Commission of the European Communities Organisation for Economic Co-operation General Directorate XII and Development Joint Research Centre Nuclear Energy Agency Ispra Establishment Committee on the Safety of Nuclear Installations S.P./I.07.C1.86.62 CSNI No. 121 Evaluation of the PISC-II trials results PISC II Report No. 5 - September 1986 FINAL ISSUE Programme for the Inspection of Steel Components
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Commission of the European Communities Organisation for Economic Co-operation General Directorate XII and Development Joint Research Centre Nuclear Energy Agency Ispra Establishment Committee on the Safety
of Nuclear Installations S.P./I.07.C1.86.62 CSNI No. 121
Evaluation of the PISC-II trials results PISC II Report No. 5 - September 1986
FINAL ISSUE
Programme for the Inspection of Steel Components
FOREWORD
In assuring light water reactor safety it is vital to have confidence that no leaks or breaks will develop in the reactor pressure vessel and associated piping which together constitute the primary coolant circuit. The possibility must be considered tha initially small defects in these thick steel components may grow under the stresses arising from repeated pressure and temperature changes and the embrittlement of the metal caused by the radiation emanating from the reactor core.
Ultrasonic testing is widely used for detecting, locating and sizing flaws in primary circuit elements at various stages of plant life. The successive PISC projects have constituted the most notable, sustained, international effort to assess the effectiveness of these inspection techniques.
The Plate Inspection Steering Committee (PISC-I) programme (1976-1980) was intended to establish the capabilities of the 1974 ASME Code Section XI ultrasonic procedure. The Programme for the Inspection of Steel Components (PISC-II, 1981-1986) constitutes a more detailed evaluation of the best performance obtainable by modern ultrasonic techniques under optimal conditions.
The evaluation is conducted at the global and procedure level and does not consider the detail of the techniques constituting the procedures. Analysis at the level of techniques will be performed during the second stage of data analysis commencing in September 1985.
in
GLOSSARY
ALOK - Amplitude Locus Curves CEC - Commission of the European Communities CSNI - Committee on the Safety of Nuclear Installations of the OECD Nuclear Energy Agency DAC - Distance Amplitude Cuoie (orrtvt><~* Acjt**~u LevtA. DOT - Defect Detection Trials (United Kingdom) DG XII - General Directorate XII (Science, Research and Development of the CEC) EMAT - Electromagnetic acoustic transducer FBH - Flat Bottom Hole HAZ - Heat Affected Zone IWB-3000 - ASME Boiler & Pressure Vessel Code -Section XI - Standards for examination evaluations ISI - In-Service Inspection ISPRA - Ispra Establishment of the CEC Joint Research Centre JPVRC - Japanese Pressure Vessel Research Committee JRC - Joint Research Centre (of the CEC) KTA rules - Regelwerk des Kerntechnische AusschufS LWR - Light water reactor MBSP - Multiple Beam Satellite Pulse MITECH - Multibeam Inspection Technology NDE - Non-destructive examination NDT - Non-destructive testing NEA - Nuclear Energy Agency of the OECD OECD - Organisation for Economic Co-operation and Development P-Scan - Projection Scan (Inspection instrument developed by the Danish Welding Institute) PSI - Pre-Service Inspection PVRC - (United States) Pressure Vessel Research Committee RL - Referee Laboratory (J.R.C. Ispra Establishment) RRT - Round Robin Testing (or Trials), in which samples are circulated to various locations for
inspection by different teams RT - total roughness SAFT - Synthetic Aperture Focussing Technique SDH - Side Drilled Hole SEL - Dual beam probes : emitter-receiver (also called T.R.L.) TOFD - Time-of-flight diffraction VAP - Variable angle probes
ASHE Code : American Society of Mechanical Engineers Boiler and Pressure Vessel Code, - Section XI Rules for Inspection of Nuclear Power Plant Components (1974 Edition), - Section V Nondestructive Examination (1980 Edition)
Ligament : band of metal remaining between an internal crack and the surface
Plate : large welded test assembly
ASME Procedure (50% PAC) : As defined in ASME Section V, Article 4 this procedure requires that a weld be scanned in two directions parallel and perpendicular to the weld direction with a minimum scan line separation of 90% of the transducer element size. Two angle probes are required 45 and 60 or two other angles separated by a minimum of 10 . Generally, in inservice inspection three angles are used; 0 , 45 and 60 and the recording level is set at 50% of the distance amplitude correction (DAC) curve (ASME Section XI, Appendix I).
ASME 20% DAC : The ASME Procedure but with the recording level set at 20% of the DAC curve.
ASME 10% DAC : The ASME Procedure but with the recording level set at 10% of the DAC curve.
Procedure in the spirit of ASME :
A procedure that approximately corresponds to the ASME procedure. For example an ASME procedure complemented by a 10 angle probe or working at an intermediate recording level.
Special Procedure :
A procedure using a complex combination of standard and advanced techniques which cannot be classed as an ASME procedure or a procedure irft he spirit of ASME.
Advanced sizing techniques :
Advanced techniques such as focused beams, holography, spectroscopy, tip peak diffraction, TOFD and SAFT used singly for the specific purpose of sizing defects.
Best results for each team
The set of results for each team showing the best detection and correct rejection rate, identified for a team's only data sheet, the best data sheet where several were submitted, or a computer combination of partial procedures or techniques.
V
TABLE OF MAJOR PARAMETERS
For full definitions of the parameters see PISC-II Report No. 4, sections 4 and 5.
Procedure Description Parameters
A = Automatic scanning procedure B = Examination from both sides of plate or nozzle C = Examination from clad side only % OAC = Signal amplitude threshold level M = Manual scanning procedure S = Surface or high-angle probe techniques (70 S, 70 L, 70 SEL) TD » Tandem technique U = Examination from unclad side only
Calculated Descriptive Parameters
CAF = Correct acceptance frequency for one team on all acceptable defects CAF' = CAF calculated on a limited number of defects CAP « Correct acceptance frequency for one acceptable defect by all teams CRF = Correct rejection frequency for one team on all rejectable defects CRF' = CRF calculated on a limited number of defects CRP = Correct rejection frequency for one rejectable defect by all teams DDF = Defect detection frequency for one team or procedure on all intended defects DDFR = Defect detection frequency for one team or procedure on all rejectable defects (DDF)T = Defect Detection frequency for one team or procedure on all defects DDP = Defect detection frequency for one defect by all teams ELX(ELY.ELZ) = Absolute error in location of defect in the X(Y, Z) direction (mm) ESY(ESZ) = Absolute error in sizing in the Y(Z) direction (mm) ESL = Absolute error in locating edge of defect relative to plate or nozzle surface (mm) MCAF.MCAP... = Any of the above parameters preceeded by M indicates the mean value over a group of defects or teams SELX.SESY... = Standard deviation of location and sizing errors % TND = Percentage of all defects in a plate measured by teams using sizing procedure only
Parameters used on Graphs and Histograms
DX(DZ.DY) = Absolute size of defect in X (Z,Y) direction (mm) DZC = Z dimension of defect at which detection approaches certainty M = Mean value of location and sizing errors S = Standard deviation of location and sizing errors o = Rejectable defect according to section 1KB 3510 of ASME XI (1974) o = Acceptable defect according to section IWB 3510 of ASME XI (1974) HSLN = Crack aspect ratio (DZ/DY) HSTN = Ratio of crack depth (DZ) to plate thickness HSLS = Crack aspect ratio for surface crack according to ASME XI, IWB 3510 HSTS = Ratio of surface crack depth to plate thickness (ASME XI, IWB 3510)
VI
TABLE OF CONTENTS
1. INTRODUCTION
2. IMPORTANT REMARKS
2.1. Base material defects in plate No. 1
2.2. Small natural defects in plate No. 2 2.3. Particular aspects of plate No. 3 2.4. Defects No. 6 and No. 12 in plate No. 9 2.5. Inner radius cracks 2.6. Parameters 2.7. Procedures "in the spirit of ASME" and number of manual procedures 2.8. Calibration on ASME holes 2.9. ASME Procedure at 35% DAC 2.10. The codification of the procedures 2.11. Statistical Aspects 2.12. Tolerance on Destructive Examination Results 2.13. The Codification of the Procedures
3. TOLERANCE OF DETECTION AND DDP - CORRECTION OF TEAM DATA
4. EVALUATION OF RESULTS AT THE LEVEL OF THE TEAMS
4.1. Global results
4.2. Teams results
5. EVALUATION AT THE LEVEL OF THE PROCEDURES
5.1. Groups of procedures 5.2. Evaluation of the results of manual inspection procedures
5.2.1. Reporting level as a parameter 5.2.2. Supplementary techniques as a parameter 5.2.3. Scanning surface as a parameter 5.2.4. Comparison of conclusions with individual team results
5.3. Evaluation of the automatic scanning procedures in the spirit of ASME 5.3.1. Global results 5.3.2. Evaluation at the level of teams or procedures 5.3.3. Discussion on plate No. 3 results
5.4. Evaluation of special procedures 5.5. Sizing techniques
5.5.1. All advanced sizing techniques 5.5.2. Comparison of the performance of the advanced sizing tech
niques with that of the procedures in the spirit of ASME 5.5.3. Effect of the cladding on sizing performance
VII
6. SPECIFIC CASE OF PLATE No. 3
6.1. Procedures in the spirit of ASME, manual examination 6.2. Procedures in the spirit of ASME, mechanized scanning 6.3. Special procedures at low reporting level 6.4. Advanced sizing techniques
7. DETECTION AND SIZING OF INNER RADIUS CRACKS IN PLATES No. 3 and 9
7.1. Introduction 7.2. Parameters used for the evaluation 7.3. Global evaluation of the results 7.4. Evaluation at the level of procedures
8. DISCUSSION Of kl.: I'LTS AS A FUNCTION OF THE TOLERANCE ON SIZING
8.1. Introduction 8.2. Tolerance on size as declared by some teams 8.3. Sizing tolerance as a parameter
9. ANALYSIS OF RESULTS AS A FUNCTION OF SUBGROUPS OF DEFECTS
9.1. Introduction 9.2. Importance of defect position
9.2.1. Effect of the defect location on defect detection 9.2.2. Effect of the defect location on defect evaluation 9.2.3. Effect of the location on sizing errors 9.2.4. Conclusion of the effect of defect position
9.3. Importance of defect size 9.3.1. Detection as a function of defect size 9.3.2. Defect evaluation as a function of defect size 9.3.3. Conclusion of the sizing error as a function of the defect
size 9.4. Combination of size and position 9.5. Importance of the defect characteristics
9.5.1. Families of defects in plate No. 3 9.5.2. Limitat ions 9.5.3. Defect detection probability as a function of defect height 9.5.4. Evaluation of the results for the ASME procedure at 20%
DAC for three plates 9.5.5. Conclusions on the influence of defect category on defect
detection 9.5.6. Importance of defect category for the correct defect rejec
tion 9.5.7. Conclusions on sizing as a function of defect category 9.5.8. Conclusions
9.6. Underclad cracks 9.7. Particular combination of defect position and sizes 9.8. Importance of the orientation of the defect
VIII
10. EVALUATION OF THE RESULTS AT THE LEVEL OF INDIVIDUAL TECHNIQUES
10.1. Introduction and limitations 10.1.1. Data produced by teams 10.1.2. Quality of detailed data on each plate 10.1.3. Variables used
10.2. Techniques performances at 20% DAC 10.2.1. Importance of the scanning surface 10.2.2. Influence of the defect position in depth 10.2.3. Influence of the defect size 10.2.4. Influence of the defect category 10.2.5. Complementarity of techniques 10.2.6. Verification of the conclusion on some individual teams
results 10.3. Evaluation of techniques performances as a function of the DAC
Cut off level 10.3.1. Introduction 10.3.2. results at 10% DAC 10.3.3. Results at 50% DAC 10.3.4. Comparison of techniques performances as a function of
the cut-off level
10.4. Influence of the orientation of the defect
10.5. Influence of the Probe Frequency.
11. OPTIMIZATION OF PROCEDURES
12. DETECTION OF THE BASE MATERIAL DEFECTS IN PLATE No. 1
12.1. Introduction 12.2. Detection of the base material defects by full procedures 12.3. Detection of the base material defects by individual techniques
at 20% DAC and 50% DAC
13. RESOURCES USED BY TEAMS
14. CONCLUSIONS DRAWN FROM THE ANALYSIS
15. MAJOR CONCLUSIONS FROM THE PISC II RRT
16. RECOMMENDATIONS OF THE PISC II MANAGING GROUP TO THE PISC III MANAGING BOARD
REFERENCES
Appendix 1 : Procedures Code
Appendix 2 : Reference defects used for the evaluation of results on plates No. 1, No. 9, No.2 and No. 3.
Appendix 3 : Results for each plate, defect by defect for each procedure family
Appendix 4 : Detailed results of each team
IX
1. INTRODUCTION
1.1. The complete evaluation of the results of PISC-II is conducted following the general scheme described in Reference No. 4. The present report, however, considers three levels of evaluation :
- a comparison between teams, - a comparison between procedures, - a comparision of the individual techniques and components of the procedures.
1.2. The evaluation of the NDE results for plate No. 2 is based on reference defects determined by destructive examination. Satellite defects were numerous in this plate, and in several cases two envelopes are proposed for one defective zone. The first envelope, generally used for subsequent evaluation, is limited as much as possible to the main defect in the defective zone; it often corresponds to the intended defect (Ref. 3). The second envelope includes all the (unintended) satellite defects in the particular defective zone.
1.3. Plate No. 3 was assembled in a more elaborate way than the other three PISC-II plates. It contains several types of defects and also represents a real PWR vessel geometry. This curved plate, which incorporates a real nozzle, was designed principally to evaluate the performance of specific techniques. However the evaluation of the test results starts with an examination of the results as a whole, followed by an assessment with respect to the various procedures employed.
Plate No. 3 was designed also to bring out the importance of defect characteristics. Remarks concerning the analysis to be performed on this aspect are made later in the report.
In contrast to plates No.l, No. 2 and No. 9, reference defect locations and dimensions for the evaluation of the NDE results of plate 3 have not been determined by destructive examination. Instead, as reported in Ref. 3, four steps of non-destructive verification have led to a set of reference defects which appear sufficiently well-defined for the present evaluation. This fact must be kept in mind during the discussion of the results. This plate will be further tested in a third phase of PISC investigation.
2. IMPORTANT REMARKS
2.1. Base material defects in plate No. 1
One part of plate No. 1 contained base material defects in extended zones near to the heat affected zone (Figure 2.1). It had been hoped not to have such defects in PISC II. They appear to be manganese sulphide inclusions and as indicated in chapter 12, most teams detected the defects but did not declare them as such. Final inspection results often showed large defective zones, frequent combined with detection of intended defects. The PISC II terms of reference did not consider base material defects for the evaluation of results. Moreover, teams declared that this situation was not representative of reality except in the case of old plants. As a consequence, results on plate No. 1 are distorted and errors distribution tails are often due to these inclusions : large "over-sizing" of real weld defects is frequent.
2.2. Small natural defects in plate No. 2
The Managing Group of PISC-II asked that all defects with equivalent diameter equal to or greater than 3 mm be considered in the evaluation. Plate No. 2 contains several welds (Ref. 2, 3) with many natural defects in them. As a result, 10 natural defects have been added to the 18 intended defects in the evaluation.
Moreover, the technique of inserting strips containing intended defects into the plate has introduced many small defects which for some sensitive techniques appear as long defects. Only the 10 defects mentioned above are considered reportable, and the test data have therefore been "cleaned" for comparison with the 28 reference defects.
Plate No. 2 also contained natural defects in the base material. These segregations were not considered in the evaluation of results, even though several teams reported indications clearly corresponding to these natural defective zones.
Analysis of defects 3 and 4 as well as of defects 9, 18 and 8 in plate No. 2 could be criticized : some proximity rules would call for such defects to be combined (ASME Code, Section XI, 1974 or 1977). The analysis has been conducted separating all defects, for the interest of the resulting exercise, and because most of the teams did not in fact combine these defects.
2
T
LU E te £
CM
o Q.
E E
o
I -o LU CO
Figure 2.1. : Base material defects in Plate No. 1.
3
2.3. Particular aspects of plate No. 3
JDejid_zones in plate No. 3 _
Because of reduced dimensions for some procedures that would be applied, it was decided to create two dead zones in the plate No. 3 weldment as shown on Figure 2.2. Near these two zones, composite defects were therefore introduced (defects 6, 7, 3, 9).Several teams did not report defects 6, 7, 8, 9, perhaps because they overestimated the size of the dead zone.
Several teams declared that they had not examined any surface defects. For the evaluation of the results for these teams, defects near the clad side have been eliminated (unless detected and reported by the team, which was often the case).
Several teams did not use the proposed coordinate system and reference benchmarks. Simple corrections were generally possible. However errors were often made in relating the selected reference system to that proposed for PISC-II. In two cases doubt remained as to whether the corrections introduced by the Referee Laboratory were sufficient to compensate for the errors thus introduced.
The early decision of the Managing Group to fix the origin of the Z-axis at the top of the nozzle made drawings, computer work and understanding of results difficult. It was also incompatible with the most logical destructive examination or complementary (X-Ray) non destructive examination.
A new definition has been chosen corresponding to the Z direction remaining parallel to the nozzle axis but with the origin on the clad surface. This transformation made it easy to prepare drawings of the results, as shown on Figure 2.3.
The Z-coordinate transformation by the teams introduced errors as great as 7 mm, depending on the reference chosen for measurement of the height of the nozzle. The Referee Laboratory tried to compensate for obvious systematic errors. In any event, only errors larger than + 5 mm are considered to constitute location errors. Results were reexamined using the measured profiles of plate No. 3 (Ref. 3). However uncertaini-ty regarding ligament dimensions remains high, +_ 3 mm or more if the team made co-ordinate transformations. In most of the cases, such errors were corrected in a second iteration.
4
o <o /
- 4
0
jR>jj^^1^
(
-2300
\ \ o,
Ejlgî D
\
«»
o rg
t
Figure 2.2. : Dead zones in PISC-II plate No. 3.
500.
400.
Z 300.
100.
la]
Figure 2.3.
D
NOZZLE PLATE No. 3
LBQ04I97
,=-Jl 60 30 120 150 180 210
CIRCUMFERENTIAL POSITION
240
D
270 300 360
YD
Example of PISC II Plate No. 3 drawing representating the developed weld central cylinder. The Z = 0 value is referred to the clad surface.
2.4. Defects No. 6 and No. 12 in Plate No. 9
Two defective zones of plate No. 9, containing defects No. 6 and No. 12 respectively, were badly detected by teams. Only some techniques, specifically applied for near surface defects, showed good detection and sizing of these defects. However, these defective zones present particular cases : transverse defects near to the clad' surface are connected by sets of small porosities or thin cracks (Figure 2.2.). The envelopes of these defective zones, determined by destructive examination were(Ref. 3) :
Defect No. 6
Defect No. 12
Defect No. 21
4
8
6
DX (mm) DY (mm)
53
67
10
DZ (mm)
18
23
4
Defect No. 21 was a set of porosities and was considered as non intended.
For defects 12 and 6, envelopes corresponded to rejectable defects according to the criteria set down in ASME IWB 3510 (1974) and used as a general reference for evaluation of results. Several individual transverse cracks of these two defective zones are also rejectable defects due to their through thickness dimension.
Thus, the evaluation of results was in accord with the general procedure but it could be argued that in the case of defects 6, 12 and 21 this approach was overly severe.
6
2.5. Inner radius cracks
The evaluation of the detection and sizing results for the inner radius cracks forms part of this report, but plates 9 and 3 are considered together rather than separately.
2.6. Parameters
All the parameters used are explained in Reference 4. It is important to stress the illustrative character of statistics that have been computed and of the regression lines drawn; the samples involved were often very small.
Results on one defect for all teams were characterized by the parameters DDP, CRP, and CAP. Results on all defects were characterized by the parameters MDDP, MCRP, and MCAP. Results for a team or procedure were characterized by the parameters DDF, CRF, and CAF. Results should be judged as a function of a set of three numbers or three graphs; thus :
- a procedure giving DDF = 1 , CRF = 1, and CAF = 0, was biassed on the "safety" side : it rejected all defects, even the acceptable ones;
- on the other hand, a procedure giving DDF = 1, CRF = 0 and CAF = 1 was biassed on the economy side : it accepted all defects, even the rejectable ones.
2.7. Procedures "in the spirit of ASME" and number of manual procedures
Throughout this report an "ASME procedure" or "a procedure in the spirit of ASME" implies a procedure that approximately corresponds to the requirements of article 4 of ASME V (1980). Essentially, such a procedure requires that a weld be scanned in two directions parallel and perpendicular to the weld direction, with a maximum scan line separation of 90% of the transducer element size. Angle probes of 45 and 60 are required, or two other angles separated by a minimum of 10 . Distance amplitude calibration in PISC-II also refers to the requirements of ASME V article 4 and uses a series of 1/4 T, 1/2 T and 3/4 T side drilled holes 3/8" (9.5 mm) in diameter.
In PISC-II an ASME sizing procedure implies a procedure in which probe movement and amplitude drop techniques are used to establish defect dimensions in the Y and Z directions. However, different criteria are used (50% DAC, 20% DAC etc.) to determine the extremities of the defects, than those specifically recommended by ASME. Similarly, the reporting threshold of 50% DAC for indications recommended in ASME V is varied in the procedures "in the spirit of ASME". Small departures from ASME occur with some teams but in the opinion of the PISC-II Data Analysis Group it is reasonable to include many teams in groups of procedures generally referred to as ASME or ASME type procedures.
7
PISC-II had to put the emphasis on ISI procedures. It thus expected that a majority of the results received would be obtained by mechanical scanning systems. Due to the many procedures in the spirit of ASME which were applied on the PISC II plates, however, the majority of PISC II results relate to manual scanning. It was decided to consider all these manual procedures as well, as they have provided the evaluation with large samples of data on various classic techniques.
2.8. Calibration on ASME holes
Calibration on ASME holes leads to the DAC cut-off level used to group ASME procedures : 10%, 20%, 35%, 50% DAC. This calibration method is the one used for standard probes and is not to be considered for 70 Lor SEL probes for which side drilled holes and flat bottom holes are used. When an ASME procedure including 70 angle probes is qualified as 20% DAC, it does not mean that the 70 angle probe was calibrated on ASME holes. Cal ibrat ion speci f icat ions are given in appendix 1 of PISC-II Report No. 2.
2.9. ASME Procedure at 35% DAC
To increase the data samples, procedures using cut-off levels of 20% DAC and 10% DAC were used to create results of hypothetical procedures at 35% DAC. Of course, these new data sets were used for DDF evaluation only.
2.10. Sizing techniques in the ASME type procedures
In the groups of procedures in the spirit of ASME at 10% DAC and 20% DAC, the sizing procedures were often different but all of them were based on "Amplitude Drop" methods. Some considered the 6dB drop, others detailed the method as follows, as a function of the maximum amplitude level :
- 20% of the maximum amplitude, - 40% of the maximum amplitude, - 80% of the maximum amplitude.
In the analysis no difference was made between these techniques within a group. Trials were performed to evaluate these technique results separately but the dispersions obtained covered the total dispersion of the group of procedures in the spirit of ASME at 10% or 20% DAC.
8
2.11. Statistical aspects
PISC-II report No. 4 (Ref. 4) describes the analysis procedure and defines the parameters used in evaluating the results of the RRT. It should be emphasized that only simple descriptive statistical parameters normally used to describe experimental results have been used, such as mean values and standard deviations. Conclusions were drawn based on trends in the results seen in tables, graphs or histograms and no attempt was made to use statistical tests of significance to qualify the reported trends. In many cases the use of such tests was inappropriate because of the high dispersion of results and the low population sample. However, every attempt was made, where trends were observed among groups of teams, to confirm the observation by reference to the results of specific teams who performed many inspection in a systematic manner. Inevitably, in some cases, it was necessary to use what is often called "engineering judgment" to decide whether some results were worth reporting. This was done mainly where observed trends have a real engineering value to NDT and were thus considered too important to suppress.
On the use of PISC II results it is perhaps important to repeat paragraph 1.8. of report No. 4 :
The results presented here should be interpreted as simple descriptive statistics. That is, no confidence bands have been added to imply that the results can be extended to some wider population of reactor components. It is most important that, before broader conclusions are drawn from the PISC II results, first, the parameters used are clearly understood and second, the relevance of the PISC II component and defect types to any other components be carefully assessed.
2.12. Tolerance on destructive examination results
The final sizes and location co-ordinates of defects given were based on the results of the destructive examination. These were performed by progressively cutting the original plates into smaller and smaller blocks. From the initial plate reference datum it was necessary to re-establish new datums on each smaller block. After several such operations it was inevitable that an accumulation of errors in measurement occured and it was estimated that the final sizes and location errors are between +_ 1 mm and +_ 2 mm. More precise data are given in reference 3. The errors on defect sizes and locations, however, are always much smaller than the dispersion in the NDE results (Fig. 4.13).
