Finite Difference and Machine Learning Simulation of Magneto-Tetra-Hybrid Nanofluid Transport in Time-Dependent Constricted Arteries

Abstract

This research examines the unsteady Magnetohydrodynamic (MHD) tetra-hybrid nanofluid blood flow in a time-dependent stenosed artery, focusing on momentum transport, augmentation of heat transfer, enhancement of nanoparticle concentration, and entropy generation. By suspending four kinds of nanoparticles, namely copper Cu, aluminium oxide (Al₂O₃), titania (TiO₂) and gold (Au), the residence time is enhanced to form a Cu-Al₂O₃-TiO₂-Au/blood tetra-hybrid nanofluid in which blood acts as a non-Newtonian Jeffrey dendritic structure fluid. The issue is of great biomedical relevance, as highly stenosed arterial flow has a direct correlation with cardiovascular pathologies, while advanced nanofluid-based transport models can provide important insights for localized thermal control, drug delivery, and hemodynamic modulation in diseased vessels. The governing nonlinear dimensionless equations describing the velocity, temperature, concentration, and entropy characteristics are derived under suitably chosen physiological assumptions and solved numerically using an implicit Finite Difference Method (FDM). Furthermore, the Levenberg-Marquardt training algorithm-based Artificial Neural Network (ANN) model is applied for predicting the transport responses and examining the ability of data-driven approximation in comparison to numerical results. The numerical results are evident that the enhancement of the Jeffrey fluid parameter hastens the flow, and the volume/space effect of a stronger magnetic field hinders velocity subject to Lorentz forces with a rising temperature field. The ANN predictions agree well with the FDM solutions, demonstrating the robustness of the proposed hybrid computational framework. The novelty of the present work lies in the simultaneous evaluation of transient stenosed arterial transport under the MHD tetra-hybrid nanofluid model approach with a coupled FDM-ANN framework, advancing the prior studies by integrating unsteady arterial constriction effects, enhanced thermal transport characteristics, and entropy behavior as well as machine learning-enabled prediction in a consistent biomedical-oriented assessment. The current results could aid in the theoretical design of advanced hemodynamic and thermal management strategies across pathological arterial beds.