Report on six-year structural integrity study28 July 2000
A project to assess the integrity of a cracked structure was started six years ago. The NESC Network has now reported on the results of that assessment.
The assessment of the integrity of a cracked structure is an interdisciplinary activity. Different countries have different practices to inspect and assess the fitness for purpose of cracked structures such as pressure vessels and piping.
The Network for Evaluating Structural Components (NESC-1) project was set up to improve international practice throughout the entire process of structural integrity assessment. It evaluated the interaction between the different disciplines, and the ways by which technical decisions and judgements are made. This approach makes it possible to identify where the uncertainties in the structural integrity process lie, and validate the findings with a large-scale experiment.
Unfortunately, the cost of research using large-scale specimens is prohibitive. Organisations need to collaborate to carry out such research, so that the organisations can compare information.
Collaboration in NESC-1 was through ‘in-kind’ contributions, with specific areas of work such as the test, reference laboratory and administrative management being provided by the key sponsors, the Health and Safety Executive of the UK and the Institute for Advanced Materials of the European Commission.
NESC-1 was the first large scale project to evaluate the reliability of the entire process of structural integrity assessment. It examined all the key components including inspection accuracy and reliability, material properties data requirements, test measurement techniques, and the appropriate levels of complexity for thermal/structural analysis and fracture assessments. There were over 80 contributing partners to NESC-1.
The objectives of the project were to identify sources of weakness and inaccuracy in current analytical techniques, to discover the extent of likely inaccuracies, and to assess the adequacy of existing structural integrity codes.
The project centred on a simulated pressurised thermal shock experiment, which used a unique spinning cylinder system. The experiment simulated selected conditions associated with an ageing, flawed reactor pressure vessel (RPV) subject to severe pressurised thermal shock (PTS) loading. The main aim of the project was to validate the combination of non-destructive inspection and structural mechanics assessment procedures for evaluating the integrity of such an aged structure containing the postulated flaws.
The test piece was an internally-clad 7t steel cylinder, with an outer diameter of 1395mm and a total wall thickness of 175mm. A total of 18 defects, differing widely in fabrication method, size and location were introduced into the test specimen. The defect set was designed to produce a challenge to both the inspection teams and the structural/fracture analysts responsible for designing the PTS transient. It included a range of sub-clad defects and a large through-clad defect dimensioned such that a cleavage fracture event might be expected. The test therefore represented a more complex and typical situation than previous PTS experiments.
The pre-test structural integrity analyses were based solely on defect sizes supplied by the inspection group, which had no prior access to the details of the defects. These inspections were conducted using procedures typical of current practice. Seven European laboratories carried out a comprehensive material-testing programme, resulting in the test component becoming one of the most thoroughly characterised in the nuclear industry. The spinning cylinder test was successfully performed in March 1997, and gave the planned cleavage run-and-arrest event at the large through-clad defect. All the analysts accurately forecast the defect growth, and the best prediction of the time in the transient for cleavage fracture was within ten seconds of the actual event recorded during the twelve-minute transient.
The NESC-1 test may only be a single datum for evaluation of inspection performance and fracture mechanics, but the number and size of the defects, the world-class inspection and assessment teams, the comprehensiveness of the materials properties database and the accuracy of the data give the results special significance.
NESC-1 will also benefit non-nuclear industries, by providing results to a wide spectrum of structural integrity applications. For example, findings concerning the inspection process, the structural/fracture assessments, and the synergistic relationship between these are applicable to any steel structure or component.
OUTCOME OF THE SPINNING CYLINDER TEST
The NESC-1 test demonstrated that, for these specific conditions, defects up to 74mm depth in material related to that of an ageing RPV would not propagate to cause catastrophic failure under a severe PTS-type thermal shock. This outcome was fully in line with the pre-test analysis forecasts.
The large through-clad defect produced several millimetres of ductile tearing, which preceeded the triggering of a local cleavage event at one end of the defect just below the heat-affected zone at 216 seconds into the PTS transient. This demonstrated that stable tearing consistent with small scale testing behaviour can occur in thick component prior to cleavage fracture in the transition regime.
