Hybrid computational and ANN-based analysis of heat transfer and bioconvection in Sutterby nanofluid flow across a stretched surface

This study presents the application of computational fluid dynamics in conjunction with artificial neural networks to analyze the heat and mass transfer characteristics of bioconvective Sutterby nanofluid over a two-dimensional stretching sheet. The Darcy-Forchheimer model evaluates porous media resistance in the presence of chemical reactions. By applying suitable similarity transformations, the governing equations are transformed into a non-dimensional form and solved numerically using the bvp4c approach. Additionally, an ANN model is developed and trained using the Levenberg–Marquardt Backpropagation algorithm (LMBP) to accurately predict skin friction, Nusselt number, Sherwood number, and the concentration of motile microorganisms. It can be concluded that the Darcy and Deborah numbers exhibit a similar increasing trend within the velocity profile. The Brownian motion parameter has the opposite effect on thermal distribution and the mass transport rate. The ANN predictions and numerical results for heat and mass transfer showed excellent agreement. The optimized ANN model accurately predicted critical parameters with a variance of $\pm 2\%$ and a maximum error of $1.8\%$ in all scenarios. This demonstrates the efficacy of the hybrid computational and ANN framework in simulating the complex flow and heat transfer properties of nanofluids on stretched surfaces.