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

ROUND ROBIN TEST RESULTS ON ULTRASONIC TESTING OF AUSTENITIC WELDS M. Bièth, J. L. Monjaret European Commission JRC, IAM Petten, The Netherlands F. Hardie, Mitsui Babcock, Renfrew, U.K. P. Krarup Force Institute, Brøndby, Denmark E. B. Pers-Anderson ABB TRC, Täby, Sweden Corresponding Author Contact: Email: bieth@jrc.nl, Web: http://www.jrc.nl

Abstract

Reliable ultrasonic inspections for austenitic stainless steel welds are mainly hampered because of severe attenuation of the ultrasound. The sound is scattered and mode-converted at the boundaries of the columnar grains of the austenitic weld metal. This has several effects for evaluation of ultrasonic signals and then for detect, characterization and sizing of defects which could cause weld failure:
The amplitude of the ultrasonic echo is influenced by ultrasound scattering, phase-changes and mode-conversion in a characteristic way.
The spectrum of the ultrasonic pulse is distorted depending on the transfer characteristics of the columnar grained weld metal.
Ultrasonic grain boundary backscatter is measured as noise decreasing the signal-to-noise ratio during ultrasonic testing.
Four groups of materials comprising the full scale of industrially relevant stainless steels were investigated: austenitic stainless Cr-Ni steels, fully austenitic stainless steels with increased Ni-content, Nickel-base-alloys, and Duplex steels (Ferritic-austenitic steels).
Test specimens with reference reflectors and realistic defects in the weld area, were prepared and inspected using various ultrasonic testing techniques. Manual and automated inspections were carried out to assess and compare the intrinsic capability of different ultrasonic testing techniques currently used by the three industrial partners participating to this project.
The Round Robin Test results demonstrate that:
• the transducer frequency is important on the weld inspectability,
• a probe/procedure performance evaluation can be applied successfully to a reduced population of material and sharp edged notches,
• through wall thickness sizing results are poor because tip notch diffraction echoes in scattering material are not reliable.

Introduction

This paper summarizes the work carried out by JRC/IAM, in evaluating the ultrasonic testing data provided by ABB-TRC, Force Institute and Mitsui Babcock collected during the round-robin test in the scope of the Work Package 5, of the 3 year European Commission Contract SMT4-CT95-2012, entitled “Effect of Ultrasonic Scattering on Inspection of Austenitic Welds”.

The entire project started in February 1996 and was carried out by 5 partners: ABB-TRC, BAM, Force Institute, JRC/IAM and Mitsui-Babcock under the co-ordination of BAM (1). Seven Work Packages are constituting the project, with each partner co-ordinating at least one of these packages and making a contribution to some of the others (Table 1).

Preparation of test welds and weld metal samples

Four industrially relevant material groups have been selected:

• Group 1: Austenitic stainless Cr-Ni steels
  Austenitic steel grades of this group are the commonly most used high alloy steels. These steels generally contain at minimum 12% chromium to improve the corrosion resistance. Sufficient Ni, Mn, C, and N stabilise the austenitic structure. Austenitic stainless steels consist of an austenitic matrix, which may contain small quantities of ferrite. The microstructure depends on the content of elements stabilising ferrite and austenite and differs between fully austenitic and austenitic-ferritic.
• Group 2: Fully austenitic stainless steels with increased Ni-content
  In contrast to the steels grades of group 1 the austenitic steels with increased Ni-content consist of only one metallic phase with less complicated microstructure.
• Group 3: Nickel-based alloys
  Nickel based alloys differ from high alloy Cr-Ni steels by increased Ni content. They get more and more the pure Ni properties, e.g. low thermal conductivity. It may be assumed that this feature influences the grain growth by influencing the cooling rate. Nickel based alloys are alloyed with Mo, which forms segregations of Mo carbides and intermetallic phases.
• Group 4: Duplex steels
  Duplex stainless steels have a balanced ferritic-austenitic microstructure obtained by controlled chemical analysis and heat treatment with a limited ferrite content of 40 to 60 %. The ferritic matrix of the base material contains lathy or also globular austenitic grains. The so-called duplex structure exists also in the weld metal and therefore columnar grain-growth as observed in austenitic materials can be avoided.

