Geometric optimization of two-stage thermoelectric generator using genetic algorithms and thermodynamic analysis

Abstract

Multi-objective genetic algorithms are used to optimize the structure, assignment of configuration and load resistance of a two-stage thermoelectric generator, where Skutterudite and Bi2Te3 are chosen as upper stage and lower stage TE leg materials, respectively. Heat convection and radiation are considered on the top of the upper substrate. In the optimization process, the specific power and entropy generation rate are considered synchronously as objective functions to maximize the power output per unit area and to minimize the irreversibilities. The FEM is adopted in the simulation model, and the Seebeck effect, together with the Peltier effect, Joule heating, Thomson effect, and Fourier heat conduction phenomena are all considered in the simulation process. Shannon's entropy method is applied to select the best solution from the Pareto Frontier. Besides, the exergy destruction rate is analyzed, the results show that the exergy destruction rate increases as the load resistance increases. In addition, the different relationships between the load resistance and the voltage, power output, efficiency and entropy generation rate are presented. The principle of performance enhancement is also explained by comparing the ZT value along the TE legs. The optimization is important to the development of more compact and high-efficiency thermoelectric generators.