Coupled magneto-bioconvective dynamics of Eyring–Powell ternary hybrid nanofluids through porous media: a neural network-based predictive approach

This study introduces the two-stage analysis in fluid dynamics by investigating the heat and mass transport behavior of the non-Newtonian Eyring-Powell fluid model. It scrutinizes the synergistic effects of a trihybrid nanofluid– comprising Molybdenum disulfide (MoS₂), magnetic iron oxide (Fe₃O₄), Uranium dioxide (UO₂), and blood– with magnetic field effect, mixed convection, viscous dissipation, heat source, and thermal radiation, alongside the bioconvection phenomenon over a permeably elongating sheet. This research adopts the local non-similarity approach, transforming and solving the system of equations to evaluate the velocity, temperature, and concentration profiles. The results include graphical representations of these profiles with MATLAB’s bvp4c scheme. The tabular data showcases that the shear stress values are increased with the magnetic parameter, and decreased with permeability, mixed convection and material fluid. As Eckert number rises, the heat transfer rate also rises. But with the values of radiation, permeability and magnetic field increasing, the rate of heat transfer declines. Both Sherwood number and the mass transfer rate increase when their corresponding dimensionless parameters (i.e., Lewis, Peclet, Schmidt numbers and chemical reaction) are increased. The Levenberg-Marquardt scheme from Artificial Neural Networks was employed, and the accuracy of the ANN-LMBPS model is evaluated by comparing with the bvp4c results of the engineering parameters– namely, shear stress, heat and mass transfer rates, and the Sherwood number. Error reductions for the model vary from E⁻⁰³ to E⁻⁰⁴ for the shear stress, Sherwood number, and mass transfer rate. For heat transfer rate, they range from E⁻⁰³ to E⁻⁰⁵. Additionally, the effects caused by the physical parameters on the momentum, thermal and concentration boundary layers are exhibited via mean squared error plots, training state, error histogram and regression analyses, under eight different scenarios. This study is unique in its fusion of a blood-based trihybrid nanofluid with a non-Newtonian Eyring-Powell framework, under multi-physical influences, which has not been thoroughly explored before. Additionally, the hybrid two-phase research methodology demonstrates the computational benefits of the ANN model in predicting intricate bio-convective transport systems in addition to confirming the accuracy and consistency of both approaches. This approach is particularly significant for advancing computational modeling in biomedical and industrial cooling systems, where accurate control of heat transmission is critical.