Accelerated creep testing for high temperature design

Components in-service at temperature are most likely designed with a creep strain limit achieved in a time frame as end of life criteria. In applying a negligible limit of 0.2% creep strain at 200kh recent work within European Standards has found the traditional ’rule-ofthumb’ that ferritic martensitic (F/M) materials do not creep below 375°C, and austenitic materials below 425°C may not be the case. Creep testing is normally carried out at temperatures well above the limits and there is simply no experimental evidence of creep behaviour at temperatures around the limits with little to justify the expensive and extensive long-term creep testing required. This work focuses on the models and techniques available to establish

those limits for individual materials.

The recognised creep models available for extrapolation have been found to be limited in their usefulness simply not extrapolating to low enough temperatures or simple not representing the physical reality adequately being unstable at lower stresses. The models that can extrapolate to lower temperatures have been applied to German VDEh and NIMS data sets with results finding the limit is too broad with potential differences of 100°C between C-Mn and Grade 91 steels. The Wilshire model and equations have been reviewed and applied to the NIMS data for Grade 91. The Wilshire equations in their original form appear to provide conservative estimations of the no-creep temperature. An Evans-Wilshire modification is reviewed and found to more appropriate for the Grade 91 material with a no-creep temperature around 410°C at a design reference stress of 2/3 tensile strength at temperature. The use of an artificial neural network confirms the Evans-Wilshire model results for the NIMS data. This work also addresses the gap in experimental data around the limit and provides a methodology for defining the temperature limits. Temperature limits for P265GH, Grade 91 and 316L(N) are presented from both experimental and historical data sets. Finally a novel stress relaxation method has been developed and described to estimate long-term creep properties with limited specialist equipment and skills.