Performance analysis of thermoelectric generator mounted chaotic channel by using non-Newtonian nanofluid and modeling with efficient computational methods

Abstract

Performance features of a thermoelectric system mounted in a chaotic channel with non-Newtonian power law fluid are numerically explored with finite element method. The analysis is performed for different values of Re number of the hot and cold fluid streams (250 <= Re <= 1000), power law indices (0.75 <= n <= 1.25) and solid volume fraction of alumina (0 <= phi <= 4%) in water. It is observed that the fluid type with different power law indices significantly affected the electric potential variations and power generation of the thermoelectric system. Impacts of Re number on the power generation enhancement amount depends upon the power law index. The power rises by about 123.78%, 94.13% and 52.30% at the highest Re for different power law index combinations of (0.75,0.75), (0.75,12.5) and (1.25,1.25), respectively. Thermoelectric power reduces by about 39.71% for shear thinning fluids in both channels while it rises by about 43.48% for shear thickening fluids in chaotic channels. The potential of using nanofluids is more when both channels contain shear thinning fluids. Nanofluids rise the power of thermoelectric system by about 31%, 29% and 28% for the case when the hot side fluid is shear thinning, Newtonian and shear thickening fluid types while the cold side chaotic channel is shear thinning. When constant and varying interface temperature configurations are compared, there is at most 3% variations in the generated power while the trends in the curves for varying parameters are similar. The computational cost of constant interface temperature and computations only in the thermoelectric domains are much cheaper as compared to high fidelity coupled computational fluid dynamics simulations. The temperature field in the whole computational domain is approximated by using POD based approach with nine modes. A polynomial type regression model is used for POD-modal coefficients while fast and accurate results for interface temperatures are obtained. (C) 2021 THE AUTHORS. Published by Elsevier BV on behalf of Faculty of Engineering, Alexandria University.