presented in the TOFDPROOF recommendations are being requested by training centre. Some examples are given below:

• An eye test must be passed (near vision and color) to be eligible for the PCN/SNT Levels 1 to 3 examination.

• The supervised work experience is to be obtained either before the examination or within 12 months after passing the exam.

• The TOFD Level 1 & 2 examination includes a theory paper as well as specific paper and a practical exam.

Overall, the recommendations for TOFD training and certification presented by the TOFDPROOF project provide an acceptable compromise and are considered more practical.

23

2.3.4 Work Package WP4: Acceptance criteria definition

2.3.4.1 Objectives

The TOFDPROOF objectives of this package were: to develop justified acceptance criteria; design severity levels to achieve an acceptable repair rate and integrity level and to validate the acceptance criteria on real structures.

The integrity level of a structure when using the developed TOFD acceptance criteria were planned to be equal to or better than the achieved integrity level when utilising the currently applied NDT methods.

The work package was divided into three tasks: a literature survey, the design of acceptance criteria and the validation of the acceptance criteria.

This package has the most safety issues.


a Literature survey

The first task was to carry out a literature survey to review all existing projects and studies relevant to the development of acceptance criteria for TOFD. The literature survey was carried out by TWI.

The main reviewed projects were:

• “The development of acceptance criteria for time-of-flight diffraction examination method” KINT project in Netherlands.

The concluded limits of detection provided in the TOFDPROOF final report (although it is not clear where in the source literature these figures come from) were the following:

o h=0.5 mm for thickness t=7mm,

o h=1mm for a thickness t=100mm,

o h=1.5mm for a thickness t >200mm.

• “Evaluation of acceptance criteria for the ultrasonic time-of-flight diffraction (TOFD) technique” HSE project in UK; this project included an independent assessment of KINT proposal based on TOFD responses from existing TWI data base – thickness range 10mm to 93mm. TOFD rejects a similar number of flaws to radiography according to British standards and UT according to EN standards.

• ASME B & PV Code case 2235 [10] in USA, this code case allows the replacement of radiography by ultrasonics for thickness above 12.7mm. The code case requires manufacture of a calibration block containing at least two planar defects. Acceptance criteria are tabulated and specify maximum acceptable height as a fraction of the thickness for a given range of flaw aspect (height/length ratio),

The acceptance criteria developed under TOFDPROOF are intended to be integrated into the existing European standardisation scheme (see Table 9).

24

Table 9 European standard

NDT Method NDT procedure for welds

Acceptance levels for welds

VT EN 970 EN 25817

RT EN 1435 EN 12517

UT EN 1714 EN 1712

PT EN 571-1 EN 1289

MT EN 1290 EN 1291

TOFD XPCEN/TS 14751

The European standardisation is organised in different Technical Committees (TC) entrusted to write standards. To perform NDT of welds for pressure vessels, it is in fact necessary to use the standards issued by CEN/TC 138 (NDT), CEN/TC 121 (Welds), CEN/TC 54 (Pressure vessels). CEN/TC 138 provides general requirements for the application of a given NDT method. CEN/TC 121 adds complementary instructions for testing welds giving the detailed methodology to follow for detecting the flaws and if required to perform their characterization. CEN/TC 121 proposes various testing and acceptance levels. CEN/TC 54 adds specific requirements for testing welds on pressure vessels and specifies the acceptance level to apply for pressure vessels.

The general philosophy of EN ISO 5817 [11] has been adopted.

EN 25817 [12] contains three quality levels B, C and D where B is the most stringent. Quality level C can be used for pressure vessel not submitted to fatigue, creep, and design with no reduced safety factor, otherwise quality level B shall be used.

However, the physical flaw descriptions of EN ISO 5817 do not equate to the capabilities of different NDT techniques to detect and discriminate certain types of imperfections.

So CEN/TC 121 and CEN/TC 54 have agreed that there is a fundamental difference between Quality level and Acceptance level. The former limits the size and the number of imperfections in a given weld, including their physical size, while the latter limits the size and/or number of indications in a given weld using a given NDT technique. Thus acceptance levels are established in accordance with the capability of individual techniques to detect and discriminate certain types of imperfections and to determine their size in relation to the quality requirements.

EN 12062 [13] is used as a starting point or as a method transfer function. This standard defines the interface between quality levels in EN ISO 5817 and acceptance levels of indications as per the EN standards.

25

hh

The main conclusions from the literature survey were:

• The acceptance criteria developed in the KINT project “Development of Acceptance Criteria for TOFD examination method” are a good starting point to compare the reject rate between TOFD and conventional NDT method

• The types of flaws rejected by applying TOFD acceptance criteria should be compared to those rejected by applying conventional NDT techniques.

• ASME Code Case 2235 is an important alternative and should be compared to the KINT acceptance criteria proposal.

• An acceptance criteria should be included in the European standardisation scheme.

b Design of acceptance criteria

Acceptance criteria in accordance with EN 25817 were designed and proposed tables designed by the TOFD experts of the consortium, meeting the various levels as defined in EN 25817 and EN 12062 have been produced. The proposal consists of TOFD acceptance criteria for three acceptance levels related to the quality levels according to EN 25817 (see Table 10).

Table 10 Acceptance levels

Quality level according Acceptance level to EN 25817

B (Stringent) 1

C (Intermediate) 2

D (Moderate) 3

The proposed acceptance criteria are linked to the capability of TOFD to classify defects as specified in DD CEN/TS 14751 and illustrated by Figure 9.

h

h

p

l

t

hh

h

pp

l

tt

Surface breaking discontinuity

p

p

l

p

p

p

l

ppp

ppp

l

pph

Embedded discontinuity

Figure 9 TOFD capabilities concerning defect classification.

26

For single discontinuities, the proposed acceptance criteria approved by the TOFDPROOF consortium are given in Table 11 to Table 13.

Where: l: length of an discontinuity h: height of an discontinuity dd: nominal wall thickness

Table 11 Acceptance criteria for acceptance level 1

Maximum allowable length (Lmax) Maximum allowable if h < h2 or h3. height (h1) when

Surface breaking Embedded L > Lmax

Thickness range Lmax [mm] h3 [mm] h2 [mm] h1 [mm]

6mm< dd ≤ 15mm 0.75 x dd 1.5 2 1

15mm< dd ≤ 50mm 0.75 x dd 2 3 1

50mm< dd ≤ 100mm 40 2.5 4 2

dd > 100mm 50 3 5 2



Surface breaking Embedded L > Lmax


6mm< dd ≤ 15mm dd 2 2 1

15mm< dd ≤ 50mm dd 2 4 1

50mm< dd ≤ 100mm 50 3 5 2

dd > 100mm 60 4 6 3



L > Lmax Surface breaking Embedded


ddx1.5 (max. 20) 2 2 16mm< dd ≤ 15mm

ddx1.5 (max. 60) 2.5 4.5 215mm< dd ≤ 50mm

60 4 6 350mm< dd ≤ 100mm

dd > 100mm 75 5 8 4

27

c

Rules for how to treat groups of discontinuities are also provided in the proposed acceptance criteria. These rules depend on the acceptance level.

Additionnal information on “groups of discontinuities” was presented during the TOFDPROOF seminar in Villepinte. The information is provided below:

• Indications shall be considered as a group if:

o The distance between two indications along the weld is less than the length of the longest indication.

o The distance between two indications in the thickness direction of the weld is less than the height of the highest indication.

