Combining machine learning with thermochemical modeling of reactive liquid flow over a two-phase cone to improve the cooling of chemical processes

Magnetohydrodynamic (MHD) flow plays an important role in chemical processing, polymer extrusion, thermal storage, and advanced cooling systems, where controlled heat and mass transfer is essential. Motivated by these applications, this study examines the MHD mixed convection flow of an electrically conducting Ostwald–de Waele fluid over a rotating conical surface subjected simultaneously to linear surface temperature (LST) and linear surface heat flux (LSHF) boundary conditions. The coexistence of these conditions generates a thermal transition zone that significantly alters boundary-layer behavior. The model incorporates thermal radiation, viscous dissipation, buoyancy effects, and a first-order homogeneous chemical reaction. Using appropriate similarity transformations, the governing nonlinear partial differential equations are reduced to ordinary differential equations and solved numerically using MATLAB’s BVP4c solver. To enhance computational efficiency, a feedforward artificial neural network (ANN) is developed as a surrogate predictor for velocity, temperature, concentration, and stream function profiles. The ANN shows excellent agreement with numerical results and very low prediction errors within the trained parameter range. The results indicate that the magnetic parameter suppresses fluid motion via the Lorentz force, while surface rotation enhances near-wall momentum transport. Thermal radiation and viscous dissipation increase the temperature field, whereas stronger chemical reaction and higher Schmidt number reduce species concentration. The proposed framework provides an efficient and reliable tool for analyzing and optimizing heat and mass transfer in reactive rotating non-Newtonian systems.