SIGNAL PROCESSING AND IMPROVED QUALIFICATION FOR NON-DESTRUCTIVE TESTING OF AGEING REACTORS (SPIQNAR) N. Cameron 1, W. Daniels 2, R. Martinez Oña 3, H. Vandreiscche 4, R. Chapman 5, O. Roy 6, T. Stepinski 7, K.J. Langenberg 8, X. Gros 9, L. Horácek 10, R. Sundberg 11, B. Neundorf 12 1) Mitsui Babcock, Renfrew 7) Uppsala University 2) Serco Assurance, Risley 8) University of Kassel 3) Tecnatom, Madrid 9) EC-JRC/IE, Petten 4) AIB-Vinçotte International, Brussels 10) NRI, Rež 5) British Energy Limited, Gloucester 11) SQC, Taby 6) CEA, Saclay 12) E.ON Kernkraft, Hannover Summary SPIQNAR addresses two related areas of potential improvement in the ultrasonic inspection of pressure vessels and piping in reactor systems. Improving the reliability in the interpretation of inspection data by exploiting signal processing techniques and improving the reliability of inspection qualification by gaining a better understanding of the ultrasonic response from defects of the types arising in practice. The project started in October 2000 and brings together partners representing nuclear utilities, consultants, inspection service vendors and signal-processing specialists. Originally scheduled to take three years, recently a request has been granted for a 9-month time only extension. The project scope is closely related to work of the European Network for Inspection Qualification (ENIQ) and several members of the network are project partners. To date the following has been achieved: Development of a detailed specification for signal processing; identification and ultrasonic characterisation of 5 austenitic specimens affected by stress-corrosion cracking; initial destructive examination of the most promising austenitic specimen; investigation of realistic defect growth techniques and a call for tenders to manufacture specimens containing synthetic defects; production of signal processing toolboxes and InaSAFT programme; draft issue of integrated CIVA software (a common system for data processing and display); a study of virtual defects and the development of a defect library A Background In-service inspection programmes require volumetric inspection of pressure boundary components of reactor systems and ultrasonics is usually the preferred inspection technique. Inspection of components constructed from austenitic steel or high-nickel alloys is difficult due to the scattering of the ultrasound by the material grain structure. Experience has shown that austenitic steel components are vulnerable to stresscorrosion cracking, which has a complex morphology and can be difficult to detect and characterise by ultrasonic methods. Where these cracks occur within or close to a weld, the risk of defects being obscured by signals scattered from grain structure is increased.
The PISC III programme on inspection capability (ref [1]) included various round-robin trials including inspections of testpieces representing safe-end welds joining austenitic steel piping to ferritic steel nozzles (typical of PWR/BWR primary coolant loops). Results (ref [2]) showed that the detection rates achieved by industrial inspection procedures were in the range 40 90%, false-call rates were up to 40% and that errors at 70% confidence in measuring the throughwall dimensions of defects were 5mm or more. Signal-processing techniques can potentially reduce grain-structure signals relative to defect signals and hence improve inspection performance. The recent evolution of computers and the development of more sophisticated algorithms now allow for the practical implementation of signal processing techniques. Recognition of the difficulties of ultrasonic inspection led to the development of inspection qualification. In Europe, ENIQ has developed a methodology for inspection qualification and has conducted a pilot study on austenitic piping. This study (ref [3]) highlighted the difficulty of qualifying the capability of an inspection to detect and characterise service-induced defects. Qualification for inspection of austenitic piping often relies to large extent on experimental demonstration on test specimens containing simulated defects. The ultrasonic response from such defects may be different from that of real stress-corrosion defects, and there is a risk that a successful qualification may not guarantee reliable performance in practice. Availability of samples containing service-induced stress-corrosion cracking provides an opportunity to gather data on real defects. The availability of new techniques for low-cost simulation of stress-corrosion defects provides the potential for more realistic test specimens. A quantitative comparison with real defects is required. The establishment of the ENIQ methodology (ref [4]) for inspection qualification provides a framework for a systematic approach in tackling the above problems. B Work Programme The project comprises a total of twelve task which interact with each other as illustrate in figure 1. The tasks (task leader in brackets) are defined as: Task 1 Compilation of specimen data and specifications (Serco Assurance); Review of the test specimens, containing real service-induced cracking, available to the project for comparison with synthetic defects and trials of signal processing techniques. An outline programme for the inspections of the test specimens to be developed. A second aspect of this task is the preparation of an end-user specification for the signal-processing activities; including performance targets. Task 2 Inspection procedures (Tecnatom); Design and production of procedures for the inspections of the various testspecimens defined in the first task. The aim to ensure that data collection conditions are fully defined. Task 3 Testpiece manufacture (JRC Petten); Review of available methods to simulate service-induced cracking in austenitic steel. Development and issue of a tender for the design and manufacture of a set of specimens containing synthetic defects. Some of the synthetic defects to simulate the service induced cracking of the specimens identified during task 1.