• The sum of the lengths of the individual indications measured along the weld over a length of 12 dd shall be less or equal to:

o Level 1: 3.5 dd with a maximum of 150mm

o Level 2: 4 dd with a maximum of 200mm

o Level 3: 4.5 dd with a maximum of 250mm

• Indications that do not fulfill the requirements mentioned above should be considered as single indications.

A maximum accumulated length is given for discontinuities: 10% of the total weld length with a maximum of 500mm.

Rejection rate comparison

In addition, work has been carried out to compare the TOFD rejection rate when applying the present acceptance criteria proposal to RT and UT rejection rates when applied according to CEN standards . To this end the different NDT techniques were simulated in a probabilistic model and fracture mechanics assessments were performed for different sample geometries, loading conditions, material data and defect configuration.

In these simulations a structure with a population of defects is inspected with TOFD and other NDT techniques. Depending on the performance of the applied NDT technique and on corresponding acceptance criteria, a certain percentage of the total number of defects will be rejected. The number and types of rejectable defects largely depend upon the applied NDT technique.

By comparing the calculated reject rates and calculated failure probabilities, the acceptance criteria for TOFD were justified in such a manner that a structure inspected with TOFD would have the same safety level (or probability of failure) as a conventionally inspected structure.

The methodology used in the probabilistic comparison and the fracture mechanics assessment was based upon the results from the KINT project

With use of the ALIAS software the probability of detection (POD) and the population distribution of flaws (PDF) were computed from the available TOFDPROOF database collected during the Round Robin Trials. The parameters determined from the TOFDPROOF database were used to validate the proposed TOFD acceptance criteria.

28

It was demonstrated that the final proposal for TOFD acceptance criteria meets the stated conditions:

• A better or equal probability of failure compared to conventional NDT.

• An equal or lesser percentage of rejections for TOFD compared to conventional NDT.

In order to validate the acceptance criteria, an “on site” validation was scheduled. To complete this task, a TOFD inspection was carried out on pressure vessels at the manufacturing stage and the results were compared with conventional NDT methods. “Unfortunately (for the TOFDPROOF project) during this exercise, the quality of the welds was found to be high (by each of the NDT methods). Therefore, due to the lack of a representative population it has been concluded that it was not possible to perform a realistic “on site” acceptance criteria validation.

2.3.4.3 Comments

The acceptance levels tables (Table 11 to Table 13) for acceptance level 1 to 3, have not been modified since the last TOFDPROOF acceptance criteria draft report. The comments raised by MB on the draft, at the time, are therefore still applicable:

• The criteria assume that surface breaking cracks height measurement can be achieved with a resolution of at least ±0.5mm which is doubtful.

• It would be difficult in practice to resolve edges or measure defects accurately enough to apply the criteria proposed.

• It is doubtful whether the criteria make much sense from a structural integrity viewpoint, e.g. if one can accept 2mm x 38mm in 50mm thick weld then why not 3 x 50mm in weld thickness ≥100mm.

Discontinuities as a group are briefly mentioned in the TOFDPROOF acceptance criteria: “a maximum accumulated length is given for discontinuities: 10% of the total weld length with a maximum of 500mm”. However, it is believed by Mitsui Babcock that the maximum accumulated length of 500mm is not justified. The maximum value should only be related to the weld length as very long welds could be under investigation.

2.3.5 Work Package WP5: Economic analysis

2.3.5.1 Objectives

The objective was to carry out an economic analysis of the TOFD technique compared with conventional NDT methods. The economic analysis was carried out using the TOFD acceptance criteria proposed for each level.

The comparison was to take into account:

• The direct cost of the inspection (day or night working) based on an average rate for the inspector, the equipment, consumable, etc;

• The indirect cost of the inspection caused by for example: false calls and time to analyse the results;

29

• The cost of purchase and maintenance of the equipment and the cost of training of the personnel;

• The manufacturing disturbance caused by the NDT applied;

• The productivity gains.

2.3.5.2 Cost Comparison

The economic analysis of TOFD was carried out by TÜV and reported to the TOFDPROOF partners. The following paragraphs were extracted from the report.

Case studies

Six case studies were distributed to the partners. For each case study, direct cost and indirect cost were required for manual UT, TOFD and γ-ray inspections.

Each partner involved in the consortium was asked to supply the estimated average costs in their own country for inspection of several industrial components by TOFD, RT and UT. For this purpose representative case studies were defined:

1. Vessel of 2m diameter; thickness 36mm, 50mm, 100mm, length 10m; number of welds to be inspected: 3 circumferential welds and 2 longitudinal welds. The cost per meter of weld inspected was to be supplied for each of configuration.

2. Piping of 500mm diameter, thickness 6mm, 15mm and 36mm; number of welds to be inspected: 10 welds. The cost per circumferential weld was to be supplied for each of configuration.

The costs of a TOFD inspection was compared with the direct and indirect costs generated when using:

• Manual UT applied according EN 1714 and EN 1713 (acceptance criteria according EN 1712 [14] – level 2 + no planar imperfection accepted)

• X-rays applied according EN 1435 [3] (acceptance criteria according EN 12517 – level 1 and 2 [15] & table 6.5.3.2.1 from Pr EN 13445 [16]).

The effect of the POD and FCR on the resulting cost of an inspection was quantified using “decision trees”, taking into account the rejection rate and the probability of failure. An example of probability of failure using this method was studied. The decision tree input are given in Table 14. The calculation was carried out using a real model with cost for the inspection, repair and consequential loss of a failure. The results are given in Table 15 and illustrated by Figure .

30

Table 14 Decision tree input

POD FCR Cost [C1] TOFD 82% 11% 1000

Automated PE 84% 14% 10,000 γ-ray 60% 11% 500

Manual UT 52% 23% 250 No inspection 0% 0% 0

Probability of defect present 5% Price of repair 1K Consequential loss 500K Probability of failure if defect not 50% detected Remark Maximum of two repairs allowed * Price [C1] = Inspection price

Table 15 Comparison of decision tree results TOTAL Cost Probability of Failure

ActualActual TOFD = 1 TOFD = 1(x10-3) TOFD 3,543 1 5.1 1

Automated PE 11,287 3.2 4.9 1 γ-ray 6,279.5 1.8 11 2.2

Manual UT 8,331 2.4 16 3 No inspection 12500 3.5 25 4.9

0 ion

Pri [ ]

l

li il

2000

4000

6000

8000

10000

12000

14000

TOFD PE Meander G-Ray Manual UT No Inspect0,0E+00

5,0E-03

1,0E-02

1,5E-02

2,0E-02

2,5E-02

3,0E-02

ce C1

Tota Costs

Probabi ty of fa ure

Figure 11 Comparison of cost and probability of failure

As illustrated by Figure 11, TOFD inspection appeared to be the cheapest inspection method and not the most expensive one as it is generally assumed.

31

Conclusions from the cost comparison report

• TOFD is more competitive for high amount of inspections to be performed (per meters of welds) especially for medium and high thicknesses.

• The economic advantages of applying TOFD increase with the thickness of the material inspected.

• TOFD is even more competitive when taking into account the better performance in PoD resulting in fewer problems during in-service inspection.

• During recent years, due to the better availability of TOFD, the ever increasing speed of computing power and experience being built up, the price for inspection with TOFD has come down. Due to increased emphasis on radiation safety, the price of radiography has still increasing direct and indirect costs. These factors have pushed the break-even point in relation to the material thickness even further down.

• If all the factors are taken into account, it can be shown that TOFD can be cheaper than conventional NDT methods.

