Numerical and ANN-Based Study of Mixed Convection Flow of Casson Nanofluid in a Chamfered Square Cavity with Multiple Fins

Desktop computer power supply units’ ventilated heat-sink chambers, where air travels through fin arrays to remove heat from electronic components, IGBT power modules in EV, exhibit thermal-fluid behavior like the existing vented finned cavity design. Solar air heater ducts with internal fins and polymer extrusion cooling chambers with non-Newtonian fluids have similar layouts. This computational investigation of mixed convection heat transport of a Casson nanofluid in a square, vented, chamfered cavity with multiple fins is inspired by this real-world usage. The left vertical wall and fin are consistently hot, whereas the right wall is thermally insulated and the horizontal walls are cooled to T c . Nanofluid enters and leaves the compartment through the input and exit holes. COMSOL Multiphysics solves non-dimensional equations using finite elements. Simulations include Reynolds numbers, Hartmann numbers, Casson parameters, and nanoparticle concentration parameters. Streamlines, isotherms, and local and average Nusselt numbers explain how these properties impact flow structure and heat transfer. Higher Casson and Hartmann numbers impede fluid motion and reduce heat transfer rates, whereas higher Reynolds numbers and nanoparticle concentrations improve thermal conductivity and convective circulation, increasing heat transmission. The chamfered cavity transmits heat better than the square cavity because it has less stagnant spots and more effective heat transfer areas. Using numerical data, an Artificial Neural Network model predicts thermal performance. The ANN model is a reliable surrogate predictor for thermal-fluid systems, as it matches numerical data with a mean squared error of 10 −6 and correlation coefficients close to unity (R = 0.999).