Task 4 Acquisition of ultrasonic data (Mitsui Babcock Technology); Scanning test specimens acquired during tasks 1 and 3 by the following techniques representative of current inspection practice: Conventional pulse-echo inspection Immersion focused-probe inspection TOFD technique Phased-array techniques Task 5 Development of signal-processing methods (Uppsala University /University of Kassel); Development of robust signal-processing algorithms for enhancement of signal-tonoise levels and defect resolution (Uppsala University). Development of a Synthetic Aperture Focusing Technique (SAFT) algorithm for imaging data for a given inhomogeneous, anisotropic weld grain structure (University of Kassel). An emphasis is to be placed on testing on real data collected from task 4 of the algorithms selected. Task 6 Interface signal processing methods with ultrasonic systems (CEA); Production of a PC version of the CIVA package for interpretation, display and processing of ultrasonic data from all project partners. This will include integration of developments from task 5. Task 7 Library (JRC Petten); Establishment of a library of test data to include specimen geometry, defect details - including destructive examination reports and ultrasonic data from task 8 and 4. Work Package 5 Development of signal processing methods Work Package 6 Interface signal processing methods with ultrasonic systems Work Package 11 Final trials Work Package 1 Compilation of Specimen Data and Specifications Work Package 2 Inspection Procedures Work Package 3 TestpieceManufacture Work Package 4 Acquisition of ultrasonic data Work Package 12 Dissemination of results Work Package 7 Library Work Package 8 Destructive Examination Work Package 9 Comparison Work Package 10 Virtual defects Figure 1Interaction between the SPIQNAR tasks Task 8 Destructive examination (NRI Rež); Destructive examination and metallography of selected specimens from tasks 1 and 3 to establish defect morphology and weld grain structure. Both real and simulated defects will be sectioned. Task 9 Comparison (CEA); Comparison of the data from the real and the simulated defects (tasks 1 and 3) to establish a transfer function. Allowing a qualification body to estimate the significance of results from simulated defects in terms of the performance from real service-induced defects.
Task 10 Virtual defects (Serco Assurance); To use the ultrasonic data and analyses generated in the tasks 4 and 8 to establish the feasibility of using simulation methods in qualification. Task 11 Final trials (JRC Petten); Evaluation of the industrial applicability of the signal-processing methods and the interface software (CIVA). A comparison of assessment of the inspection data from task 4 is to be carried out with and without signal processing. Task 12 Dissemination of results (Serco Assurance). Dissemination and planning of exploitation of the project results in order to improve inspection capability for austenitic welds through signal processing and to ensure more cost-effective qualification with real and simulated reflectors. ENIQ will be a key vehicle for dissemination. C Achievements C.1 Test specimens JRC Petten hold a set of stainless steel 12 and 28 diameter piping assemblies originating from US BWR plants. These were obtained originally for PISC III. Comparison of the reports from the various inspections carried out for PISC III identified five specimens for which there is good reason to believe that Inter Granular Stress Corrosion Cracking (IGSCC) defects are present (Task 1). In addition SQC have made available to the project a small diameter pipe sample containing IGSCC, while British Energy have made available two specimens containing reheat cracks. All of the specimens were extracted from nuclear plant. The JRC assemblies were shown to contain levels of ionising radioactive isotopes (Co 60). The transportation of these specimens between the partner states has highlighted differences in requirements for shipping of contaminated goods and in addition there have been associated difficulties, exacerbated by the political climate of the last two years. The JRC specimens have been inspected by pulse-echo, TOFD and phased array techniques. The presence of significant defects has been confirmed in two of the assemblies, and one has been identified for destructive investigations. (Task 4). JRC Petten have investigated the available options for the manufacture of realistic defects as part of task 3. A call for tender was issued during June 2002 to manufacture defects within spare (12 and 28 ) assemblies similar to those identified in task 1 and inspected as part of task 4. However following difficulties with the contamination levels of the assemblies the tender was withdrawn. Following this a final decision on the split of defect types and the construction of defective specimens was held off until further inspection data was available. A new call for tender was submitted to the companies which responded to the original call. A contract for the manufacture of specimens was placed in June 2003. C.2 Inspections Inspection procedures have been developed for 3 inspection methods (Pulse-echo, TOFD and Phased array) and inspections have been performed on all 5 of the JRC assemblies identified under task 1. From the inspection results two of the assemblies were
