Abstract
This paper introduces a computationally efficient extremum condition-based Reduced Order Modeling (ROM) approach for the probabilistic predictions of creep in finite element (FE). Component-level probabilistic simulations are needed to assess the reliability and safety of high-temperature components. Full-scale probabilistic creep models in FE are computationally expensive, requiring many hundreds of simulations to replicate the uncertainty of component failure. In this study, an extremum condition-based ROM approach is proposed. In the extremum approach, full-scale probabilistic simulations are completed in 1D across a wide range of stresses, the data is processed and extremum conditions extracted, and those conditions alone are applied in 2D/3D FE to predict the mean and range of creep-failure. The probabilistic Sinh model, calibrated for alloy 304 stainless steel, is selected . The uncertainty sources (i.e. test condition, pre-existing damage, and model constants) are evaluated and pdfs sampling are performed via Monte carlo method. The extremum conditions are chosen from numerous 1D model simulations. These conditions include extremum cases of creep ductility, rupture, and area under creep curves. Only the extremum cases are simulated for 2D model saving significant computational time and memory. The goodness-of-fit of the predicted creep response for 1D and 2D model shows satisfactory agreement with the experimental data. The accuracy of the extremum condition-based ROM will reduce significant computational burden of simulating complex engineering systems. Introduction of multi-stage Sinh, stochasticity, and spatial uncertainty will further improve the prediction.