Specimens containing test welds and weld metal samples have been prepared for:

• experimental validation through attenuation measurement,
• microstructure characterisation of austenitic stainless steel weld metal,
• capability assessment of ultrasonic testing for austenitic welds.

Description of the test pieces used in the round robin test

The test pieces were designed and fabricated in Work Package 1 under the leadership of JRC Petten. The test pieces, which form the basis of this activity, started to be circulated among the participating partners in November 1997. All the test pieces contained sharp edges PISC type A notches (2).

Four groups of material have been investigated.
Group 1:
The base metal was a wrought austenitic stainless Cr-Ni steel (test piece SMT 1.1), with a thickness of 80 mm. Eight notches were machined, roughly in the middle of the weld on the inner side, of 5, 10, 20 and 40% Through Wall (TW), four were straight TW and four were 35° tilted.
Group 2:
The base metal is a centrifugally casted and fully austenitic stainless steels with increased Ni content (test pieces SMT 2.2, 2.4, 2.6, 2.8), with a thickness of 65 mm.
Test pieces SMT 2.2 and 2.4 have fully equiaxis texture, while test pieces SMT 2.6 and 2.8 have 50% equiaxis texture and 50% column texture.
Test pieces SMT 2.2 and SMT 2.6 contain tilted notches, while SMT 2.4 and SMT 2.8 contain straight notches.
Twelve notches were machined roughly in the middle of the weld on the inner side, of 10, 20 and 40% TW, six were straight TW and six were 35° tilted.
Group 3:
The base metal is nickel-based alloys (test piece SMT 3.2), with a thickness of 20 mm.
Eight notches were machined on the inner side, 10 and 20% TW, four were straight TW and four were 45° tilted.
Group 4:
The base metal is duplex steels (test piece SMT 4.1), with a thickness of 32 mm.
Three notches were machined on the inner side, 10, 20 and 40% TW, straight through wall.

Principles for the assessment of the round robin data

Probe efficiency through notches
The target of this study was to try to set a procedure for performance evaluation of transducer/techniques in view of qualification. For each probe, all the results on all the notches were processed group per group and the UT inspections were done from both sides of the notches.
All the UT results generated in the RRT represent a large amount of data. Display priorities in parameters were defined to summarize the results and to show on graphs easily readable trends:

• The first priority is obviously the Flaw Detection Frequency (FDF). If nothing is detected then obviously nothing can be processed.
• The Through-Wall Size (TWS) of the defect is, in an industrial situation, the second main parameter. In this study, because all the notches are open at the inner surface, the ligament parameter, which is also important, was not taken into account.
• The location error of the defect in the weld cross section is the third main parameter. We have used the difference in mm between the real location and the UT declared location.

Notches detectability
For each group, the results obtained on all notches by each probe. The results obtained on all notches by each probe are shown for each group. The three main parameters are illustrated in “bubble type” graphs as shown hereafter. The Y axis is dedicated to the FDF, the X axis shows the trend of “under or over” TW sizing and the location error is displayed by the radius r of the bubble.

Results of the round robin test

Probe efficiency through notches
For Group 1 (graph 1), the global results are not very good. LW 45° probes did not give good results, particularly for FDF, LW 60° probes gave better results except probe (6). For TW sizing, results are scattered. FDF and TW are spreaded for LW 70° probes. It is necessary to remark the particular case of the Shear Wave 60° probe, which gave very good results even though this was not expected.

For group 2 (graph 2), as obviously expected, LW 0° probes are not at all efficient for straight notch detection. On this test piece, FDF is really poor due to nearly zero detection on test pieces SMT 2.4 and SMT  2.6. This zero detection is due to straight notches and not to wave scattering differences of materials. Both SMT 2.4 and SMT  2.6 contain straight notches and are respectively made of fully equiaxis texture material and mixed 50% equiaxis and column textures.

