21 The Effect of the Annealing Schedule on Simulated Annealing for Function Optimization and Fuel Cell Design

This paper reports an empirical investigation into the effect of different annealing (cooling) schedules on the performance of the simulated annealing algorithm for (i) proton exchange membrane fuel cell stack design optimization, and (ii) parabolic function minimization. The performance metric is the solution quality of the optimal or near-optimal solution obtained at a predetermined number of function evaluations. Five different annealing schemes are considered: linear, exponential, geometric, Cauchy, and Boltzmann. Computational results show that the exponential and geometric schedules outperform the three other schedules on both the problems. One-tailed t-tests are used to study the statistical significance of the results. The results reported in this paper are better than the solutions for PEMFC stack design published by Mohamed and Jenkins in Journal of Power Sources 131 (2004).