Non-Fourier heat and mass transfer analysis of MHD Oldroyd-B hybrid nanofluid: An artificial intelligence assisted overlapping spectral quasi-linearization approach

Abstract

The present study develops a high-fidelity numerical analysis of magnetohydrodynamic (MHD) transport phenomena in an Oldroyd-B hybrid nanofluid composed of blood as the base fluid with copper (Cu) and aluminum oxide (Al₂O₃) nanoparticles. The flow is induced by a nonlinear stretching surface situated inside a porous medium under the combined effects of thermal radiation, viscous dissipation, Joule heating, and finite thermal and solutal relaxation. The resulting nonlinear boundary-layer equations governing momentum, heat, and mass transfer are transformed into a coupled system of ordinary differential equations using similarity variables. The solutions are obtained with high accuracy using the overlapping spectral quasi-linearization method (OSQLM), which combines the advantages of quasi-linearization with the spectral accuracy of Chebyshev collocation on overlapping domains. A detailed parametric analysis is carried out to examine the effects of viscoelasticity, porous medium resistance, magnetic damping, nonlinear stretching, and nanoparticle concentration on the velocity, temperature, and concentration distributions. The hybrid nanofluid demonstrates enhanced thermal and mass transport characteristics compared with the conventional nanofluid due to its improved effective thermophysical properties. In particular, thermal relaxation significantly suppresses convective cooling, whereas increasing nonlinear stretching enhances the cooling effect. Mass transfer is found to depend strongly on the solutal relaxation parameter, Schmidt number, and power-law index. Furthermore, the hybrid nanofluid requires approximately 30%–40% lower nanoparticle volume fraction to achieve equivalent thermal performance, which is important for reducing biomedical toxicity concerns in practical applications. To complement the physics-based solver, an artificial neural network (ANN) surrogate model is developed using high-fidelity OSQLM data to rapidly predict engineering quantities. The ANN demonstrates excellent predictive performance with minimal error in estimating the skin friction coefficient, Nusselt number, and Sherwood number. The hybrid OSQLM-ANN framework provides an efficient computational tool for analysis, optimization, and real-time control of hybrid nanofluid transport in biomedical systems, porous media, and advanced thermal management applications.