Comparative Physics Informed Neural Network Study of Tetra and Ternary Nanolubricant Systems in MHD Flow

This study investigates heat and mass transfer characteristics of Casson hybrid nanofluids flowing over a stretching surface in the presence of a magnetic field, Joule heating, and thermal radiation. Two advanced nano-lubricant configurations are examined. The first is a ternary hybrid nanofluid consisting of Al_2O_3, ZnO, and SiC nanoparticles suspended in engine oil, while the second is a tetra hybrid nanofluid formed by adding graphene nanoplatelets (GNPs) to the ternary mixture. The objective is to compare how nanoparticle composition influences thermal conductivity, flow behaviour, and energy dissipation mechanisms. The mathematical formulation incorporates effects of chemical reaction, thermo-diffusion, Joule heating, and radiative heat flux. Through suitable similarity transformations, the governing nonlinear partial differential equations are reduced into a coupled system of ordinary differential equations. These equations are solved using a Physics Informed Neural Network (PINN) approach developed specifically for nanofluid lubrication systems. The proposed PINN architecture embeds the governing physical laws directly into the loss function, allowing simultaneous minimization of equation residuals and boundary condition errors. This two way optimization enhances solution stability and computational efficiency. Numerical results indicate that the tetra hybrid nanofluid provides significantly higher thermal enhancement and reduced entropy generation compared to the ternary formulation. This performance improvement is mainly due to the exceptional thermal conductivity and large surface area of GNPs. However, the ternary hybrid nanofluid exhibits relatively stable viscous behavior with moderate temperature gradients. Overall, the PINN framework shows strong convergence and predictive accuracy, offering a reliable computational tool for complex nonlinear thermal-fluid problems in lubrication applications.