2.3.5.3 Comments on the cost comparison report

It is not clear to what extend the report took into account the results of the case studies.

Although most of the comments presented in the report (summarised in section 2.3.5.2 above) are to some extent true, the cost of TOFD is still calculated in respect to the company facility and expertise. Most companies would not be specialised in TOFD inspection and would therefore perform the inspection at higher cost to cover for new equipment and specific transducers. Specialised TOFD companies, however, would have an advantage as they would already own the TOFD equipment with a variety of probes and scanners as well as have appropriate knowledge and experience and would therefore be able to provide inspection at a lower cost. The case studies certainly highlight this point. Most of the companies who participated in the case studies provided higher total cost for TOFD than conventional UT, with X-ray being the most costly inspection out of the three inspection techniques. The lower cost for the TOFD inspection was provided by a specialised company.

However, even if the TOFD technique is flexible, it is more suitable for smooth surfaces with straight forward geometry. Rough or uneven surfaces make it difficult to maintain both probes on the scanning surface and can generate poor data quality. Difficult geometry can be scanned but the interpretation of the data becomes more complicated and may require complicated software.

The preparation time for the TOFD technique should not be much different to the preparation time needed for the manual inspection.

32

2.3.6 Work Package WP6: Exploitation and dissemination

2.3.6.1 Objectives

The objectives were to:

• Supply the results obtained, the proposed guidelines and the proposed acceptance criteria to all parties involved in NDT inspection in the European Union;

• Get feedback through specific seminars, publication of papers and Web site, in order to obtain the widest consensus on the guidelines and proposed acceptance criteria;

• Convey to the relevant CEN Technical Committees the revised guidelines and proposal of acceptance criteria taking into account the observations received.

The main deliverables of the programme which required exploitation were: the acceptance criteria, the analysis of TOFD performances, guidelines for TOFD interpretation (CD-ROM), recommendations for training and certification and recommendations for TOFD application.

The results were transmitted to CEN and related European networks. Some members of the consortium were involved in National, European and International standardisation activity. Three partners were member of the CEN/TC 121 ad hoc group dealing with TOFD standardisation for welds.


• Specific National (Lisbon, 03/2005) and International workshops (Paris, 04/2005) were organised both having attracted numerous participants from all Europe (about 70 attended the Paris workshop).

• The TOFDPROOF project specific Web site hosted by MPA (http://www.mpa-lifetech.de/tofd) was created.

• Maintaining permanent link with CEN - TOFD ad hoc group CEN/TC 121 through several project partners and permanent observers.

o RIMAP and FITNET networks respective chairmen were kept informed about TOFDPROOF progress (Annual progress reports).

o Dr Hecht, chairman of the CEN/TC 121 TOFD ad “hoc group”, Dr Ewert (BAM) and Dr E. Zeelenberg (Lloyd Register DK) were permanent Observers.

• CEN/TC 121 adopted WI 00121377 “Welding – Use of time-of-flight diffraction technique (TOFD) for testing of welds as a Technical Specification” including parts of the TOFD procedure written in task 2.1.

• Various publication have been presented:

5

o ”In-Service Inspection and Life Management of Pressure Equipment”, TOFDPROOF project presentation at the EPERC workshop, Didier Flotté,

th October 2001;

33

o “Treatment of Uncertainties in Determination of Acceptance Criteria for Ultrasonic Testing”, by A. Jovanovic, D, Balos, MPA Stuttgart, R. Kauer, TÜV Süddeutschland München Paper presented at the UNCERT-AM conference on Management of Uncertainties in Mechanical Testing & Inspection on October 8, 2003, Stuttgart, Germany;

o Communication on TOFDPROOF objectives and achievements at the International Conference on Pressure Vessels – ESOPE 2004 – Paris, October 2004, D. Flotté, D. Chauveau and al.

o TOFDPROOF workshop, 14th March 2005, Lisbon, Portugal, organised by ISQ.

o Communication at the national COFREND conference (May 2005) "Description and results of the European project TOFDPROOF", D. Flotté, D. Chauveau - Institut de Soudure

o TOFDPROOF workshop, 27th May 2005, Villepinte, France, organised by IS and all partners.

TOFD acceptance criteria was on the agenda of the CEN/TC 121 Berlin meeting in May 2005. The previous year, the Dutch Normalisation Committee submitted the Dutch National standard NEN 1822 (Acceptance Criteria for the TOFD technique). However, the document prepared under the TOFDPROOF project was more in line with the latest developments and with the other European Standards. It was suggested by Jan Verkooijen (Sonovation), that the TOFDPROOF document should be used as an European Standard as a starting point.

The Acceptance Standard is now accepted as a new Work Item. Five countries are participating in the work, that is: France, Germany, Netherlands, United Kingdom and Finland.

This proposal is the starting point for the CEN standardisation. The first meeting of the ad hoc group was scheduled in autumn 2005.

2.3.6.3 Comments

Two out of the main deliverables of the programme were not disseminated as public documents through the TOFDPROOF web site, that is:

• Analysis of performances and reproducibility of TOFD inspection with conventional NDT

• Design and proposal of acceptance criteria with several severity levels.

The information from the TOFDPROOF project was fed to a number of TC groups and it is clear that the results from the TOFDPROOF project were taken into consideration. Draft reports were submitted as references for new EU standards. Work still needs to be carried out to finalise these documents.

34

2.3.7 Work Package WP7: Data storage, analysis and exchange

2.3.7.1 Objectives

This work package concentrates on exchanging information between the partners (reports, minutes from meetings, etc.) through the web site created under this work package. The web site also enables access to the data collected during the project for analysis.

A wide collection of data was to be analysed by specific tools developed within the work package in order to highlight weaknesses and strengths of the TOFD inspection.


This work package concentrates on exchanging information between the partners including reports, minutes from meetings, etc. the exchange was carried out through the web site created under this work package: http://www.mpa-lifetech.de/tofd/. The web site also enables access to the data collected during the project for analysis once the data collection was completed.

The existing tools (MPA’s ALIAS system) were used as initial platform for results analysis. Based on the literature overview and on the results of similar projects, it was decided to perform the hit-miss data analysis using the Weibull three parameter cumulative distributions (as suggested by KINT report).

It was concluded by the work package leader that for the purpose of data analysis and development of acceptance criteria, the applied data analysis methods gave more than reasonable and feasible results, comparable with similar results obtained in other related projects.

Furthermore, the TOFD training tool was made available through the TOFDPROOF web site to the general public free of charge. In order to have a better overview of the general public interest, it was decided by the consortium to require registration for the use of the training tool. After approximately 3 months after the training tool was made available, the number of registered users has reached 50, ranging mostly from European countries, nevertheless participants from country such as US, Canada, China and India have also been registered, thus showing that the interest for the project results is not only limited to EU countries.

2.3.7.3 Comments

Although the website was useful to access information from the project, the information was fed to the website very slowly and some documents were not made available until the very end of the project (well after the round robin trials were completed). Some of the newer issued reports are still not available through the site, at the time of writing this current report.

35

3. SAFETY RELATED ISSUES

3.1 INTRODUCTION

Mitsui Babcock previously reported significant limitations of the TOFD technique in a previous project “Critical evaluation of TOFD for search scanning” carried out for the HSE. Mitsui Babcock highlighted situations where TOFD limited for detection and it was concluded that, even if TOFD is a good complementary NDT technique for defect detection, there are circumstances where it should not be used as a standalone search method.