2.13. The codification of the procedures
The codification of the procedures used is explained in Appendix 1 and the reference defects are listed in Appendix 2.
g
3. TOLERANCE OF DETECTION AND DDP - CORRECTION OF TEAM DATA
The first step in evaluating the results was that of defining limits; in particular a tolerance limit in location error was necessary to avoid confusing defects. To guide the choice of the tolerance limit the graphs in Figures 3.1 and 3.2 were drawn utilizing all the best procedures of each team.
It is clear that when the tolerance (as defined in Ref. 4) increases, detection increases. The graphs show that for plates 2 and 9 a high detection value is reached very quickly. For most of the defects, one could stop at tolerance t = 15 mm. To avoid any misunderstanding of the teams' results, when a mismatch appeared between the reference defect and a teams declaration at 25 mm tolerance in detection, the team result was carefully compared on diagrams of findings in order to decide whether the reference defect had been detected or not. Such situations frequently arose for plates No. 3 and No. 1. For plate No. 3, a tolerance of 100 mm was often necessary (Figure 3.3 ; for several defects, DDP increases with tolerance up to 120 mm).
On the basis of these rules, the data were "cleaned". From the general data set (declarations of teams, with corrected coordinates), which was called the "BOAT" and which will not be altered for any reason, a cleaned data set, the "TRAIN", was created. The "TRAIN" omited all indications not corresponding to a reference defect determined by the destructive examination. Comparison of diagrams led to many operations for plate 3 results; subjective conclusions were thus introduced by the Referee Laboratory.
It was because of this cleaning phase that most of the corrections were asked for by the teams and accepted by the Referee Laboratory. In particular :
- a defect was detected but very badly located; - a defect was badly assembled from its corresponding indications by the Referee Laboratory or by the team itself; and
- some distant spot indications were added to the defect increasing its dimensions too much.
The Referee Laboratory adopted a favourable attitude to such requests because :
- there was clear documentation on which the request was based; and - the objective of the PISC II programme was to evaluate the performances of procedures and techniques, and not to emphasize the human errors that were made.
In all cases justifications were given for all corrections performed.
Figure 3.3. : Detection rate of defects in plate No. 3 as a function of the tolerance in defect location.
12
4. EVALUATION OF RESULTS AT THE LEVEL OF THE TEAMS
4.1. Global results
The best results for each team were identified either by:
- using the results from the only data sheet (DS 6.2.) submitted by the team (e.g. AB000699);
- or selecting the best DS6.2. when several were submitted by one team (e.g. a team might have reported inspection results for inspections from the unclad side, from the clad side and from both sides);
- or combining partial procedures (using the computer code) to obtain a global estimation (producing a DS 6.2.) : for example inspection from both sides when results were given separately for inspection from the inside and the outside (e.g. AB005397),
** Figures 4.1. to 4.4. show the results for the four plates when considering the defect sizes (DZ and DY) as a paramètre.
Figures 4.5. to 4.12 show the rejectable (•) defects and acceptable (o) defects in the plates, according to the ASME XI IWB 3150 Criteria, 1974 Edition. The results are globally of good standard, when these ASME criteria are applied.
Figures 4.13 to 4.16 show the overall cumulative diagrams of errors in location and sizing (in Y and Z directions) for each of the plates. Dispersion is high in general and the distributions have distinct tails.
Some remarks are necessary on plate No. 1 results :
a. Due to the large zones of manganese sulfide inclusions that teams often confused with real intended defects in the weld (cf. § 2.1) several defects were oversized considerably in the Y and Z directions.
Thus tails of the distributions were due to oversizing.
b. Defect No. 1 in plate No. 1 was a long (1 m) crack which was detected only in parts.
The location in the Y direction of this detected defect and its sizing in the Y direction were thus often biased. (This explains the tail(s) of the distribution : error of location and of sizing (undersizing in Y).
Plate No. 2 appeared easier to inspect but no significant difference in results seemed to occur between the plates : a similar dispersion of sizing errors is shown in Table 4.1.
See Section 9.5. for discussion of composite defects in Plate No. 3 and possible alternative positionning of points on Figure 4.4.
The graphs of DDP as a function of DZ show how detection approaches certainity as defect size approaches a certain value DZC, which is typical for each plate (Table 4.2.) :
Parameter
DZC
Plate No. 1
16 mm
Plate No. 2
8 mm
Plate^ No. 9
16 mm
Plate No. 3
32 mm
Table 4.2.
It is often difficult to interpret CRP as a function of DZ because :
- it is difficult to fit a regression line, and
- the most likely line could be one parallel to CRP = 1, at the CRP value given in the table : (that is MCRF).
Such a global evaluation should be used as a basis for the comparison of specific procedures. It is unwise to draw too many general conclusions from a mix of the different procedures used by different teams.
o. o r u s Q u i s i : c s te e a e> •/> M> <£ in P> M 0> r-ea o> M is e* f* en r> c n u c* n fe Qtvotuttfi^ii/iu>,9i£f»u>i£<£mfwk/tr*)tt OIUNOD v c o o *
eu +-> (O
a.
\tf\j(\jrv<iffuru<uf\ E 10
O
D-
'
\ 8 ^ , ,
: •
* 0 ° « . o
>
^
8 è
i
°
0
s
S>
0
0
0
'
• •
•• • ' , n
• ~—
R
Jf
?
0
0
1
1 ,8
__ _ _ _ ^ _ °
»
O
^^ __ ___ ^ _ _ °
.. ^ . . . ^ 2
«=2 c! ^ °
(J rO (D
4 - O O
to
3 to a> s-+-> to (V
CO
• r— to
+-> u 0)
•*-a> a
<d-
a i
8 55 2 55555!: s 5 5 ;55:555 5
11 e
E ro a>
t
• :
:SsT
8 î
•
•
0 ^ \ ^ 0
0 ^ »
S
O
• \
\l >
•
S
^ *
\ -
s
0
0
0
'.% 8 S
0
8
O
a
eu 4 -O
to + J r— 3 l/l eu i -
+ J to Cl)
CQ
r-J 0
a i M
• 1 —
to
+J 0 (U
'fra i
0
a> s -en
16
CO
o
< c
T\ Î "
O 0 8 * o
O 0
s
D o »
DO
s 3
g
3
?
•s.
1
s?
î
° °
s§
c c
t:
c IX
o
o
S. C Q i _ _a oj
• X<~** s
— • . \ : —
• •
o „ o o
o ° o o
o
0
a
o o
o
o
o
o
o o
c
o
•
o
o
eu
o ID eu
l / l
0) i-
+ J l / l
ai co
•i— i / i
-!-> U ai u-a> o
X3 *3"
a>
CSl
o
ai
o
1
o
• \
\ o \ o
\ \
" °
O
o
o
o
o
o
D
O
O
O
O
V Y
• • A i
D
O
O
Q
Q
O
O
o
r .
o
o
• •
°
o
°° •%
D
O
•
O
Q
O
O o o
o o
o
D
«»
E
ai
ai
1 —
3 l / l ai S-
+J i / i C1J
CQ
ai NI
• i— l / l
+J u ai
4 -0)
CJ
ai
17
PLATE N. 1
H '.'/
SURFACE DEFECTS
H/K7.) 10
H/TC/.)
HSlNi
SUS-SURFACE OEFECTS
20
HSLS
PLATE No. 2
HSTNV.
SURFACE OErcCTS
HSTS*/.
Figures 4.5., 4.7.
Figures 4.8., 4.9.
: Defects compared with the criteria ASME XI, IWB 3510; subsurface defects.
•— u M- Q. (O o ai to en t*_ E c o s_ N o Cn •»— 4_> (O en ,—
•r- 3
ai a s.
> V - C + J 4-» O (/) n3 «i— a»
•— -4-> _Q
e u .— 3 O .—
O .— <S
n n o
Cv U1 CN
^r u> M
•-n M
on n •-
r» LU
*~ f i n »•
09 CD
U> (V <*> r> a>
s -3
CJ o
I
a.
i —
_ | 1 | L .
^
|:f
gHUTTsT^ a» ci
= 9 • ii s: ai
i
<L'-> LU
1
r1 1
.'
r^ S
O s-s-a i
4 -o
m f-03 i_ Cn (O
"O
a i
> .,— + J IO
•3 t -rs
<_)
• o * i
a i + J 10
r— a .
• O l c
M -i— l/î
- o c: (O
c o
•r~ +->
• H
a i 4->
i : o (D a i
4 -o
i/i + J
< — ^ en a i i_
4-> CO a»
- O
3 en
20
ro O
et
Q.
g a i tî M « « ° %'A èi i H^ M * °
ELY
M=
-16.
1
s=
3a.i
* 1
1 1
^ <n I D n tsi w CN CN «- * - * -
* - OJ m
O)
4 -O
U l
H rO S-
en rO
•i—
" O
a i
> • r—
+-> <0
i — =3
H - 1
C_)
(O i — Q_
CT> C
• (— M • i— CO
-o c ro
c o
• I—
+•> rO ( J C )
» —
_c l J rO
eu 4 -O
LO
+-> i
73 l / l
ai i -
+-> </> ai .o i — • —
<s.
*3-
ai
N r- »- »-
en
o
_ l Q.
a i* 11 , . . . i , — „ > , — , , i ,
2 LU
1: Ol 09
O
i .o
• 1
II
I I.
• u
en LU
71.
(À
• E
£l
S=
21.1
'
1
g * U «
i r - 1
i l
6" j< i °
S-
o 1 -i -O l
4 -
o l / l
fcr (D 1 -
cn <o
• i —
" O
O l
> • i —
*-> ta i —
=J
fc 3
<_>
^ O l
+-> rO
i — Q_
en c
• i — rsl
•r— c/l
" O
c <o
c o
• r—
+-> n3 O O '—
03 0 1
*-> c~
U
ro eu
4 -
o t /1
- l - > 1 —
=J ( / l
eu (_
+ J i / i <IJ n
i — i —
<
eu s -
21
4.2. Team Results
Tables 4.3, 4.4, 4.5 and 4.6 show the best results of each team in the four plates. Several teams achieved high values of DDFR, CRF and CAF. Some teams achieved small errors in sizing (parameters MELY, MELZ). It thus appears that on all plates it is possible to identify procedures capable of good detection and sizing. It appeared clear during "cleaning" of the data for plate 3 that the large location errors were due to the difficult geometry of the plate.
Tables 4.7 to 4.10 show the complete location errors data in the three directions for the three plates. These tables as well as table 4.11 show the errors made by the teams in evaluating the defect ligaments in the three plates. It is important to remember that for plate 3 there was an uncertainty of + 3 mm in the estimated ligament values even after the detailed profile measurement by the Referee Laboratory (Ref. 3). For most of the defects of plate No. 3 near the clad surface (ligament less than 20 mm) the uncertainty was lower (_+ 2 mm) due to the process used to implant the defects.
For plate No. 2 several sets of references were used which include or exclude most or all satellite defects; these different references did not modify substantially the CRP and CAP values, even if the smallest boxes were used around the defects.
5. EVALUATION AT THE LEVEL OF THE PROCEDURES
5.1. Groups of procedures
Several groups of procedures were identified from the data sets :
5.1.1. Within the manual inspection procedures in the spirit of ASME, following the guidelines of the D.A.G. the first grouping parameter was the recording level (cut-off level : CO). Most procedures were (or were considered to be ) modified ASME procedures. Four groups of recording levels were selected :
a. 10% DAC for which CO < 20% DAC, b. 20% DAC for which 20%<.C0 ^ 30% DAC, c. 35% DAC for which 30%<C0 < 40% DAC, d. 50% DAC for which CO J^ 40% DAC.
In each subgroup, supplementary techniques were grouped as follows:
a. (70 L, 70 S, 70 SEL ...) were classed as "surface probe": S, b. tandem probes : TD, c. both S and TD techniques : S, TD.
Table 4.6. : All the best results of each team for Nozzle plate No. 3.
27
o i r o ' f l ' L O O j ^ û Q C D t D t û ' - m o ^ o o L n o o n c ^ L n o o ^ - ^ - ' - o j o UJ * - -«- ^- -r- , - , - , - , - , - , - , - * - ^- „_ fy „_
_J c n v ^ r ^ " O J O J t ^ - ^ ^ ' L O L i ï ' r - c T ) - » - u > ^ - 0 ) r ^ c n o j < j ) ^ - ' ~ ' f l - m n j b J « - l l l I *~ I l I I I •»- | | T - O j T - ' - f U I
^ i > i i i i i
I\I
(Si
N j < D i / ) T r * - T - r - ( \ j K i f o a j ( i i r m ( a y ) 5 ) L f t f u a j r u T - w ( a o o f u y i i i i i i i i i i T- i
> _iOTift^^v'<rLntum*-L/)«)<DCocuLA*-u}*-njc7>r^ru»-aj y i e u t i eu i i i i -«— T-E. i N i
x tn *~ *~ *~ *~
X — J c u - i - ^ - c n a j o o s i r ^ r r ) C T ) C D C 3 v a ) a j [ ^ c 7 ) r - ' ^ - ^ - f o s ) < D a j o o y * - i i i i i i •»- * - a j i i I I I I I i i Z. i i i i
p " - 0 > o > C T > o > r - r - o > o > o > C T > o i O ) O > [ s - r , - r - i s - O ) r ^ o > < 7 >
r a u u u n n o i / i u ) œ u i - h u u ^ û u a : i D . u < c c û û û u u u H H - i j j î : E c a û £ U ) u i i < : i - i - >
Oi (9 O» (7) LO O) ^T V U> L/) Ol m I - I - »-o Q o ^ 3 X r w 3
UJ ai »- ai *- w(n<û ' - *- <T *- m *- ai ai »-
0 1 C n O » 0 0 Œ ) L O l O ( I Ï U > < i ï r - ^ " ( S ) O > * - « - < D ' ^ " ' - C r l 0 0 ' - Q 0 f U I O U J * - I i i - a j i i *- i ai * - ^ i i f i n i i *- ai »- *- ai C i i i i i i i i i i i i i i
_ i r ' - i n ^ - ^ - y ) ^ r m o o a i œ ^ - n - o o a > Œ ) ' ^ t U ' « - o o i s > * - ® o , » ^ r u " * - — « - c u m *- T a i * - • « - * - •«-
_ i a i s > ^ - v a i a i ( n t o a i ^ - , * - « - < s > a t u > * - a i u ) r > - t n r - * - * - v u i t - i I I I I i « - « - l i a i i i i (u 2 i i ' «
_ j O ] < f l ^ r o > t » ) i / > Œ > o > r - u > h - V ' » - o > i D r s - < s > r - a j ' « ' » - t s - a i r > -u T »- m <D en <- a i « - u j a i e n o o u ï C D a ï u ï ^ - a j a i * - * * * -0 1 * - »- CU OJ
s _ 1 0 l < D Q O » - C O a i L f > ^ - U Ï C O O O O i < I > V C n o O , n ' 0 > < i > 0 » a i l O Œ ) C n U a i •*- i ^r *- *- • « - < i ) i * - T a i ' - « - I O I C D E i i i i *-
to
s:
_ j u j o » m ^ - « - T * - « - ( U ^ - o i v o o a i * - [ s - a i a i « - ( r ) ^ - ^ - ( n ( n u * - * - i i T - i i S i »
a u > N ( o u ) u > i / ) Œ ) ^ " O U ) ^ , < r ' - r ' ) U ) ^ , û . L f t r - * - r - u i i f t O )
| - S ) l 9 I S < - Q S ) < S I S S > I O l S ) l i J S S ) f i ) I S G ) I S Q l 9 I O O Û Û
( i a a i u û u u j u j " i ^ j j £ z c u ) i / ) ( / ) ( n h h p r ( / )
u j * - * - * - •»- , - , - , - , - , - « - , - , - , - , - . - «- i - f t j « - n j n j f f u m » - eu ru to
_ j t n c n L f t m c u ( U O j m i j > u > r - r ï ' - c T î < i ) t û r o Œ ) ( n ' - u , i ' - a j a j ' - o > c 7 » * - v r ^ c j o ( ^ L n v t ^ U I I I I I I l l l l « — I I I « r - I I I I O J I r i i
N j i n w n j Œ a j o n w o o o i f t a j r u T f n u i m f n o m f o u i c o ' - n j n j n j s i D ' - T m w I L I I I I I i i l I I i l l i * - i l l O J i Z i t
> ^ ^ u ) ( n i ^ o o u i u » i j ) < s o o r - o > r < - ( i j a ) a ) ' 9 - u > a > r - ' 4 " 0 ) ( n a j c » o o ( 9 a i ^ - r - C D o o r - ( * > tn
> _ j w a j v a j w * - v » - < D - r - * - a i ' * - m n j n j ' " - T r ( U ( u n - m m w » - a j « - * - * - v s ) - « - ' « - * -> v j « - l • ' ( U l I I I I l l i — l i — I l I I l £ i i
x _ i o > œ u ) ( û U > ( » » i / » u ) N « ) < D < 7 > o > T N O U ) n v T ( » u ) ( n o o n v y ) o o u > N o o * - o o i / ) UJ ro * - * - «- ru * - * - » - * - m eu •»- * -tn
X ^ c D ^ e n ^ u ) v ^ r u r ^ ^ ^ G D U > r n G 3 ^ ^ r o r > - u 5 r o a D r u r n * - c j o c D ^ r * - - a D * - * - e u u ï U J > < < • * i i i i i l * —
z Œ u n m u L i n c o 3 u m o ) i i ï a u 2 : ( : u « i u o u E H ï 3 l i i . Y 3 i u a c a a u o u j u j u u i U H H - ) ^ ^ j _ i _ i E z z û : u ) U ) u ) U i o i i n i - i - r u i j > -
X I
00
x > IO
tn
_ J c n u > u > a j u ) L n L n « - c D a j ^ - ( D * - U ï o o r ^ c D a j i s f n m r u i s i t o c u r ^ n j a j c ^ o o o o o r ^ r ^ s > U| I I I I I I l l l l l l l r- l I I « - l « - l l v - I O J v -E i i i i i
N • j * - - ^ « - [ ^ r ~ c o œ m » - a i r u * - « ) ' - o c o - ^ ( 7 > ' - a j ^ ' f o u > o O L D L n c » i v j L n c 7 ) o o o > o o c D L U ' . - . - . - , - , - , - » - . - » - » - . - , - » - i - . - * - * - » - — • - , - , - n i « - » -
i -
I N <r J U i a j ( u c o n j c D a j u ï C D r ^ L o a j r u ^ r n j L o t r ) m c D u î o j i j > r ^ * - a » f n a j * - a j C D n - ^ - ^ - v o UJ * i i i i i i I « - I I I ( v i i n j i •-* C • • • _i
z : > a ^ ^ i n m ^ o o i > L A i A ^ i > i > 0 ) U c u u > 0 ) m u ) 0 » u ) ' « - a > ( u n j o > i > C D a j t > o o c D i > u ) m z w *~ *~ "~ *~ "~ *~ , - , - , - *-i- i- *-o i-( i -<r > u _ t ( * ) a j u > < - ^ - c u m c D i > n j a j ( n a j ' r ( u a j a j r n ^ - c T } i > a j a i ^ - C D n j < - ( u c D n * - c u c D L i ) o UJ * - i i i ru i i i i i i «- i •*- i i i _ i T i i 2= i-t
X oc LU m * - * - »- r u » - - r - t - ^ - m a j , - , -o tn oc o:
" X g J O ^ o o e D ^ m ^ a j « > ^ ^ ^ v a j c D c u c D a j t f l ( f l m ^ ( u c O ' - L j > C D f n ' - e D « - e D r u u >
LU I I I I I I I 1 I I I 1 1 * -
<r E
O>OT0>OTC»OTO>O>OïC7)O>O)C7)a>C7>C7>17>C7>0iO>O>O>C7)C7)O>O>C7)0>a>0>O>C7tC7>C7> E u > i j > u ) v w v a j r ^ v < D U ) r n i V f u r u < i > m ^ < o u > i i > c n c D ' - r u c v ^ - o > o D [ ^ , « - ^ - u i C D Œ u ^ a ^ u ) U ) u ) l S 5 l ^ m a J ^ s ) U ) ( Q n J U ) a > ( n ^ ) u ) l / > • s ) ^ ( u n J U > ^ * - ) j ) ( 7 » u > o o Ui CD CD AJ CD CD CD OD OD OD CD CD CD CD OD QD CD GD OJ CD GD GD OD CD OD OD OD OD OD GD OD r~ l~ K~ GD H** GD G3 CD *— GD OD S) (S CS CD CD CD CD CD CD O CD UJ CD OD CD OD GD CD CD GD QD GD CD GD O O Û GD
r Œ u c û a i u u . ( n u ) 3 u u ) u ) < i ( i m u r t ' i J œ u û u C r t t 7 x Q . ^ 3 i u a n c D u a u u t i j t j y H H - ) ^ ^ j j j E z z o c ( n ( f ] ( / ) ( / ) u i u i i - i - z ( n ' > -
- J f U N ( U M ' ~ L / ) i n f n i n i - ^ - N O > û 0 C 7 » n ( U M \ J ' - O > C U V
<r «3
U ' - n j T ^ , c f t f ^ » - u ) p i N K i N f | ) a j * - T s ) a j i L c u f l - c o o j
o u j r u * - r o * - * - c o o r r > * - L r > c u ^ - f - [ ^ T r a j - 3 - * - ' q - o > « - r o r o
o — J ' 3 * * - c o r ^ « - o c u « - s ) c r ) t - m ^ - o o T - T o o s ) ' ^ - n j < û L r > c o LJ I I » -— < — -•— I T I I * - •«- I ST i i I N
OS o a. 0L
_ ) Q r ^ c o L o o o o o r ^ * - v c o o o c u L r ï v c ï ) r u L f > r ^ c û « ï c u c u r o UJ ru ru * - * - n j « - « - r o a i * - r - * - r u r o r u c u c u m r u
_ i L n ^ ( u ^ - ^ - L / i o o [ ^ - C f l r ^ ' - - f l - r O ' - T - r ^ « - ( X i ' - m c n ^ - u ) W 7 * - i «- cu i i «- «- i i r o CUCUOJ * - ' I I I I
cftcnc,»OiO>c7>c,)i^ci>c,)r^o>c1)t7)c,)r^r^o»o»t7)a>a>o> . ° ! O i n o ) ( 7 ) O I C i > ^ 0 ) a i C ' I O I C 1 > Q ) C l i a o > O C i > 0 ) 0 ) ( 7 ) l 7 ) g s > L n u > i n " T U > c o * - c r > < o * - » - « - c o < D * - r - t t j L Q r ^ c a ^ - c » ^ L ^ r r - O D L O L o r o c u ^ r r ^ Q v ^ - o r o f o v o c r Q t - o o L o r o U j r a o a J S S > S S S G Q S ) O S > C 0 6 3 S ) — G) G3 O Q t— f-l - O S I S I Q S l O S I Q O O O Q O Q Q j Q ^ Q m g j Q Q Q
Z a u o o L u i u u i en ÛÛ u z h y ^ û a 5 û . u ^ i < r a i m c û i i j L i j L L i i - i > - i - ) _ i _ i _ i E z : Q : L o i o L o i - > z : 3
In each of these groups, the surface which was used for scanning was also differentiated :
a. unclad surface (U), b. clad surface (C), c. both surfaces (B).
5.1.2. There were several mechanized (or automatic) scanning inspection procedures, but it was difficult to separate them into classes. An automatic procedure was often a combination of several techniques, both traditional and advanced. In fact it is perhaps better simply to consider the individual procedures as reported in Tables 4.3 to 4.6.
5.1.3. Sizing techniques were difficult to group as classes. They were considered separately.
5.1.4. When possible, the size of each group sample was increased by using data at different cut-off levels (for detection evaluation only) (cf. § 2.8.).
5.2. Evaluation of the results of manual inspection procedures
5.2.1. Reporting level as a parameter
a. Results obtained on each plate using a 50% DAC reporting threshold gave average DDP values over all defects (MDDFT) of between 0.28 and 0.40. Considering only intended defects these figures rise (MDDF) to between 0.33 and 0.56. Figure 5.1 shows the results obtained on plate No. 3 at 50% DAC.
b. The average correct rejection rate on all plates, at a 50% threshold varies between .07 on plate No. 3 and 0.67 on plate No. 2.
c. Figures 5.2 and 5.3 show results at 20% DAC. Higher values than those at 50% DAC were obtained for both detection and correct rejection.
d. Figures 5.4 and 5.5 for 10% DAC show results only slightly better than those at 20% DAC. It thus appears that, whilst reducing the recording threshold from 50% to 20% DAC increases detection rate, no significant improvement is observed by a further reduction.
e. On the other hand use of a 35% DAC procedure did not appear to produce as good results as a 20% DAC procedure. This conclusion is based partly on generated results (§ 2.8.) (Figure 5.6).
f. To summarize the conclusions, Figures 5.7 and 5.8 show how the detection rate and the correct rejection rate vary as a function of the reporting level for the four plates.