In the case of the large sub-clad defect, no cleavage occurred, although it had grown along the entire initial crack front. Fractographic investigations determined that a major part of this growth was by an unanticipated – and as yet unexplained – intergranular fracture mode. This will be further investigated to understand the conditions that gave rise to this mechanism in NESC-1 sub-clad defects.
The small amount of defect growth found in the three smaller sub-clad defects was also found to be intergranular in nature. The remaining defects, all below 12mm in depth, showed no evidence of extension.
The results demonstrate the effect of the cladding in inhibiting cleavage initiation in the near-surface regions of the NESC-1 cylinder. Comparison of the results with those from a previous spinning cylinder test, which did not contain cladding, demonstrates that it requires relatively higher loading to obtain cleavage fracture with cladding present.
Several conclusions have emerged from the project.
The inspection trials were undertaken without knowledge of what defects were present, and showed that the detection and sizing performance with ultrasonic techniques were good. In general, the detection performances achieved were better than the results obtained in the comparable PISC II trials a decade earlier, demonstrating that good progress has been made.
Fourteen out of seventeen inspection teams detected all the defects. Three teams missed some defects with a through-wall extent of 2.5-13.5mm.
The good NDE performance is considered to be due to three factors:
• The selection of optimum techniques that are well adapted for the purpose.
• The ability of the inspectors.
• The ability to reduce the occurrence of human error.
Most teams were able to profile the deeper contours of the two large fatigue cracks; however, one team using focused probes was capable of accurately contouring the profile of the defect extension developed during the test.
In the post-test inspections, the change in stress and material conditions produced by the thermal shock did not significantly affect the inspection performance with regard to both detection and sizing.
For ultrasonic inspection, the tip-diffracted wave from a defect should be employed for accurate sizing of through-wall extent. Four teams using this approach in the NESC-1 study performed outstandingly well, with a RMS error of 2.4mm or less. However, since all inspection teams using this technique did not perform equally well, it is essential that the inspection system and personnel should be suitably qualified.
In applying inspection procedures, precautions should be taken to reduce the incidence of human error as far as possible. This can be achieved by well-written, unambiguous procedures for both data acquisition and data analysis, good quality control and training of inspection personnel.
There was a comprehensive materials properties test programme which looked especially at the dynamic and static fracture toughness tests and crack arrest tests. These demonstrated that the toughness of NESC-1 cylinder base material corresponds satisfactorily to an aged reactor pressure vessel condition.
Fracture tests on specimens machined from the sub-clad heat-affected zone (HAZ) showed that the toughness was increased. This effect proved crucial for refining the pre-test analysis predictions in terms of location and timing of the cleavage event.
There was limited available data for the sub-clad HAZ region. The master curve approach allowed this data to be assessed. The increased toughness of the HAZ cannot be adequately represented in the RTNDT approach since the small material volume precludes generation of the test data required.
Ideally, fracture toughness data should be used to determine the reference temperature for the ductile-to-brittle transition behaviour. It may be the only means to adequately describe behaviour in critical locations such as the thin near-surface region. Here the ASME RTNDT parameter cannot be applied because the appropriate test data are unobtainable.
Structural analysis and fracture assessment
Simplified engineering methods produced relatively conservative results. Refined analyses with detailed 3-D finite element modelling of the critical defect allowed accurate prediction of the actual time and location of the cleavage event in the transient, as well as a precise estimate of the extent of prior ductile tearing. The anticipated cleavage event at the largest sub-clad defects did not occur, but the significance of this is mitigated by intervention of an integranular growth mechanism.
The crack driving force estimates varied by +/-10% in detailed 3-D finite element analyses. There was considerable spread in maximum predicted CDF values in simplified analyses. Some estimates were in line with those generated by the 3-D finite element analyses, while more basic approaches produced conservative predictions, by a factor of up to 50%.
