Abstract

This study investigates the heat and mass transfer characteristics of a Casson hybrid nanofluid composed of TiO<inf>2</inf> and Al<inf>2</inf>O<inf>3</inf> nanoparticles dispersed in water over a stretching sheet. The model incorporates key physical effects, including magnetic field (MHD), thermal radiation, porosity, heat generation/absorption, and activation energy. To enhance the accuracy and efficiency of modeling complex heat and mass transfer in hybrid nanofluids by integrating realistic physical effects with a data-driven ANN approach. The governing nonlinear partial differential equations are reduced using similarity transformations. The results reveal that increasing the Casson parameter enhances velocity while reducing temperature and concentration. Magnetic fields and porosity influence the boundary layers and flow resistance, while thermal radiation facilitates heat dissipation. Heat generation raises fluid temperature and solute mobility. Activation energy significantly impacts chemical reaction rates and concentration levels. Results reveal that a 40% increase in the Casson parameter enhances velocity while reducing temperature and concentration by approximately 18% and 12%, respectively. A rise in magnetic field strength decreases velocity by 22% and thickens thermal and solutal boundary layers. Heat generation raises temperature and concentration by up to 20%, while thermal radiation reduces temperature by 15%. Activation energy increases solute concentration by 25% due to slower reaction rates. The results show that a higher Casson parameter increases velocity but decreases temperature and concentration. Magnetic fields and radiation reduce velocity and temperature, respectively, while porosity enhances flow and heat transfer. Heat generation and activation energy raise solute concentration, whereas stronger chemical reactions reduce it. These findings provide useful insights for designing thermally efficient systems in energy, manufacturing, cooling systems, solar thermal technologies, and chemical processes. The ANN results show excellent agreement with benchmark Runge–Kutta solutions, validating the method's reliability.