Design impacts of stochastically-varying input parameters on advanced thermoelectric conversion systems

Advanced, direct thermal energy conversion technologies are receiving increased research attention in order to recover waste thermal energy in advanced vehicles and industrial processes. Advanced thermoelectric (TE) systems necessarily require integrated system-level analyses to establish accurate optimum system designs. Past system-level design and analysis has relied on well-defined deterministic input parameters even though many critically important environmental and system design parameters in the above mentioned applications are often randomly variable, sometimes according to complex relationships, rather than discrete, well-known deterministic variables. This work describes new research and development creating techniques and capabilities for probabilistic design and analysis of advanced TE power generation systems to quantify the effects of randomly uncertain design inputs in determining more robust optimum TE system designs and expected outputs. Selected case studies involving stochastic TE .material properties and coupled multi-variable stochasticity in key environmental and design parameters are presented and discussed to demonstrate key impacts from considering stochastic design inputs on the TE design optimization process. Critical findings show that: 1) stochastic Gaussian input distributions may produce Gaussian or non-Gaussian outcome probability distributions for critical TE design parameters, and 2) probabilistic input considerations can create design effects that warrant significant modifications to deterministically-derived optimum TE system designs. Magnitudes and directions of these design modifications are quantified for selected TE system design analysis cases.