ANN-based modeling of porous MHD mixed convection in tri-hybrid nanofluids with thermal radiation

This study develops a novel computational-intelligent framework for modeling magnetohydrodynamic mixed convection in a porous medium filled with a morphology-tuned tri-hybrid nanofluid, incorporating the joint effects of nanoparticle shape and interfacial nanolayers. Unlike existing models, this work uniquely integrates morphology and nanolayer-dependent thermophysical properties with thermal radiation and cross-diffusion mechanisms, then solves the system using a hybrid numerical-ANN approach. The physical scenario involves a stretching sheet with uniform suction, which is relevant to cooling applications in the coating, extrusion, and thermal processing industries. Similarity transformations reduce the governing PDEs into nonlinear ODEs, solved via the BVP4C method. A Levenberg-Marquardt-trained artificial neural network (LM-ANN) is then employed to predict thermal and mass transport under varying parameter regimes rapidly. Results show that increased nanolayer thickness and spherical nanoparticles enhance thermal transport, while magnetic fields and porosity reduce velocity due to Lorentz drag and Darcy resistance. The LM-ANN model achieves excellent agreement with numerical data regression coefficient R = 1 and mean squared error MSE <10 −9 , validating its predictive accuracy. This dual-solver strategy enables fast and reliable simulation of complex thermal-fluid systems, making it applicable for optimization in microfluidic cooling and energy systems.