Thermoelectric generation in bifurcating channels and efficient modeling by using hybrid CFD and artificial neural networks

Abstract

Thermoelectric power generation within TEG mounted branching channels is considered with finite element method. In the heat transfer fluid of bifurcating channels, nanodiamond + Fe3O4 binary particles are used for further system performance improvement. It was observed that when compared to non-bifurcating channels, TEG power will be reduced with the use of branching channels while branching location also affects the interface temperature variations. At (Re1, Re2)=(1000, 200), TEG power is reduced 34.7% when both channels are branching while it is 9.9% for only upper channel branching case as compared to non-branching channel case. Up to 18% variation of power is obtained when location of the upper branching channel varies. Highest powers are achieved when both channels are filled with hybrid nanofluid while at (Re1,Re2)=(1000,200) TEG power rises by about 33% and 15.5% with nanofluid in both channels and with nanofluid in only one channel cases when compared to fluid in both channel configuration. The computational cost of electric potential and power generation in TEG device is drastically reduced from 6 hours with fully coupled high fidelity CFD to 3 minutes by using hybrid CFD and artificial neural networks. The proposed approach will very helpful in the efficient design and optimization of TEG installed renewable energy systems.