2nd International Conference on NDE in Relation to Structural Integrity for Nuclear and Pressurized Components, New Orleans May 2000.

TABLE OF CONTENTS

Abstract Background Evaluation Method Results References

Abstract

A research programme sponsored by the German utilities (VGB - Technical Association of Large Power Plant Operators) was performed at MPA Stuttgart between 1993 and 1999. Objectives of the programme were to gain information on the feasibility of evaluation methods and on influencial parameters on the results of the non-destructive testing. Circumferential welds of typical austenitic pipes or pipe-elbow connections in nuclear power plants were investigated in a round robin test. Teams of different NDT vendors and research institutes performed blind tests using
• ultrasonic testing,
• radiographic testing ( film technique and evaluation of digitized films),
• digital radiographic testing.

The test pieces consisted of cut-out austenitic weld specimens containing service induced cracks. After the non-destructive testing metallographic analysis was performed in order to verify the NDT results. With these data the performance of the NDT could be assessed using POD and ROC curves.

Background

Pipes in nuclear power plant with boiling water reactors or pressurized water reactors are often made of austenitic steel. To assure the integrity of a component, mainly ultrasonic testing or radiographic testing is carried out. The testing of austenitic welds or of dissimilar welds is still a demanding NDT task. Reasons for this are the geometric conditions in the weld root or the root undercut resulting from shrinking of the austenitic material. The geometry of welds produced in the 1970's is often characterized by
• a weld crown, that has not been ground,
• excessive root bead,
• misalignment of the inner diameters.

Welds produced since the 1980's normaly do not have these geometric constraints. In new power plants or in those, where pipes have been replaced it is far more easy to produce good NDT-results. Except for geometric constraints, the ultrasonic inspection is also affected by the dendritic structure of the weld material which may lead to skewing or tilting of the ultrasonic beam, to scatter or high attenuation. The reliability of the NDT-result is influenced by these factors. This has been a topic in many investigations and research programmes in the past. Examples that should be mentioned here are performance demonstration methods developed in the USA and the international Programme for the Inspection of Steel Components (PISC). The European Network for Inspection Qualification (ENIQ) has developed a methodology for qualification that shall lead to standardized NDT qualifications in the future. In Germany the requirements for non-destructive testing of austenitic and dissimilar welds are listed in the KTA-code. The testing techniques have to be according to the state of science and technology, which is envigilated by independent technical experts like TÜV ( Technischer Überwachungsverein i.e. Organisation for Technical Envigilation). The qualification of non-destructive testing methods in Germany is done by

• theoretical physical considerations,
• Practical assessment of test methods using test mockups with artificial reflectors like flat bottom holes, side drilled holes and notches,
• realistic or real reflectors for special test problems and
• personnel training and certification according to EN473, even though this is not required in the KTA code.

Furthermore, service experiences with the components are also considered in the qualification. The qualification of NDT methods including the written procedures is envigilated by technical experts of the regulatory bodies. This method has been applied with good results in the past. In the future, in Germany qualification of NDT-methods will be done according to the ENIQ methodology. A research programme sponsored by German VGB and performed by MPA Stuttgart (Materialprüfungsanstalt , Institute for Materials Testing) had the objective to evaluate methods to assess the performance of non-destructive testing methods. For this programme, material with real IGSCC originating from replacement of pipe systems in several German boiling water reactors could be used.

The research programme consisted of the following steps:

• discussion of the national and international state of testing for austenitic and dissimilar welds among envigilators, plant owners and inspection vendors
• procurement of test specimens with real IGSCC and implantation of artificial reflectors (Fig 1)
• non-destructive testing
• verification of the real reflectors by metallographic investigation
• comparison of destructive and non-destructive results

Fig 1: Drawing of typical weld used in the round robin test

Fig 2: Typical shape of a weld produced in the 1970's

Fig 3: Typical shape of aweld produced in the 1980's

After the evaluation of the test results of welds made in 1970's ( s. Fig. 2) a second part of the research programme was launched, that assessed welds made in the 1980's (s. Fig. 3) and with state of the art production. With this, the results of the round robin test are representative for the actual weld state in most of the German power plants.

The round robin test was performed as a real blind test, i.e. the inspectors did not have any information about the flaws in the test specimens and did not get any feedback about their performance until all teams had finished their inspection task. The following NDT-methods were applied:

• ultrasonic testing
• electrical four probe resistance method
• radiographic testing
• liquid penetrant testing

A comparison of radiographic film and digitized films was also performed. This paper concentrates on radiographic and ultrasonic testing.

