COMPARISON BETWEEN DIFFERENTIAL EVOLUTION AND SIMULATED ANNEALING ALGORITHMS APPLIED TO THE CONSTRUCTAL DESIGN OF DOUBLE-T SHAPED CAVITIES
ResumoIn this paper it is analyzed the application of two meta-heuristic methods, DifferentialEvolution and Simulated Annealing, for the geometric optimization of a complex cavity intrudedinto a heat conducting solid wall. The geometric evaluation is performed with the ConstructalDesign method, which is used to define the performance parameter, constraints and degreesof freedom of the problem to be optimized. The main purpose of this work is the comparisonbetween the results of these two meta-heuristics, mainly for reproduction of the effect of degreesof freedom over the thermal performance of the studied problem. The experiment consistson the simulation of thirty runs for each algorithm, with different values for the configurationparameters, and also four versions of Differential Evolution and five versions of Simulated Annealing.The optimization results show that the meta-heuristic algorithms and their parameterconfigurations are important for proper prediction of the effect of degrees of freedom over thethermal performance and definition of system design. Results also indicated that one of theDifferential E volution algorithms led to the best and most robust performance. Therefore, thesignificant contribution here is the recommendation of the more reliable meta-heuristic, and itscorrect parameters for the studied problem.
Fenômenos de Transporte