Neural and neuro-fuzzy simulation of mixed convective Carreau blood nanofluid flow with magnesium oxide nanoparticles

In recent years, non-Newtonian nanofluids, particularly Carreau fluids, have gained attention in engineering and biomedical applications due to their heat transfer characteristics and controllable flow behavior. The shear-thinning nature of Carreau nanofluid facilitates modeling of complex industrial and biomedical flow phenomena, involving non-Newtonian behavior and enhanced thermal transport. This study investigates the dispersion of magnesium oxide nanoparticles within a blood-based Carreau nanofluid, aiming to enhance thermal conductivity and support applications such as targeted drug delivery, bio-thermal regulation, and artificial circulatory systems. The governing partial differential equations of momentum and energy, describing the flow and heat transfer characteristics of the blood-based Carreau nanofluid containing magnesium oxide nanoparticles, are formulated based on conservation laws. Local similarity transformations reduce these equations to nonlinear ordinary differential equations, which are solved using a collocation-based boundary value problem approach in Python to obtain reference dataset for velocity and temperature profiles. A machine learning framework is developed using the Keras sequential application programming interface to predict velocity and temperature distributions within the nanofluid flow. A neuro-fuzzy inference system estimates key engineering quantities, including the Nusselt number and skin friction coefficient, providing insights into flow characteristics and parameter sensitivity analysis. The findings show excellent agreement between machine learning predictions and reference solutions, demonstrating the robustness of the proposed hybrid modeling approach. This study presents an efficient framework for analyzing and designing advanced bio-thermal systems involving non-Newtonian nanofluids.