Final project report - ATLAS+ (Advanced Structural Integrity Assessment Tools for Safe Long Term Operation): D6.9-21

The European Horizon 2020 project ATLAS+ (Advanced Structural Integrity Assessment Tools for Safe Long Term Operation) focuses on developing:
• innovative quantitative methodologies to transfer laboratory material properties to assess the structural integrity of large piping components,
• enhanced methods for treatment of weld residual stresses when subjected to long term operation,
• advanced simulation tools based on fracture mechanics methods using physically based mechanistic models,
• improved engineering methods to assess components under long term operation taking into account specific operational demands,
• integrated probabilistic assessment methods to reveal uncertainties and justify safety margins.
The work was divided into four technical work packages.
WP1 focuses on materials characterisation of ferritic and austenitic pipes, and dissimilar metal welds (DMW). The tearing resistance curves (J-R curves) were determined with compact tension (C(T)) and single-edge notched tension (SE(T)) specimens. The tensile properties were characterised with smooth round bar specimens and notched tension specimens. The pipes were tested in four-point bending with elliptical cracks on the outside of the pipe, and through wall cracks. The testing was mostly done in room temperature, but for one case the testing was carried out in 100 °C. The data was applied in WP3 for development of numerical models.
WP2 primarily focuses on modelling and measuring residual stresses. Thick and thin-walled (manufactured with low and high heat-input) narrow-gap gas tungsten arc-welds (GTAW) (AISI 316L), fully circumferential and 120° patch overlay welds, and thick walled thermally aged NG-GTAW were manufactured, and the residual stress profiles were measured with different techniques vital to minimize uncertainty. The FEM 2D and 3D residual stress predictions agreed with the experimental results. The austenitic pipe girth welds are important new benchmarks for RS simulation of engineering weldments. The austenitic pipe girth welds provide valuable new data for improving the RS profiles used in engineering structural integrity assessments. WP2 focused also on improving the assessment of residual stress profiles, assessing residual stresses in operation conditions and investigating the effect of residual stresses on fracture.
The first task in WP3 was to investigate crack formation and growth through fatigue caused by mixing of fluids with large temperature differences at piping T-junctions. The problem was modelled with simple 1D models and 3D models including computational fluid dynamics with a higher computational cost. New tools and recommendations were developed.
The objective of sub-WP 3.2 is to benchmark selected Local Approach (LA) models on WP 1 experimental data dealing with the transferability of ductile fracture toughness properties from a specimen to a component. The partners used already established LA models (e.g. Gurson-Tvergaard-Needleman, Rousselier) or improved energetic approaches (such as Ji-Gfr or the cohesive zone model). The models were calibrated based on small scale specimen data and validated based on the large scale experimental data.
Most of the participating organizations were able to develop a reasonable approach for prediction of ductile fracture in large and mid-scale mock-ups that are representative of real nuclear components. Performed benchmarks revealed a robust implementation of GTN type LA models in different FE codes and their capability to take into account both the constraint and transferability effects. The achieved results have potential to improve the qualification of key mechanical components relevant for safety of nuclear power plants. Based on the results of LA analyses the following recommendations are given: 1) develop appropriate optimization tools for material parameter selection. 2) The results indicate that, if possible, it is beneficial to calibrate the material parameters with both low and high constrained fracture mechanical specimens. 3) In case of the application of LA modelling to welded joints, it is recommended to calibrate material parameters to the weld, base metals and heat affected zones (HAZ). 4) It has been shown that good results can be expected with a mesh size in the range of 0.1 to 0.2 mm. 5) It was concluded that the stress-strain curve needs to be optimized. Especially the transition region from the linear elastic to elastic plastic behavior has a large impact on the prediction results. 6) most of the analyses were conducted with ¼ FE models. Post-test analyses of mock-ups FP1 and FP3 with 1/2 and full FE models, allowing out of plane crack propagation, interestingly did manage to predict the large out of plane shear driven crack propagation. The work was complemented by proposing two methodologies for generating component J-R curves.
Sub-work package WP3.4 is associated with improving the understanding and methodology of Leak-Before-Break (LBB) assessment of piping and associated components. The first report provides an overview of the development and use of advanced tools for computing crack opening areas and leak rates. The conclusions led to various recommendations being made for estimating COA (crack opening areas) and leak rates in LBB assessments. These included the fact that WRS (weld residual stresses) should be considered in the calculation of COA, especially for small cracks where elastic conditions (small scale yielding) prevail, and that best estimate leak rate tools such as LOCI or SQUIRT should ideally be used.
The second report is a best practice guidance document on LBB based on outcome of the ATLAS+ project and previous experience of the authors. The best practice guidance document takes into account and highlights LBB methodologies from various codes and assessment procedures developed and utilised in different countries. This enables the user to be well informed when performing a LBB assessment and effectively provides a template for carrying out LBB assessments. The guidance contains two types of LBB methodologies. The two types of methodologies and the strong emphasis on undertaking sensitivity studies are considered to be an enhancement on some of the current LBB methodologies and practices. The studies undertaken on probabilistic LBB, under this work package and under WP4, will form a good basis for extending the various deterministic LBB methods which are based on conservative assumptions and prescribed margins in some cases, to methods which enable the probability of LBB to be evaluated on a case-by-case basis.
In WP4, the first task consisted of selection of suitable benchmark cases including of analyses of pipes with cracks in welds (ferritic, austenitic and dissimilar metal welds), in addition to definition of methods and models to evaluate subcritical crack growth (fatigue and stress corrosion), and evaluation of instability for: ductile tearing, brittle propagation or plastic collapse (limit load), and leakage area and leakage rate estimation for LBB assessment. After that the uncertainties were defined 1) crack existence and size and the probability of detection at an in-service inspection, 2) loadings covering the global stresses dominated by the pipework as well as the local stresses due to thermal transients and welding residual stresses, and 3) the material properties required for the integrity assessment. Finally, a deterministic assessment of the selected benchmark cases was carried out considering the best-estimate values and compared with probabilistic assessments.
The deterministic round robin analysis indicated large differences between the partners for the leakage probability analyses due to differences in the limit state definition and formulation. In case of deterministic analysis, the advanced crack shape evaluation method during crack growth provides leakage at an operating time that is almost twice as long as with the simple method. The probabilistic round robin analyses had initially large differences, but after refining of the parameters acceptable agreement between the partners was achieved despite the different fracture assessment methodologies applied and differences in the limit states. Probabilistic safety margins were assessed based on the ratio between leakage and failure probability, and they were influenced by inspection intervals (no significant of several early inspections), probability of detection, material, and leakage detection. The results show that no probabilistic margin exists for the DMW benchmark case, while there is a distinct margin for the austenitic pipe benchmark case (larger than a factor of 100), resulting from differences in POD. The results have made it possible to better understand how different assumptions and parameters influence a probabilistic assessment. When evaluating results from probabilistic analyses using a specific tool, it is important to have a good knowledge of how this tool define different limit states, how the deterministic fracture assessment methodology is defined and how the re-characterization of surface breaking defects at “Snap-Through” is defined.