There is a tendency to favour TOFD as a replacement to more traditional NDT methods, especially radiography (e.g. Code case 2235 of the ASME Boiler and Pressure Vessel Code). One of the main disadvantages of radiography is the potential hazard to health associated with the ionising radiations which are the basis of the method. Ultrasonic inspection does not have any significant inherent safety issues and therefore can be more attractive to apply than radiography. However, it is important that the associated safety and economic advantages are not gained at the expense of reduced confidence in weld integrity.

Although it should be possible to minimise the deficiencies of TOFD using optimised parameter settings it might not always be considered as the inspection could become costly.

When considering TOFD as a stand-alone technique for the inspection of safety critical components, the following limitations should be kept in mind:

• The diffracted tip wave is relatively small in amplitude so the sensitivity of the NDT needs to be high which can then lead to false calls;

• Other techniques should be applied to cover the near and rear surface regions.

Near surface defects could be missed due to the presence of the lateral wave. If such defects are of concern, then additional techniques should be performed e.g., ultrasonic pulse echo using shear wave at full skip, especially to investigate the cap area of an undressed weld.

Rear surface defects could be missed due to the presence of the backwall echo. This is even more likely when the root bead is still present. It was demonstrated during previous research (including previous work for the HSE [17]) and during the TOFDPROOF project that defects up to 4mm height could be missed when using the TOFD technique. Due to the uncertainty of defect detection in the weld root, it was concluded that complementary methods should be use.

• As the weld thickness increases so does the number of probe separations which are required to cover the inspection volume.

• The technique requires optimization for the defects of concern.

• Skilled operators are required to operate the equipment and interpret the images.

36

• Resolution between the defect tips may be difficult to achieve which can lead to misinterpretation of defect.

The results from the TOFDPROOF project round robin trials confirm some of the TOFD limitations and highlight the need for an appropriate procedure, a skilled data analyst and for realistic acceptance criteria.

3.2 ACCEPTANCE CRITERIA

An acceptance criteria document is clearly required if TOFD is to be used as a standalone inspection method. Carefully specified acceptance criteria are required to ensure component integrity without unnecessary rejection e.g. due to innocuous defects or false calls.

The application of the TOFD standards (BS 7706:1993 and DD ENV 583-6:2000) has been slowed down by the lack of appropriate defect acceptance criteria. The need of such criteria is apparent.

The criteria prepared under the TOFDPROOF project have now been presented to CEN/TC121/SC5B and CEN/TC54/WGE for adoption as a work item in order to integrate them in EN standards.

The acceptance criteria document presented is considered to provide some unrealistic targets as it assumes a very high capability of detection for search scanning which is not always achievable mainly at the “dead zone” area. Small surface breaking defects may be just detectable or/and wrongly sized.

The results from the TOFDPROOF trials indicate that the height of upper surface defects is difficult to determine and in many cases these are inaccurately sized. Where the apparent extent of the lateral wave is greater than or equal to the acceptable height of a surface defect, it may be appropriate to regard all detected upper surface defects as possibly rejectable regardless of their measured height and then use a complementary technique to determine acceptability.

The accuracy of TOFD depends on defect location, probe arrangement and specimen geometry. The rule of thumb which Mitsui Babcock are aware of and which was concluded from independent trials is that an accuracy of 2% of the specimen thickness is obtainable for dedicated sizing scans and 3% for search scans under ideal conditions.

Whilst this may be the case, errors previously calculated by Mitsui Babcock (shown in Appendix 2) are less optimistic. Taking the example in Appendix 2, a 3mm high backwall defect in a 15mm thick specimen. Adding up the errors, as shown in Table 1 of Appendix 2, provides a worst-case estimate. In practice worst-case errors for all aspects at the same time are unlikely so that there is a reasonable prospect of justifying an error tolerance of around 2mm.

Acceptance levels should be established in accordance with the capability of individual techniques to detect and discriminate certain types of imperfections and to determine their size in relation to the quality requirements. However, the proposed acceptance levels, do not differentiation between defect types. TOFD can not easily differentiate between planar and non planar embedded defects. In practice resolving defect tips may not always be achievable and small cracks could be misinterpreted

37

as slag. Similarly, slag has been reported as embedded linear defects when diffracted signals were observed from the upper and lower edges of the indications.

It is therefore important to consider the maximum allowance height and identify whether it as an achievable value. It is why Mitsui Babcock have reservations over the acceptance level criteria proposed under the TOFDPROOF project. The reservations are mainly over the following points:

• The criteria assume that a surface breaking flaw height measurement can be achieved with a resolution of at least ±0.5mm;

• The criteria assume that surface breaking flaw can be resolved from the lateral/backwall echo with a resolution as little as 1.5mm (for thin samples).

• The criteria assume that embedded defect edges can be resolved within 2mm.

• The criteria do not appear to make much sense from a structural integrity viewpoint, e.g. if one can accept 2mm x 38mm in 50mm thick weld then why not 3mm x 50mm in weld thickness ≥100mm.

• According to the tables, the maximum allowable height when the length of the defect exceeds the thickness of the specimen is 1mm for specimens up to 50mm thick for acceptance level 1 and 2. Since it is not considered possible to measure height less than 1mm by TOFD, this implies that most of the defects detected (if not all) would be rejected for welds of thickness below 50mm if defect length exceeds the wall thickness.

Moreover, the criteria presented under the TOFDPROOF project are for in-manufacture inspections, however there were discussions on extending their applications to in-service inspection. The criteria would be even more inappropriate for in-service inspection as factors such as surface finish/irregularity, accessibility, mismatch of parent material and environment would have even more influence on the image quality achievable. Further, the TOFD technique is often used as a monitoring technique for defects previously detected and analysed by another NDT technique using the relevant acceptance criteria. These defects can be in some circumstances much larger than those specified in the acceptance criteria proposed.

38

4. CONCLUSIONS

Mitsui Babcock was a partner in the European Collaborative Project “TOFDPROOF”, which was intended to provide guidance on the scope and capability of TOFD procedures and acceptance criteria for the manufacturing inspection of pressure vessels. Since the conclusions and recommendations could influence the approach adopted in the UK, Mitsui Babcock has been reviewing the safety implications of TOFDPROOF for the HSE. The work has involved keeping the HSE aware of TOFDPROOF progress, reviewing TOFDPROOF results and recommendations which could impact on component safety, and providing feedback to the TOFDPROOF consortium on HSE/Mitsui Babcock views.

The TOFDPROOF project was aimed at promoting TOFD and to create documents which could be used as references for future standards. The project achieved its main goal, the results of the project being communicated to parties involved in NDT inspection and a series of documents produced during the project were used as a base or contribution for recent and future European Standards.

However, it is not clear how/if some of the conclusions were directly produced from the data collected during the project and how much the results from traditional NDT were used to compare the capability of the TOFD. Nevertheless, the TOFDPROOF project has increased awareness of TOFD capabilities and limitations and provides useful guidelines on the technique.

It is likely that TOFD will continue to be used as a stand alone search technique. However, the TOFDPROOF project highlighted the danger of missing defects, in particular in the “dead zones”, or misinterpreting the results.

The technique suffers from limited coverage resulting from two inspection dead zones: the first dead zone at the near surface resulting from the lateral wave and the second at the backwall resulting from the width of the backwall reflection. Coverage of the whole inspection zone must be achieved. Specific investigations in these areas should be applied. The best approach could be to use the TOFD technique combined with the pulse echo technique in order to eliminate the uncertainty of a lack of coverage.