assessed has having significant levels of throughwall cracking. Thus one of the 28 assemblies was selected and shipped to NRI Rež in the Czech Republic for destructive examination. Depending upon the outcome of the destructive investigations an alternative 12 assembly may also be destructively examined. C.3 Signal Processing A functional specification for signal processing has been prepared (Task 1). This details the range of applications and defines targets for improvement in signal-to-noise levels, reduction in false call rate, and improved defect detection, sizing and location accuracy. The specification calls for robust processing methods which are not unduly susceptible to the examination conditions. In particular, it is vital to provide processes which ultrasonic inspection technicians can set up and apply without requiring specialist knowledge. This specification was developed in consultation with all the partners. Following issue of the functional specification Uppsala University has established two MatLab toolboxes based on the development algorithms of the project (Task 5). These are, a noise reduction toolbox, and a deconvolution toolbox. At the same time the University of Kassel developed the InaSAFT algorithm for anisotropic inhomogeneous media (Task 5). The InaSAFT programme is now in a form suitable for integration with the CIVA software platform to be developed under task 6. Trials of the InaSAFT software using data from the inspection of an homogeneous anisotropic testblock are currently underway. An additional piece of work, carried out for the project by Uppsala University, is a report on the compression of ultrasonic inspection data and the development of a data compression toolbox. This additional work concluded that the JPEG 2000 Image compression standard is suitable for the compression of ultrasonic B-scans. C.4 CIVA data display and processing software A PC version of CIVA has been established which can be used by the partners (Task 6). The package is capable of displaying ultrasonic data in A, B, C and D scan forms, both before and after application of signal processing. The package provides an operator interface to the signal processing algorithms developed under task 5, and is capable of reading all the ultrasonic data formats of the partners of SPIQNAR. C.5 Defect information JRC Petten have started to create the layout and access details for the defect library, to be in the format of web pages that can be browsed using the internet and a conventional web browser. Ultrasonic data and inspection reports from task 4 will be included within this library along with information from the destructive examinations of task 8. C.6 Dissemination and Exploitation As coordinator Mitsui Babcock have presented two papers on the SPIQNAR project at ECNDT, Barcelona 2002 and BINDT Southport 2002 (ref [5, 6]). Uppsala University have submitted several papers to IEEE which refer to the work performed as part of the SPIQNAR project (ref [7 9]). Dr R Hanneman of the University of Kassel has published her PhD thesis, which is heavily based on work performed for the SPIQNAR project, on the internet at http://www.dissertation.de/englisch/buch.php3?buch=1260.
A SPIQNAR website has been developed and is being reviewed for approval, the location is http://safelife.jrc.nl/eniq/projects/spiqnar/index.php The status of the project is being reported and discussed at the 6-monthly meetings of the ENIQ steering committee. D Conclusion & Benefits Significant progress has been made in creating an integrated ultrasonic data analysis platform capable of assessing a variety of UT inspection data formats and also of applying signal processing to enhance data interpretation. A substantial amount of data will be available within the library of defects (real, realistic and artificial), along with destructive examination details. If successful, the project will both enhance the reliability of inspection, particularly of austenitic steel and high-nickel alloy piping welds, and also make inspection qualification more secure by providing means of relating results on affordable test specimens to performance on real, service-induced defects. In particular it is envisaged that the information to be obtained on the relative ultrasonic responses of real and simulated stress corrosion cracking will be suitable for incorporation into ENIQ recommended practices providing guidance on the conduct of inspection qualification. References [1] Crutzen, S, Nicholls, R & Miller, A; PISC III: A status report ; p. 219 proc. 12th Int. Conf. On NDE in the Nuclear & Pressure Vessel Industries; Philadelphia, October 1993. [2] Crutzen, S; Dombret, P; Main conclusions of the PISC III action on safe-end welds ; p 203 proc. 12th Int. Conf. On NDE in the Nuclear & Pressure Vessel Industries; Philadelphia, October 1993. [3] Lemaitre, P; Eriksen, B, Whittle, J; Lessons learned from the ENIQ pilot study ; p. 1998, vol 2, proc. 7th European Conference on NDT; Copenhagen, May 1998. [4] Lemaitre, P, Crutzen, S & Champigny, F; ENIQ: A status report ; p 2082, vol 2, proc. 7th European Conference on NDT; Copenhagen, May 1998. [5] Cameron, N, Dikstra, B, "Signal Processing in Qualified Inspections of Stainless Steel welds: The SPIQNAR project", 8th ECNDT, Barcelona,June 2002. [6] Cameron, N, Dikstra, B, "Signal Processing in Qualified Inspections of Stainless Steel welds: The SPIQNAR project", p 193 Annual conference of British Institute of Non-destructive Testing 2002, Southport, September 2002. [7] T. Olofsson, "Deconvolution and model based restoration of clipped ultrasonic signals", submitted to IEEE Transactions on Instrumentation and Measurement, 2003 [8] Guang-Ming Zhang, T. Olofsson, T. Stepinski, A Comparative Study of Transform and Subband Coding for Ultrasonic Image Compression submitted to Ultrasonic, 2003 [9] Guang-Ming Zhang, T. Olofsson, T. Stepinski, Simultaneous Denoising and Compression Using Wavelet Transform and Vector Quantization, submitted to IEEE Transactions on Image Processing, 2003