Multi-objective genetic algorithm optimization of thermoelectric heat exchanger for waste heat recovery

A model is developed to simulate a cross-flow heat exchanger, including fins, in the wall of which thermoelectric generators are sandwiched. Such a system could be used for waste heat recovery. The model is used to optimize the device based on several objective functions: total volume, total number of thermoelectric modules, power output, and pumping power. The design variables are the local distribution of modules and of current, the shape of the fins, and the division of the heat exchanger in sub-channels. Pareto fronts are achieved with a multi-objective genetic algorithm, and are presented here. The results show that the number of sub-channels in the heat exchanger has a larger impact on the overall performance than the fin geometry for this particular problem. Also, the net power output is mostly correlated to the number of thermoelectric modules, and less to the heat exchanger volume. Various relations between the different competing objectives are shown and analyzed. Copyright (c) 2011 John Wiley & Sons, Ltd.