31
PLATE No. 3
DETECTtCN
OOP CUP CUP
-• l >4ôw4i*ïw (iJ CO oz<nto
CORRECT REJECTION ASHE,IV8-35IO)
V * 3o COKlECT ACCEPTANCE
1o so SO
1 2 3 4 S S 7 t 9 18 11 12 13 14 15 IS 17 IS 19 28 21 22 23 24 25 2S 27 28 29 3» 31 32
Figure 5.1. : Detection rate, correct rejection rate and correct acceptance rate of defects in plate No. 3 for the procedures in the spirit cf ASME XI at 50% DAC.
32
«. te r~ r--<« a ^ N C C C s* r-« e o » r-c f-
O S z s
£ i «£ _ J a .
!£££!C
1 ^M^"^M5^!n!^n^^!!^!!!:^:^! :
•
• •
1—l'
. i B i ; i
o
1 h o
•^—]» o
• . •
- r " a
o
ft
o
o
s
o
a» i/ï +-> •!-> (O (J C S_ Ol -r-
4-c a> i/> o -D a<
•r- S-4-> 4 - 3 O O X I OJ CU
•r-5 a> O ai +-> o o ! . « I . <
s_ a. o +-> u a; a; s« a> o . c o s_ c +-> CSJ s_ <o O +-> S_ LLJ o Q . o s :
QJ 4 - </} - u < t
a i o r o +-> (O 4 -io • o t . -i-) o
c_> z + J e c u •«-o &- at s-
•• - s_ 4-> •<-4-> O O Q-(J O •— </> ai Q . +-> TD a; <u c a jz
Q (O -r- +J
m
ai
_ _ > a.
"2 2.TÏ ttf I*ITU>OJO ^ V
- < U « - « S
o o s o n n s a w A i Bsni i>uiwiaaoi*ut i )vo) D S a n t u i A s O * ï o s i i ) 0»
8 S
| 1 ! • •
- •
i S3
o
R
2 i a.
I 1 1
o •• • ' — ^ ^ ^ - _ — «
g
?
o
—3= "
a ï
-~~9 • ' I ' J»
o> +-> l O
s-c ("> • 1 —
+-> t >
<\> '<-> <\> <-+-> o o> s-i _
o <) « (1)
+-> <a i -
c o
+ J ( J tu
•» ->
(1) Q
U l
+-> U o>
4 -O)
T 3
4 -O
O l 4-> (O V .
OJ u c I O
+-> a. OJ
u u n3
+ J o 0) i-t .
o () -o c-t o
1 _
• ' "
(/> ai i -13
-o a; o o <-a. ai
s-+-> s_ o
4 -
C M
O i £
ai -t->
<o • —
u.
c • 1 —
(_) ct o c>&
o I M
+-> I O
U J
^ (/) <. 4 -
o
+-> • 1 —
1 -• 1 —
Q . i / l
ai - C
+->
33
SS SSSSÏS SÏSSS8ÏSSSSSSS3
I U M ^ w n M U M M ^ M ^ * ! ! M ^ ^ -
1 1
: f . J 5 ,
8 ï
r ! o
j I s
g
?
i L 4^ L
• B
' 3°
* i I u
i
f 1 '
• •
•
. 18
o
o
1 ft
T i *
o
o
•
— 8 ^ 1 '
• •
a
90
8
O
D
ft
?
O
a
a
CU +-> IO i .
c o
• 1 —
• ! - >
1) a> ••- a i t -
+-> o a> s-t . o o
#t
cu + J lO i_
c o
• 1 —
+ J
«> cu +J (1) o
in
+-> u CU
a i T3
l * -o cu
+ J rO s_ a i o c ro + J n. <u u u ro
+-> ( ) a i s_ s-o o
• o c ro
C
'*~ 10 a i s-3
X ) at o o s-a .
cu _c -•-> S-o
4 -
CO
• o 2 :
(U • ! - >
ro c ^ O .
c • 1 —
• <_) < t O
^•5 o r—
+ J ro
LU
s: t o <4
M -O
+ J -r-i -
• i —
a . to
a i j = + J
LO
ir>
cu
s S5S5 : 5:52 53 : i !!:!5!!5S5!5!52!!!522:5255555 : 5 -«.n»».r-..srs=z;::s:sssisi!ssss;s k
-
I
1
•
00 B-B —
*
S
o
s
fl
?
o
o
o
•
_u •
! • 1
r S
°
a
f,
o
g
o
o
a
£ _ s
i 7 F = — 2
cu + J ro s_ c o
• i —
+-> O cu
• f - » ai
<-+-> o <u s-S-O U
#» CU - l - > ro S-.
c o .(— +J
( ) cu +-> a i
</> •<-> i .; cu cu -o 4 -o cu +-> ro 4-
<D O c ro +-> O -CU O <J ro
+J U CU i-
*_ o o -o c
c •I—
i n cu u_ 3
X> cu u o 4-o. cu .c +-> i -o
M-
CVJ
• o i £
cu 4-> ro
r— ra
t i
• t_)
•* o S-Ï o •—
+-> ro
LU s: t o «a-
M-O
+ J • i — S_
•r— U .
(/) CU
- C O ro -r-
«3-
ir>
cu
34
o f u m e & ' o o w n o o s i S N w oc _ «i *** O ( K l A o S I t A O W S t t O I A ' - O I D S I I S S g ^ r-
- a J n , r l / > « > ^ - o » o ^ •
)BOSOU&ls s (D
r 1
i as « - • - • - * - a 3 S) O B S
ô * - n j n v u ) W N w a ; s * - r u u > r > c a n s >
•
ft \ o
o
•
°
O
I
Lf)
<r>
S I
O
O
— ^ -
8 ^
O
O
QJ x : +->
a. S I / ) B O I D I D O O O S O S ) S > I D < D < D < D 0 l > t 0 (O o u i n i e u i i A o ï A e s o u i s o o s i o i B B n L/I
f ' l > « * B S S I B ' * , * ' ' t l l B f i ( I S t l B I D S O * T - v ^ d C I « - v < v - T - < B ( D a D S S S ) S ) S > 0 ) S)
c o
* l —
+-> o c 3
4 -
<o
l/> ra
a» +-> 03
s-c o
• r -
+-> O ai
4->
a> Q
o z
a> +-> m
( — Q-
S_ O
4 -
i —
a> > ai •— en c
• r -T3 S-O O
a> s-
• m n T u i i g N n n .
ai s-
'
1 \ ^
o
I >
o
8°
Q
O
'
s
I
1
1
0 ^"N-
'
m
•
0 °
8
O
R
o
Q
O
35
MDDF
9 - i . ^ _ — j — •-
0 . r 50V.DAC 35V.DAC 20V.DAC
1 ^ 10V.DAC
CUT-OFF LEVEL
Figure 5.7. : Detection rate as a funct ion of the report ing level fo r a l l four plates
MCRF
1 - -
. 5 -
50V. DAC 3 S V. DAC 20V.OAC 10 V. DAC CUT-OFF LEVEL
Figure 5.8. : Correct re jec t ion rate (ASME XI IWB 3150) as a funct ion of the report ing level of a l l four p la tes.
36
5.2.2. Supplementary techniques as a parameter
a. It was possible to identify several groups of procedures at the 20% DAC reporting level. Table 5.1. illustrates the results obtained on plate No. 2 for these different procedures.
Procedure Code
(a) 20% M,U, (b) 20% M,U,S, (c) 20% M,B,S,TD,
MDDF
.76
.94
.91
MCRF
.79 1. .88
MCAF
.80
.65 .75
Table 5.1 (Plate No. 2)
Addition of the supplementary S (surface probes) technique clearly increased the quality of results. The contribution of the tandem technique proposed in (c) is not clear. The best procedure of the group would simply be ASME at 20% DAC, applied through the unclad surface, with the addition of 70 angle probes (often the 70 SEL probe).
These trends are confirmed on the other plates (Figures 5.9 and 5.10). The results for nozzle plate No. 9 also illustrates the importance of supplementary techniques:
Procedure
20% M,U,
20% M,B,S
MDDF
.74
.90
MCRF
.50
.72
MC
.9
.8
Table 5.2 (Plate No. 9)
5.2.3. Scanning surface as a parameter
There is some indication (which may not be significant) that when scanning is carried out on the clad surface, the cladding makes sizing slightly more difficult (see Tables 5.3 and 5.4).
37
.
i . ;
g
t ssss s ssss o . . . . . j ^ . . | £ £ E E S £ £ E S £ £ £ S £
ss s
SSSSSSKSSSSSSS^ %
* i
h 8 otn
.i
» i
M : i ' b ! Si
dtO
.1
o
ï
i I
8 s . s
i
s
D
R
o
rect
re
ject
ion rat
ice
rate of pro
cedu
of AS
ME XI at 20
ntar
y pr
obes (70
) on the clad
side
„ QJ -r- Q. C
r j <J wi 3 o
ecti
on ra
corr
ect e
in the
, with
s
manu
al s
ai c a> «=c c o <o s_ o ra
CD
e 5.1
C*vJ
O
a S £ £ £ m £ £ £ i » m £ £ £ m m £ Ë 2 !
» SîS? ? S2!S!S 5
1 i
'
«
8 .
O
* T
P
- » - . ,
| •
•
J
S
p
O
O
o
c o • ï -
• ! - >
O ai
• o ai i -
+-> u ai s_ s-o u
•* a> + J m s-
c o
• 1 —
-l-l u ai
4 J ai
c=i
proc
e
4 -O
ai +-> IO s-
ai o c ra
+ J
o. ai o o ra
+-> u ai i_ *-O U
•* ai +-> ra s-
20%
4-> ra
i—i
X
LU y . t o <t
<+-o
•>->
*|— S-• 1 —
CL m
ai J= •!->
c • i —
in ai s-3
T3
ai sz -t->
R o i -
4 -
cr c •«— c c-ra u m
r— ra "3 e ra f-
c + J •c—
* « o <r a
* ai •o • i —
« •o ra
i —
o c 3
ai s -3 en
38
Procedure
(a) 20%, M,U (b) 20%, M,C (c) 20%, M, B (d) 20%, M,U,S (e) 20%, M,C,S
MDDF
. 76
. 81
. 83
. 94
. 98
MCRF
. 79
. 77
. 82 1. .97
MCAF
. 80
. 79
. 71
. 65
. 69
Table 5.3 (Plate No. 2)
Plate No. 9 shows a more noticeable difference between the surfaces with a higher detection and correct rejection rate through the clad surface. The result is probably due to the distribution of defects in plate No. 9 with many defects near to the clad surface. Insufficient results are available on plate No. 3 to form meaningful sub groups of procedures through the clad and unclad surfaces. In plate No. 3 the majority of defects lie near to the clad surface.
5.2.4. Comparison of conclusions with individual team results
Using results from individual teams it is possible to confirm the trends suggested in paragraph 5.2.1., 5.2.2. and 5.2.3. This is possible because some teams (EC, NK, SE, SD, LB, IC, RK, BC) submitted results on a series of independent examinations using different procedures. Thus, using the multiple sets of results from a single team, it is possible to investigate the effect of varying recording threshold, use of supplementary techniques, etc. Results of the EC team are shown in Table 5.4 for plate No. 2. The results confirm the above conclusions as follows :
i. Decreasing the recording level from 50% to 20% DAC improves detection and correct rejection of defects. However, further reduction to 10% DAC has no significant effects.
ii. Supplementary techniques significantly improve the correct rejection rate, when complementing the 50% DAC ASME procedure.
iii. The difference between the contribution of the surface probe and tandem techniques is not clear. However it has been demonstrated in Table 5.4. that using both improves results.
iv. No significant difference appears to exist between inspection through the clad and unclad surface of plate No. 2. Although comparing the 50% M, U and 50% M» C results in Table 5.4. suggests that undersizing of rejectable defects occurs through the cladding as CRF decreases from .5 to .42 through the cladding. Assuming that the Tandem technique does not reduce sizing performance, the same conclusion may be reached by comparing the 20% M, U and 20%, M, C, TD results.
Table 5.5 illustrates the results of team LB on plate No. 3, confirming several conclusions but showing a counter example of the trend previously identified on the effect of the cladding characteristics.
. The inspection through the unclad surface without supplementary probes showed that the defects near to the clad surface are difficult to detect even at 20% DAC. In these conditions, a possible effect of the cladding was hindered by the location of the defects (majority within the first 60 mm near to the clad).
. The addition of S probes improved detection and correct rejection
rate. . On plate No. 3, inspection of both sides was often of benefit for detection because of the nature of the defects introduced.
Results of team RK, also shown in Table 5.5 would confirm the possible negative influence of the cladding on sizing performance.
Team
LB000699 LB002599 LB002699 LB004199
RK000699 RK002697 RK004197
Procedure
20%, M,U 20%, M,C 20%, M,C,S 20%, M,B,S
20% M,U 20% M,C,S 20% M,B,S
DDF
.23
.55
.74
.84
.53
.60
.77
CRF
.0
.43
.64
.64
.36
.21
.57
CAF
.92
.85
.73
.58
.92
.8
.72
Table 5.5 (PI ate No. 3 )
5.3. Evaluation of the automatic scanning procedures in the spirit of ASME
5.3.1. Global results
The group of mechanized procedures was split into groups corresponding to those defined for the manual procedures. However, the limited number of procedures available made it possible to define only a few groups :
a. all mechanized scanning procedures, b. automatic scanning at 10% DAC with supplementary probes, c. automatic scanning at 20% DAC.
Results are shown in Table 5.6.
Peocedure Code
Mechanized
All mech.proc.
10SDAC,A(A11)
20«DAC,A(A11)
Manual
10«DAC,M(A11)
20%DAC,M(A11)
MDDF
.61
.60
.40
1.
.73
Plate 1
MCRF
.68
.86
.57
.71
.71
MCAF
.64
.63
.75
.50
.50
MDDF
.9
.98
.86
.94
.91
Plate 2
MCRF
.87
.91
.73
.97
.93
MCAF
.84
.66
.73
.69
.65
MDDF
.89
.91
.83
1.
.84
Plate 9
MCRF
.63
.61
.67
.94
.64
MCAF
.72
.72
.5
.56
.83
MDDF
.84
.90
.67
(1.)
.75
Plate 3
MCRF
.85
.90
.73
(1.)
.49
MCAF
.71
.69
.74
(.67)
.78
Note : Values of MDDF, MCRF and MCAF for Plate No. obtained from one team only.
Table 5.6.
3, 10% DAC M, are
These global results show that no clear conclusion can be drawn but that manual procedures at 20% DAC for plates No. 1, No. 2 and No. 3 appear to be better for detection (MDDF values). Defect evaluation with manual scanning was better on plate No. 2 than with mechanized scanning (MCRF values) in these complete procedures at 20% DAC.
The results of the mechanized procedures used at low cut-off level (10% DAC) are good but they consider both surfaces for scanning as well as surface probes and tandem techniques (see also section 5.3.3.). This conclusion deals only with ASME type procedures conducted in both ways : manual and mechanized. In practice, mechanized procedures are more complete.
5.3.2. Evaluation at the level of teams or procedures
Some of the teams parametrized their results as shown on Table 5.7 for plate No. 2.
The individual team results that were used in groups of procedures show that ASME type procedures at 20% DAC conducted automatically show again a slight reduction in detection performance when compared with the corresponding manual procedures (Table 5.8).
Procedure
10%, B,S,TD
20%, U,S
Ma MDDF
1.
.94
nual MCRF
.97
1.
MCAF
.65
.65
Automatic MDDF MCRF
.98
.80
1.
.78
MCAF
.66
.72
Table 5.8 (Plate No. 2)
This is even better illustrated by the results of individual teams such as those for team "ES" which correctly used the same procedure in both scanning modes :
For plate No. 3, as shown in Table 5.6, the results are clearly in favour of the mechanized procedures. However, care must be taken in concluding that the improvement was due to the introduction of mechanized scanning : this group is composed of procedures which combined techniques for detection at low cut-off level, with the introduction of tandem as well as other supplementary techniques such as T0FD.
The present global evaluation does not allow a strict conclusion. A further study is necessary to know if, in the present case, the benefit gained is due to the mechanized scanning or to the supplementary techniques applied at low cut-off level.
42
5.4. Evaluation of special procedures
Several special procedures, some of which are currently used in plants for actual inspection and which cannot be put in the group of ASME procedures (or procedures "in the spirit of ASME), are considered separately. These procedures usually use a low recording level or complex combinations of techniques. Most of these procedures were applied on the PISC-II plates with mechanized scanning. Table 5.9 shows the results obtained by the teams using these special procedures with a mechanized scanner; an indication of the techniques is given. The group contains many good results on each of the plates. In particular, on plate No. 3.
TEAM
BA007599 BDOOS599
DB005599
EF003699
IS007399 KR00S299 LN009199 SR002499
SM007297
SUDT9450 TH007099 YC008097
PROCEDURE
Phased array Complex comb. of techniq. Complex Comb. of techniq. ASME,Complex Comb, of techn. MBSP VAP TOFO Complex Comb. of techniq. Contact Foe. probes EMATS Tandem only Immersion
Foe.probes
DDF
.7
-
.93
---.80
-
.53
.60
.70
~
Plate 1
CRF
.4
-
.86
---.80
-
.43 -.80
~
CAF
1.
-
.14
--
.40
-
.88 -.80
-
DDF
.89
-
1.
-1. 1. 1.
1.
.75 1. .94
1.
Plate 2
CRF
.91
-
1.
1. 1. 1.
1.
.60 -.90
1.
CAF
.57
-
.69
-.82 .56 .46
.65
.88 -.63
.35
DDF
.75
-
-
-.75 .92 -
-
-.92 .73
1.
Plate 9
CRF
.22
-
-
-0. .67 -
-
-.67 .88
.71
CAF
1.
-
-
1. .67 -
-
-.78 .71
.88
DDF
.50
.97
-
.94
.43 -1.
.97
---
.8
Plate
CRF
.2
.86
1. .60 -.93
.86
-
.79
3
CAF
.85
.96
-
.62 1.
.38
1.
-
.76
Table 5.9.
5.5. Sizing techniques
Discussions of the Referee Laboratory with several teams led to the conclusion that it was unwise to group advanced techniques into categories or families. Each technique or procedure has to be considered separately. However, the global average results of all the advanced techniques together were compared with the results obtained with ASME procedures at 20% DAC (or "in the spirit of ASME").
43
5.5.1. All advanced sizing techniques
All the advanced techniques involved are detailed in Table 5.10. and Figure 5.11. Some of these sizing techniques obtained high CRF' values on several plates as shown by MCRF', average value of CRF' for the three plates. The average CRF' and CAF' values for all advanced techniques taken together, plate by plate are shown in Table 5.11.
Plate 1 Plate 2 Plate 9 Plate 3
MCRF' MCRF' MCRF' MCRF"
= .85 = .82 = .78 - .60
Table 5 11
MCAF' MCAF1
MCAF1
MCAF'
= .61 = .67 = .55 = .72
A better measure of the performance of these sizing techniques is the error in sizing obtained. Such values are available for the individual techniques in Tables 4.3 and 4.6. The average values and the corresponding dispersion are shown in Table 5.12. These should be compared with the global results of Table 4.1.
Plate 1 Plate 2 Plate 9 Plate 3
MESY (mm)
* - 72.7 - 8.5 - 15 - 15
SESY (mm)
154* 35 30 36
Tabl
MESZ (mm)
** 15.8 3 0 1
e 5.12
SE (m
21 20 12 22
Average sizing in Z is often exact but there is an important dispersion in all cases; undersizing is frequent in Y (length of the defect).
5.5.2. Comparison of the performance of the advanced sizing techniques with that of the procedures in the spirit of ASME
The diagrams in Figure 5.12 relating to the results achieved by the ASME procedures should be compared to the ones in Figure 5.11. Although the meaning of CRF' is different from that if CRF (Ref. 4) it is clear that no substantial improvement is obtained when all advanced sizing techniques are considered together; even if a trend appears which is in favour of these advanced sizing techniques.
* Unrealistic average value due to partial detection of defect No. 1 ** Confusion of defects with base material defects.
44
CO
o
OJ +J (O
eu
e n
o
a i
+-» to
Ou
C M
O Z
a» •4-» «O
Cu
O z
at + J (O
Û -
UJ =3 O *
Z 3= O LU 1—
~- UJ
oc => o LU CJ O
oc CL
E
<c LU h -
LU
< CJ
U -OC CJ
U .
CJ
Liât: CJ
Lu < CJ
Lice: CJ
U -
< CJ
Liât: CJ
N N W r - O • r - r»«. * * - r-. co i n • •
i— . . . . | , , . ,— ) 1 r— 1 1 1 1
i n i n i— o r-* m N N u > m . v o
i <— i • « — t i i — i i i t
co r». • r o i o i o m o o o o o o o
^ - m co r o m m r o co i n i - oo co . c o > co N • r o . a o • r -
o «— r » m i n * r • i n • i— m o N • i n • • «—
r^ oo o r o o r*- K O u> co • m e n r v • # • oo v o oo •
' " " " " " '
r o o O O r o . o CD
i i — i i t t i . . . • i <— • i i
r - r * o m • i o • * n i n r-.
i •— i i i t • •— •— - i •—
ncen
tric
Ar
ray
.Hol
o.Cl
ad Si
de
logr
. Un
clad
es 0, 45°
oc.
prob
es
, U
+C
wave
s e
Holo
gr.,
Cr
eep W.
O + J C J O - Q Lu « t t= C J * « i •> m o + - T -
l/> 3 LO C. . >— Q 1». r - » CJ • O O O • Q . > , D L . 1 - L. o < D « ) O U ) P .— - < 3 * + o t / i «a: co irt • c z ( / i » i / i • ^ •
- i n m 3 o o + • s~ • c. e * : E « * >> «a- o o » • se - o c. L. * e n O <C U U L L , t / i O S - C j a i < C C n c n O CM U > , C > » < O I - C C T 1 U J + J O >> <—
• • 0 . 0 ) 0 . • . o o o x t . - o <— * o o o 3 o • <c o L U r— a i A i o c ? :
= ) • u c r o c r • c. o • u v i ï L * > c j i / i a i i / > a > : 3 o . ^ X 3 : c f f i <c L. J *
E O O L. O C • A3 • i-> O SI • O) «ï Lu t . Cn U M - L. E 01 4J h - O CL -*-> x> +-> 01
•• p i - * j T c n i i i a . w + ui a» 3 o . h - - 0 4 J O + J O T 3 3 ^ 0 1 - 3 trt O-Lu 5= U r - Ol *— *— C O O O - O L u O «O E CJ. « t > - Q . 3 D . 3 O «J -r- (J —J O O O . C O T -t o o t o x L O ï : a : i — i— < c « c z i — < a . c j t —
e n t o • ^ vo e n i— r o ^
o c n c o t n c n i n c n c n c n o i c n c n c n o i o c n c n <— < T i r - . t o r o ^ o c r * C T > c f t C h < T » c \ J C T i c n i ^ c r > c r > ( 0 * N N < î t f ) | N 0 0 1 U ) W i - (\J r-> 0 0 N <T> C h l — l / ) C O C O N N N I O O > l i ) r s O ) N C O N l O N N U N N N N N N N N N N N N N N L O t / ) U J t / ) l / l < / l < / ) L O L 0 L O L 0 l / ) L O t / l 0 O L 0 L 0 I - I U U U U 3 3 0 H < O Û 3 < O H Z ) I C O U U J L J I i J U Z : £ û . O ^ O H £ C t H S : |
c s o
1/1 E n3 O)
O 1 —
• LT)
OJ
-X I ro h-
o ro d)
>» JD
T3 OJ S -
a j
-a • ï —
i / i
c o o
o eu
M-a> -o 4 -o i -(U
J3 E ^ c ai
- E +->
OJ +-> O z:
<£>
• <d-
LD
> «3"
<3" •
«3-
CO
• *d-
i / i r—
OJ J3 ro
+J
C o
45
o z ai
• •
»
•
•3-
O
O Z ai -t-j
(O
O -
-*—a
.—— »
. — 9
| - 3
• _ 8
cn «3-
C£ C_>
ai +-> «o
4
•
_ 3
J fr
CO
C-J
O z ai
s
« 1 \
•
• "
8
È
CO
ia
4 -o
..—. -Li_
cc <_> —-
CU +-> (O
s-
c o
.f—
+-> o ai
•<-> CU S_
+•> U CD S-
s_ o o
'— 10
s_ o
4 -
w ai 3 cr
• i —
c - C o ai +->
cn c
• i — rsi
•i—
cn
T D CU CJ
c: IO
> -a lO
• t o
ai +-> to
i —
a . t—• i—i
i
o 1/5 i—i Q -
in a> 3
a i
o z
1-
(U S
4
! 1"
i
* * ?
