Determination of material parameters for a unified viscoplasticity-damage model for a P91 power plant steel

Societal pressures are mounting on electricity operators to operate traditional fossil-fuel power plants in an efficient and flexible manner in conjunction with renewable power plants. This requires the uses of high frequency start up – shut down load profiles in order to better match market demands. As such, high temperature/pressure components such as steam pipe sections and headers experience fluctuating mechanical and thermal loads. There is therefore an industrial need for the accurate prediction of fatigue and creep damage in order to estimate remnant component life. In the present work, a continuum damage model has been coupled with a Chaboche unified viscoplastic constitutive model in order to predict stress–strain behaviour of a P91 martensitic steel (a material used for power plant steam pipes) due to cyclic plasticity and damage accumulation. The experimental data used here are from the previous work [1]. Cyclic fully reversed strain controlled experiments (±0.4%, ±0.25% and ±0.2% strain ranges) and cyclic test with a dwell period (±0.5% strain ranges) for a P91 martensitic steel under isothermal conditions (600 °C) are utilised. The physically relevant material parameters are determined and optimised using experimental results. Although many material parameter identification procedures can be found in the literature [1], [2], [3], [4], [5] and [6], there are uncertainties in determining the limits for the parameters used in the optimisation procedure. This could result in unrealistic parameters while optimising using experimental data. The issue is addressed here by using additional dwell test to identify the limits for stress relaxation parameters before using Cottrell's stress partition method to identify the limits for strain hardening parameters. Accumulated stored energy for damage initiation criterion and damage evolution parameters are also extracted from the experimental results. The estimated failure lifetimes for ±0.4%, ±0.25% and ±0.2% cases are 1600, 4250 and 9500 cycles, respectively, as opposed to 1424, 3522 and 10512 cycles as given by experiments.