A Novel Paradigm for Predicting Thermal Analysis of Electro-magnetohydrodynamic Ternary Nanofluid Flow Using Machine Learning Approach

Abstract

This investigation addresses an important gap in the current literature on nanofluids, particularly regarding their interactions and associated computational models. The aim of presented study is to integrate artificial intelligence (AI) and machine learning (ML) techniques with computational fluid dynamics (CFD) to analyze the thermal attributes of a ternary nanofluid. The analysis looks at how electro-magnetohydrodynamics affects a porous material surface while following different heat flow conditions. This research uses the Levenberg-Marquardt algorithm along with a back-propagation artificial neural network (BPA-ANN) after adjusting the settings through various AI training methods. We apply similarity transformations to derive ordinary differential equations from non-linear governing partial differential equations. We then use modified finite difference discretisation to numerically evaluate the resultant equations. At elevated levels of the Forchheimer parameter, the velocity profile exhibits a marked decline attributable to heightened frictional forces. The Biot number augments the thermal state and boundary layer. A 28:3%, 18:7%, and 8:8% rise in thermal transfer efficiency is reported for the ternary, hybrid, and mono nanofluids. The proposed investigation improves the understanding of CFD issues through the development of an innovative computational structure that amalgamates ANNs with numerical simulations, thereby providing superior precision and effectiveness in modeling and forecasting fluid behavior within intricate physical frameworks.