Metodologia de otimização de redes malhadas através da programação não linear.

Abstract

The design complexity of looped networks has forced to use traditional trial and error methods to attain a solution for the problem. Those methods, where the Hardy-Cross method is the most known among them, only carry out energy and mass balance of the network without dealing with the system's cost, that is, nor estimating neither improving the system's cost. In this work, a method for designing economical looped networks based on nonlinear programming is presented. To reduce the number of variables and, probabily, improve the performance of the solution procedure, this method is composed of two stages to reach an optimal solution. In a first stage, a nonlinear programming technique is applied to determine the flows and diameters of the pipes connecting two nodes of the network. In a second stage are chosen which are the upper and lower values of the results attained at the first stage for each pipe segment, and a nonlinear programming technique is applied once more to determine the length of each diameter for each pipe segment along with its flow. In both stages the objective function was related to the cost of the pipes and pumping requirements. This method has been applied to two examples of looped networks, which have been used in the literature to illustrate the application of other optimization methods developed by other authors. The optimal solutions attained from the method presented herein have shown to be better than the ones resulting from the application of any other method, which were taken into account for comparison in this work.