' *
« J t
cn
a: o
.— x to
o </>
fO s-
c o
•(— +-> o CU
•<-> CU
s-
+-> u CU
s-s_ o <_>
U -l/>
ai . c +->
c: • i —
i / i CU
s-3
-o ai u o s_ a .
# t_) «£ Q
<î^ O CNJ
+-> <D
3 cn
O z a i
- e
s
« 9
s r
IT) 00
(X. c_>
o z CU
I
i
?
s
3
! r
o U3
46
A more stringent comparison is made in Figure 5.13, showing the cumulative diagrams of error^ in sizing. Dispersion have to be considered as shown in Table 5.13.
advanced
1E 20% DAC
1E 10% DAC
P la te No. 1 MSESZ (mm)
21
50
50
Pla te No. 2 MSESZ (mm)
19
23
22
Pla te No. 9 MSESZ (mm)
12
15
14
Pla te No. 3 MSESZ (mm)
22
23
19
(DT procedures are excluded : e.g. NK DT 5497)
Table 5.13
Dispersion of sizing errors is often slightly better but of the same order of magnitude for the advanced sizing techniques as for the procedures in the spirit of ASME. Advanced techniques for sizing have to be considered individually as shown in Table 5.10. Several of them show high potential performance, but such techniques often lack experience on real defects (as has been recognised by developers).
5.5.3. Effect of the cladding on sizing performance
The results obtained by one team, EC, illustrate the possible influence of the cladded surface on the sizing possibilities with advanced sizing techniques. When exactly the same procedure was applied in examining both sides of the three plates separately, the cladded surface appeared to cause a reduction in performance in several cases as shown in Table 5.14.
Team
ECSZ8265 ECSZ7665 ECSZ8439/16 ECSZ7839/69
Procedure
Mu l t i f . Hologr., Clad Mu l t i f . Hologr.. Unclad Spectroscopy 45 ,60 ,Clad Spectroscopy 45 ,60 .Unclad
Plate No. 2
CRF'
1. .90 .50 .88
CAF'
.0
.71 1. 1.
Plate No. 9
CRF'
1. 1. .83 .87
CAF'
.33
.67
.6
.8
Plate No. 3
CRF'
.71
.33
.42
CAF'
.50
.83
.67
Table 5.14.
* Table 5.13 contains results of teams using advanced sizing techniques only. Such results are identified in Tables 4.3 to 4.6 and 5.10 by the characters "SZ" in the Code (e.g. BT SZ 9310).
47
ADVANCED SIZING ASME 20% DAC
CO
o
_ J
IO
•
1
CO L U
CM
CM CM
II
1
r i
i t
o
Q .
O co
- 1
a L C N
CD
>
1
c o LU
CM
O CM
II 0 )
' ' ' 1
1 1
1
\
O C 0 t 0 « - C M O C 0 t D ' « - C M C M o m o o r ^ t o ^ - p j . r M —
I I I L. m CM 3
c o
o CM
œ o
(S o>
• • CD
CM tv
O to
CO
•»" ce o *-CM
M
—
I I I I ' '
t/1 t | _ •—< 5_ O X O S- - o U J
îr, c ^ Cl) (O OO
c (U M -• i - 3 O M O "
'£ 'c £ M - O O CU
- M co E en <o c S- - i -C7> M (O - r -
• i - </)
-o eu ai > <-> •r- c
fO > r— " O 3 (O E 3 M -
C_> O
CO
CU
C J
o
o CM • a
eu u . e o -t-» S- - i -Q. S
. a co
-120
a CM
~ O Cû
— ir>
eu s-3 C7>
U_
Pi CO
10 Irt
01 • • •
CM
*-IO m
CO CM _ CM
48
6. SPECIFIC CASE OF PLATE No. 3
Plate No. 3 incorporated a real PWR nozzle; it contained several families of defects and thus represented an important test plate for the evaluation of NDT performance.
Results are shown in Table 4.6, but it is important to extract from them typical figures describing the capability of the procedures on this sample with a real geometry.
6.1. Procedures in the spirit of ASME, manual examination
The contribution of supplementary techniques (e.g. 70 L, SE) is always identifiable but with low statistical significance. It is to be noted
o that the 70 L, SE probes used on plate No. 3 often limited their inspection to the first few centimeters of the wall thickness under the cladding. The effect of the cladding is not identifiable but it must be remembered that two thirds of the defects were in the first 12 cm near to the cladding. Team EC again has parametrized its procedure as shown on Table 6.2.
This table is the only one in which an important contribution of the tandem probes is clearly shown when compared with Table 6.1.
49
6.2. Procedures in the spirit of ASME, mechanized scanning
Table 6.3 should be compared with Table 6.1
MDDF MCRF MCAF
ASME 10% DAC, (all) ASME 20% DAC, (all)
.90
.67 .90 73
.69
.74
Table 6.3
The above results of procedures using mechanized scanning with low reporting level (10% DAC) generally involve complex combinations of techniques. This is not the case for the other plates where it was possible to compare the performances of the same procedure conducted with both manual and mechanized scanning.
6.3. Special procedures at low reporting level
The group of special procedures demonstrated high performances. Parameter values my be compared with those of tables 6.1 and 6.3 for procedures at 10% DAC in the spirit of ASME, manual or mechanized. Results are shown on Figure 6.1. at the global level for both parameters DZ and DY.
This group relates to results from 6 teams as shown on Table 6.4.
Complex combin. of techniques ASME 10% DAC+suppl. techniques T0FD Complex combin. of techniques Complex comb, of tech. (AL0K) Immersion Focusing Probes
DDF
.97
.94 1. .97 .81 .80
DDFR
.93
CAF
.96
.62
.38 1.0 .31 .76
CRF
.86 1. .93 .86 1. .79
Table 6.4
50
i i » s
D a
0
B
o o
o
o o 1
o in ' .
on
i i s
p»
B o o
o
o
o
o
o
>
i
o
t c
» s 3
?
« i
s s o
° o
o
0
o
*8 o o • o
o
o
s 6 s s ?
s s
o
o
?
3
3
S i » 3 o F : 5 5 S S B s s B o g a a o t t ss:
• • • • « 3SSSSSS8S8SSSS8SSSS: S D ! 3 S S S S ! î 5 S S ' : t ! ? t S 5 t ! ! ï 5
» » h M A — * ' C ï ! î a t ! ! 2 5
9
o
• S 8 °
...
r
C3
O
9
•
a
a
o
1 °
o
D
D
ft
•
9
•
a
ai i_ 3
"O O) o o s-CL
r— rO
• r -O ai a. m at .c + J
4 -O
l/l
+•>
<) a» 4 -O)
• o
.c +-> o
-O
s-o 4 -
co
>-o "O c rt>
M O
B
o 8
8 00
r D
7
a
D
O
r>
o
I
s- a i + J
•— <o (O ,—
JD O . O
•— C en o
a i
51
Table 6.4 shows that the objective of PISC-II was achieved: high-performance procedures have been identified. At a later stage, however, it will be necessary however to differentiate, between the procedures which are applied daily in nuclear power plants and those which are still proposals. For example, it is interesting to note that the team applying procedure YC008099 scanned plate No. 3 in its entirety in less than one hour and produced the results of their inspection in a few hours. The question of inspection time and team resources is treated in chapter 13.
6.4. Advanced sizing techniques
As a complement to Table 5.10, which could give an incorrect, pessimistic impression of the possibilities of sizing, Figure 6.2. illustrates how well some teams were able to inspect plate No. 3.
7. DETECTION AND SIZING OF INNER RADIUS CRACKS IN PLATES No. 3 AND No. 9
7.1. Introduction
Plates No. 9 and No. 3 of PISC-II both contained artificial defects in the inner radius corner of the nozzle. In plate No. 9, three deep and long fatigue cracks were introduced, one of which was partially repaired according to the ASME procedure. This left a large part of the inner radius unclad, thus introducing geometric reflectors for the ultrasonic inspection. As explained in Ref. 3, radiographic examination has confirmed the existence of these cracks. Some cladding imperfections also appear to exist in plate No. 9, but these defects will be considered later. Three small fatigue crack simulations were introduced into plate No. 3 (Ref. 3). These defects have also been confirmed by radiographic examinations.
7.2. Parameters used for the evaluation
The Z-coordinate recommended by the Managing Group was used. Dimensions were thus defined as shown on Figure 7.1. The reference dimensions of the inner radius cracks were as follows :
No team used more than one inspection procedure in searching for the inner radius corner cracks. Only three groups of procedures can be identified :
a. surface probes (70 S or L or SEL), b. examination from the outer surface (e.g. with 60° S probes or 45° S
probes), and c. special techniques such as surface waves, T0FD, focusing probes,
phased arrays, and line holography.
The results obtained by the teams are shown in Table 7.2. The results can be summarised defect by defect as shown in Table 7.3. Figure 7.2 shows the cumulative error diagrams for all teams.
Some remarks can be made on the basis of these tables.
a. Detection
Average detection appears good : MDDP = .89. It must be pointed out that several teams did not apply techniques to detect the defect around Y = 90 in plate No. 9 because of the large repair without clad. Taking this into account, an MDDP calculated on a reduced set of accessible defects would be .94.
b. Location
Errors of location are often within 25 mm, both for plate No. 9 and plate No. 3. Some large errors are difficult to explain. Generally, location appears more accurate and exact for plate No. 3 than for plate No. 9. The average error of location in angle (polar coordinate) is MELY = 4° for plate No. 9, and MELY = 0.1° for plate No. 3.
90°L C . W . p l a t e 3 ; 7 0 ° S , U , A , p l a t e 9
S e v e r a l 70 L m u l t i p l e x e d , A,C
7 0 ° L , T . R . . M,C
7 0 ° L , A,C
7 0 ° L , A,C
14° w i t h 0°L T . R . , M,C
7 0 ° L , T . R . , A.C+U
7 0 ° L , H,C
TOFD, A,C
60S, M,U
7 0 ° L , T . R . , H.C
7 0 ° I . , T . R . , M , C p i . 3 ; 7 0 ° S , M,U, p l a t e
7 0 ° L , T . R . , M.C
ASME X I , 2 0 , M.U.S
21L, 60S, H.U
7 0 ° L , T . R . , A.C * TOFD
TOFD,Phased A r r a y s , L i n e H o l o g r . , C + U
7 0 ° L , T . R . , M.C
Immersion Foe . P r o b e s , 70 L, A,C
EMAT's, H , C+U
9
Table 7.2 : Teams results
Defect
9 . 1 .
9 . 2 . .
9 .3.
3 . 1 .
3.2.
3.3.
DDP
. 7
1 .
0.84
0.95
0.88
0.95
•
MELY
deg
-2° 4.5
6.4
- . 4
- . 8
1.4
SELY
deg
4.3°
11
15
4.8
4.2
2.8
ME SX
nun
- 86
- 26
- 17
5
5.5
7
SESX
mm
9
26
15
14
14
14
MESZ
mm
- 87
- 33
- 27
- 8
- 6 . 5
- 2
SESZ
mm
24
20
24
11
12
11
MDDP = . 8 9 P l a t e 9 : MESZ = - 49 mm P l a t e 3 MESZ = - 5 . 5 mm
MESX = - 43 nun MESX = 5 . 8 mm
Table 7.3. : Statistics for each defects
D e f e c t
9 . 1 .
9 . 2 .
9 . 3 .
3 . 1 .
3 . 2 .
3 . 3 .
DDP deg
. 75
. 5
.5
MELY
deg
2
7.5.
24
- 3
- 2.5
3
MELX
mm
- 2 4 . 5
- 92
- 3 . 2
- 7 . 5
-3
MELZ
mm
- I l l
- 50
- 52
- 13
1 5 . 5
6
Table 7.4. : Sta t is t ica l for each defect for procedures based on the external surface.
56
The "envelope" of a defect was introduced to represent the size of the defect. Such a representation is not appropriate for inner radius cracks. However data were furnished in this way by the teams, who followed the format proposed by the PISC Managing Group. There are two important dimensions of the inner radius cracks : the length and depth. In the PISC coordinates one has the dimensions DX and DZ, but for plate No. 9, neither DX nor DZ represents a measure
of the crack depth. For plate No. 3 the envelope (DX, DZ) better defines the defect dimensions. The overall average sizing errors values appear to be negative :
MESX = - 18 mm; MESZ = - 27 mm
There is thus a systematic undersizing of DZ for both plates and this appears to be very large for plate No. 9 (Table 7.3). This raises the question whether several teams found the defect dimensions in plate No. 9 unbelievable. For dimension DX, different conclusions must be drawn for plate No. 9 and plate No. 3 :
. teams tended to undersize defects in plate 9 : MESX = - 43 mm; and
. teams tended to oversize defects in plate 3 : MESX = 5.8 mm
Table 7.2. shows that exact results were obtained by several teams :
. team BD (using a complex combination of techniques);
. team AN (using creeping waves, 90 L probes, 0 angle probes, inspection of the clad side on plate No. 3 and unclad side on plate No. 9);
. team EF (with 70 surface angle probes, inspection of the clad side); and
. team YC (immersion focusing probes, inspection of the clad side).
There may exist errors of interpretation, for example confusion between the defect depth and DZ.
7.4. Evaluation at the level of procedures
a. Procedures based on 70 angle probes (70 L, T.R.).
No important remark can be made except that good results were obtained that belong to this group.
b. Procedures based on scanning the outside surface.
Such procedures gave rather large errors as shown by a comparison of results in Table 7.4. with those in Table 7.3. It is to be noted that defect 9.1. was well detected from the unclad side. The techniques were not disturbed by the repair of this inner radius crack.
8. DISCUSSION OF RESULTS AS A FUNCTION OF THE TOLERANCE ON SIZING
8.1. Introduction
The ASME criteria IWB 3000 of Section XI often appears to teams to be too restrictive. Several defects are near to the border line between acceptable and rejectable defects areas (Figures 4.5. to 4.11). A small sizing error (e.g. of 1 mm in undersizing) has given several teams a CRF value which does not represent the quality of their results. Some examples are clear on plate No. 3; for example :
BD005599 : CRF = .86
whereas the average sizing error on Z is only .25 mm.
A discussion of the CRF parameter values is thus necessary when taking sizing tolerance into account.
8.2. Tolerance on size as declared by some teams
Several teams declared an uncertainity on their results for the size of the defects. This declared tolerance or error band was generally between +_ 3 mm and +_ 5 mm.
These declarations were taken into account is a positive way : when a team accepted a rejectable defect, the CRF value was computed again, increasing the defect size by adding the declared tolerance to all three dimensions of the defect.
Table 8.1. shows how the results of these teams were modified where it was fair to use their declaration of error band or tolerance on size. The procedures considered here are full detection and sizing procedures.
Team
BD005599
LN009199
SR00AZ99
DDF
.97
1.0
.97
DDFR
1.0
1.0
1.0
CRF without toi.
.86
.93
.86
with toi.
1.0
1.0
1.0
CAF without
toi.
.96
.38
1.0
with toi.
.96
.38
1.0
Table 8.1. (Plate No. 3)
58
High performances could thus be obtained by these procedures when the ASME IWB 3000 criteria was modulated by the sizing error declared by the team : DDFR = 1., CRF = 1., CAF = 1.
8.3. Sizing tolerance as a parameter
a. Genejra]_ £e su_[t£
On the basis of the above results which show the potential performances of three special full procedures, it appeared that the tolerance on sizing could be used as a parameter to evaluate which average sizing error was characteristic of a particular procedure. This was done both for full procedures and advanced sizing techniques. Figures 8.1. and 8.2. show how CRF and CAF vary as a function of the error in size. The best team's curve is included as a reference, see section 4.1.). Observations can readily be made on :
. the average increase of size necessary to reach the best possible CRF for each procedure;
. the upper limit of CRF which could be reached by a particular procedure (corresponding to the DDFR value); and
. the consequence that a high CRF can lead to a lower CAF value and, thus, unjustifiable repairs.
b. £ajltj£iulair_cas£ of_tjTe_S£e£iaJ_p£0£edure^
Figures 8.3 and 8.4 show the results of the six special procedures considered on plate No. 3. Most of the procedures reach CRF values of 1 for a small increase of the average defect size. If the detection of rejectable defects is not perfect, the CRF value cannot reach 1.
c. £a£tj£ujj ir_cas£ £f_tjie_advan£ed^ £i^i£g_t£c]inj_que
Figures 8.5 and 8.6 give the results of these techniques as a function of the tolerance on size. It is to be remembered that most of these techniques considered only few selected defects (Table 8.3) and the imperfect results of technique EC have to be balanced against the fact that many defects were considered by this technique on plate No. 3.
d. £o£cJ_usi£ns_
Major observations are grouped in Table 8.2 and confirm earlier conclusion on the best possible selection of cutt-off level. For defect evaluation, 20% DAC is often better than 10% DAC. If a pessimistic assumption is made : real life could induce inspection teams to under-size defects to be declared, figure 8.1 to 8.6 show how CRF values could drop, in average, for all procedures. Table 8.3 gives all the parameter values for the results judged to be best for each team and at the sizing tolerance values - 10, - 5, 0, +5, + 20, and + 30 mm.
59
NOZZLE N. 3 1.0-4
Figure 8.1 - MCRF values for groups of procedures as a function of the tolerance on sizing
NOZZLE N. 3
M C 0 . 5 ft F
0.0
- I t iy i i ri'i PTT *j I'M i | t i I i M i i r 1 1 1 i i M i i i [ t T f i 1 1 r I I j 11 i i T t 11 qw
10 15
TOLERANCE(MM)
E0 25 30
BEST TEAMS SIZ .PROC.
SPEC.PROC. ASME 50 flSME 10
flSME 20ALL
Figure 8.2 - MCAF values of groups of procedures as a function of the tolerance on sizing SZ.
60
NOZZLE N. 3-SPEC.PROC. 1.0-1
C 0 . 5 R F
0 . 0 -
-10 M i " M i " i'
5 10 15
TOLERANCE CMM)
I ' T ^ ^ I i i j i i i i i i
20 25 T
30
BO EF LM SR UH YC
Figure 8.3 - CRF values for the Special Procedures as a function o of the tolerance on defect sizing.
NOZZLE N. 3-SPEC.PROC.
1.0-
C 0.5 ft F
.0-
-5 5 10 15
TOLERflNCE(HM)
20 25 30
BD EF LN SR UH YC
Figure 8.4 - CAF values for the Special Procedures as a function of the tolerance on defect sizing .
61
NOZZLE 3 - SIZ.PROC. 1.0
C 0 . 5 R F
0 . 0 -
TB
/
/
&•• y - ~ - s ; / _ . _ . - - j C
:z / /
JbT
T"t ,•^•«'^^^^p•*•»^^^^T ,^»*
-10 5 18 15
TOLERflNCE(MM)
20 25 30
BT • — E C
,' . PO OJ
CH GI MU RO TB
Figure 8.5. - CRF values for advanced sizing procedures or techniques as a function of tolerance on defect sizing.
NOZZLE 3 - SIZ.PROC.
c ft F
I.B-j
9.5-
8.8-
T B v y*
V Vi. X
G l / \ » V
\ \HW
\
t
X
"V
1 1 1
O J
\
i i i »
V.
i
X »
1
\
\
V \
\ '
\ " \ \ •. \?° \
EC \ \\>V-\ \ \ \ • •
x \ \ 'RD vA
Y&T
CH
" " ' " • * \
1
lEi ; 13 15
TOLERANCE (tlîl)
28 25 38
BT EC
" PO ÛJ
CH Gl MU RD TB
Figure 8.6. - CAF values for advanced sizing procedures or techniques as a function of tolerance on defect sizing.
62
Firm conclusions should not be drawn from this parametrization of results as only average values are considered for such an evaluation. Limitations observed in the general evaluation of results for several procedures are, however, confirmed, as well as a ranking between the ASME type procedures.
Special procedures appear to be capable of achieving the level of defect detection necessary for safety purposes if inspection teams take account of the uncertainities in their observations of defect size.
The general conservative attitude adopted toward defect size in fracture mechanics assessments would in any case ensure that uncertainities in defect size are always taken into account.
9 9 0 0 > 9 ff»o*r**caer» p*o»co—ca f - * o c o r « c * N N N N N W W W W W 3 0 0 0 « C h c C O H
CO
00
eu
64
9. ANALYSIS OF RESULTS AS A FUNCTION OF SUBGROUPS OF DEFECTS
9.1. INTRODUCTION
Groups of defects were selected for a particular analysis of results. These are defined as a function of their size, position and category :
a. defect near to the surface and : with ligament size SL between 0 and 18 mm (0018);
b. defect with ligament SL between 18 and 50 mm; c. defect with ligament SL between 50 and Z max; d. defects near to the external surface; e. defects ranging in size from 0 to 10 mm; f. defects ranging in size from 10 to 25 mm; g. defects ranging in size from 25 to 80 mm; h. defects of the fatigue crack type (family "A"); i. defects of the volumetric type (family "C"); j. defects of the intermediate or unrealistic type (family "B"); k. undercladding cracks (CLAD); and 1. combinations of size and position : specifically, defects with sizes
from 10 to 25 mm if located near to the clad surface (SL <c 50 mm) and defects with sizes larger than 25 mm if located far away from the clad surface.
Due to the base material defects and particular aspect of most of the defects of plate No. 1, this plate has not been considered for the present analysis of results.
Table 9.1. gives the sample size of all these defect groups.
As always, variables used for the evaluation of results were :
. detection (DDF),
. location, ligament (MELZ, SELZ, MESL, SESL),
. sizing (MESZ, SESZ), and
. ASME XI rejection criteria (CRF, CAF).
The evaluation was performed for the major groups of procedures :
a. ASME at 50% DAC, b. ASME at 20% DAC, c. ASME at 10% DAC, d. advanced sizing techniques, e. special procedures (group of ), and f. separately for the special procedures.
9.2. IMPORTANCE OF DEFECT POSITION
Defects of groups a, b, c and d (9.1) were analysed.
65
Defect group
Defects near to the cl*d surface
Defects located between 18 and 50 ea In depth
Defects located between 50 «• and 25 «a In depth
Defects near to the unclad surface
Defects with s lied (07.) smaller than 10 M
Defects with sties (02) ranging f ro . 10 a* to 25 «JB
Large defects 02 25 •»
Underclad cracks
Particular set :
a. DZ> 10 •>; SLOOea b. DZ>25aa; SLfSOna
Nuaber of defects
Plates No.
3
9
9
22
5
23
9
8
9
16
2
9
5
IS
3
14
3
11
7
14
9
(3)
0
15
1
e
4
5
(3)
-
Rejectable defects Plates No.
3
5
2
«
0
23
8
t
5
11
2
5
2
5
2
0
2
10
6
"
9
(2)
0
8
0
1
3
5
(2)
-
Acceptable defects Plates Ho.
3
4
2
I t
5
0
1
2
4
5
2
4
3
10
1
14
1
1
1
3
9
(1)
0
7
1
7
1
0
I I )
-
Table 9.1. : defects 6 and 12 of plate No. 9 which have not been used for the present study.
PLATE NO 3
Defect Subgroup
Procedure
ASHE 50% DAC
ASHE 20X DAC ( A l l )
ASHE 10X DAC ( A l l )
A l l special proc.
BD005599
EF
LN
SR
VH
YC
0018 (DDFT)
. 44
.70
.56
.79
1 .
.78
1 .
. 89
.67
.89
1850 (DDF)
. 22
.60
.72
.83
.89
.89
1 .
0 .89
.67
.78
50 MX (DDF)
. 27
.67
.70
.87
1 .
.82
1 .
.86
.91
.64
M 20 MX (DDF)
. 33
.72
.80
.87
1 .
1 .
1 .
1 .
. 8
.4
A l l defects (DDFT)
. 31
.66
.66
.83
.96
.83
1.0
.88
.75
.77
A l l I n t . def . (MDDF)
. 35
.75
.76
.90
.97
. 9 4 J i
1.0
.97
.81
.80
Table 9.2. : Detection as a function of the defect position
66
All defects were considered, even the non intended ones. The MDDFT values thus take account of the 40 defects in plate 3, for example, and not just the intended defects that have been considered in chapters 1 to 8 of this report. Columns 5 and 6 of Table 9.2 show the difference.
9.2.1. Effect of the defect location on defect detection (MDDFT)
Both for ASME type procedures (20% and 10% DAC) and special procedures:
- detection of defects near to the internal surface often appears to be slightly more difficult than the detection of defects in the central part of the plate thickness;
- defects near to the external surface are often well detected;
- such conclusions have to be reconsidered taking further account of the influence of the size and category of defects.
These observations are illustrated by Figure 9.1.
