Programação matemática e evolução diferencial para a otimização de redes de dutos

Abstract

The optimization of an pipeline network is a complex problem and addressed in the current literature. The mathematical modeling of this problem proposed in this paper creates a problem of combinatorial optimization. Methods for solving this problem using linear mixed integer programming and heuristic algorithms of differential evolution (Binary Differential Evolution and Discretized Differential Evolution) are proposed using binary variables. The results obtained with the linear programming have optimal values for the benchmarks with small search spaces and sub-optimal for large values. Results using the differential evolution are also presented as an alternative low computational effort. The application of these methods provides alternatives for transporting different products in a defined time horizon and compare heuristic methods with continuous and binary encodings. Such results encourage the use of heuristic algorithms with continuous coding and the point discretization methods as effective for solving problems discrete alternatives.