For Group 3 (graph 3) as it could be expected, the results are much better than on the previous groups, FDF is for all probes over 85%. The TW sizings are accurate and not spread.

For Group 4 (Graph 4) FDF results are generally good, except for probes (1) and (3). TW sizing is also good. These good results are due to the low scattering of ultrasonic beam by the material.

graph1:

graph2:

graph3:

graph4:

Notches detectability
For this analysis, only the results provided by the testing of specimens SMT 1 and 2 are taken into account, because the good properties of the materials of SMT 3 and 4 made all their notches obviously fully detectable. Graph 5, made of the data from specimens SMT  2.2, SMT  2.4 and SMT  2.8 shows that a 35° tilted notch (including all heights of the defects) is more detectable (including all probes) than straight notches. Tilted notches are better detectable on one side, due to the wide interface for the beam. In contrary, straight notches are seen with the same FDF and amplitude on each side, which was expected. Also for tilted notches, as expected, it was found that FDF and amplitude increased with the notch height.

In graph 6, the analysis is made taking separately into account both the angle and the height of the notches.

graph5:

graph6:

It shows that, for the centrifugally cast stainless steel, straight notches with a height of 15% to 25% of the TW thickness are selective enough to sort ultrasonic probes efficiently.

Conclusions

The intention of this project is to improve ultrasonic examination procedures for austenitic stainless steel welds.
Four groups of materials comprising the full scale of industrially relevant stainless steels were investigated: austenitic stainless Cr-Ni steels, fully austenitic stainless steels with increased Ni-content, Nickel-base-alloys, and Duplex steels (Ferritic-austenitic steels). Automated and manual ultrasonic inspection was carried out by the three industrial partners on the test specimens representing the four groups of stainless steel under investigation in this study. Capability of the ultrasonic testing was determined using the pisc-type a sharp notches, which were introduced in the weld area.
Results of the round robin test show that:
• for flaw detection, the frequency is a very important parameter to be considered when inspecting austenitic welds. There is a slight improvement in detection rate when using a 1 MHz transducer rather than a 2 MHz transducer. The detection rate improves dramatically when using a 0,5 MHz transducer instead of a 1 MHz transducer. This was clearly demonstrated, when inspecting welds of the groups 1 and 2,
• for inclined longitudinal wave probes, straight and tilted notches in the weld area, with different through wall size are highly selective for probe efficiency evaluation,
• there is a wide spreading of length sizing, which could also be partly due to the progressive shape of the notch ends,
• the 3 mm side drilled holes, which were introduced in the weld, the base metal and in the along the weld-base metal interface, are essential for conducting an efficient ultrasonic inspection of austenitic welds.
It is possible to apply a probe/procedure performance evaluation with reduced population of material and reflectors. The material has to be representative of inspection conditions with the same kind (or more absorbent) of material to be tested. Some straight notches of some TW sizes are enough selective. On thick and highly absorbent materials, the LW 45° low frequency probes give better results than higher angled probes. The thickness is a relevant parameter. Through wall thickness sizing results are poor but that was expected because tip notch diffraction echoes in scattering material are not reliable.

Acknowledgement

The research project, referenced as Project No. SMT4-CT95-2012, is funded by the Commission of the European Communities under the Standards, Measurements & Testing Programme, 1996 - 1999, DG XII, Brussels.

References

1. M.Bièth, C. Pecorari, T. Seldis, E. Neumann, P. Krarup, F. Hardie, E.B.Pers-Anderson, “Effect of ultrasonic scattering on inspection of welds in austenitic steels”, 7^th International Conference on NuclearEngineering, ICONE 7, 19-23 April 1999, Tokyo.
2. Nichols,R. W., Crutzen, S: “Ultrasonic Inspection of Heavy Section Steel Components”,The PISC II Final Report, Elsevier Applied Science Publ. Ltd. London (1988)ISBN1-85166-155-7.

© NDT.net - info@ndt.net

|Top|