The cleavage event that occurred in the large surface-breaking defect, 16.5mm below the inner surface of the cylinder, was best interpreted using standard, high constraint fracture toughness data. Detailed finite element analysis showed that the constraint parameters are close to small-scale yielding conditions at this location, so constraint loss was unlikely to have influenced the cleavage behaviour.
Detailed finite element analysis was able to rationalise the absence of cleavage in the HAZ region, by considering loss of constraint, higher toughness of the HAZ, the reduced effective crack-front width and cladding effects.
Strict application of nuclear safety codes for in-service assessment of the defects in the NESC-1 spinning cylinder would typically allow defects with a maximum depth of 1-9mm. The test outcomes included an apparently large margin, attributed to substantial safety factors and a lower-bound representation of material fracture toughness.
Simplified integrity assessment procedures must be adequately calibrated, and comparing their predictions with detailed 3-D finite element simulations is useful for assessing safety margins.
Personnel undertaking structural analyses and fracture assessments should be adequately trained, and software tools should be developed to ensure that expertise is systematically maintained and developed.
Detailed 3-D finite element analyses capable of modelling local variations in materials properties, clad residual stresses and defect-tip constraint, can obtain reliable best estimates of defect behaviour in the complex near-surface region.
However, further work to establish quantitative procedures relating fracture test data on specific specimen geometries to local constraint parameters estimated from 3-D finite element simulations is needed if such results are to be widely applied to real structures. Analytical methods capable of simulating the micro-mechanisms in ductile tearing and cleavage are the most promising route for obtaining such improvements. An approach based on Weibull probabilistic calculations of the NESC-1 test has produced reliable results for cleavage fracture.
For specific conditions of the spinning cylinder test, there is an order of magnitude difference in critical defect size estimated by existing safety code procedures and state-of-the-art analysis.
THE INTEGRATED APPROACH
By combining defect sizing information from blind inspections trials, the comprehensive materials data set generated for the NESC-1 component and a range of structural analysis tools, it was possible to accurately predict the occurrence of the cleavage event in a large surface-breaking defect.
The major source of uncertainty in the assessment process was the variability of the material toughness data, and the intrinsically statistical nature of cleavage in the transition range. Statistical representation of actual fracture toughness data, especially the master curve approach, proved to be the best in interpretating test results. Variations in the estimate of driving force had a lesser impact on the analysis result. The uncertainty in reported defect sizes was relatively uncritical, in particular for the larger defects. Blind inspection trials showed that the defect sizing capability of the inspection teams was sufficiently accurate for structural integrity analyses.
The results of the NESC-1 test gave strong support to using statistical approaches to describe the risk for local cleavage events in RPV integrity assessment. These approaches can be further developed as part of detailed simulations of defects.
Further work in materials characterisation will benefit the integrated assessment process. There is also scope to improve fracture analysis procedures. As far as improving inspection goes, focus should be directed to:
• Accuracy for small defects.
• Measurement reliability.
The success of NESC-1 emphasises the benefits of the network approach to tackling multi-disciplinary problems. The network structure should be exploited for disseminating the project results and for further developing the integrated structural integrity approach via successor large-scale projects.
|Results of destructive examination of the spinning cylinder|
|Destructive examination of the NESC-1 cylinder determined the precise position and dimensions of each defect in the cylinder. The examination was carried out by the Reference Laboratory and the Destructive Evaluation Advisory Group, consisting of 12 national experts. Each defect was located using both X-ray and ulrasonic analysis, and then cut out. The 6 tonne cylinder was cut into over 200 sections. Intensive fractographic investigation were carried out to establish the extent and mode of growth of the defects that had occurred during the PTS. Four defects produced evidence of growth during the test: The largest through-clad defect of initial depth 74mm. A sub-clad defect of initial depth 77mm. A sub-clad defect EDM notch of initial depth 24mm. A sub-clad defect EDM notch of initial depth 14mm. The first of these produced limited ductile initiation and tearing, followed by a significant local cleavage event at one end of the defect in the lateral position just below the HAZ. At the deepest point, small amounts of crack extension were found.|