9.2.2. Effect of the defect location on defect evaluation (MCRF, MCAF)
No clear trend was apparent for the three plates assessed here when the parameters MCRF and MCAF were considered.
9.2.3. Effect of the location on sizing errors (MMESZ)
There was a large oversizing of defects near to the external surface of plate No. 3 (Figure 9.7.). Oversizing appeared to increase the further defects were from the internal surface (Figure 9.3.).
Plate No. 3 represented the best sample for detecting this trend sample as in plate No. 2 defects were located mainly near to both surfaces. Moreover, nearly all procedures scanned plate No. 2 from both sides. Nearly all defects in plate No. 9 were in the central depth zone.
9.2.4. Conclusion on the effect of defect position
Excluding the ASME procedure at 50% DAC, dominant trends are :
- there is a slight tendency for defects near to the clad surface to be less reliably detected than defects in the middle or outer parts of the plates; and
- sizing errors in height increase towards the outer surface of plate No. 3 with defects near the outer surface being oversized.
67
PLATE N.3
poof. ASMC HTL M, i.S
P
««ME 20% N.B.S
M H PLATE THKKMCSS(mm) m 1W ta M PLAIE THICKNESS (mm) l i t I M
POOfT ASMC SOX MOOfX MX SPECIAL PROCEDURES
y PLUE IHKNt t» (mm)
m-mm* m-y^m
" PLUl THICKNESS (mm) * " " •
Figure 9.1. : MDDFT as a function of defect location (position in the thickness of the plate).
MESZ
Ttmm)
r imm)
10-
10
ALL SIZING PROCEDURES
VhVAttWk HI
T(mm)
MESZ
BO
PLATE N.3
I (mm) Tlmm)
Hmm) 210 m
rimm)
Figure 9.2. : Errors of sizing as a function of defect location. Plate No. 3.
68
M(MESZ)
• 10
MIMESZ), T(mm)
PLATE N . 2
^
ijb. \ T (mm)
M(MES2)»
10-
10
20
t
A
PLATE N.9
MX T (mm)
• SPECIAL PROCEDURES
• 20V. OAC
û 10V. DAC
o 50V. DAC
* SIZING TECHNIQUES
Figure 9.3. : Errors of sizing as a function of defect location in the three plates.
69
9.3. Importance of defect size
The parameter "size" is, as always, represented by DZ; the through thickness dimension of the defect. Three groups of dimensions were chosen : DZsg 10mm, 10 < DZ 25 mm and DZ :> 25 mm.
9.3.1. Detection as a function of defect size
From the results in Figure 9.4. it is apparent that :
- small defects (3 DZ«?10 mm) were detected with a DDP of about 0.5;
- defects between 10 and 25 mm were generally well detected; - defects > 25 mm were also generally well detected; and - special procedures did not reach perfect detection for defects smaller than 10 mm.
9.3.2. Defect evaluation as a function of defect size (Figure 9.5)
From Figure 9.5. it appears that : - ASME procedures at 10 and 20% were better at evaluating defects between 10 and 25 mm than for defects> 25 mm; and
- the performance of special procedures did not depend on the defect size.
9.3.3. Conclusion on the sizing error as a function of the defect size
Figure 9.6. shows the mean sizing errors (MESZ) as a function of defect size for plate Nos. 2, 3 and 9. The observed trend is similar on all three plates with small defects being generally oversized and large defects being on average slightly undersized. Note that in plate No. 9 all small defects less than 10 mm are unintended defects, small pores etc.
It is important also to consider the dispersion in sizing errors. In table 9.3. the standard errors of the means for MESZ are shown as a function of both defect size range and inspection procedure. In general the ASME type procedures all give larger dispersion than the special or advanced sizing procedures. This same result can also be seen in the sizing error histograms of figure 9.7.a. where the higher standard deviations occur for the ASME 20% procedure in both the sizing ranges. Figure 9.7.a. also includes the results for sizing error in the Y dimension.
The trends are similar to those for the defect height but with considerably increased dispersion for the group of smaller defects. An alternative method of presenting these results is shown in the graphs of figure 9.7.b. where the measured size is plotted as a function of the actual defect sizes in plates Nos. 2 and 3. All defects, both intended and unintended are included in these graphs. The graphs show clearly the general oversizing of small defects and the approximately correct sizing, on average, for larger defects. However, several defects greater than 25 mm are undersized, on average by up to 10 mm by the special procedures and 15 mm by the ASME 20% procedure. The results should be compared with the performance of team BD in figure 6.1.
70
MDOFT
MODFT ASME 10% OAC
OZlmml
MDDFT ASME 20 V. DAC
10 M « 0 Z l „
i SPECIAL PROCEDURES
„ O J ^ , io is »s 0 Z l m m l
PLATE N. 2
0 10 25
10% DAC
DZlmm)
,1
0.5-
0 0
M j
10
-1 2
25 a n
20% OAC
DZlmm)
50*/. DAC
DZlmml
PLATEN.9
MDOFT
14
ASME 10% M.B.S MOOFT
, 0 DZtmn.1
ASME 50%
ASME 20% M.B.S
10 25 <0 H 7 i i
,, ALL SPECIAL PROCEDURES
•• oz lM> ,0 " PLATE N . 3
OZlmml
Figure 9.4 : Detection as a function of defect size
71
MCRF ,,
ASMÊ 10*/. M.B.S
01 ui
•MMMË
ASME 20V. M.8.5.
MChf , * DEFECT SIZE , 0 "
(mm) MCRF DEFECT SIZE „
(mm) ALL SPECIAL PROCEDURES
±à\ ii ri11 ' »ii i
MCRF,
•» DEFECT SIZE (mnt)
PLATE N. 3 DEFECT SIZE
(mm)
Figure 9.5 : Correct rejection as a function of defect size in plate No. 3.
MIMESZ)
M ( M E S Z ) ,
M(MESZ)
PLATEN. 2
DEFECT SIZE
( m m )
10 25 6!>
DEFECT SIZE (mm)
• SPECIAL PROCEDURES
• 20 V. DAC
c 10 V. DAC
o 50V. DAC
. SIZING TECHNIQUES
Mote : small defects in Plate No 9 are unintended sets of pores; detection is weak and not to be considered here.
DEFEq SIZE (mm)
Figure 9.6. : Sizing error as a function of the defect size for three plates.
72
\
^ \ ^ Defect s i z e \ ^
Category ^ v .
From 3 to 10mm
from 10.5 to 25mm
from 25.5 to 80mm
Plate No. 3
ASME 50% DAC
9
5
11
ASME 20% DAC
7
7
11
ASME 10% DAC
8
12
20
Spec. Proc.
5
4
9
Sizing
Proc.
4
5
8
(mm)
Plate No. 2
^ s .
^ \
Defect size \. category ^ \
from 3 to 10mm
from 10.5 to 25mm
from 25.5. to 80mm
ASME 50% DAC
13
15
17
ASME 20% DAC
6
5
6
ASME 10% DAC
10
7
9
Spec. Proc.
7
4
6
Sizing Proc.
5
4
4
(mm)
Table 9.3. : Standard errors onthe means MESZ for categories of defects.
Note : The trend observed on ESZ as a function of the defect size is also appearing for ESY.
Figure 9.7.a. : Distribution of sizing errors for different classes of defect size. From the sandard deviation values, error bars on points in figure 9.6. can be computed.
74
C\2
2 -us E - U J
a,
*£ V
\
\
m
ttD m
ir> n j
ts ru
<S3
t_»
O f L J
LâJ Q £ U
OL
unct
C u a t ^ s a e u O tn—-isiuj -EUT x^ujŒi/i=>teLtja tn«—«rsjuj
\
\
• \
CO
O Ï
US C - H LU
OH
\
v
• " • \ M M M M M t h Œ)
M M » • 4M M •
• " i i '
\
\
\
< 4 \
..s M « M M >•»
CuauiSQciiJO m>—«ISILU C C
" ' I 1 ' " I " " I " ' ' I ' • • • I
tj> irt •v en t\j
• M M > -
,,,,,,,ft-
• o 01 s_ 3 l/> rO a i fc a> a i IO s_ O)
> <£
X I
r~
a> CD s_ 3 O )
( J CD
<«-CD -o r— lO CD i -
a> sz -!-> M -O
75
9.4. Combination of size and position
The conclusions on the influence of defect position (cf. 9.2.) could be biased by the distribution of defect size (cf. 9.3.). Figure 9.8 shows the distribution of defect size as a function of defect position in plate 3.
It appears that the defect size is not well distributed across the thickness for the important size group 10 mm < DZ 25 mm. Figure 9.9 shows, however, that for this size group considered alone, the influence of defect position corresponds to that seen for the whole set of defects.
The conclusions in 9.2. and 9.3. above would seem to be unbiased by the distribution of defect size.
9.5. IMPORTANCE OF THE DEFECT CHARACTERISTICS
9.5.1. Families of defects in Plate No. 3
Plate No. 3 was designed and built particularly for the parametric studies on the influence of the defect characteristics on detection and sizing. The defects inserted can be classed in categories which are characterized by surface roughness and crack tip aspect. Eight groups of defects can be identified in plate No. 3 :
a. The smooth cracks are either smooth thin disks (e.g. 25 mm in diameter) or smooth thin rectangles (e.g. 10x50 mm). Both are caracterized by sharp "crack tips", generally with an equivalent tip radius of the order of 0.02 mm. The surface roughness corresponds to RT = .035 mm (total roughness). These defects are numbered : 2, 19, 21, 27 and 30 (see Appendix 2).
b. The rough cracks are disks or rectangles. Their roughness corresponds to RT = .35 mm. Such defects are numbered : 5, 12, 18, 20 and 28 (see Appendix 2).
c. Smooth cracks with starters where the starter is a long defect 3 mm in diameter representing a slag inclusion.
Two such defects, numbered 4 and 11, were introduced in plate No. 3. The cracks were as sharp and smooth as those in a.
d. Lack of fusion defects were introduced using the classic method of the small plate welded on the chamfer flank. The welding process did not produce a sharp edge, but such defects are very thin. Four such lack of fusion defects were introduced into plate No. 3; numbered 23, 24, 25 and 26 (Appendix 2). Such defects have smooth surfaces (RT = .035 mm) when considering the ultrasonic wave length at 2.25 MHz (1,4 mm).
76
6z UOOEF.PL. 3)
SIZE DEFECTS REPARTITION
PLATE Ne 3
Si - 16.1 mm
ÔZ, 0 < l «18 « I A3
DJ!.I8«L*50 » 20.3
DZ,50*L«M» . 15.1
ÔZ, 221«L*M» = 8
Tlmml
Figure 9 .8 . : : D i s t r i b u t i o n of defect s ize i n p la te No. 3 as a f unc t i on of the l oca t i on i n depth.
POSITION OF THE DEFECT IN DEPTH
(mm)
. SPECIAL PROCEDURES
• 20*/. DAC
A 10 V. OAC
o 50V. DAC
* SIZING TECHNIQUES
Figure 9.9. : Error in size as a function of defect position for the only group of defects with sizes between 10 mm and 25 mm. Plate No. 3.
77
e. CETIM cracks
Forged defects were produced by CETIM, with initial characteristics near those of the smooth cracks in (a) above. However, the dimensions of all these defects increased during the process of welding in the implant. These defects ended up as cracks with irregular surfaces and "crack tips" which are, in two cases, very different from a real crack. Four of these defects were introduced into plate No. 3 as near-surface defects, close to the clad side. The roughness corresponds to RT = .07 mm. These defects numbered 13 and 15 are comparable to defects 14 and 16 in category B and to defects in category A.
f. SljK£
Five slag defects were introduced during welding, having equivalent cross-section dimensions to a weld pass : 3 to 4 mm. These volumetric defects are numbered 1, 10, 17, 22 and 29 (Appendix 2).
g. Composite defects
Composite defects were introduced at the request of some participants in order to identify techniques able to deal with shadow effects or having enough resolution to distinguish between several defects in a defective area. Two types of composite defects were therefore introduced.
g.l. The "shadow defects" There were two shadow defects, numbered 8 and 9 (Appendix 2), present in plate No. 3; (one of them appeared to have opened as a consequence of the welding-in process). All components of such defects were of the smooth crack type (as in A).
g.2. The "cloud defects" The cloud defects were sets of smooth circular cracks (as in a.) of diameter varying between 6 and 18 mm. They are numbered 6 and 7 (Appendix 2).
9.5.2. Limitations
Most of the defects in plate No. 3 were introduced using the implant technique. Thermal stresses arising during the welding-in process have opened some of them in a manner which is understandable from the fabrication process of the defect in the implant (Ref. 3). Such opening (and thus variation of dimension) were verified in two steps : by ultrasonics immediately after the implant was welded and with X-ray of the assembled plate.
78
Destructive examination has not been performed as yet on plate No. 3, and doubts can therefore always persist even if evidence can be given of the reference defects as proposed for the evaluation of results.
Introducing defects by implant techniques also causes satellite defects. Some of the defects described above (e.g. defect 28 and 3) were clearly in the neighberhood of satellite defects. This may have modified their probability of being detected particularly when the implanted defects were smooth cracks, difficult to detect with traditional techniques.
It was not possible to introduce more than 31 defects into the plate No. 3 weldment. Now that results are available, the absence of more defects can be regretted, particularly smooth rectangular cracks without a starter.
Plates No. 2, 9 and 1 presented only few defects of category A and in the evaluation here under, only examples of defects from these plates are used to confirm the conclusions.
9.5.3. Defect detection probability as a function of defect height
The curve of DDP as a function of DZ is shown in Fig. 9.10. for each family of defects in plate No. 3.
The most general sample of procedures was used that is all teams' best results including the high-performance procedures which work at high sensitivity. Although it is recognised that this is a heterogeneous , positively-biassed group of results, several observations can be made.
a. Smooth cracks (group a) have the lowest detection rate for all sizes of defects.
b. Slags (volumetric defects) have the highest detection rate. It is to be noted that such slags were, in this case, all of small through-thickness dimension.
c. Rough defects, lack of fusion and smooth defects with starters give the same order of magnitude of DDP as a function of size.
d. Composite defects yield very poor detection results. This could be due to the limitation expressed in chapter 2. However, detection could be referred to the dimensions of the separate components (from 6 to 18 mm) rather than to the overall envelope. When this is done, the points fit the regression line on Figure 9.10 rather wel1.
e. The observation in c. the further point that the crack tip aspect of the defect is as important as its surface roughness.
79
o CO
(MM
)
O NI r»
o u>
o m
o
O O
O CM
O
O
a i • o
a i - C 4->
>) JD
• o 01 IM
S-4-> a i B ta s-ta CL
a i 4->
s-
c o
4->
o a i 4-> a i o
ai •o
o
a i
-B ta
c o
i n u
•r-4-1 m
•t— ai 4-1 ta
IO
u
4 J U a i
t -
•r-
dure
s ro
ce
o .
<+-o
o. 3 O s-
<> o
o r\i
UJ o> s :
a i
4->
o < 4 -
(/) 4-> <J a i
<+-
i/>
< *4-O
4-> *r— S-
a . m
a i - C 4->
CO
# o
a i 4-> <a
r— CL.
—
O o
o l/ï
o o
ai s-
ai o -o
lA i—
h. O
O a :
UJ 2 D _ i Q
• • U J
l/>
FE
CT
UJ o ce o 1/1
IU
T 10
-> n ce
•* -
GES
n
X u
< ce
(•) rr
3 T
s
a .
VHS
i
ï
</> 2C U
< ce o X t r
O i
</> "•
*
O a UJ
< j
< ce u
>, JD
• o Ol Isl
• 1 — S-
4-> a i fc S-lO u .
n i 4->
s-
c o
• i — 4-> U a i
4 J a i o
a i i_
4->
a i CD
. m u
4-> m
•t—
i-tu
4-> U <a s-ta
. c u
4-> U eu
>*-
• T J a i i_ a i
X>
i/> c o u
a i s-lO
f -IO a i 4->
JC O m a i
CO
• o ; £
a i 4-> m
i — CL.
•
doa ai
80
f. Finally, it is possible to divide all these defects into three categories as a function of the crack tip aspect and surface roughness :
A. smooth cracks with sharp edges (fatigue cracks);
B. rough cracks, and some cracks which were strongly modified during the implantation process and have unrealistic crack tip aspects; simulations of lack of fusion; and
C. volumetric defects (slags and porosities).
It should be noted that this definition of defect categories A, B, C has no correlation with the IIW nomenclature of defect types.
9.5.4. Evaluation of the results for the ASME procedure at 20% PAC for three plates
All results at 20% DAC, yielded a diagram similar to the one seen in section 9.5.3. Figure 9.11. should be compared to Figure 9.12. On comparing the results obtained for plates No. 2 and No. 9 with those for plate No. 3 (Fig. 9.12.) it can be concluded that Figure 9.11. also provides a good summary of the results for plates No. 2 and No. 9.
9.5.5. Conclusions on the influence of defect category on defect detection.
The following general conclusions on defect detection can be drawn from Figures 9.11. - 9.15.
a. Detection of category A defects is always more difficult than any other type of flaw.
b. Even at 50% DAC, small volumetric defects are, on average easier to detect than planar, smooth sharp defects (Figure 9.13.).
c. The defect critical size (size for D D P — 1) is a function of the defect category and of the procedure type (Figure 9.14.).
d. A key parameter appears to be the crack tip aspect (Figure 9.15.). The total surface roughness is important too.
9.5.6. Importance of defect category for the correct defect rejection
Figure 9.16. illustrates the following conclusions on defect rejection.
a. Correct rejection of rejectable defects is more difficult for family A defects than for others.
b. ASME 50% DAC procedures have no chance to reject any rejectable smooth sharp crack oriented perpendicularly to the surface Figure 9.16.
c. Defect critical size for good rejection depends on the procedure. For ASME 50% DAC it appears to be near to the wall thickness (T) (cf. PISC I, defect 1 in plate 50/52). This confirms results from defect No. 1 in plate 50/52 of PISC I (Ref. 7).
Plate No. 1 PLATE No. 2
• ">
^ J
r H 3 d 5 5 55 60 7 0 â o
7 .7 •• ••
, ir
i • i
0 I
•
<«
(+>
0 30 ^ 0 50 GO 1 0 00
ranri)
Note : uncertainties exist on Plate No 2 : a. satellite defects modify defect characteristics b. defects 12 and 15 were transverse defects , ( ).
PLATE No. 9 PLATE No. 3
7
•
/
/
~ni M 15 Î0 DZtrm
&
c
> 1
•
• • fc < é •
t
t
mi
> ut 1 W
1 0 *
• 4 i rj/
/ / /
/ r
0 0 < to :
<
SO E 0 1 0 ( OZCWIJ
u
• : VOLUMETRIC DEFECTS
• : ROUGH DEFECTS OR DEFECTS
WITH LARGE CRACK EDGES
• : SMOOTH CRACKS WITH SHARP
CRACK EDGE
Figure 9.12. : Detection rate for the three plates for the procedures in the spirit of ASME (20% DAC).
82
CL û O
I.UU
.50
J10
j \
^ 9
/
/
1 3 3 333 i
1
2 3
3 / 12)
/ /
/ (2:
3
(2)
10 20 30 40 50 60 70 80
DZ(MM)
Set of "transverse defects" .
Figure 9.12.b. All category A defects of the 4 PISC II Pltes
83
ASME 20*/. DAC MDDF
0-5
0
M00FA
MDDF
0.5
0
ASME 20"/. MRS
Bjgpl
MDOF
A B C
MDDF
MDDF
PLATE N.3
Figure 9.13. - Detection as a function of defect category
OOP
i
.5
ASME 10V. DAC
Bi 1 I 1 1 1
, A ; / / i i
OOP 4
i PLATE N.3
10 20 30 «) 50 60 70 D Z m m
DDP^
1
.
c!
'
ASME 5(
/ t
i i
B /
/ A / / /
) ' / . DAC
10 50 DZmm
ASME 207. DAC
-i—r
i
10
DPP
m .*&
5 0 DZ mm
ALL SPECIAL PROCi
A/ / / /
50 DZ mm
Figure 9.14. - Detection as a function of defect size and defect category.
84
ALL SPECIAL A - CRACKS SIMULAT.
B . I : ROUSH CRACKS
8 . 2 • S.C. WITH STARTER
B.3 = LACK OF FUSION
B.4 = FORGED CRACKS (OPENED)
C s SLAGS
Figure 9.15. - Defect detection as a function of the equivalent radius of the defect tip. Q (mm)
C R P + ASME 10V. DAC
W
ï—1-
i , •',
1
PLATE N. 3
10 20 30 AO SO 60 70 DZmm CRP
.5
i
y#
v . v
ASME
8 s
/ -
50 V. DAC
A
• « - • > — • " » — —*-10 50 OZ mm
.5
ASME 20V. DAC
/ / i - /
8 | A /
/ /
CRP,,
M m 11 IT
50 DZmm
ALL SPECIAL PROC.
K) 50 DZmm
Figure 9.16. - Resectable defect correct rejection as a function of defect size and defect category.
85
9.5.7. Conclusions on sizing errors as a function of defect category
a. The following observations can be made from Figure 9.17. a and b. There is a systematic trends to undersizing of planar defects (category A) and oversizing of volumetric defects (category C).
b. Special procedures, however, nearly always produce correct average sizing.
Conclusions of chapter 8 could be recalled : perfect results can be obtained by the special procedures for an increase in the declared size by a few mm.
9.5.8. General conclusions on defect category
The defect category appears as a parameter which dominates all the others considered for the evaluation of the PISC II RRT results.
9.6. Underclad cracks
Table 9.4. indicates that several special procedures are well designed for the detection and sizing of the underclad cracks identified in Table 9.1.
In general, for ASME type procedures, performance was higher than that obtained on category A defects, probably due to the use of 70 angle probes.
In particular, results regarding plate No. 2 were high because scanning was often conducted from both sides. This is interesting to consider the results obtained when scanning the clad surface only. Table 9.5. shows the results for plate No. 2.
ASME 10% (clad side)
ASME 20% (clad side)
ASME 50% (clad side)
MDDF
.65
.94
.71
MCAF
1.
.57
.5
MCRF
.65
.9
.75
Table 9.5. : (inspection from the clad surface only)
Under clad cracks appear better detected than general defects.The results at 50% DAC are surprising. They were probably affected by satellite defects and the characteristics of defects in plate No. 2. and by the performance of the 70 degree angle probes as discussed in chapter 10.
86
10 UJ o
i a
< :!:!:::;?: <
S -
o (U
+-> (O O
O ai a>
-o
c o
u c 3
m ui X °
< o
Ui
z m <
f\j UJ S Z *
DA
C
1
g r*
ASM
E
I N ' 1 : u
u
AS
ME
50*
/. D
;
L_
ZM
o o a.
SIZ
ING
_i - i
«
im h^gr
1
AB
C
1 A
SME
20V
..DA
C M
3.S
&à=sn§
1 < <
(/l (O i_ o S-i-
a> en c: M
•r— C/0
I
ai s -
cr>
87
ASHE 10t DAC
ASHE 201 DAC, all
ASHE 50X DAC
All special proc.
All sizing proc.
BD
KR
LN
SR
VHDT
YCFL
MDDF
.92
.96 1 . \.71
.90
.95 "^1. "\.67
-
1.
1.
1.
1.
1.
1.
MCAF
.25
.45 ^ \ .
\o.
.67
.36 1 .
.43 ^ - 1 . ^ . 0
0.
0.
o. ,
1.
0.
0.
HCRF
.76
.92 ^ 1 . \.5
(.89)*
.85 ^ - 1 . \.4
.92 s 1.
(other special proc.) were applied on plate No. 2.
* High values because of the many procedures applied from the unclad side (corner effect)
PLATE No 2
ASHE 10% DAC
ASHE 20% DAC, all
ASHE 50% DAC
All special procedure
All sizing procedure
BD
EF
LN
SR
VHDT
ÏC
HDDF
.61
.89 • 7 4<.56
.37
.80 <'• .67
-
.89
.78
1.0
.89
.67
.78
HCAF
.75
•"<!«
.83
• " < • 1.
•79 <.o
1.0
.50
0.
1.0
.75
.75
MCRF
.30
.58<'-
0.
93. K
•3 .8C
1.
•75<.o 1.0
1.0
1.0
.80
1.0
.80
Table 9.4.
PLATE No 3
MDDF, MCAF, MCRF for underclad cracks
88
9.7. Particular combinations of defect position and size
A particular family of defects which could be of special interest and importance would be composed of those defects > 10mm in size (DZ) situated within 50 mm of the clad surface and those ^ 25mm in size further away from the clad surface. The results of an analysis of such a family is presented in Table 9.6. This table shows that the reliable results were not obtained on plate No. 3 using ASME type procedures. For this important group of defects, in plate No. 3 :
- ASME 20% : MDDF =0.7 MCRF =0.44
- Special procedures : MDDF = 0.93 MCRF = 0.91
Better parameter values where obtained on Plate No. 2 but it should be remembered that only 1 defect is of family A in this plate. From the preceding paragraphs it is clear that this particular group of defects, if of category C, would give MDDF = near to 1 and MCRF = near to 1.