Moreover, if indications are detected, an additional technique should be applied to allow for defect characterisation, as a misinterpretation of a defect could lead to wrongly accepting the defect.

The project showed that missing a defect may not only be due to the capability of the technique but can be linked to the use of the wrong/inappropriate TOFD setups or/and poor analysis. The project highlighted the need for specific and full training and experience.

TOFD can be regarded as a cheap technique mainly because it appears to many as a “one scan” inspection, though this can be misleading as different probe types and arrangements should be used to enable minimise lack of coverage (additional scans may also be needed to investigate suspicious indications). The reduction of recommended scans should not be justified on the basic of cost reduction if this reduces component integrity for safety critical components. The result of the survey carried out on the cost comparison of the NDT techniques still identified TOFD as the cheapest technique (for those specialising in the TOFD technique).

39

The main area of concern, however, was considered to be the proposed acceptance criteria. Until now, one of the most recognised needs for the TOFD technique was acceptance criteria. The final issue of the TOFDPROOF acceptance criteria document has recently been accepted as a starting point for CEN standardisation. The document is considered to present some unrealistic targets as it assumes a very high capability of detection for search scanning which is not always achievable mainly at the “dead zone” area. In general, It would be difficult in practice to resolve edges or measure defects accurately enough to apply the criteria proposed and it is doubtful whether the criteria make much sense from a structural integrity viewpoint, e.g. it accepts 2mm x 38mm in 50mm thick weld but not 3 x 50mm in weld thickness ≥100mm.

40

5. REFERENCES

The documents listed below are directly referred to within this report.

[1]. BS EN 1714:1998 – Non-destructive testing of welded joints – Ultrasonic testing of welded joints.

[2]. BS EN 1713:1998 – Non-destructive testing of welds – Ultrasonic testing – Characterization of indications in welds.

[3]. BS EN 1435:1997 – Non-destructive examination of welds – Radiographic examination of welded joints.

[4]. DD ENV 583-6:2000 - Non-destructive testing – ultrasonic examination – part 6: time-of-flight diffraction technique as a method for detection and sizing of discontinuities.

[5]. NS Goujon & BW Kenzie, “Recommendations for Applying TOFD (Field of Application, Strengths & Weaknesses)”, Issue 1, Report No. 2-31-D-2004-02-1, report for the TOFDPROOF project, January 2005.

[6]. DD CEN/TS 14751: 2004 - Welding – Use of time-of-flight diffraction technique (TOFD) for examination of welds, technique specification.

[7]. BW Kenzie, ‘Recommendations for TOFD Certification’, issue 2, TWI report for the TOFDPROOF project, March 2005.

[8]. BW Kenzie, ‘Recommendations for TOFD training’, issue 1, TWI report for the TOFDPROOF project, February 2005.

[9]. “Specific requirements for the certification of personnel engaged in ultrasonic time of flight diffraction testing of linear butt welds in ferritic steel”, PCN/GEN Appendix C4.1 Issue 1 Rev B, 25 January 2002.

[10]. Cases of ASME Boiler and Pressure Vessel Code – Code Case 2235-4: Use of Ultrasonic Examination in Lieu of Radiography, Section I and Section VIII, Divisions 1 and 2.

[11]. BS EN ISO 5817:2003 – Welding – Fusion –welded joints in steel, nickel, titanium and their alloys (beam welding excluded) - Quality levels for imperfections.

[12]. BS EN 25817:1992 – Arc-welded joints in steel – Guidance on quality levels for imperfections (superseded).

[13]. BS EN 12062:1998 – Non-destructive examination of welds – General rules for metallic materials.

[14]. BS EN 1712:1997 – Non-destructive testing of weld – Ultrasonic testing of welded joints – Acceptance levels.

[15]. BS EN 12517:1998 – Non-destructive testing of welds – Radiographic testing of welded joints – Acceptance levels.

41

[16]. Pr EN 13445 – Unfired pressure vessels – Inspection and testing.

[17]. NS Goujon & BWO. Shepherd ‘Critical evaluation of TOFD for search scanning’, Report No71/00/037, 13 December 2004 (HSE research report).

42

APPENDICES

43

APPENDIX 1: RECOMMENDATIONS FOR APPLYING TOFD (FIELD OF APPLICATION, STRENGTHS, WEAKNESSES)

44

TOFDPROOF

Document Name: RECOMMENDATIONS FOR APPLYING TOFD (FIELD OF APPLICATION, STRENGTHS, WEAKNESSES)

Document Date: 2005-01-11 Document Owner Mitsui Babcock / TWI Document Author/s: NS Goujon & BW Kenzie Document approved by: Task / Deliverable Number: WP2 / 31 Issue: 1 Status Document Reference Number: 2-31-D-2005-02-1 TOFDPROOF project n° G6RD-CT-2001-00626

Institut de Soudure F IS Service F Sonovation NL TWI Limited UK Mitsui Babcock Technology Centre UK Staatliche Materialprüfungsanstalt Stuttgart D Tecnatom S. A SP VTT FIN Instituto de Soldadura e Qualidade PT TÜV Süddeutschland Bau und Betrieb GmbH D

© COPYRIGHT 2005 The TOFDPROOF Consortium

This document may not be copied, reproduced, or modified in whole or in part for any purpose without written permission from the TOFDPROOF Consortium. In addition, to such written permission to copy, acknowledgement of the authors of the document and all applicable portions of the copyright notice must be clearly referenced.

All rights reserved

45

CONTENTS

CONTENTS ........................................................................................................................................................ 46

1. INTRODUCTION ................................................................................................................................. 47

2. RECOMMENDATIONS BASED ON THE TOFDPROOF RRT RESULTS ................................. 472.1 Procedure ........................................................................................................................................ 472.2 Identification of reference marks .................................................................................................... 482.3 Set-up.............................................................................................................................................. 482.4 Classification of indications............................................................................................................ 502.5 Evaluation of indications ................................................................................................................ 512.6 Personnel qualification ................................................................................................................... 562.7 Acceptance criteria ......................................................................................................................... 56

3. ADDITIONAL RECOMMENDATIONS............................................................................................ 57

4. CONCLUSIONS.................................................................................................................................... 57

5. REFERENCES ...................................................................................................................................... 58

46

1. INTRODUCTION

Following the round robin trial (RRT) exercise and the reporting of the results, the data were collated and a review focusing on the causes of discrepancies in the results was carried out by MBEL and TWI. This was in order to highlight the strengths and weaknesses of the TOFD technique. Report No 2-18-D-2004-01-1 [1] presents the discrepancy analysis of the round robin trial results.

In order to complete work package WP2, recommendations were written to explain when a complementary NDT technique is recommended. This report proposes recommendations for applying TOFD. The recommendations are based on the TOFDPROOF project round robin results and the difficulties identified during the study. Additional considerations must be taken into account when different material and component geometry are under study. These recommendations will be transmitted to CEN TC121, TC54, TC138 and EPERC.

2. RECOMMENDATIONS BASED ON THE TOFDPROOF RRT RESULTS

The following recommendations are based on the TOFDPROOF project round robin trials results and the discrepancy analysis report No 2-18-D-2004-01-1. They are provided for seven categories, including:

• Procedure; • Identification of reference marks; • Set-up; • Classification of indications; • Evaluation of indications; • Personnel qualification; • Acceptance criteria.

For each group, comments and recommendations are provided.

