Application of genetic algorithms for optimal chiller selection

Abstract: Genetic Algorithms are employed to optimize the selection of chiller size in a multi-chiller central plant for arbitrary cooling load profiles. The objective of the optimization is to minimize life cycle (capital + operating) costs while keeping all chillers running in good part load conditions to avoid any surge of chillers. The approach has the capability of solving the combinatorial optimization problem with both discrete variables and continuous variables. All the variables are encoded to form the chromosomes. The objective function is then utilized to evaluate the fitness of each chromosome to give the evolution direction. According to the direction, genetic operations, such as crossover and mutation, are used to find the optimal solution. Penalty functions are added into the fitness function to constrain the optimal solution in the feasible region. Comparisons between conventional and the new methods for different cooling load profiles are provided in application examples to illustrate the advantages of the method