Multivariate Stochastic Optimization Approach Applied in a Flux-Cored Arc Welding Process

Abstract

One of the main goals in flux-cored arc welding processes is the optimization of bead geometry, in which multiple geometric characteristics of the welding bead are important; therefore, multiobjective optimization programming is often applied. However, several optimization problems that use stochastic programming do not consider the impact of the correlation between the output variables on their probabilistic constraints. In this context, this paper aims to present a multiobjective optimization method based on multivariate stochastic programming. To demonstrate the applicability of the proposal, we conducted a design of experiments to optimize a flux-cored arc welding process for stainless-steel claddings. The weighting-sums method was applied to formulate the multiobjective optimization problem. It was possible to formulate a multivariate probability distribution for the penetration and dilution. In addition, a 95% probability to meet the predefined specification limits of the geometric characteristics was achieved.

Keywords

• Bead geometry
• bivariate normal distribution
• flux-cored arc welding
• multiobjective optimization programming
• stainless steel claddings
• stochastic programming