2.1 Procedure

• A specific procedure shall be written in accordance with the guidelines given in ENV 583-6 [2] and PrCEN/TS 14751 [3] for each individual type of inspection.

• The TOFD procedure written under the TOFDPROOF project ‘Procedure for TOFD Inspection of Welds used for the Round Robin Trials report No. 2-21-Q-2002-01-4 [4] can be used as an example.

• The procedure shall include the following information:

SCOPE

GENERAL REQUIREMENTS o Description of the component o Restrictions o Surface preparation o Weld profile o Coverage o Possible defects o Inspection conditions (temperature, lighting, etc.) o Examination level

47

REFERENCE DOCUMENTS

PERSONNEL REQUIREMENTS & QUALIFICATION

EQUIPMENT REQUIREMENTS o General requirements o Equipment o Scanning mechanism o Probes o Reference block o Couplant IDENTIFICATION OF REFERENCE MARKS

CALIBRATION & SETTINGS o Choice of probes and Probe Centre Separation (PCS) o Sensitivity setting o Time calibration o Time window o Scan resolution setting o Verification of the setting

WELD INSPECTION o longitudinal defects o transverse defects DATA ANALYSIS o Interpretation and analysis of TOFD images o Assessing the quality of the TOFD image o Classification and evaluation of indications DATA STORAGE

REPORTING

2.2 Identification of reference marks

The identification of reference marks (including datum on the component and reference point on the inspection probes array) is critical to allow repeatability of the inspection and results comparison.

The reference point on the probe array (usually the middle of the back face of one of the probes) shall be clearly defined in the procedure and the reference marks on the component should be clearly visible.

2.3 Set-up

• Care should be taken to choose appropriate combinations of parameters.

• The capability to cover the thickness range of interest must be demonstrated on a reference block.

• Sensitivity setting

Setting of an adequate sensitivity is essential to enable the detection of weak diffracted signals and at the same time avoiding overloading the system with non-relevant signals. The inspection teams must make sure that the lateral wave and the BWE is not saturated to investigate for possible surface breaking defects.

• Selection of probes and probe configuration

48

- Always use the most suitable probes for the component and for the type of defects under investigation. The choice of the type of probes (including: frequency, crystal diameter and angle) to be used for an inspection can be critical (especially for defect characterisation). Selection of probes and probe configuration for full coverage of the complete weld thickness should follow the recommendations provided in Table 1. The values given in Table 1 relate to the reviewed values used for the TOFDPROOF project round robin trial.

Thickness t (mm)

Minimum number of

TOFD set-up(s)*

Depth-range (mm)

Frequency (MHz)

Beam angle (°)

Crystal size (mm)

Beam intersection

6-10 1 0-t 15 70 2-3 2/3t >10-15 1 0-t 15-10 70 2-3 2/3t >15-35 1 0-t 10-5 70-60 2-6 2/3t >35-50 1 0-t 5-3.5 70-60 3-6 2/3t >50-100 2 0-t/2 5-3.5 70-60 3-6 1/3t

t/2-t 5-3.5 60-45 6-12 5/6t for 60° or t for 45º

* Note that the number of TOFD set-up(s) given in Table 1 is the minimum number of TOFD set-up(s) recommended.

Table 1: probes set-up versus thickness

- Other probe types and configuration than those given in Table 1 can be used after demonstration on an appropriate calibration/reference block, see Appendix 1 of reference [4].

- The probe frequency used has to be high enough to achieve the best possible resolution taking care of achieving the required sensitivity setting.

- High frequency, small crystal diameter probes (15MHz, 3mm) are preferable for the inspection of thin samples (up to 15mm) especially when the weld surfaces are as-welded. An alternative choice (e.g. 10MHz) may not be appropriate.

- The probe frequency used has to be high enough to achieve the best possible resolution. However frequencies at the lower end of the bands defined in Table 1 may be used if the required sensitivity setting cannot be achieved with higher frequencies.

• Consistency in the inspection method has to be adopted. It is essential that the inspection team follows the procedure provided, or provides a justification for any variation between the work carried out and the procedure.

• TOFD images are commonly represented by a grey scale. This is to allow for better contrast and to permit the identification of indications. It is recommended that this approach be used to allow for repeatability and consistency between inspection teams (although other approaches can be used as long as they are well understood).

49

• The time window for data collection shall be extended to at least 1µs beyond the first mode converted BWE, in order to study possible defect mode converted echoes. Note CEN/TS 14751 states that the time window ‘shall at least cover the depth range covered in Table 1’, however, defect information may be provided by mode converted echoes (e.g. transverse cracks) and therefore it is important to extend the window to allow collection of mode converted echoes, when appropriate.

• Additional scanning

- For wide welds (especially for as-welded and double-V weld preparation), at least two offset scans must be considered to achieve the whole weld body inspection coverage, one at each side of the weld centre line.

- A separate root scan should be considered. The resolution of the backwall echo and the defect tip of a small root defect is difficult, especially when the root bead is still present. The presence of a weld root can prevent a clear break of the back-wall (with the exception of the larger defects). Indication of the presence of defects may be given by the waves arriving after the back-wall echo. However, small defects can still be missed, or misinterpreted as point reflectors.

- Inspection from both surfaces is recommended when access permits. This is especially important for thick as-welded components where satisfactory coverage of the near surface is not achievable. When inspection from the internal surface is not possible, it is recommended to use a complementary NDT technique.

- Inspection for transverse indications can be limited especially when the weld cap is present. The normal TOFD configuration is not optimised for transverse defect inspection. When transverse defects are expected and the weld is as-welded, additional NDT technique(s) should be used.

2.4 Classification of indications

The indications shall be classified into categories clearly defined in the inspection procedure. The following categories are recommended:

• Surface breaking indication (at scanning surface, at opposite surface and 100% through-wall)

Surface breaking at the scanning surface: this type of indication shows up as either a weakening, deviation or loss of lateral wave (not always observed) and an elongated pattern generated by the signal from the lower edge of the indication. The lower edge can be hidden by the lateral wave, but generally a pattern can be observed in the mode-converted part of the image. For small indications, only a slight shift of the lateral wave towards longer time-of-flight may be observed.

Surface breaking at the opposite surface: this type of indication shows up as either a weakening, deviation or a loss of the backwall signal (not always observed) and an elongated pattern generated by the signal emitted from the upper edge of the indication.

Through-wall indication: this type of indication shows up as a loss or weakening of both the lateral wave and the backwall signal.

50

• Embedded indications (point-like, elongated with a measurable height or without a measurable height)

Embedded point-like indication: the most common pattern characterised by a single arc shaped curve fitting the theoretical hyperbolic curve corresponding to the depth of the indication. This pattern is mostly produced by a pore, but it can also be generated by the edges of a transverse crack.

Embedded elongated indication with non measurable height: the indication appears as an elongated pattern corresponding to an apparent upper edge signal (approximately in phase with the backwall).

Embedded elongated indication with a measurable height: the indication appears as two separate elongated patterns located at different positions in depth, corresponding to the upper and lower edges of the indication.

• Transverse indications

Transverse indication: can be surface breaking or embedded. The signal from the upper and lower edges of a transverse crack may appear as a point-like defect.

• Uncategorised

Uncategorised indications: all indications that cannot be properly classified into one of the above categories.

2.5 Evaluation of indications

Any feature, which is not due to geometry and appears as an indication on the TOFD image, shall be investigated to the extent that it can be evaluated in terms of acceptance criteria.

