Surrogate modelling for multiscale thermal simulation of powder-bed additive manufacturing

Abstract

In this master thesis, a surrogate modelling strategy for temperature prediction in powder-bed additive manufacturing is presented. The goal is to reduce the computational effort needed to run a multi-scale finite element simulation for selective laser melting developed at Empa. The reduction is achieved by substituting part of the finite element simulations by a surrogate model that is cheap to evaluate. The proposed surrogate model uses principal component analysis to reduce the dimensions of input and output space of temperature fields and polynomial chaos expansion to model the dependence between reduced input and output data. For various different geometries, evolution of the temperatures during the printing process and peak temperatures in the build part are modeled and compared to corresponding finite element simulations for surrogate model validation.

Keywords

Surrogate modelling, multi-scale finite elements, additive manufacturing

BibTeX cite

MSCTHESIS{FKellerThesis,
author = {Keller, Fabian},
title = {Surrogate modelling for multiscale thermal simulation of powder-bed additive manufacturing},
school = {ETH Zurich and Empa, Zurich, Switzerland},
year = {2020}
}