Geometry design for maximizing output power of segmented skutterudite thermoelectric generator by evolutionary computation

This study focuses on the design of a segmented thermoelectric generator system to maximize the system output power. Skutterudite, (Sr, Ba, Yb)(y)Co4Sb12 with 9.1% In0.4Co4Sb12, is used as the n-type element, while DD0.59Fe2.7Co1.3Sb11.8Sn0.2 is used as the p-type one. They both are employed as the hot side segments. Alternatively, hydrothermal synthesized nanostructure thermoelectric material (Bi0.4Sb1.6Te3) is chosen as the cold side segments. To achieve the optimum design, a numerical method is developed, while the multi-objective genetic algorithm is adopted. The evolutionary computation processes during seeking the optimum combination of the segments are visualized, and it is found that four generations are enough for reaching the target. With the leg length of 3 mm, the optimum n-type and p-type cold side segment lengths are 0.5 mm and 0.78 mm, respectively. Compared to the equal-segmented thermoelectric couple, the optimized couple at a temperature difference of 400 K can increase the output power by 21.94% and its efficiency is 14.05% which is much higher than conventional thermoelectric generators. The theory of impedance matching does not apply to the segmented thermoelectric generator. The heat flux distribution in the couple is dependent on the temperature difference. Overall, the segmented elements with evolutionary computation design is a promising tool for intensifying thermoelectric generator performance.