2.5.1 Interpretation of TOFD images

Initial analysis has to be carried out on unprocessed data. Straightening and removal (for lateral wave and BWE) tools can be use for subsequent analysis e.g. confirmation of presence/absence of surface defects.

Surface defects

It is well known that one of the limitations of the TOFD technique is the surface inspection (upper and rear surfaces). The presence of the lateral wave and the backwall echo restrict the inspection zone. Small defects in these zones can be missed. Surface defects are more difficult to detect especially when the weld is in the as-welded state. Sizing errors are also more likely to occur.

Care must be taken when surface indications are observed. Additionnal scanning may be needed with more appropriate probes type and arrangements.

The RRT results indicate that the height of near surface defects is difficult to determine and in many cases these are inaccurately sized. Where the apparent extent of the lateral wave is greater than or equal to the acceptable height of a surface defect, it may be appropriate to consider all detected upper surface defects as rejectable regardless of their measured height and confirm results with another NDT technique.

Embedded defects

51

A number of defects can be wrongly reported as linear if the resolution of the defect tips cannot be achieved. The probe frequency used has to be high enough to achieve the required resolution.

Where a large number of point-like indications have been detected that creates a cluster of indications, that could mask the presence of a more serious defect, the inspection should be supported by another NDT technique.

Transverse defects

The presence of some mode converted echoes associated with a point like indication may suggest that transverse defects could be present. The normal TOFD configuration is not optimised for transverse defect inspection. When transverse defects are expected or/and when indications on the TOFD image suggest the presence of such defects (especially if the weld is as-welded), additional NDT technique(s) should be used.

2.5.2 Determination of indications dimensions

Determination of length

Indications with a length equal to or less than the probes beam width will appear as a single hyperbolic shaped arc (point-like discontinuity).

For elongated discontinuities with or without a measurable height, depending upon the type of indication, a technique for length sizing shall be selected from the following:

• Length sizing of linear indications:

This type of indication does not have length measurement characteristics which change significantly in the through-wall direction, i.e. embedded defects like slag and lack of fusion.

A hyperbolic cursor, shaped to fit the arc produced by a point-like flaw, is fitted to the indication. The cursor is fitted at both extremities of the indication and the difference between the measured positions of the turning points on the cursors provides the length of the indication, see Figure 1.

52

Figure 1: Length sizing by fitting arc-shaped cursors

If the hyperbolic cursors do not fit the extremities of the indication, the 6 dB drop method shall be used. The maximum amplitude (where the reflector extends across the full width of the ultrasonic beam) shall be determined using the cursor. The extremities of the indication shall be identified where the amplitude provided by the cursor has fallen by half, see Figure 2.

Figure 2: Length sizing by the 6 dB drop method.

53

• Length sizing of extended parabola-like indications:

This type of indication has length measurement characteristics which change significantly in the through-wall direction, e.g. surface breaking defects like cracks.

A cursor, shaped to fit the arc produced by a point-like flaw, is positioned at either end of the indication at a time delay of one third of the indication penetration. The distance moved between the cursor positions at each end of the indication is taken to represent the length of the indication, see Figure 3.

Figure 3: length sizing of extended parabola-like indications.

Determination of depth and height

The depth and the height of the indication shall be determined as follows:

• Assuming that the ultrasonic energy enters and leaves the specimen at the probes index points and that the discontinuity is mid-way between the two probes, the depth of the defect can be given by:

d = [¼ c2(t - to)² - S² ] 1/2

Where: c is the ultrasonic velocity t is the transit time to is the total time delay in the probe shoes d is the depth of the tip of the discontinuity S is the mid-distance between the ultrasonic probes index points

• To prevent errors that may arise from the estimation of probe delay and probe centre separation distance, the depth d shall be calculated, where possible, from the time of flight differences, ∆T, between the lateral wave and the diffracted pulse or between the backwall echo and the diffracted pulse. Moreover, in order to reduce the error related to time measurement, the measurement shall be done

54

from the A-scan and by choosing a consistent position on the waveforms. It is recommended to use one of the methods described below (see Figure 4).

Method 1: by measuring the transit time to the rising signal.

Method 2: by measuring the transit time to the first maximum.

Method 3: by measuring the transit time to the peak amplitude.

Positi i i

ll or ip

l

ons for measur ng the trans t time

Backwadefect upper-t

Latera wave or defect lower-tip

Method 1 Method 2 Method 3

Figure 4: Position of the cursor for time measurement – Methods 1, 2 & 3.

• All depth measurements should be performed after straightening of the position of the lateral wave or the backwall echo.

• The height of a near surface breaking discontinuity is determined by measuring the distance between the near surface and the lower-tip diffraction signal from the indication.

• The height of a rear surface breaking discontinuity is determined by measuring the distance between the rear surface and the upper-tip diffraction signal from the indication.

• The height of an embedded discontinuity is determined by the difference in depth between the upper-tip and lower-tip diffraction. For indications displaying varying depth along their length, the height is determined at the position along the length of the discontinuity where the difference is the greatest.

• Other measurement methods such as those proposed by ENV 583 part 6 can be used as long as a justification is provided.

55

2.5.3 Sizing errors

Errors in reported height and length measurements for the RRT were reported.

• In general, the discrepancies in length measurement related to the intended values were significant. The standard deviation of the errors in the reported length measurements for the RRT was 11.7mm. The errors may have been partly related to differences between the intended values and the real values. However, sizing errors also varied between inspection teams. On this basis, length measurement from TOFD techniques should therefore be treated with caution.

• The variations in reported height measurement may have been related to the teams choice of the variables used to linearise the TOFD results (such as: the reference time to the lateral wave, the reference time to the backwall echo, the velocity and component thickness). The standard deviation of the errors in the reported height measurements for the RRT was 2.0mm. It is important to use a defined measurement technique.

The sizing method used to determine the defect dimensions (height and length) should be clearly defined in the inspection procedure. This is in order to provide a repeatable measurement technique and to allow comparison between inspection teams and repeat inspections. The measurement techniques used for the calibration and on the actual component should be consistent.

2.6 Personnel qualification

• As the detection and sizing performance were highly dependant on the inspection team, it was concluded that the training and experience of the inspection personnel is critical.

• In addition to a general knowledge of ultrasonic weld inspection, all key personnel should be experienced in TOFD inspections.

• At least one of the inspection personnel should be familiar with preparation of written test instructions, final off-line analysis of data and be qualified to approve the final inspection report. This inspection personnel should be certified as a

]minimum to level 2 in accordance with EN 473 [5 or equivalent in ultrasonic testing for the relevant industrial sector.

• In cases where the above minimum qualifications are not considered adequate, job-specific training should be carried out.

2.7 Acceptance criteria

Carefully specified acceptance criteria are required to ensure component integrity without unnecessary rejection e.g. due to innocuous defects or false calls.

56

3. ADDITIONAL RECOMMENDATIONS

Some TOFD technique limitations and recommendations were identified under ]another project [6 :

• A better response is obtained when the included angle between the probes is 120°. Experimental results confirms the theory.

• Offset-scans, that is, scans parallel to the weld-axis, where the beam intersection point is not on the centre-line of the weld, should be carried out (especially for thick X-shaped welds to ensure detection of toe cracks at the surface opposite the scanning surface). Omitting offset-scans could lead to depth position errors, e.g. indications will tend to be plotted deeper than their true through-wall location.

