Convergence refinement of stochastic optimization by coupling a genetic algorithm and a simulated annealing procedure

Numerous trials have to be carried out to refine the convergence of Simulated Annealing (SA) and Genetic Algorithms (GAs). A procedure that couples both GAs and SA algorithms using the ability of GAs to reach quite optimal solutions, and SA procedure for scanning the neighboring solutions to efficiently improve the objective function value. The probability of reaching the true optimal solution, can be significantly increased by this coupling. This approach is illustrated by the problem of synthesis of distillation sequences of mixtures involving 10 to 16 hydrocarbons.