Consideration of Thomson effect in multi-objective optimization of thermoelectric generator using metaheuristic techniques

Abstract

This paper proposes the performance estimation of a multi-parameter multi-objective optimization problem of a thermoelectric generator (TEG) considering the Thomson Effect using metaheuristic techniques. The hot and cold side temperatures, load resistance, length of thermoelectric legs, and fill factor of the TEG are varied simultaneously to optimize the output power and efficiency of the TEG. Thirteen-metaheuristic algorithms have been used to perform single and multi-objective optimization of TEG system with multi-variable space and obtained the best performing algorithm among them for TEG model with and without considering the Thomson effect. In terms of accuracy and speed, moth flame optimization (MFO) and cuckoo search (CS) algorithms are found to be the most suitable for the models with and without considering the Thomson effect. Cuckoo search, genetic algorithm (GA), and moth flame optimization (MFO) algorithms gave consistent results for the model considering the Thomson effect. The TEG performance of both models with and without considering the Thomson effect is also compared. For single-objective optimization, CS and MFO produced optimal power output and efficiency of 16.99325 W & 2.0696 % and 19.20837 W & 2.4427 % respectively for models with and without considering the Thomson effect respectively. For multi-objective optimization, the optimal power outputs of 16.993 W & 19.20837 W and efficiencies of 1.9583 % & 2.25932 % are obtained on the Pareto front for models with and without considering the Thomson effect respectively. These optimal points are determined using the Shannon entropy decision-making approach which is found to be the best technique.