Aplicação de algoritmos genéticos para otimização do projeto de trocadores de calor do tipo Air Cooler

Abstract

Air cooler heat exchangers use ambient air as the cooling fluid in processes involving the cooling or condensation of hot streams. As the air is a natural resource, there are no environmental limitations on its application, air coolers have become increasingly an alternative to using heat exchangers that employ as cooling utility. Despite of its relevance, a number of works about the optimization of projects of this equipment in the literature is still reduced. In this way, the present dissertation is based on the use of genetic algorithms in order to optimize the design of these equipment with the aim of reducing their total annualized cost. This is composed of the capital, maintenance and operating costs. First, the search for the total minimum cost is investigated in the following design variables: tube length and external diameter, tube pitch ratio, number of passes per row, number of rows, number of tubes per row, number of bays, number of bundles per bay, number of fans per bay, fan diameter and surface characteristics (number of fins per length, fin height and thickness). Subsequently, we investigated an inclusion of the outlet temperature, as a design variable, which result in an even lower objective function value. The constraints that ensure the technical availability of the equipment are represented by a minimum excess limit of area and by the check if the size of the fan accords the clearances and the minimum coverage percentage of fans. The results obtained using the Genetic Algorithm were better than the conventional methods and a comparison with the literature showed equivalent results for an approach based on a trial-and-error procedure with human intervention, but it is worse than the global optimum found using mathematical programming. However, the expansion of the problem including the outlet air temperature allowed a further reduction of the value of the objective function, even obtaining better results than the original global optimum of the original problem.