The welding process produces strains and residual stresses that must be taken into account to evaluate the final quality of the assembly. For that, the simulation of the process by a thermomechanical computation is very widely used. Numerical simulation uses many parameters (materials, heat source, boundary conditions) whose effects must be studied. But as the compuation becoming very expensive, massive parametric studies quickly become unusable. To reduce the computation time, we propose an approach based on the spatial and temporal similarity of thermal results, by a method of hyper-reduction on a slice of the domain.
|ScienceOpen disciplines:||Applied mathematics, Applications, Statistics, Data analysis, Mathematics, Mathematical modeling & Computation|
|Keywords:||welding, POD, DEIM, finite elements|