This study presents a hybrid Artificial Intelligence–Computational Fluid Dynamics (AI-CFD) framework for optimizing heat transfer in magnetic bio-nanofluids subjected to external magnetic fields. Magnetic bio-nanofluids, composed of biocompatible base fluids containing superparamagnetic nanoparticles, exhibit tunable thermal and flow behavior, making them promising for biomedical and micro-cooling applications. Conventional optimization methods based on experiments or brute-force CFD are computationally expensive and limited in exploring the full design space. To overcome these challenges, an Artificial Neural Network (ANN) surrogate model was developed to predict two key performance indicators, the Nusselt number and the friction factor, with high accuracy (R² > 0.997). The surrogate was then integrated with the NSGA-II multi-objective genetic algorithm to generate a Pareto front capturing the trade-off between maximizing heat transfer and minimizing pressure drop. Sensitivity analysis revealed the Reynolds number as the dominant factor, followed by the Hartmann number and nanoparticle concentration, which improved thermal performance but increased viscous losses. Representative Pareto-optimal designs demonstrated up to 25% enhancement in heat transfer with only a 10% increase in friction compared to baseline conditions. The results highlight the scientific value of combining physics-based magnetohydrodynamic modeling with AI-driven optimization, providing both methodological contributions and practical insights for cancer hyperthermia, targeted drug delivery, and compact biomedical electronics cooling. Limitations related to single-phase modeling and lack of experimental validation are discussed, with future work recommended in multi-phase modeling and Physics-Informed Neural Networks (PINNs).
Hasan Attyah Shaboot (Wed,) studied this question.