Design and optimization of a shell-and-tube heat exchanger using the univariate marginal distribution algorithm

Abstract

Abstract Estimation of Distribution Algorithms (EDAs) are interesting evolutionary methods; they have the characteristic of using an explicit distribution model, instead of mutation and crossover operators. In particular, Univariate Marginal Distribution Algorithms (UMDA) are some of the simplest distribution-based methods. In this work UMDA is used to design and optimize a shell-and-tube heat exchanger. To model the heat exchanger, Kern method and the Bell-Delaware method are employed. It has been found that UMDA has an inferior performance when compared with other metaheuristic algorithms. Nevertheless, other EDAs may lead to better results, which is part of future research.