Intelligent computing through neural networks for entropy generation in MHD third-grade nanofluid under chemical reaction and viscous dissipation

This study explores Artificial Neural Network with Back Propagated Levenberg Marquardt (ANN-BPLM) for entropy generation in magnetohydrodynamic third-grade nanofluid flow model (MHD-TGNFM) with chemical reaction and heat sink/source effect. The nonlinear ODE system for MHD-TGNFM is obtained after simplifying the presented mathematical model in PDEs through a suitable transformation system. The dataset was constructed from the effective modifications in the physical parameters of MHD-TGNFM with the Homotopy Analysis Method (HAM). To interpret the approximated solution testing, validation and training sets are used in ANN-BPLM. The comparison with a standard solution is investigated by the performance of MSE convergence, Error histogram and regression studies. Moreover, the impacts of physical variants on temperature, Entropy production rate, velocity, Bejan number and concentration are also analyzed. The result reveals that velocity gradient f'(eta) inclines for rising values of Re, B-1 and B-3, whereas the converse behaviour is seen for magnetic parameters. Increment in values of M, Br, Nt, Nb and Q enhances the temperature gradient theta(eta). Concentration gradient phi(eta) increases, whereas the opposite behaviour is seen for Nt and Sc. N-G is elevated for increasing values of Br, whereas Be declines for greater values of Br. Entropy and Bejan number are increased for L.

Automated summary

This study investigates entropy generation in magnetohydrodynamic third-grade nanofluid flow model using Artificial Neural Network. The impacts of physical parameters on temperature, entropy production rate, velocity, Bejan number, and concentration are analyzed.