Overview

Laser shock peening (LSP) is used to improve the strength and fatigue resistance of components by hardening their surface. In our lab we have developed a framework for LSP simulation, which is based on the finite volume method. The critical component of the framework is the dynamic simulation of the elastoplastic shockwave that subjects the component material to a strain rate of the order 1E+7 1/s. Under such high strain rates, the material properties need to be simulated as rate-dependent, which requires more complex material models. Unfortunately, these models are usually calibrated using strain rates only up to 1E+4 1/s, which means that for the LSP conditions, extrapolation is required.

The goal of the proposed stay is to propose and perform an analysis of the identifiability of the parameters of the used material models and of the LSP simulation itself. Our starting point for the analysis will be the Morris screening method. However, it may happen that this method, used directly on the LSP simulation, will prove too computationally demanding. In such a case, it will be necessary to construct and calibrate a suitable surrogate model and to analyze the combined uncertainties of the approaches.

Prior knowledge of numerical mathematics (numerical solution of partial differential equations), and C++ and Python programming is welcomed.