9.8. Importance of the orientation of the defect
In plate No. 3, several defects had different orientations in terms of their tilt and skew angles. Table 9.6. lists these parameters as well as the performance of different procedures on each defect. No clear conclusion can be drawn on the influence of the orientation of the defect because of the strong influence of "defect category" which appears to dominate other parameters. From the ASME 20% DAC and special procedures a trend appears (which might not be significant) that small tilt angles generally correspond to higher defect detection rates. The systematic study of these parameters is being addressed by the parametric studies on the effect of defect characteristics in PISC III.
89
ASME 101
ASME 20%
ASME 501
All special procedure
All sizing procedure
BO
KR
LN
SR
VHDT
YCFL
ASME lOt
ASME 201
ASME 50%
All special procedure
All sizing procedure
BD
EF
LN
SR
VIIDT
YC
MDOF
•'«'J .91^1.0
^.79 .64
.93 -1.0 -.75
-
:.
l.
I.
' l.
i.
l.
•«<:S •»<:;? •- < \ l î
•»<-i -
i .
.94
1.
1.
.88
.81
HCAF
••'<'.£-.55^1.0
^0.0
.78
.48^1.0 ^0.0
.71
.67
0.
.33
1.
0.
0.
-,„ /-80 .70 < ,„
^ .60
- < o . o °
•« <]i •«<£ .68
.80
.60
0.
1.
.10
.60
MCRF
,s < -
.92 ^ 1 . 0 ^ . 6 4
.67
.91 <-"1.0 ^ . 6 0
.04
1.0
1.
1.
1.
1.
1.
•»<:;? - <Joao
•« < o29o
.91 <'-K0-V.73!
.60
.91
1.0
.91
.91
1.0
.73
Table 9.6 : MDDF, MCRF, MCAF, for particular combination of defect position and defect size
( DZ > 10mm for Zl 50mm DZ > 25mm for Zl> 50mm)
90
u>
ï •g
°" (O
r. ë Q .
i n
S 3 «
S
z S!
o *z a *« S
z: co «t
O . QC O
CL. Q a
u. ce
I X
o
u.
U . OC c_>
o . o
o
^ + 1 X 01 3 <X
</>
o • • "
+1 •*-»
i—
• ? * > 01
o -o ? F. • ë o S
+» o
«
0 0
1
CO CO
o
,
CO CO
a» CO
1
* * •
* t
o
i n
i
§
o
* <c
CM
1 0
CO CO
•
o
1
r*
en 0 0
i n
•M
y
CM
*» «C
en
\o
i
CO GO
-
1
O
CO
i
r-» 1 0
o
i n
i
£ +J
£ r-.
* « ï
,— CM
1
-
O
CO CO
1
m CM
o
o CM
g CO
o
* <
r-. CM
•
1
CM
-
O
-
1
CM CM
O
KO CM
- C + J O
CO
** <c
o CO
1
1
o
o
1
i n r^
O
o i
V .
*H X
i -U l
« t
-*
•
1
1
o
o
1
r*
0 0
o
o
V
* £
t -LU <-> «C
KO
'
O
o
1
CO
i n CM
o
«* +
CJ
a . u
<_> «c
l O
1
o
o
o
CO
r-. CM
O
CO
+
o
L.
«_» «C
r-
,
u»
•
o
o
1
CO
CO CO
o
-1
CO
S.
o t o « ï
CO
1
1
o
o
1
i n
i n CM
o
CO
+
CO
i~
o «c
en
i
i
o
i
o
KO
1
i n 00 i n
3 o os OÛ
i n
i
•
o
o
1
0 0 CO
o
-
o i n
o
3 O
OC
CD
CM
1 0
-
CO CO
KO
1
i n
o CM
*
i n
H oc CO
0 0
KO
1
-'
1 ^ 1 0
1 0
t
o
i n CM
i
r^
*3 oc CÛ
CM
i n
i
CO CO
t o m
o
1
fe» oc
CO
CM
t
1
o
CO CO
1
o
0 0
r^
CO
+ CO CO
*r
**
i
~
CO CO
o m
i
o
1
i n
+ CO CO
,_
91
10. EVALUATION OF THE RESULTS AT THE LEVEL OF INDIVIDUAL TECHNIQUES
10.1. Introduction and limitations
10.1.1. Data produced by teams
Several teams furnished detailed inspection results corresponding to the contribution of each technique composing their procedure. Often, however, these detailed data sets corresponded to the standard probes generally used in the spirit of the ASME procedures. It was fortunate that many manual procedures were included in PlSC II so that results on techniques were available but, unfortunately the contribution of individual techniques were never detailed in the results of special procedures. The procedure TOFD, however, consisted of one technique only which often was applied as part of other special procedures such as BD and SR (complex combination of standard and advanced techniques). The techniques on which a detailed evaluation has been possible are, therefore :
The largest quantity of data related to the 20% DAC cut-off level. At 10% DAC, data was very scarce and for the evaluation of techniques at 50% DAC, some data sets were created from the 20% DAC results.
10.1.2. Quality of detailed data on each plate
The quantity of data available on techniques depended on the plate and the degree of usefulness also depended on the general importance of the defects in the plates.
- Plate No. 1 showed so many indications that \jery little detailed data was available. Results on plate No. 1 were thus used only for parameter influence evaluation.
- Plate No. 2 was rich in detailed data at 20% DAC.
- Plate No. 3 gave several data sets on techniques for procedures in the spirit of ASME but because of important corrections which were necessary on coordinates and human errors, a large part of the detailed data is unusable. As far as possible, because of the variety of defects present in plate No. 3, it has been used for the evaluation of technique performance.
Often, however, samples consist only of the results of two teams.
92
- Plate No. 9 also produced some data sets on techniques. Results were used to confirm conclusions drawn on plates No. 2 and No. 3 but the reduced variety of defects limited the usefullness of plate No. 9 for a full evaluation.
10.1.3. Variables used
To evaluate the performances of individual techniques, emphasis was put on some variables which did not give clear answers during the evaluation of results at the level of procedures. These variables were :
- scanning surface : clad or unclad side, - position of the defect in depth, - frequency of the probes (1 MHz, 2.25 MHz, 4 MHz), and - angular position of the defect.
The following variables successfully used in the assessment of procedures were also considered again :
Figures 10.1., 10.2. and 10.3. demonstrate contradictory results : on plate No. 3, the clad surface appears to disturb the evaluation of defects (CRF) by the 45 and 60 angle probes; on plate No. 2 the contrary could be concluded. The type and position of the defects in plate No. 2 explains this result. Most of the defects were easy to detect because they were often surrounded by satellites. Moreover, defects in plate No. 2 were concentrated near to the surfaces and corner effects must have played an important role. In plate No. 3, where defects were concentrated near to the clad surface, the corner effect could have been beneficial when scanning the unclad surface. In the zone near to the clad surface, the 70 angle probes appeared to fulfil 1 their role. Tandem probes gave partial results, generally equivalent in all situations, but the near surface zone was excluded for such a technique. T0FD is very good but it must be remembered that the technique does not claim tor detect and size defects very near to the clad surface or entering the cladding. Such defects were thus not considered for the evaluation of the performances of T0FD (mainly on plate No. 3).
93
MDDF,,
0.5-
. MODFn
JL^^L^^LJ^ B B TECHNIQUE 0* 45" 60* 70" TD TOFP
Depth zone : 0 to 18 mm Depth zone : 0 to 250 mm
TECHNIQUE
MCRF M C R F n
, u C #> u C ,. U C , El B TECHNIQUE 0" 45' 60* 70* TD TOFD
TECHNIQUE
U = UNCLAD S IDE
C = CLAD SIDE
MCAF Depth zone : 0 to 250 mm
HT1BI
M m U C „ U C , , U C „ TD TOFD
0 ' 45* 60". 70* TECHNIQUE
Plate No. 2
20% DAC (but TOFD)
Figure 1 0 . 1 . : Technique performance on p la te No. 2 as a f unc t i on of defect pos i t i on and scanning sur face.
94
MODF
M 45* 60"
TECHNIQUE O* 45* 60" 70" TO TOFD
Depth zone : 0 to 18 mm
MDDF
Depth zone : 0 to 250 mm
TECHNIQUE
MCRF , MCRF
7 0 - TD TOFO TECHNIQUE TO TOFD TECHNIQUE
MCAF
Depth zone : 0 to 250 mm
P •âa
I 4 É W
u c,
«•
U
C
^ v 60
= £
C , , U C ,
70'
UNCLAD
TO TOFD
SIDE
CLAD SIDE
TECHNIQUE
PLATE N. 3
20% DAC
Figure 10.2. : Techniques performances in plate No. 3 as a function of defect position and scanning surface.
95
MDOFn
F?
n JLLML U C „ U C , . U C j 10 TECHNIQUE
MCRF
_IZL U C . . U C , > U C , TO
45* 60' 70* TECHNIQUE
MCAF
0* .45* 60" 70*
U = UNCLAD SIDE
C = CLAD SIDE
TECHNIQUE
Plate No. 9 20% DAC
Depth zone : 0 to 200 mm
Figure 10.3. : Techniques performances on p la te No. 9 as a f unc t i on of the scanning sur face.
96
10.2.2. Influence of the defect position
Figures 10.1. to 10.3. demonstrate the enhanced performance of 70 angle probes for thenear surface defects. On plates No. 3 and No 2 few conclusions can be drawn on the influence
o o of the defect position in depth other than that 45 and 60 angle probes are weak for near surface defect evaluation when the clad surface is used. Figure 10.4.
10.2.3. Influence of the defect size
From the CRF values in Figure 10.5. it appears that there was a systematic undersizing of large defects when 45 and 60 angle probes were used on the clad surface, and was also observed from the assessment of procedures. The performance of the 70 angle probe appeared to be independent of defect size.
10.2.4. Influence of the defect category
Table 10.1. presents an important set od PISC II results. It introduces the defect category and shows the performance of the different techniques as a function of these categories of defects. The difficulty to detect family A defects with the standard techniques is clear. The contribution of the 70 angle probes is shown to be important when used for defects near to the clad surface.
The table also shows that TOFD performed well, remembering that this technique does not claim to detect near surface defects or clad defects. Tandem probes did not appear to solve the detection problem. They also do not claim to detect near surface defects.
10.2.5. Complementarity of techniques
An important question is to understand the possible complementarity between different techniques as well as the redundancy obtained when all techniques are used. Answers to such a question can be obtained by considering detailed tables of detection and evaluation by each technique of all individual defects. Tables 10.2. and 10.3. present this information for plate No. 2 and No. 3 respectively.
For plates No. 2 there was a complementarity between 45 , 70 angle probes and tandem probes or between 60 , 70 angle probes and tandem probes.
The plate No. 3 results showed the complementarity of 45 and 70 angle probes or of 60 and 70 angle probes. The TOFD results from a single team could not be introduced in these tables which are based on the statistics of several team's results. For the same reason tandem probes were not included for plate No. 3.
Both tables show redundancy between the 60 , 45 and 0 angle probes.
97
MOOFl UNO.AO M 0 0 F " CLAD 1.0
mmmm
.5
1 ia so
MCRFi
EOT (mm) » SO
MCRF,,
1.0
Ttmm)
Tlmm)
JL 220 T (mm)
L Ttmffl)
T(mm)
Plate No 3. Technique : 45 s
HOOP- MOD F CLAO
Ttmm)
Plate No 3. Technique 60 s
Tlmm,
MODFA UNCLAO MODFA C L A D
MCRF, i
T(mm)
MCRF,,
230 T(mm) 18 50
Ttmm)
L • M Tlmm) Ttmm)
230 Tlmm>
Plate No 2. Technique 45s
MDDF l 1.0
UNCLAO
M
M0DF,, 1.0
1 MCRF
.5-
i
1 '
MCRF
• 1.0
.5 I
Ttmm)
Ttmm)
Plate No 2. Technique 60s
Figure 10.4. : Importance of the defect position in depth for the 45 and 60° angle probe.
98
o X
< • - I
8 SE
•il
K "•'•"•"ttttK'KWfc-K-K-:-:
E
I ••->,-,-,-,v;y,m-:v,-.v.<
8 i E
I a u z
O
PO
a> . 2 s z o-LU "c t— £ «c o —i ai a . t—
ZZŒBB s 3 *
s l •I
u. a a
a l * O
1 I
M
•'•'T'T'I I'I i I'I'I I'I'IV" " " "
u X
n •
' • ? E E N
8
e
o
o z LU 1— « - J fi.
l / l O VO
.. ai 3 CT
-^ C . C u a> i—
+ J
efec
"O
4->
4 -o ai u c ai 3
i —
>•-C
»—i
U Z
o < -J
4=fl I
§
ft -J :
mill
nu nu 3
t l m slïl 3
SE E
ai s -en
LO O LT>
•1 s. M O
S
« *r LU • • •— <c ai —1 3 o. o-
* * c J Z < j
a>
99
Clearer indications on complementarity between 0°, 45°, 60°, 70° tandem probes and TOFD techniques are to be found inthe analysis of the results of separate teams in the following section 10.2.6.
o°u o°c
45° U
45° C
60° U
60° C
70°(0-18mm)
70° U
70° C
TD U
TD C
TOFD(C)
Category A
MDOF
.05
.27
.27
.36
.36
.90
MCRF
0.
.29
.14
.14
.0
.50
HCAF
1.
1.
1.
.88
1.
1.
only one defect on the unclad side
.77
.73
.18
1.
.50
.57
.29
.86
.88
1.
1.
.0
Category B
MDDF
.31
.50
.50
.81
.50
-
.58
.77
.38
.23
1.
MCRF
.08
.25
.33
.58
.42
-
.25
.42
.50
.17
1.
MCAF
.86
1.
.79
.64
.93
-
.86
.93
1.
1.
.33
Category C
MODF
.50
.50
.70
.60
.50
-
.30
.60
.40
.20
1.
MCRF
-
-
-
-
- « u
def(
1
ptab
le
-i ---
MCAF
.90
.60
.90
1.
1.
1.
.40
Table 10.1. : Contribution of the different techniques at 20% DAC (except for TOFD) as a function of the defect category in plate No. 3.
100
101
l/J <D 3 cr
• i —
c x: u <u +-> _
(O 3
• o
> -o c * '
O
</) QJ O c (O E V. o
M-s-a>
a.
0.3
.
. CO
o •*£
(X) -t-> (O
r^ Q .
c •r—
-t->
u (D «4-CU
T3
x: o <o ai
i_ o
<+-
a;
102
10.2.6. Verification of the conclusions on techniques from some indivi
dual team results
As was the case for the evaluation of procedures, results from team EC can be used for the verification of the conclusions on techniques. These results are presented in Tables 10.4. and 10.5. to which the TOFD results of team LN are also added. As each result is that of only one team, defect detection or correct rejection is either 1 or 0 (success of failure). The complementary of the 45 (or 60 ) angle probes with 70 angle probes is clearly shown by the results for plate No. 3. The good performance of the TOFD technique should be note.
10.3. Evaluation of technique performance as a function of the PAC cut off level
10.3.1. Introduction
As has been shown earlier, 10% DAC and 20% DAC results were often comparable and 50% DAC results were often poor. A verification of such statements can be made utilizing the few detailed data available on the performance of techniques.
10.3.2. Results at 10% DAC
The results of team NK shown in Table 10.6. give the best indications of the performance obtained with the different techniques conducted at the low cut- off level of 10% DAC. The results were not better than those obtained at 20% DAC (for example Figure 10.2.). The high performance obtained by procedure NK was clearly due to the correct combination of results of all technique. Tandem techniques as used by NK proved to be efficient.
10.3.3. Results at 50% DAC
Results of team EC were available on plate No. 2 to illustrate 50% DAC performance and are shown in Figure 10.6. The results of team EC were always near to the best achieved by ASME type procedures, it is surprising to see the good performance at 50% DAC on plate No. 2 of the 70°L probes.
The explanation is that most of the defects present in plate No. 2 were of category B or C.
103
~ " \ M F E C T N .
TECHNIQUES—-^
O'L,
45* S
60* S
70" S. L.SEL
TO
TOFD
•
• ;
! ' • • •
1
2
- •
3
2
i
•
li
5
OOP
6
' : ?
|
7
-
b ^
8 9
.•
1
10
Ï' i
u
;
â
12
"
il
13
'
'"'
W
i
PLATE N . 2 CLAO
15
!;-
k;
16
; . • •
- • .
Is
17
- • • •
V
.: •
18
v .
i, *
ft
ONE SIDE
1
',*•
;.' • » . -
2
•.'•
i
T E A M
3
! ; - • •
1-4*
t.
1. ' " ;
1*'' !
i •
É
5
•
:FEC
LU
' a «a
LL (_ C
u
3 C
] } }
7 8
> —
,< _
CRP
9
=3
-U
TABL
_
3 --U
_) <
,0
-»'
ii
P'
E
i. Pi
l_L.
12 13
!•
: • .«
;-
, •
K 15
I AC
CEPT
ABLE
DE
FECT
"
16
: •
••
17
m
d
y
1 AC
CEPT
ABLE
DE
FECT
Table 10.4. : Performances of techniques as applied by teams EC and LN. Plate No. 2.
~"--~^DEFECT N .
TECHNIQUES'*-^
O 'L
45* S
60* S
70* S.L.SEL
T D
TOFD
0* L
45* S
60* S
70* S.L.SEL
T D
TOFD
2
;
(—
ACCE
PTAB
LE DE
FE(
6
il
7
, •;
; J,
1
CATEGORY /
8
'•?-
9
' ;:
I !
§1
u,
:';: • '
-_•
%
a*. I
16
-'• = ;"
- f i l
If;
19
..
T:' i 1
21
' •
.!
#~
27
X
PTAB
LE DE
FECT K
ACCE
30
X
u
*
)
/ i
MS . • u .
*
"S
5
\;
' 7
"•• i
1 1
1
CATEGORY B
11
LE D
EFEC
T'-
\CCE
P:AB
12
f
13
'.
I è
S
15
'
l II
1
18
'•'
_
20
BLt
ACCE
PTA
23
!'.
"
2A 25
k«-
:FEC-
Q
26
• .
?
...
? i
!
28
X CC
EPTA
B1 E
DEF
ECT.
**.
CATEGORY C
)
-
10 17 22
'. î
.. . . J
29
-••
':
CRP
DDP
Table 10.5. : Performances of techniques applied by teams EC and LN. Plate No. 3.
104
Depth zone 0 to 18 mm
D F j
1.0-
0 5 -
L
I •iH.. U SL^rvJL _Ç^vJL _C^ TD
0* «5* 60' 70" TECHNIQUE
MDDF
Depth zone 0 tO 265 mm
0* 45* 60* 70* TECHNIQUE
Depth zone 0 to 18mm
MCRF
TECHNIQUE 0 ' «5* 60" 70*
Depth zone 0 to 265 mm
MCRF
1.0 4-
0.5
TECHNIQUE
MCAF
Depth zone 0 to 265mm
TECHNIQUE 0* « * 601 70"
U « UNCLAO SIDE
C : CLAD SIDE
Plate No. 2 50% DAC
Figure 10.6. : Example of technique performance on plate No. 2 50% DAC (Team EC),
105
10.3.4. Comparison of techniques performances as a function of the
cut-off level
Again, due to the lack of data at 10% DAC and 50% DAC the results of one team only can be used for comparison purposes when the cut-off level is used as a parameter.
Tables 10.7., 10.8. and 10.9. give the results of individual techniques as a function of the DAC level for three plates.
The information is consolidated in Figures 10.7. and 10.8. from which the following conclusions are clearly apparent.
a. 70 angle probes were efficient (particularly as applied here by
team EC) even at 50% DAC.
b. 45 and 60 angle probes produced similar results and were very weak
at 50% DAC.
c. the contribution of tandem probes appears poor as conducted by team EC; as shown on Table 10.6. however, tandem was very efficient at 10% DAC as conducted by team NK.
Figure 10.7. confirms Figure 5.8. based on the evaluation of procedures. The importance that the 70 angle probe can have for the near clad surface defect is enhanced.
NOTE : The low DDP values at 50% DAC on plate No. 1, for 45°, 60° and • o
70 angle probes was due to the fact that the near clad surface defects in plate No. 1 were small diameter slags or pores.
10.4. Influence of the orientatin of the defect (Techniques at 20% DAC)
Table 10.10., like Table 9.7., indicates defect characteristics in plate No. 3. Tilt and skew angles have to be noted.
The performance of 45 techniques appears to be dependent more on the defect category (A or B) than on the orientation of the defect.
60 angle probes appear less sensitive to defect category in this table. Also, the overall statistics shows less efficient detection for category A defects (Table 10.1.).
70 angle probes are of course sensitive to defect position in depth. As already shown by table 10.1., their detection performance is rather indépendant of the defect category.
No simple observation can be made for the importance of tilt and skew angles of the defects. Parametric studies deal with these specific parameters.
106
TECH 10%
NK
Plate No. 3
Plate No. 2
Plate No. 1
HDDF HCRF HCAF
MDDF HCRF HCAF
HDDF HCRF HCAF
0°
U C
.26
.07
.96
.50 .44
.18 .09 1 . 1 .
.73 .47
.43 .43 1. 1.
45°
U C
.29
.14
.88
.83 .44
.36 .18 1. 1 .
.73 .33
.29 .43
.88 .88
60°
U C
.23
.14
.96
.06 .56
.0 .18
.1 1 .
.33 .27
.14 .43 1. 1.
70° 0-18
U C
.40
.0
1.
.80
.0
.0
1. 1. .33
70°
u c
.10 .0
1.
.33
.0 1 .
.47
.29
.75
TD
U C
1. .40 1.
'. -94 .27 1.
.60 .47
.29 .14 1. 1.
Table 10.6.
MDDF
10%
20%
50%
MCRF
10%
20%
50%
MCAr
10%
20%
50%
0°U 0°C
.67
.33
.20
.0
.14
.0
.88
1.
' •
45°U
.73
.43
.20
.57
.57
.0
1.
.86
1.
45°C
.67
.33
.20
.43
.29
.0
.88
1.
1.
60°U
.53
.50
.20
.57
.67
.29
1.
1.
1.
60°C
.47
.55
.27
.71
.33
.0
1.
1.
1.
70°C 0-10
1.
1.
.40
.0
.50
.0
1.
.33
1.
70°C
.33
.33
.20
.0
.14
.14
1.
.75
1.
TD U
.33
.31
.20
.17
.43
.43
1.
1.
1.
TD C
-
.31
.27
-
.29
.14
-
1.
1.
Plate No. 1 Note : At 10% DAC,only spot ind icat ions are
Table 10.7. reported by the team.
107
Plate No. 2
MDDF
101
20X
sot
MCRF
10X
20X
501
MCAF
10X
201
501
0°U
.56
.39
.06
.09
.09
.0
1
1.
1.
0°C
.20
.50
.06
.0
.0
.01
1
1.
1.
45°U
.89
.61
.47
.27
.64
.5
1
.87
.87
45°C
.78
.82
.33
.18
.36
.09
.88
.88
.99
60°U
.78
.83
.39
.45
.45
.18
.88
.88
.92
60°C
1.0
.94
.44
.73
.73
.28
.76
.94
1.0
70°C 0-18
1.
1.
1.
.40
1
.80
1.
.75
1.
70°C
.44
.28
.28
.18
.45
.36
1.
.94
1.
TO U
-
.94
.78
-
.82
.91
-
.36
.50
TD C
-
.72
-
-
.82
-
-
.50
-
Table 10.8. 10% DAC Procedure correspond to Fab. Control. Only spot indications are recorded.
Table 10.10 (Two teams only) Techniques at 20% DAC.
110
10.5. Influence of the probe frequency
The PISC II Data Bank does not allow the evaluation of the importance of the probe frequency : too seldom, teams used the same technique with different frequencies.
10.6. Inspection of the nozzle to shell weld through the nozzle base
Several teams used as part of their procedure ot as their only procedure a set of techniques based on simple scanning of the nozzle bore for detection, location and sizing of defects present in the nozzle to shell weld. Transducers were generally designed for longitudinal waves at 0 , 10 or 15 and some of them were focusing ones.
Although the whole set of such techniques has not been considered, trends are clearly apparent which favour this technique when the geometry of the nozzle makes its application possible.
a. Plate No. 9 : Technique in the spirit of ASME at 20% DAC
After excluding the defects near to the clad surface (Z = 0 to 2 2 = 25 mm) which cannot be reached due to the nozzle corner geometry, the detection is as good as 90% (DDF = .90).
b. Plate No. 9 : Special procedures
Team YC using immersion focusing probes reached DDP = 1., when exclusion was made of the defects near to the clad surface as in a. above.
