Optimization of thermoelectric generator integrated recuperator

Abstract In this paper, an optimization method combined computational fluid dynamics model (CFD), back propagation neural network (BPNN) model and genetic algorithm (GA) is proposed to optimize the performance of a novel concept of thermoelectric generator integrated recuperator (TEGR). Three geometrical parameters of TEGR including fin thickness, fin pitch and fin height are set as optimization variables. The CFD calculations using thermoelectric-hydraulic coupled model are used to obtain a large number of samples to train the BPNN model. It is found that the ‘3-5-2’ BPNN model is suitable for dealing with the present sample data. And genetic algorithm is used to search for the optimal solution in consideration of the power output and cost of TEGR. The predicted values of the optimization method are in good agreement with the numerical values of CFD.