Algoritmos estocásticos para otimização de layout de plantas de processo

Abstract

The optimization of process plant layout is an academic and industrial interest problem, presenting complexity in formulation and solution. The papers in this area presents the problems using deterministic, stochastic and hybrid algorithms.This work objective to solve this problem using stochastic algorithms and understand the solution steps. This work investigates three algorithms to solve the Process Plant Layout Problem in a single-floor.The first algorithm is a Genetic Algorithm (GA) with a uniform discrete approach. The second algorithm contains a non-uniform discrete(NUD) approach. The third algorithm treats the problem using a continuous approach (CA). It is also proposed, bodily, a versatile approach for the objective function in order to attend any criteria.The results suggest that the algorithms are efficient in solving the problem achieving benchmarks problems. The GA is efficient achieving the optimal solutions. The NUD showed promising results, especially in computational time. The CA is efficient in generate solutions without an optimization method.Investigations in GA shows that the crossover operation works better forbidding the generation of infeasible solutions