Enhanced thermal transmission and flow dynamics of nanofluids based on soft computing optimization approach subject to heat source and slip effects

In thermal engineering, Casson nanofluids with induced convective transport have several applications, including cooling technologies, heat exchangers, and magnetohydrodynamic energy systems. In this paper, the three-dimensional Casson nanofluid flow is analytically studied in the presence of a uniform magnetic field acting on a stretching surface embedded in a porous medium. To accurately portray the underlying transport processes, the model accounts for velocity slip, Brownian diffusion, chemical interactions, and heat source/sink effects. An artificial neural network based on the Levenberg-Marquardt Backpropagation (LMB-NN) algorithm is used to solve the resulting set of dimensionless nonlinear ordinary differential equations. Increasing the Casson fluid parameter results in drastically reduced thermal, nanoparticle concentration, and velocity profiles, according to the computational results. On the other hand, the parameters for magnetic and chemical reactions enhance mass and heat transmission. Stronger magnetic forcing increases temperature through Lorentz-forced heating. The Artificial Neural Network architecture shows remarkable precision with mean squared errors on the scale of e^-10-e^-11. The overall absolute errors vary from e^-03-e^-08. The gradients are close to e^-08. The learning statistics demonstrate excellent convergence and generalization with best validation of MSE is 4.1764e^-11 at epoch 219. The suggested Artificial Neural Network surrogate reliably captures complex nonlinear fluid dynamics, as shown by these performance metrics. As an effective alternative to traditional solvers, this study develops a robust Artificial Neural Network-assisted methodology for simulating nonlinear boundary-layer flows in Casson nanofluid.