• The operators must ensure the proper coverage of the area of interest. TOFD can be limited by the geometry of the sample or by an obstruction limiting the scanning area. For example: at the ends of long seams adjacent to circumferential seams (require grinding); inspection of mismatch pipe to pipe weld and material of small wall thickness t such as t ≤10mm thick. When the required coverage is not achieved by TOFD, additional NDT techniques are required to complete the inspection.

• Existence of a dead zone of the order of 2-3mm below the scanning surface. This problem also occurs at the back-wall but the extent of dead zone may vary. TOFD is not reliable for detecting surface defects of height less than 4mm. Experimental results showed that root defects with a depth of less than 4mm are easily missed or misinterpreted. Moreover, the difficulty of detection increases with the defect offset position relative to the weld centre line.

4. CONCLUSIONS

This document provides recommendations for applying TOFD.

The recommendations are provided for seven categories, including: procedure, identification of reference marks, set-up, classification of indications, evaluation of indications, personnel qualification and acceptance criteria. For each group, comments and recommendations are provided.

The results from the TOFDPROOF project round robin trials confirm some of the TOFD limitations and highlight the need for an appropriate procedure, a skilled data analyst and for realistic acceptance criteria.

57

5. REFERENCES

[1] NS Goujon & BW Kenzie, ‘Discrepancy Analysis of the Round Robin Trial Results’, TOFDPROOF report No 2-18-D-2004-01-1.

[2] ENV 583-6: Non destructive testing – Ultrasonic examination. Part 6: Time of flight diffraction technique as a method for defect detection and sizing.

[3] PrCEN/TS 14751: Welding – Use of time of flight diffraction technique for examination of welds.

[4] D Flotté, ‘Procedure for TOFD Inspection of welds used for the Round Robin Trials’ TOFDPROOF report No 2-21-Q-2002-01-4.

[5] EN 473: Qualification and certification of NDT personnel – General principles.

[6] JM Farley, NS Goujon & BWO. Shepherd ‘Critical evaluation of TOFD for search scanning’, 16th WCNDT 2004, Montreal, Canada.

58

APPENDIX 2 EXAMPLE OF TOFD ERRORS FOR A 15MM THICK SAMPLE

59

1. COMPARISON OF SIZING METHODS AVAILABLE

As an example, consider a 15mm butt weld containing a defect tip 3mm above backwall (hence depth of defect is 12mm from scan surface) scanned with TOFD probes such that their index points are separated by 60 mm (= pair of 63° probes). Assume velocity of 5.9 mm.µs-1.

Two TOFD sizing methods are available:

a) Measure the difference in arrival time of the defect tip echo and the backwall echo.

b) Measure the difference in arrival time of the defect tip echo and the lateral wave.

Both methods have drawbacks.

Measurement relative to the backwall echo is affected by uncertainty as to the profile of the backwall over the weld. The large difference in amplitude between the BWE and the tip echo makes it difficult to measure the time difference between the two signals accurately.

Measurement relative to the lateral wave depends on the lateral wave being successfully propagated through the weld. This should not be a problem for ferritic welds. For austenitic welds, propagation through the weld parallel to the pipe axis is likely to be affected by high attenuation. There is also a possibility of the beam axis following a curved path which could introduce significant error into the size estimate. Mismatch will complicate depth estimates relative to a lateral wave.

The effect of individual errors is presented in a series of graphs below. In each case, the vertical axis represents the depth estimate which would result as a function of the value of the error in the particular variable from its correct value.

The errors detailed in the following pages are summarised below:

Error component Measurement wrt

BWE (mm)

Measurement wrt Lateral Wave

(mm)

Probe separation +/- 2 mm 0.2 0.4

Timing +/- 1 cycle (0.2 µs) 1.7 1.7

Velocity +/- 0.2 mm.µs-1 0.1 0.2

Thickness +/- 2 mm 0.5 N/A

Total 2.5 2.3

Table 1: Summary of the errors

Hence errors are of the same order for both methods for a small defect at the backwall. The most important error component is the timing error. This can potentially be reduced particularly in the case of referencing relative to the lateral wave.

Adding up the errors as shown provides a worst case estimate. In practice worst case errors for all components at the same time are unlikely so that there is a reasonable prospect of justifying an error tolerance of 2 mm or less.

60

2. DEPTH MEASUREMENT RELATIVE TO BACKWALL ECHO

2.1 Variation in reported depth with probe separation

12.5An error of +/- 2 mm in measuring 12.5

probe separation gives a +/- 0.2 mm error in depth. Note that theseparation is measured between the 12.25

index points so the separation is unlikely to be measured more d δs

accurately than +/- 2 mm. ( ) 12

11.75

11.5 11.5 3 2 1 0 1 2 3

3 δs 3

2.2 Variation in reported depth with time of flight

15Time-of-flight errors will be of the 15

order of the cycle length. It is oftenclaimed that timing errors can be 14

reduced to a small fraction of a 13cycle but in practice there can beuncertainty as to which cycle to d δt

measure from (especially whenreferencing relative to a back wall). 11

If we use a 5 MHz probe then onecycle is 0.2 µs giving an error of +/- 10

( ) 12

1.7 mm. 9 9 0.4 0.3 0.2 0.1 0 0.1 0.2 0.3 0.4 0.4 δt 0.4

2.3 Variation in reported depth with velocity

12.2The velocity will be uncertain but 12.2

variation from standard values is unlikely to be more than +/- 0.2

-1mm.µs assuming we are dealing 12.1

with wrought piping. Errors due to this uncertainty are only +/-0.1 mm. ( ) 12d δv

11.9

11.8 11.80.2 0.1 0 0.1 0.2 0.2 δv 0.2

61

2.4 Variation in reported depth with wall thickness

15The wall thickness will be uncertain 15

since the form of the backwall closeto the weld will be uncertain. 14

Tolerances may be up to about +/- 2 mm. This would affect the depth 13

measurement from the surfacesignificantly but in practice the

d( δT ) 12

objective will be to estimate depth 11 from the inside surface. When this is taken into account the error in 10

throughwall dimension would be 9 9reduced to about +/- 0.5 mm. 3 2 1 0 1 2 3

3 δT 3

3 DEPTH MEASUREMENT RELATIVE TO LATERAL WAVE

3.1 Variation in reported depth with probe separation

12.5An error of +/- 2 mm in measuring 12.5

probe separation gives a +/- 0.4 mm error in depth. Note that the separation is measured between the 12.25

index points so the separation is unlikely to be measured more d δs

accurately than +/- 2 mm. ( ) 12

11.75

11.5 11.5 3 2 1 0 1 2 3

3 δs 3

3.2 Variation in reported depth with time of flight

15Time of flight errors may be of the 15

order of the cycle length. It is often 14 claimed that timing errors can be reduced to a small fraction of a cycle 13

but in practice there can be ( ) 12d δtuncertainty as to which cycle to

measure from (though higher 11 precision – say by a factor of 5 is often achievable when referencing to 10

a lateral wave). If we use a 5 MHz probe then one cycle is 0.2 µs giving

9 9 0.4 0.3 0.2 0.1 0 0.1 0.2 0.3 0.4

an error of +/- 1.7 mm. 0.4 δt 0.4

62

3.3 Variation in reported depth with velocity

The velocity will be uncertain but variation from standard values is unlikely to be more than +/-0.2mm.µs-1

assuming we are dealing with wrought piping. Errors due to this uncertainty are only +/-0.2 mm.

12.2

12.1

( ) 12d δv

11.9

11.8 11.8 0.1 0 0.1

0.2 δv

63

0.2

RR 433

safety implications of tofd

Documents