Desiccant-wheel optimization via response surface methodology and multi-objective genetic algorithm

A two-step computational framework based on the combination of response surface methodology and multi-objective optimization is proposed to model the outlet-air state of desiccant wheels and subsequently optimize their operation. Regeneration temperature, surface area ratio, rotational speed, and wheel diameter are considered as decision parameters in the genetic algorithm. The central composite design and response surface methods have been employed to design experiments, establish predictive empirical models, and determine interactive effects of decision variables on response variables-process outlet temperature and humidity ratio. Several experiments have been performed to verify applicability of the proposed methodology and validate obtained results. A value of the coefficient of determination exceeding 0.95 demonstrates high reliability and accuracy of the modeling process involved in the proposed methodology. Results obtained demonstrate greater dominance of the surface area ratio compared to other decision variables in terms of their influence on response variables. After successful validation against experimental data, the developed models have been considered as a combination of two objective functions. A fast and elitist non-dominated sorted genetic algorithm II-based optimization technique has been employed to simultaneously determine optimum values of decision variables. A Pareto-optimum front has been presented to select the best value of each decision parameter from available points of optimum operation, and a valuable equation for Pareto-optimal points has been deduced for each material to assist designers develop an optimum design of desiccant cooling systems.