In conclusion, such a technique appears safe for detection when geometry permits the scanning.
A further evaluation has to be made but it requires a reorganization of the PISC II data for plates No. 9 and 3.
m
11. Optimization of procedures
Individual technique results can be combined, at the level of single indications,as if a team would have applied a particular procedure made of some of these techniques. Examples of these "artificial" procedures are given in Figures 11.1., 11.2. and 11.3. for plates No. 2, No. 3 and No. 9 as a result of the data available on techniques at 20% DAC (§ 10.2.).
Table 11.1. correspond to the artificial procedures in the spirit of ASME at 20% DAC as combined on the basis of Table 10.1.
The major observation to be made is that a combination of 45 and 70 angle probes (as well as of 60 and 70 ) produce performances which often, when the clad surface is used for scanning, are of the order of quality of any other standard combination of techniques. Such diagrams could provide an optimization of procedures. The best performances are obtained by the combination : 45 + 7 0 + TD from both sides!
Comparison can be made between teams' results and artificial procedure performances as obtained by the computer code. Examples are given in Tables 11.2. and 11.3.
Team procedure
Aritificial procedure
F
MDDF
.76
.91.9
'late No.
MCRF
.79
.84
2
MCAF
.80
.70
MDDF
.53
.72
Plate Nc
MCRF
.36
.62
. 3
MCAF
.88
.75
Table 11.2. (ASME 20% DAC, Unclad side)
The same comparison can be performed for other procedures :
Team procedure
Artificial procedure
F MDDF
.98
.83
'late No. MCRF
.97
.79
2 MCAF
.69
.65
MDDF
.68
.78
Plate No MCRF
.43
.57
. 3 MCAF
.77
.62
Table 11.3. (ASME 20% DAC with supplementary probes, clad side)
112
MODF
ARTIFICIAL PROCEDURE
MOOF
ARTIFICIAL PROCEDURE
MCRF
O.S
Depth zone 0 to 18mm
U C y v U C „ U C , A.TO
A B A • B
MCRF ,,
0.5
Depth zone 0 to 250mm
MCAF,t
U C , , u C , , u C , A.TD
A B A . B
Depth zone 0 to 250 mm
ARTIFICIAL PROCEDURE
A
B
U
C
3
=
=
=
45* • 70*
0%45*+60*
UNCLAD SIDE
CLAD SIDE
Plate No. 2 20% DAC
Figure 11.1. Artificial procedures performances on Plate No. 2 at 20% DAC
113
MODF
ARTIFICIAL B ».B A.TD PROCEDURE
3DF ,,
1.0
0.5-
U C ,» U C , , U C / A.TD
A B AtB ARTIFICIAL PROCEDURE
MCRF Depth zone 0 to 18 mm
ARTIFICIAL PROCEDURE
RF A Depth zone 0 to 250 mm
U C ^ U C ^ U C , A.TO
A B A . B ARTIFICIAL PROCEDURE
MCAF,, Depth zone 0 to 250 mm
ARTIFICIAL PROCEDURE
A
B
U
C
= --s
45* • 70*
o'+^'+eo* UNCLAD SIDE
CLAD SIDE
Plate No. 3 20% DAC
Figure 11.2. : Artificial procedures results on Plate No. 3.
114
MDOF
ARTIFICIAL PROCEDURE
MCRF 11
ARTIFICIAL PROCEDURE
MCAF
ARTIFICIAL PROCEDURE
Depth zone 0 to 200 mm
A = 45* • 70*
B = O'+AS'+éO*
U = UNCLAD SIDE
C = CLAD SIDE
Plate No. 9 20% DAC
Figure 11.3. : Perofrmance of the artificial procedures on Plate No. 9.
115
DEFECT CATEGORY
(45+70) U
(45+70) C
(0+460) U
(0+45+60) C
A+B U
A+B C
A+TD (U+C)
A+B+TD U
A+B+TD C
Category A
MDDF
-
68
.55
.50
.55
.68
.73
.59
.68
MCRF
-
.43
.36
.21
.36
.43
.57
.50
.43
MCAF
-
.88
.88
.88
.88
.50
.75
.75
.50
Category B
MDDF
.72
.77
.88
.65
.88
.85
.88
.88
.85
MCRF
.47
.50
.83
.67
.83
.75
.67
.83
.75
MCAF
.83
.71
.57
.64
.50
.50
.50
.50
.50
Category C
MDDF
-
.80
.80
.80
.80
.80
.80
.80
.80
MCRF
-
-
-
-
-
-
-
-
-
MCAF
-
.60
.70
.40
.70
.40
.50
.70
.40
Table 11.1. Performances of artificial procedures
on Plate No. 3 defects.
Differences are due to the very different sample of data used for the
two evaluations :
- at the level of procedures - at the level of techniques and combinations for artificial procedures.
However, it appears often that computer results (evlauation at the level of techniques) are better than the team results as expressed by their data sheet No. 6.2. (mainly for plate No. 3). Such a negative human influence has been found also in the PISC I exercise : contributing factors are that teams tend to forget single indications not confirmed by several techniques and teams often restrict the envelope of the defective zone. Such a comparison would be fully correct, in the present programme only for one particular set of team results : taking team EC results as the major example, artificial procedures results are shown on Table 11.4. and 11.5.
From these tables one has for the artificial procedure ASME 20%, C,S+TD the results :
DDF = 1 for plate No. 2 and .99 for plate No. 3 CRF= 1 for plate No. 2 and .85 for plate No. 3.
the result of team EC declared for a full procedure ASME 20%,C,S+TD was
DDF = 1 for plate No. 2 and 1 for plate No. 3 CRF = 1 for plate No. 2 and .93 for plate No. 3
This comparison is consistent and shows that team EC introduced prudent human judgment (CRF value on plate No. 3).
117
^-^OEFECT N.
TECHNI0UÈ5~\^
70% 45*
70% 60*
0% 45% 60*
0V45V 60V70*
45%70%TO
3V45V60V70ÎT0
45** 70VT0FD
1
• .
F 1
'à i
i
2
iï;
,r*
y,-
$
3
.y.
-=?
• "»
*"Û
4
:'"
! # •
• 'V
if!
5
• f
¥ /,,
& . • i t
DDP
6
•:}
-v • * v
;."'
•.£••
.i'ï
.-£.
7
,V'.'
•»i>
• J Î "
1 f '"}.
8
' • • : l
./i -1
l i s
9
i •s4
1 ià
10
; . • ' • ; •
U S :
11
. :'
. 'V
' • ) • •
12
v
;ffl
' t . • • • • - ;
•'Ji "-;t "2 .Vf
• y 'r
•SI
il fi? i..'.
«
13
:,
.'.':!
.U
*-j
' î
14
" :'
••-•
w fy !.'1
,'fi «
PLATE N c
15
•4
li. i„>
16
4
f:'
S'
17
..t
.à. i »
:f.
1 s*
''"i' *r ' ï
# !* if
•'ï '.ft ;.-* 1
.2
18
: " i
'Ï
i i i
"M;
ONE
1
!?'-!
§ . j:
*..-« U
2
? ",rî.
' . " • *
i
ilL i ï
TEAK
3
! • • ' '
VA
!;.•
4'.
k
[1 t;
»
k
L±: ;"id
5
J 6
Defe
ct s
;
- <a -• + J
Q. ai o u «c
.:!• i—
"T
1
'
«
CRP
9
•t-> u
. ' ' 1
Acce
ptab
le De
fe
1
10 »
-. i':'. i . ,
A)
U ÏA
i ï
_i
12
4 'V-
s
13
-
;•
y
:V.
ï •ïï
'y-' • -
K
l •"S.
' r
15
Acce
ptab
le De
fect
16
r
ï
±
17
—
'**'.
\
18
u ai t -01 a Ol
"jô K +J a (L
<
Table 11.4 : Artificial procedures performances as applied on plate No. 2 (20% DAC)
"~^\0EFECT N.
TECHNIQUES"--^
70** 45*
70'«60'
OV 45% 60*
0% 45** 60V70*
45% 70% TO
0*45V60%70;TD
45% 70%T0FD
70% 45*
70% 60*
0V45%70'
0V45%60V70*
45% 70%TD
0145V 60V70VT0
45% 70%TOFD
2
'
:.'.. '.,
.:
• '
Defect"
-Acc
epta
ble
6
r '
y
• J *
' • ' •
1
7
I-' :,:;
•• t \
• •" • ;?
1
'M
1
i/;
CATEGORY
8
-
' . i
;• •; j
I « « $
9
: . j
-C:
•!•
5*
1 ê
Ht.
14
.'
••}_
• ( • (
•rf-
? w P 3s*
I
16
>_
V.
i \:
\
1-
• - ' < *
'ES fril
9 É
A
19
.'(.
y. /:;
i;
21
> • • •
«v
27
! •
^ L ' M r
"Wa
3!
1 Acce
ptab
le De
fect
s
30
• . ' >
. M
y
-n
—
4
• - :
•1
' i '
5* raft-M1
5i» K-%
5
- • • !
* • ; '
; • :
1 ^ i
ffli
ï
11
Acce
ptab
le De
fect
1
12
'J
\
5-
CATEGORY
13
.,.
il
à
f I 'Vf:
1 i;
15
i
l 1 »!
l
! ^ ^
18
• i
/
20
B
23
- .i
Acce
ptab
le De
fect
s
24
•'i:
' • ' i
-'
25 26
i
• ' • ;
' ' j
:' t
'ë st
1
28
CATEGORY C
1
N/
^
Acce
ptab
le De
fect
s;
i.;
10
• *
: i
17
•'.;
22
ï : '
t .
. - i '
29
•A;
i
i .".;
>
•
Table 11.5. : Artificial procedures performances as applied on plate No. 3 (20% DAC)
118
12. DETECTION OF THE BASE MATERIAL DEFECTS IN PLATE No. 1
12.1. Introduction
No team made a specific report on the presence of base material defects in plate No. 1. Many teams however got indications of these manganese sulfide inclusions but did not record any difference between these indications and those corresponding to real defects in the heat effected zone (cf. Report No. 3). Although the PISC II Managing Group considers that such base material defects are not to be found in modern pressure vessels it is interesting to understand which procedures and which techniques provided a clear detection of these particular "defective areas" shown on Figure 2.1.
12.2. Detection of the base material defects by teams using full procedures
Table 12.1. gives the results of the teams taken from Table 4.3. : best results of each team. In general detection was high and teams could have reported on these defects separately but they were normally reported as large defective areas.
Unfortunately, several advanced sizing techniques made evaluations of real defects combined with base material defects although CH and RD sized the weld defects separately from the base metal defects (see Table 4.3.). The sizing of all the real defects examined by MA was distorted by associating them with the base material defects.
Table 12.2. shows the detection rate of each defective layer for the whole group of full procedures.
12.3. Detection of the base material defects by individual techniques at 20% PAC and 50% PAC
Utilizing the results of team EC on plate No. 1 it was possible to evaluate the detection rate of base material defects by individual techniques.
The results shown in Table 12.3. demonstrate that a critical analysis o
of these results (e.g. efficiency of 0 angle probes) would have indicated the nature of the defects.
119
Team
BA BC CB DM EC ES JS KA LC LN MT NE NS SY TP TH YH SU NK
Procedure
Ph. Arrays ASME ASME 20% ASME 10%, STD Complex comb. ASME 10%, STD ASME 20%, STD ASME 20%, S ASME Sp. 10% (ALOK) ASME 20%, S TOFD ASME 20%, S ASME 10%, S ASME 20%, S ASME 35%, TD ASME 10%, STP ASME 20%TDonly
* Team KA made a partial reporting of its findings.
Table 12.1. : Detection of segregations by full procedures
Defect layers 1
MDDP .09
2
.15
3
.13
4
.15
5
.17
6
.14
7
.06
8
.05
9
.05
Table 12.2. : Detection rate of each base material defect layer in plate No. 1; full procedures
Technique
0°
45°
60°
70° (0-18 )
TD
50% DAC DDP
.22
.0
.22
, .0
.44
20% DAC DDP
.55
.22
.44
.0
.44
Table 12.3. : Detection of base material defects layers in plate No. 1 by individual techniques
13. RESOURCES USED BY TEAMS
The PISC II Round Robin Test was conducted in favourable conditions for the inspection of the plates. Many teams inspected the plates in their laboratory. The time available was relatively long although defects were numerous to detect and to analyse. The time for evaluation of the results by the teams was unlimited.
Furthermore, samples were readily accessible; unlike the situation when a nozzle has to be inspected in the real conditions of a reactor pressure vessel in service. Scanners were specific and manual inspection was not performed in adverse conditions of temperature, humidity, and irradiation.
The effective resources used were, however, very different from one team to the other. Some teams worked in standard industrial conditions. Others deployed resources which were never used before for the inspection of a pressure vessel nozzle. Such application of extensive resources was not inconsistent with the programme objective because the PISC RRT was an exercise aimed at the assessment of the performance of NDT procedures and not a reliability evaluation exercise. It is, however, important to have an indication on the resources needed to reach the results shown in the reports.
A simple analysis was undertaken for all four plates where the following parameters were quantified for each procedure category :
- the teams' attitude (industrial, experimental test); and - the teams' resources described by :
a. the inspection time used : one or two days (D) one week (W) two weeks (2W)
b. inspector's qualifications : industrial (I) advanced (A)
c. evaluator's qualifications : industrial (I) advanced (A)
Tables 13.1. to 13.4. give the results and permit some observations to be made. In going from high cut-off level to low cut off level and to special procedures, the following trends should be noted
- The inspection time increased from one day to one or even two weeks;
- The quantification of inspector changed from those industrial technicians or engineers for basic inspection to those of highly qualified engineers and research staff applying advanced sizing techniques.
- The people in charge of the evaluation of the recorded results for special procedures are highly qualified.
121
Another clear correlation is apparent : the qualify of performance obtained seems to be a function of the time taken for inspection and of the qualifications (tertiary training) of the people in charge of the evaluation of results.
Such observations are relevant in the context of the general trend of resources applied, performance and results but there were extreme departures. On plate No. 3, for example team YC performed the full inspection in less than one hour, evaluated the results very quickly and produces results to the standard of a special procedure. Some other teams used the full two weeks available for inspection and returned the data sheet of results several months later.
122
ASME
507. DAC
ASME
" 207. DAC
ASME 107. DAC
SPECIAL 3R0CEDURES
AVANCE D SIZING
TECHNIQUES
ATTITUDE
IND
US
TRIA
L Hi
TO
BEC
OM
E IN
DU
STR
IAL
TES
T
INSPECTION
TIME
DAY
m
WE
EK
TWO
WE
EK
S
INSPECT.
QUALIF
IND
US
TRIA
L
AD
VA
NC
ED
EVALUAT.
QUALIF
IND
US
TRIA
L
AO
VAN
CED
PERFORMANCES
U. o a z M
CR
F
MC
AF
Table 1 3 . 1 . : PISC I I p l a t e No. 1
100*/.
- 100V.
-100*/ .
- 100*/.
-100*/.
ASME
507. DAC
ASME
207. DAC
ASME
107. DAC
SPECIAL
PROCEDURES
ADVANCED SIZING
TECHNIQUES
Table 13.2. : PISC II plateNo. 2
123
SPECIAL
PROCEDURES
Table 13.3. : PISC II plate No. 9
ASME
50V. DAC
ASME
207. DAC
ASME
107. DAC
SPECIAL
PROCEDURES
ADVANCED SIZING
TECHNIQUES
ATTITUDE
IND
UST
RIA
L
TO B
EC
OM
E IN
DU
STR
IAL
BK5a
TE
ST
INSPECTION TIME
DA
Y
WE
EK
TWO
W
EE
KS
INSPECT. QUALIF
IND
US
TRIA
L
AD
VA
NC
ED
EVALUAT. QUALIF
IND
US
TRIA
L
AD
VAN
CED
PERFORMANCES
M D
DF
MC
RP
MC
AF
-100*/.
L loo*/.
- 1007.
-1007.
-1007.
Table 13.4. : PISC II Plte No. 3
124
14. CONCLUSIONS DRAWN FROM THE ANALYSIS
14.1. Conclusions listed here must be considered with the major limitations which characterize the plate defects, inspection procedures and evaluation procedure of the PISC II RRT :
- plate thicknesses were either 250mm (No. 1,2,3) or 200 mm (No.9) - plate No. 1 contained base material defects which were confused by inspection teams with the intended defects (PISC II did not intend to consider base material defects)
- defects in plate No. 2 were often surrounded by satellite defects due to the implant process which modified the detection rate of these defects
- plate No. 2 and plate No. 9 contained simulated cracks which were often adjacent to or contained small volumetric defects,
- Several defects (e.g. defects No. 12 and No. 6 in plate No. 9) were badly introduced and the envelopes considered for the evaluation could result in a too severe description of the defect.
- Due to large defect location errors made by teams in plate No. 3, the evaluation team has been obliged to judge whether some defects were detected or not; a subjective aspect has thus been introduced.
- Plate No. 3 contained several sharp, smooth planar defects but artificially produced.
- Too many manual inspection procedures were applied on the PISC plates; not enough mechanized procedures were applied to provide a useful analysis of these automated procedures. Conclusions have to be understood as trends.
- Sizing results are characterized by a very large dispersion. Strict statistical rules would conclude that no difference appears between sizing capability of several of the procedures considered. However, to evaluate trends, considered as valuable informations from an engineering point of view, several graphs were drawn on the basis of average values which show clear chances of oversizing or undersizing when the standard deviations are not strictly taken into account.
- Several groups of procedures or techniques are represented by few teams (e.g. less than 5). Again results have considered as valuable indications of trends even if stricts tests of significance would show that no firm conclusions can be drawn in many of the comparison considered.
- Finally, it has to be noted that the majority of the inspections were conducted in laboratory conditions (sometimes using ISI tools for the mechanized inspection) and not in real condition.
125
14.2. It is important to recognise that results presented here should be interpreted as beeing simple descriptive statistics. That is, no confidence bands have been added to imply that the results can be extended to some wider population of reactor components. It is most important that, before broader conclusions are drawn from the PISC II results, the parameters used are first clearly understood and then the relevance of the PISC II component and defect types to any other components are carefully assessed.
14.3. At the global level of analysis, using all the best results for each team on each plate, it is possible to conclude that a high detection rate (DDP ^ .9) was achieved for defects greater than 8 mm in height (DZ) in plate No. 2, greater than 16 mm in plates No. 9 and No. 1 and greater than 32 mm in plate No. 3, (Figures 4.1. to 4.4.).
Global results for correct rejection show a tendency to undersize rejectable defects. On plate No. 1 all defects with a height (DZ) greater than 25 mm had a CRP = 1. On plate No. 2 all defects with a height (DZ) greater than 25 mm had a CRP ^ .80, on nozzle plate No. 9 such defects have a CRP ^ .75 and on nozzle plate No. 3, CRP ^ .6 for similar size defects. Average results over all teams for the primary parameters are given in Table 4.1. (A defect of 25 mm height generally represents 10% of the wall thickness).
Global results are presented only as a reference set of figures with which to compare the results of individual teams or groups of teams using similar procedures.
14.4. The best results of individual teams show a large scatter on all three plates. Nevertheless good results (DDFR ^ .9, CRF ^ .9) were achieved by many teams on plate No. 2 and by a few teams on plate No. 1 and nozzle plate No. 3. Teams were slightly less successful on nozzle plate No. 9 with the best teams achieveing a DDFR of 1.0 and CRP of 0.89 (Tables 4.3. to 4.6.).
The full range of team parameters who used detection and sizing procedures were, when considering again the rejectable defects to evaluate the detection rate :
Plate No. 1 :
Plate No. 2 :
Nozzle No. 3
Nozzle No. 9
0.3 0.14 0.14
0.82 0.24 0.27
: 0.40 0.38 0.20
: 0.67 0.56 0.00
- ^
^ -
<
< -T" "
c: <
<
^
< :
DDFR CRF CAF
DDFR CAF CRF
DDFR CAF CRF
DDFR CAF CRF
^ *r_
^i
^ s^ ^
*-:*
^x « • ^
« ^
« ^ ^C
1.0 1.0 1.0
1.0 1.0 1.0
1.0 1.0 1.0
1.0 1.0 0.89
126
14.5. Analysis of results for groups of teams using manual procedures with a reporting threshold level show that increased detection and correct rejection rates occur when decreasing the threshold from 50% to 20% DAC. However, further reduction in reporting threshold to 10% DAC does not appear to improve detection or correct rejection rates of the defects considered in the exercise (Figures 5.7 and 5.8).
Procedures using a 50% DAC reporting threshold without the addition of supplementary techniques (for exemple, 70 angle probes) gave an unsatisfactory performance.
14.6. Significant improvements occur if a 70 (S, L or SEL) angle probe is used to supplement an ASME type procedure (* see section 2.7. for PISC definitions). Improvements occur at all recording levels but are larger at 50% and 35% DAC. It has been difficult to assess the advantage of adding a tandem technique alone to an ASME procedure because most teams tended to add either a 70 angle probe or a 70 angle probe and tandem probes. Results where both are used with a 20% DAC ASME procedure tend to be very good, see for example Table 5.4.
14.7. For plate No. 2, in which defects were located either near to the clad or outer surface there is an indication that inspection through the clad surface disturbs sizing of defects when using procedures at 20% or 50% DAC (Tables 5.3 and 5.4).
14.8. In several cases, manual scanning gave better results than a mechanized one using, in principle, the same inspection procedure (Tables 5.6 and 5.8). However, this conclusion should not be misinterpreted : mechanized scanning also permits easier data treatment and thus makes it possible to consider more indications, reducing the possibility of human errors. Supplementary techniques and lower reporting levels may also be more easily introduced. The best results on the most difficult plate, nozzle plate No. 3, were achieved mainly by automated procedures. Such procedures have produced results such as :
. Detection rate (DDP) : .9 to 1.0;
. Correct rejection rate of large defects (CRP) : .9 to 1.0; and
. Correct acceptance of small defects (CAP) : > .5.
Some of the procedures which showed good performances were in the spirit of ASME (at low or medium recording level and complemented with several techniques) but most were based on different principles or complex combinations of established and advanced techniques. For these procedures it is not possible to evaluate the contribution of the mechanized scanning. Further work has been recommended by the PISC II Managing Group to the PISC III Managing Board.
14.9. Several special procedures, different from the ASME type, were used (Table 5.9) which performed well on one or more plates. Complex combinations of standard and advanced techniques were particularly successful in producing some of the best results on nozzle plate No. 3 (Figure 6.1.).
14.10. Distributions of absolute errors in defect location in the through-thickness direction show a high dispersion (Figures 4.13 to 4.16). It is a characteristic of the nozzle plate No. 3 geometry to induce large errors in location. Some procedures were, however, systematically precise in locating defects. (Figure 6.1.).
14.11. Conclusions on sizing the through thickness dimension of the defects (DZ) appear difficult to draw at the global level. Distributions of absolute errors in sizing show high dispersion and distinct tails (Figures 4.13 to 4.16). The best results were obtained with advanced sizing techniques used within very complete procedures and by a few advanced sizing techniques used alone (Table 5.9., Table 5.10).
Good results are also given by well complemented ASME type procedures (Table 5.7).
14.12. Several advanced sizing procedures or techniques demonstrated the negative influence of the cladding on sizing when the clad surface is used for scanning (Table 5.14). Available data was however too small and not detailed enough to draw clear conclusions.
14.13. Inner radius cracks were generally well detected. Procedures using the 70 angle probe through the clad surface produced good detection results. Sizing errors in the through thickness dimension are, however, often very large and dispersion is high (Table 7.2).
14.14. Several teams declared an error band on defect sizing. When considering the upper limit of this error band (increasing the defect size of 3 to 5 mm), defect evaluation (CRP values) became perfect for the special procedures (Figure 8.1). When using such a tolerance on size as a parameter real weaknesses of procedures appear for plate No. 3. Even with an important increase (30 mm) of the declared defect size, ASME type procedures at 20% or 10% DAC will not reach MCRF values greater than .80. ASME type procedures at 50% DAC are limited to a MCRF maximum value of .35. (Figure 8.1).
14.15. A dominant trend appears which indicates that defects near to the internal surface (clad side) are more likely to be undersized than defects located near to the external surface (Figures 9.3, 9.9).
128
14.16. Defect size appears to affect both detection reliability and
sizing accuracy.
a. On average, defect detection reliability is lower for defects smaller than 10 mm. This observation must, however, be qualified by considering the defect category (see paragraph 14.17).
b. Average error in defect size also depends on the size of the defects as shown in figure 9.6.
Small defects are generally oversized but medium and large defects are, on average, correctly sized. However, because of the high dispersion of sizing errors (Figure 9.7.) the risk of undersizing defects is greater for medium and large defects than for small defects.