Evaluation Method

For all welds the circumferrence was divided in grading units of 10 mm. The inspection vendors were asked to give their results according to this divisions, but a tolerance of ±1 grading unit was acceptable. This is a sharp restriction, that is normally not used in practical NDT in a power plant. By this, for each section a NDT result was available, which leads to a high statistic safety of the overall assessment. The inspection teams had the task to perform their testing and to state for each grading unit wether it contained a relevant defect or was defect free. It was the task of the teams to define the relevant indications by themselves.

The NDT result for each grading unit as stated by the inspection vendor was compared to the results of the metallographic investigation. The detection rate as a function of the flaw depth could be displayed in a POD-diagram (probability of detection). As the number of defects is still limited, it is more appropriate to speak of a detection rate rather than of detection probability. With the POD-diagram, it is possible to compare the detection rates of inspection teams and also to compare different inspection methods. The detection rate is a good means for this purpose, but it does not state how sure and how reliable the NDT result is.

The NDT of a weld can have four possible results:

• An existing flaw is detected: this case is called "true positive" TP.
• An existing flaw is not detected: this case is called "false negative", FN.
• A defect-free region is stated as defect free : "true negative" TN.
• A defect free region is stated as defective: "false negative" FN.

With these values, the flaw detection detection rate FDR can be defined as

FDR = S (TP) / S ( all sections with defects ),(1)
whereas the false call rate FCR is defined as
FCR = S (FP) / S ( all sections without defects ).(2)

The flaw detection rate FDR can be displayed as a function of the false call rate FCR. A diagram of this type reveals the relative operating characteristics and is therefore called ROC-diagram. It allows a better assessment of NDT methods and inspection teams than a POD diagram.

In order to have a clear assessment of the NDT methods it was necessary to define regions in the ROC diagram that can be used for evaluating the curves. In Fig. 4 the respective regions in the ROC diagram are marked. An excellent performance would be given in the upper left corner of the diagram. A high detection rate combined with a low false call rate gives excellent knowledge about the integrity of a weld with reasonable costs. The regions marked 1-4 are varying degrees of acceptable NDT performances.

Fig 4: Evaluation zones in ROC-diagram

Results

In Fig. 5 the ROC curve of a good radiographic team is displayed. The left curve is the result achieved in the second part of the research project, where ground austenitic welds were investigated. Along the curve the trough wall extend (TWE) of the defect is varied from 2.5% to 25% of the wall thickness. More than 80% of the 25% deep cracks could be detected. The right curve is the performance achieved in the first phase, where welds with the manufacturing technilogy of the 1970's were investigated. The shift to better results between these phases can be explained by the better geometry conditions as well as by an increase of knowledge and experience of the NDT personnel performing the tests.

Fig 5: ROC diagram for radiographic testing

Fig 6: ROC-diagram for ultrasonic testing

In Fig. 6 the ROC curve of one of the better ultrasonic teams is displayed. The left curve again reveals the detection performance achieved at new welds. The flaw detection rate is more than 80% for defects with a TWE of 25% of the wall thickness.

For both UT and RT the improved geometry of the welds produced in the 1980's (see Fig. 3) leads to a increased inspection performance. The flaw detection rate is higher ( note that the curves for 'old' geometry cover 30% TWE instead of 25% ) and the false call rate is considerably reduced.

The transfer from the results of a round robin test to results that can be achieved when using the NDT-method in a power plant is possible and may be useful. Nevertheless, some aspects have to be considered:

• The geometry conditions have to be comparable, e.g. material, grain structure, shape of weld, root and crown.
• In a power plant, the practical approach is normally different to laboratory investigations. NDT results are evaluated by more than one expert and in Germany the evaluation process is additionally envigilated by independent technical experts e.g. from TÜV.

If there is a doubt about the intgrity of a weld, additional NDT-techniques (like using a different incident angle or focussed probes) are applied. In such a case, often diversitary NDT-methods ( e.g. eddy current or radiographic testing, when the first method was UT) are used to get a better understanding of the integrity of a weld.

References

1. Zerstörungsfreie Prüfung von austenitischen und Misch-Schweißverbindungen (Entwicklung von Ringversuchen zum Nachweis der Leistungsfähigkeit von Prüfsystemen) 5.Teilbericht : Ergebnisvergleich - Auswertung der Untersuchungsergebnisse der zerstörenden und der zerstörungsfreien Prüfung (Phase IV.3), Staatliche Materialprüfungsanstalt, Universität Stuttgart

© NDT.net - info@ndt.net

|Top|