Probabilistic design optimization and reliability assessment of high temperature thermoelectric devices

Thermoelectric (TE) devices, subcomponents of which are made of brittle materials, generate an electric potential when they are subjected to thermal gradients through their thickness. These devices are of significant interest for high temperature environments in transportation and industrial applications where waste heat can be used to generate electricity (also referred to as "waste heat recovery" or "energy harvesting"). TE devices become more efficient as larger thermal gradients are applied across them. This is accomplished by larger temperature differences across the TE's hot and cold junctions or the use of low thermal conductivity TE materials or both. However, a TE brittle material with a combination of poor strength, low thermal conductivity, and large coefficient of thermal expansion can translate into high probability of mechanical failure (low reliability) in the presence of a thermal gradient, thereby preventing its use as intended. The objective of this work is to demonstrate the use of an established probabilistic design methodology developed for brittle structural components and corresponding design sensitivity analyses to optimize the reliability of an arbitrary TE device. This method can be used to guide TE material and design selection for optimum reliability. The mechanical reliability of a prototypical TE device is optimized from a structural ceramic perspective, using finite element analysis and the NASA CARES/Life integrated design code. Suggested geometric redesigns and material selection are identified to enhance the reliability of the TE device.