An efficient computational approach for heat source optimization in numerical simulations of arc welding processes

摘要

The arc welding process involve complex physical and chemical phenomena, which makes mathematical modeling difficult. Its modeling uses the heat source concept to model the energy input, that requires geometric parameters to be calibrated. However, to obtain an adequate accuracy, some heat sources have many parameters to be defined, which demands great computational time, even using optimization methodologies. In this paper, an optimization methodology, including a new technique called Reduced Geometry, is proposed to find the optimal parameters of heat sources, by using an inverse problem approach and Genetic Algorithms. The methodology is validated to the GTAW process of butt and lap joints of AISI 1020 carbon steel and AISI 304 stainless steel, which allows to test symmetric and asymmetric welds. The thermal numerical simulations are performed using the Finite Element Method, where three types of moving heat sources are considered: a Gaussian, a Conical and a Double Ellipsoid heat sources. The fusion zone shapes and thermal cycles, obtained from experiments are compared with numerical ones. Analyses of the computational time, as well as accuracy in relation to experiments are carried out. In general, thermal results obtained by numerical simulations are in good agreement with the experiments, for all cases, and the proposed methodology strongly decreases the computational time, when compared with a standard approach, with negligible loss in accuracy. (C) 2020 Elsevier Ltd. All rights reserved.