14.17. Conclusions on plate No. 3 were difficult to draw when all defects were treated together. Plate No. 3 was conceived as being representative of real geometries, and it contained three important categories of defects. A (sharp smooth cracks) C (volumetric defects) and B (rough cracks and non-realistic defects). The dispersion of the results for plate No. 3 disappears when they are interpreted with respect to these categories : a good correlation between the detection rate and defect size for each of the three categories is obtained (Figure 9.10). Category A (smooth cracks with sharp crack tips) appears to be the most conservative case, and thus should be used in validation exercises. Even the most powerful procedures show weaknesses for some defects of this family. Category A defects are nearly never detected (detection rate, CRF = .07) by ASME procedures working at high reporting level (50% DAC) (Figures 9.14 and 9.16).
Conclusions obtained on plate No. 3 are valid for the other plates (Figure 9.12). It also should be noted that X-radiography before and during destructive examination confirmed that no deliberate closure of defects occured by compressive stresses. Defects did not contain corrosion products.
14.18. Detection and correct rejection rates are thus low for category A defects. Undersizing is also systematic for these category A defects. On the contrary, volumetric defects (category C) are generally oversized (Figure 9.17). However, some teams using special procedures did well for all three categories of defects.
129
14.19. Taking conclusions in 14.12, 14.13. and 14.15. into account, the most difficult case appears to be a category A defect of 25 mm or more in height near to the clad surface. Such a defect is not reliably detected and often undersized. Again, several special procedures achieved nearly perfect results (Figure 9.17.) Such a defect is of real concern to structural integrity.
14.20. Several procedures or specific techniques (70 angle probes) appear to be well designed for underclad crack detection (Table 9.4.). For plate No. 3, the upper value of MDDF for ASME type procedure at 20% DAC using 70 angle probes is .89 and all resectable defects are rejected (CRF = 1.0).
14.21. A particular set of defects (of possible importance) has been examined :
- those larger than 10 mm in height within 50 mm of the clad surface; and - those larger than 25 mm if located further away from the clad. In plate No. 3 neither full detection nor correct rejection is obtained by all procedures in the spirit of ASME at 20% DAC : MDDF = .7. Special procedures designed for ISI gave .95 on average for detection and correct rejection (Table 9.5.).
14.22. In trying to evaluate the influence of several parameters on defect detection and sizing, it appears that the parameter "Defect Category" was dominant. No conclusion was possible on the effect of tilt or skew angle (Table 9.6) perhaps because of so many manual procedures involved inthe evaluation. Tilt and squew angles must be important when mechanized scanners are used and the parametric studies under PISC III will consider these parameters in more detail.
14.23. The evaluation of results at the level of individual techniques has generally confirmed the above conclusions on the importance of some parameters such as defect category (or defect characteristics), defect size, recording level, and effect of the clad surface (Table 10.1., Figures 10.5., 10.7.).
14.24. The good performance of 70 angle probes (S, L, SEL) and T0FD is clear in several cases (Table 10.1.). Their contribution to the detection of category A defects located near to the clad surface is to be noted. These techniques have, however, limited ranges of application :
- T0FD did not consider defects with ligament values less than 5 mm under the clad; - 70 SEL angle probes have limited focal depth; and - 70 L angle probes are valid for the whole depth of the plate, excluding their dead zone.
130
Due to the limited number of separate results it has not been possible to identify clearly the contribution of 45 tandem probes. Their performances appears to be very different from one application to an other and from one team to an other for the same application. Tandem techniques as well as 70 angle probes have specific ranges of application and are not designed for geometries such as the one of plate No. 3. The tandem technique claims to detect defects beginning 15 - 20 mm under the surface and it requires an idealized sound path via reflection off the back wall. Therefore the tandem technique is not favourable for curved nozzle geometries with varying wall thickness. The tandem technique is a detection technique and has a limited sizing capability, altough for large defects its contribution is relevant to sizing and thus to CRP which deals only with "large" defects.
14.25. The evaluation of the performance of individual techniques shows that some of them detect the same defect and that others are complementary. Sometimes 60 and 45 angle probes detect the same defects but also miss the same defects (Table 10.5). Probes at 60 and 70 or 45 and 70 often are complementary (Table 10.5.). From this evaluation of the performance of individual techniques it was possible to conclude :
- simple combinations such as 70 SEL with 45 or of 70 SEL o
with 45 and TD, would give high detection and correct rejection rates (Table 11.1.) as would the combination of all standard techniques available : 0 , 45 , 60 , 70 , TD.
- if T0FD is used a simple and efficient combination could be 70° SEL with T0FD. Such combinations are in fact part of, or constitute the basis of several of the special procedures considered in PISC II.
14.26. In creating procedures artificially (on the computer) it appeared, by comparison with full procedures results, that teams often did not assemble correctly the indications recorded. Defect evaluation therefore was not as correct as it could have been if only the instrumentation results were used. Human influence has been demonstrated to be imperfect in several cases. Both for ASME type procedures and advanced sizing techniques it sometimes reduced the possible quality of results; such a conclusion was already obtained from the PISC I exercise.
131
On the contrary some teams such as EC did better than the computer would have done.
14.27. A supplementary discussion is necessary on the difference between inspection procedures designed for Fabrication and those specifically for IS I . On Plate No. 2 most of the teams adopted an attitude typical of fabrication, using both clad and unclad surfaces for the inspection. In most of the cases results were expressed globally. On plate No. 3 most of the teams scanned only the clad surface (with some verification from the unclad side). The 70 angle probes were nearly always used from the clad surface. Results on plate No. 3 are thus more relevant to ISI. It is also the reason why most of the examples used to illsutrate the conclusions are selected from Plate No. 3
14.28. Resources used were very different between teams and between procedures. Simple statistics show that the inspection time and team qualification (scientific level), were high for most procedures which gave high performance. Some exceptions should be noted : e.g. team YC for which the inspection time for nozzle Plate No. 3 was of about 1 hour and which achieved good results.
14.29. The primary PISC II objective was to identify procedures and techniques that could reliably detect and accurately size defects in thick section steel. This objective was mainly achieved by the group of special procedures as well as by one team using an ASME (20% DAC) procedure supplemented with 70° SEL and 70° L angle probes. In general these procedures required longer inspection and evaluation periods and used highly qualified people. (An exception is mentioned above in paragraph 14.28.). Further development of these procedures will no doubt reduce the inspection periods but highly qualified staff will probably remain essential to the accurate evaluation of inspection results. The development of "expert systems" combined with better mathematical modelling of defect response may eventually reduce the need for evaluation by highly qualified staff.
132
15. Major conclusions of the PISC II Programme
15.1. The PISC II reports have presented results on the comparative performance of NDT procedures and on the resources required to use those procedures. Together these results provide an engineering value impact assessment of the relative resources necessary to achieve a given level of reliability using various NDE procedures. Specifically, results have been presented which compare the detection reliability, location and sizing accuracy of procedures and which reveal the inspection complexity, level of training and time required for a given examination. As an example, table 13.4. shows the average performance achieved on plate No. 3 and the comparative resources expended using different procedures.
15.2. Some aspects of the PISC II progamme may limit the applicability of the conclusions.
a. A substantial majority of the tests were done manually whereas fully automated procedures are typical of ISI. Further work on the comparison between manual and mechanized scanning will be performed under PISC III on large size assemblies.
b. Most of the defects were artificial or artificially introduced and a validation of the results of PISC II is necessary on real defects.
c. Some particular aspects of the plates have probably biased the results : many satellite defects near to the intended defects in plate No. 2 and base material defects in plate No. 1.
d. Inspections were generally performed in laboratory rather than real industrial conditions. The effect of environmental conditions and human factors need to be further considered.
15.3. Parametric studies under PISC II generally confirm or explain partially several results of the Round Robin Tests :
- Limited defect detectability by several techniques; - sizing dispersion due to equipment characteristics; and - influence of the clad surface on sizing accuracy.
The group of parametric studies producing the most important results is the one on the "Effect of defect characteristics on detection and sizing"; these laboratory exercises are to be continued under PISC III.
133
In particular, the response of sharp edged planar defects has been studied by this group and the low response observed explains clearly the reason for the lower detection reliability of these defects in the RRT (see paragraph 15.4.f. below).
15.4. The following conclusions on the performance of NDE procedures are of direct significance to the safety of pressurized plant :
a. Decreasing the DAC level for an ASME type examination from 50% to 20% increases the detection rate and improves the correct rejection rate (CRP). It should be noted that a good CRP depends on sufficiently accurate sizing rather than absolute accuracy. As shown on Figure 5.8. for all four plates it is advisable to work at medium amplitude cutoff : e.g. around 20% DAC for the ASME calibration procedure. A 10% DAC would often not improve detection performance, and would increase false calls and rejection of several acceptable flaws.
b. Procedures in the spirit of ASME Section XI appear to perform better if a 70 longitudinal angle probe is added either in the simple echo technique or with dual beam, or both.
Tandem probes can also make an important contribution in some
cases. Table 5.4. on plate No. 2 shows the importance of supplementary techniques : an ASME procedure at 35% DAC complemented with 70 angle probes and tandem probes matches the detection rate achieved with the 10 per cent DAC procedure.
c. The so called special procedures that combine standard techniques such as 45 , 60 and 70 angle probes and advanced techniques for sizing (but often also for detection, e.g. TOFT) obtain nearly perfect results as shown on Figure 18 . Such procedures, generally proposed for ISI but not yet industrially used, are less sensitive to the influence of parameters such as defect position, defect size and defect category.
d. Defect detection frequency appears to be independent of defect position through the plate thickness, however sizing accuracy does appear to depend on position. Defects located near the clad surface are on average correctly sized whereas defects near the outer surface tend to be oversized (Fig. 9.3.). A contrary finding was made in PISC I for the PISC procedure (ASME 50% DAC).
e. Defect size appears to affect both detection reliability and sizing accuracy :
a. On average, defect detection reliability is lower for defects smaller than 10 mm in height. This observation, however, must be further qualified by considering the defect category (see paragraph f below).
134
b. The average error of defect size also depends on the size
of the defects as shown in Figure 9.6. Small defects are generally oversized but medium and large defects are on average correctly sized. However, because of the high dispersion of sizing errors (Fig. 9.7.) the risk of undersizing defects is greater for medium and large defects than for small defects.
f. When considering defect characteristics such as the crack tip aspect and surface roughness it is possible to divide all defects in the PISC II plates into three categories :
A. smooth cracks with sharp crack edges (fatigue cracks); B. rough cracks and crack which were strongly modified during
implantation and have unrealistic crack tip aspects; and C. volumetric defects (slags and porosities).
The dispersion of the results as shown on Figure 9.10. for plate No. 3 disappears when the defect categories are considered; good correlation is then obtained between detection rate and defect size.
Category A (smooth cracks with sharp crack tips) appears to be the most difficult case and therefore the one on which validation exercices should be conducted. Even the best procedures show weaknesses for some defects of this category. Category A defects are nearly never detected and correctly evaluated by procedures in the spirit of ASME working at a high cut-off level: 50 per cent DAC (Figures 9.14. and 9.16.).
Such plots appear to be valid for the results of all four PISC-II plates. This way of presenting the data also seems to explain some results of the PISC-I and DDT exercises.
Since clear conclusions and good correlations are obtained when one considers crack tip aspect alone, it probably explains that this parameter dominates others, such as :
- variability of equipment characteristics; - human errors; and - calibration procedure employed.
g. The most difficult case appears to be one of a category A defect 25 mm or more in height near to the clad surface. Such a defect could be unreliably detected and undersized.
Over all teams and defects a high dispersion of both sizing and location error occured. Figure 4.16. shows the results for plates No. 1 and No. 3 but similar results also occured on plate No. 2 and No. 9. In particular dispersion of sizing and location accuracy in defect length was much worse than in defect height.
135
16. RECOMMENDATIONS OF THE PISC II MANAGING GROUP TO THE PISC III MANA
GING BOARD
16.1. Specific parameters not fully considered in PISC II
Further exercises should address some parameters which were not correctly or fully addressed by the PISC II RRT. These major factors are : a. the influence of defect location parameters :
. tilt angle,
. skew angle,
. ligament size;
b. the influence of the probe frequency;
c. the influence of the scanning pattern;
d. the influence of the surface conditions;
e. the influence of compressive stresses on crack detection;
f. the specific performances of each of the so called "surface probes":
. 70 S probes,
. 70 L probes,
. 70 SEL probes,
. creeping wave probes,
g. the influence of surrounding cracks for detection of a dominant crack.
Although in several cases, the parametric studies have considered these parameters, effective validations are necessary.
16.2. Validation of the important influence of some parameters
a. Defect category
The dominant influence of the defect characteristics has been emphasized. However, PISC II plates contained artificial defects. Such artificial defects could be considered as unrealistic or overconservative family A defects. Although their ultrasonic response has been compared with real fatigue cracks, a better knowledge of the real defect characteristics is necessary.
b. Structure geometry
Plate No. 3 showed a large dispersion of location errors. It has been attributed to the difficult geometry.
Difficult geometries occur in real life and PISC II results should be assessed again for most of the difficult geometries encountered in a RPV.
136
e. Mechanized scanning
Although the best results on plate No. 3 were obtained with automated systems, the reliability of mechanized scanners has not been demonstrated in PISC II (several human errors were corrected). A lower detection rate has been observed for the mechanization of pure ASME manual type procedure when compared with the corresponding ASME manual procedure. Full scale structures have to be considered in order to verify the performance of mechanised procedures when operated remotely in real situations.
16.3. Further use of PISC II Data and Reliability Studies
PISC II was aimed at the evaluation of the performance of procedures and techniques. It excluded any study of human reliability : all human errors were corrected. The PISC II data bank can, however, be used for the assessment of the importance of several human errors such as :
data recording, transcription, assembling, - data interpretation, - equipment setting, and - mechanical scanner setting
In general, reliability studies must follow the performance assessment studies.
16.4. Extention of the PISC II methodology
From the results obtained and the interest shown for the PISC II results by Licencing Authorities as well as by Codes and Standard bodies, it appears suitable to extend the successfull PISC II methodology, maintaining the international collaboration, on other components and materials of the Nuclear Reactor Primary Circuit.
Such methodology as well as the results obtained up to now should also be of interest for several non nuclear industries.
16.5. Fabrication of test structures
From the PISC II results it is possible to make recommendation on the category, size and position of defects that should be included in test components constructed for the validation of NDE methods. Specifically it has been found that smooth defects having sharp edges close to the clad surface are the least reliably detected and most inaccurately sized. Avoidance of satellite defects is also important as this has been found to considerably enhance the detection of otherwise difficult defects.
137
REFERENCES
1. PISC-11 Report No. 1 : Summary of the PISC II Project
2. PISC II Report No. 2 : . The RRT of the PISC II programme : description of the plates and the ultrasonic procedures used.
3. PISC-II Report No. 3 : Destructive examination of the PISC II RRT plates.
4. PISC-II Report No. 4 : Analysis Scheme of the PISC II Trials Results.
7. PISC I Report No. 6. Commission of the European Communities Nuclear Science and Technology EUR 6371En Volumes I to VI EEC Brussels - Luxemburg 1979.
138
APPENDIX 1
PROCEDURES AND TECHNIQUES
CODIFICATION
PROCEDURE IDENTIFICATION SCHEME
34 (e.g. EC SZ 8499)
^Technique identification
99 : results for all techniques taken together (thus procedures) as combined by the team
97 : as above, but combination made by the computer.
Procedure identification
from 01 to 68 from 69 to A2
in the spirit of ASME special (or advanced sizing)
Remarks regarding - claims for techniques and procedures - particular features - sizing particularities
. 01 : 1/2 max for sizing
. 02 : 20% DAC for sizing
. 03 : 40% DAC for sizing
. 04 : 80% DAC for sizing
. 05 : 50% DAC for sizing
10 : Sizing on C and B scans. IR : Inner Radius Examination CO : particular frequency used for clad surface examination CC : particular frequencies used for clad surface examination PT : particular application of a well-known procedure DT : detection only SZ : sizing procedure only
Team identification : confidential
140
PROCEDURE CODES
A. In the spirit of ASME
Amplitude Examination Manual or Supplementary techniques Cut-off level Surface Automatic
r^^-turo*-infur^to«-eor')cu*- tu.o> eoo>coint7>inooint7>c7>oooo to tu o> in tu *-in h-to in in to •*-tu to co in o> tu tu to to co in tu eu *- eu «- oj «- T- T- i- cu tu COCOCOfOCOCOCDfOCDCOinCOCOU)COCOCOCOininCOCOinCOCOCOCOCO
<<ro)fo«)ooTrs)o>oooor>-'«-o»f£)'«-oorN-t7»o)fu<»rnr«-inr^«HDOo to in in to *- to *- to in tu tu T-to T- oo *- tu to to to in tu T- »- tu *- ru *- *- *- *- tu tu CO t O 0 3 S ) tO CO CO CO CO tO tO tO tO tO t O tO tO tO tO CO CO CO i n t O CO CO CO CO
tO tO tO CO CO tO tO tO tO tO CO tO CO CO LD CO CO CO CD tO CO LI) tO CO tO tO tO tO cocotococncucoincofuc»inr^«-^-coLno>oor^coc7>«-T-intocuco ineor-tDcuoo*-r^oor^inoo*-tuoor^r-r^foto^"tueoojtu»-r~in
tu to r- o> to ru to co in r - to *- to co to m *- *- •*• to to un v *• in in cococotoincotococomincocoeocotDcocococoininincococococo r^r^r-tor^-'-fDtur^u)^-<otO'«-r^fOfntn^-u)'«-«)f)^-c»coc7>'«-tutnn>u>tutO'riftfuc7>c7)tocJ)cocot)00oiftfU'ï-^coT-'-ioLf>T-'-i^r^i^r^r^r^r^r^r-totDr^tor-r^r>-r^r^r^r^cotDoor-r^r»-oooo COCOCOCOU)COCOCOCOfOL/)tOCOU)COCOCOCOCOCOU>U)U>tOtOCOCOU>
CD'-tufn'«-u>tor^ooc7>fO'«-tutr),9-Lntor^oo *-*-t-T-T-f-»-.r-aifUfUfurucurufutu *- tu to •«• in to r>- co o> •«- *- *- *- *- *- *- •«- *- *- ai tu tu ru eu tu ru tu tu o
r- TT ru m T- m tu r- in r>- to to tu *- tu - to o> to in a> in oo in o a> oo oo n tu o in tu v in r- to to in to •«- ru to to in o> ru tu to to to in 0 tu tu *- tu •«- tu •«- »- *- , - cu tu z totototocococototo(SincotointocotocoLnLntotoLntocotococo 2 •*• o> to to oo TT to a) oo T- r- oo r- to to to r- CT> CT> ru oo to t - in c- to to oo ™ toinincD'«-<T>T-toLnfu *- •»-»- oo •»- ru to to to in a. t u ^ ^ - a i i - r u » - » - * - »-fuoj totototocototocotocDtocototococototococototointototococo mrr(Dtotnoja)r^intuooo>totor ,-c7>ooinfUTi-totj)is-tO'^-in'-to •^toLn'<rr^intO'-cO'-o>ootDtoootO'-fU'<rTt-torr)fOfurotU'^-to •»-for^ooo>T-ro,^-,«-'»-inr^co,^-toc>)^r'<a-'-'-Trrr)fr)in'9-^-toin totototototototototototototoincotococotocoLncocococototo cotDtotofntutoin^r-ruin'«-'<-'«-'«-'-c7)oor^foc')'r-T-intDaico incor^toaioO'<-r^-totoinooT-fuootor>-is-mtn'fl-fUfT)fUfUi-i^m
-120 -60 0 60 120 -120 -60 0 60 120 oi , , , ,r7i,l,l,i7rV-.f7« ° l
-120 -60 0 60 120 -120 -60 0 60 120
- i i—i—i—i—
ESZ n- 4 .2
S= 12.5
ASME 20% DAC SPECIAL PROCEDURES
DISTRIBUTION OF ERRORS ON SIZE AND ON DEFECT LOCATION
184
PLATE NO. REF=AAR1REF3
2 0 , , i •
. S J E L T
ie.
14.
12.
10.
8.
6.
4.
2.
0
n= -24.2
S= 35.4
'TrfT r£L
18.
16.
14.
12.
10.
8.
e.
4.
2.
n
ELZ rt= -6 .4
S= 13.Û
n= -31.7
S= 52.8
-60-40-200 20 40 60 -60-40-200 20 40 60
20 rJ I I 1—I—
184 EST
16.
14.
12.
10.
8.
6.
4.
2.
0 a jpu-
18.
16.
14.
12.
10.
8.
6.
4.
2.
n
pi i—i • * •
ESZ r
r —
• i > — t i i i .
n* -6 .5
S= 14.8
xi -120 -60 0 60 120 -120 -60 0 60 120
ASME 50% DAC
60
54
48.
42.
36
30.
24
18.
12.
6.
0
—I i—I 1 L—J I—I L
ESY n= -15.8
S= 32.0
jd
PLATE NO. 3 REF=AARIREF3
60
54.
48.
42.
36.
30.
24.
18.
12.
6.
0
j i • • i i i i i_
ELY n= 5.2
S= 54.6
60
54.
48.
42
36J
. 30.
24.
18.
12.
6.
0
.ELZ • • 1 I I l _ l I I 1_J L_L
rt» 8.0
S= 18.4
hO - 60 -40 -200 20 40 60
60
54.
48.
42
36.
30.
24.
18
12
6.
0
J — L _ l I I I • • • ' I • •
EST M= 15.9
S= 49.5
CCC Lfl
60
54
48.
42.
. 36
. 30
24.
. 18.
. 12.
6.
OL
. ESZ
r4,T "-T^T'T
J...J..-À • • ' • l _ l I—I I—L
60
54.
48.
42
36.
30.
24.
. 18.
12
6.
.ESZ
- 6 0 - 4 0 - 2 0 0 20 40 60
_l I l_l I I I I I—I I L
n= 9.8
S= 18.8
ASME 10% DAC
xt
M= 1 .5
S= 22.2
120 -60 0 60 120 -120 -60 0 60 120
ADV SIZING TECHNIQUES
DISTRIBUTIONS OF ERRORS ON SIZE AND ON DEFECT LOCATION
185
APPENDICIES 4, 5, 6, 7
DETAILED RESULTS OF EACH TEAM
These appendicies are available under separate cover and on specific request at the end of the PISC-II programme : October 1986.
APPENDICIES 4, 5, 6, 7
DETAILED RESULTS OF EACH TEAM
These appendicies are available under separate cover on specific request at the end of the PISC-11 programme : October 1986
TABLE OF MAJOR PARAMETERS
For full definitions of the parameters see PISC-II Report No. 4, sections 4 and 5.
Procedure Description Parameters
A = Automatic scanning procedure B = Examination from both sides of plate or nozzle C = Examination from clad side only % DAC = Signal amplitude threshold level M = Manual scanning procedure S = Surface or high-angle probe techniques (70 S, 70 L, 70 SEL) TD = Tandem technique U = Examination from unclad side only
Calculated Descriptive Parameters
CAF = Correct acceptance frequency for one team on all acceptable defects CAF' = CAF calculated on a limited number of defects CAP = Correct acceptance frequency for one acceptable defect by all teams CRF = Correct rejection frequency for one team on all rejectable defects CRF' = CRF calculated on a limited number of defects CRP = Correct rejection frequency for one rejectable defect by all teams DDF = Defect detection frequency for one team or procedure on all intended defects DDFR = Defect detection frequency for one team or procedure on all rejectable defects (DDF)T = Defect Detection frequency for one team or procedure on all defects DDP = Defect detection frequency for one defect by all teams ELX(ELY.ELZ) = Absolute error in location of defect in the X(Y, Z) direction (mm) ESY(ESZ) = Absolute error in sizing in the Y(Z) direction (mm) ESL = Absolute error in locating edge of defect relative to plate or nozzle surface (mm) MCAF.MCAP... = Any of the above parameters preceeded by M indicates the mean value over a group of defects or teams SELX,SESY... = Standard deviation of location and sizing errors % TND = Percentage of all defects in a plate measured by teams using sizing procedure only
Parameters used on Graphs and Histograms
DX(DZ,DY) = Absolute size of defect in X (Z,Y) direction (mm) DZC = Z dimension of defect at which detection approaches certainty M = Mean value of location and sizing errors S = Standard deviation of location and sizing errors o = Rejectable defect according to section IWB 3510 of ASME XI (1974) o = Acceptable defect according to section 1KB 3510 of ASME XI (1974) HSLN = Crack aspect ratio (DZ/DY) HSTN = Ratio of crack depth (DZ) to plate thickness HSLS = Crack aspect ratio for surface crack according to ASME XI, IWB 3510 HSTS = Ratio of surface crack depth to plate thickness (ASME XI, IWB 3510)