Long term creep deformation and crack growth predictions for grade 91 steels and risk-based methods in their component life assessment

Abstract

Grade 91 steel has been extensively used for high temperature since it was introduced in the 1980s. However, as the 9%Cr material is relatively new, there is limited experimental data and understanding with respect to quantifying the effect of variables, for example service exposure, on life prediction and long-term behaviour, in particularly weldment. Many incidents reported about premature weld failures of Grade 91 steel suggest that the design standards and guidelines may be non-conservative for the pressure vessels and piping made of Grade 91 steel. Furthermore, Grade 91 welds are also known to be prone to Type IV cracking, leading to premature failure controlled by creep crack growth (CCG). A comprehensive literature review was carried out and extensive data is collected. This includes data for a different range of steel materials at different creep temperatures and test stress which in turn provided a range of test durations from a few hundred hours to several thousand hours (ie >60,000 hours) but with the main focus of the analyses remaining on Grade 91. The NSW based creep crack growth model is found to be capable of predicting upper/lower bounds of creep crack growth well for Grade 91 when data is obtained from relatively short to medium-term laboratory experiments (< 10,000 hours). It is also found that the extrapolation of the model fails to predict the long-term creep crack propagation rates unless the actual long-term uniaxial creep data is available. Two types of creep testing are performed in this study to investigate the creep behaviour of Grade 91 steel over a range of stresses on the basis of continuum damage mechanics and fracture mechanics: Uniaxial creep and CCG. For each type of testing, creep deformation behaviour and CCG characteristics in different microstructures (parent metal, weld metal and HAZ) are investigated. The same tests are performed on as-received and service exposed Grade 91 steel to provide the relevant data to be used for the NSW prediction model. In both uniaxial and CCG experiments, the weldment (ie weld metal and HAZ) exhibits a lower creep resistance compared to the parent material. In the CCG experiments, the ex-service material exhibits a faster growth rate compared to the as-received material in the similar range of C*. Based on the test results, the relevant parameters to describe creep strength and CCG are established. The results of the analyses of short-term and long-term creep data from the literature indicate a change in the creep exponent and failure strain at longer creep lives. The creep failure strain also tends to diminish with a decrease in the applied stress.To investigate the microstructure of the steel before and after the creep test and to also understand the failure mode and the location in the uniaxial and CCG tests, metallography was carried out. One sample from both service exposed and as-received steel before any creep test as well as one uniaxial and one CCG specimen from each material (four in total) after creep tests were selected for examination. Both service exposed and as-received steels showed tempered martensite microstructure with cavities observed in the service exposed material before the test. The analyses of creep uniaxial samples after the test exhibit Type IV failure tendency in the as-received material compared to the service exposed steel where failure occurred mainly at or adjacent to the weld. With regards to the CT specimen, in a similar way the crack path in the as-received HAZ sample followed outer region of HAZ (ie. Type IV) whereas the crack path in the service exposed sample showed tendency to grow along the fusion line. This may be due to faster ageing process of the weld metal compared with HAZ and parent material when exposed to high temperature. The experimental results obtained above together with the database compiled from the literature survey are then used, in conjunction with the NSW model, to predict the CCG of Grade 91 steel. It is found that the appropriate failure strain under multi-axial stress state can be estimated by extrapolating the secondary strain to the failure time rather than the final failure strain. Using the extrapolated uniaxial failure strain data at the lower shelf (low stress) region and a reduced creep index at long terms, conservative CCG predictions have been made by the NSW model. For C* values of 10-5MJ/m2h and less, cracking is predicted to be relatively faster compared to the extrapolated short-term CCG data trend. It is found that the assumption of lower failure strains and reduced creep indices effectively produce a safe prediction for CCG using the NSW model for long-term behaviour. However, the level of conservatism increases depending on the use of uniaxial failure strains or the Monkman-Grant (MMG) failure strain or in the extreme 0.2% failure strain as the lower shelf data. The correct level of conservatism can only be determined once actual long-term CCG data are available. Following the analysis, a prototype model was developed based on risk-based life assessment (RBLA) integrity management concept using data presented in this thesis for Grade 91 piping material. Estimating remaining life based on the material limit state equation, assessing the risk associated with creep damage and making recommendations on the most effective mitigation plan were the main area of assessment treated in the RBLA model. In this study, four reported Grade 91 weld failure investigations have been compared to the assessment result from RBLA. The result was found to be similar to what the failure investigation reported, however it should be noted that when applying the risk concept, the target risk (or target probability) plays a crucial role compared to a deterministic analysis. The work done on failure cases and the comparison with models developed in this thesis demonstrated that to achieve a reliable conservative result for longer-term CCG assessment, the effect of lower creep ductility due to long-term exposure together with MMG approach to determine a lower band failure strain should be taken into consideration. Where crack growth data was available from the failure investigation report, a comparison with the finding from the investigation was made with the long-term NSW model which was developed in this work. The result indicated a good agreement with finding from the failure investigation report and falls in the scatter band predicted from the model. It could be concluded that the optimum approach to predict the short-term and especially long-term creep and CCG behaviour of Grade 91 weldment is a probabilistic method as the level of uncertainty is relatively high in material properties those contributing to failure.