Using Metamodeling and Fluid-Structure Interaction Analysis in Multi-Objective Optimization of a Butterfly Valve

Abstract

Along with the increase in computing power, new possibilities for the use of parametric coupled analysis of fluid flow machines and metamodeling for many branches of industry and medicine have appeared. In this paper, the use of a new methodology for multiobjective optimization of a butterfly valve with the application of the fluid-structure interaction metamodel is presented. The optimization objective functions were to increase the value of the KV valve’s flow coefficient while reducing the disk mass. Moreover, the equivalent von Mises stress was accepted as an additional constraint. The centred composite designs were used to plan the measuring point. Full second-order polynomials, non-parametric regression, Kriging metamodeling techniques were implemented. The optimization process was carried out using the multi-objectives genetic algorithm. For each metamodel, one of the optimization candidates was selected to verify its results. The best effect was obtained using the Kriging method. Optimization allowed to improve the KV value by 37.6%. The metamodeling process allows for the coupled analysis of the fluid flow machines in a shorter time, although its main application is geometry optimization.