Machine Learning Investigation of Marangoni Convection in Hybrid Nanofluids with Darcy-Forchheimer

Abstract This research utilizes machine learning to investigate Marangoni convection in a hybrid nanofluid (MnZnFe₂O₄ +NiZnFe₂ O₄⁄H₂ O) within a Darcy-Forchheimer porous framework. We conduct both qualitative and quantitative assessments of heat transfer, mass transfer, and viscous dissipation irreversibility during the flow. Numerical results are obtained using a Python finite difference algorithm, after which MATLAB is employed for AI-based analysis. Additionally, the Levenberg-Marquardt neural network algorithm is trained and utilized. Our findings show that fluid velocity diminishes as the inverse Darcy parameter, Marangoni ratio, and Forchheimer parameter increase. Moreover, the temperature rises with the Eckert number and Prandtl ratio. As concentration increases, activation energy and Schmidt parameter also grow. Mean Square Error (MSE) for the results reaches up to 10⁻¹¹ across various impacts.