Optimizing microelectronic module cooling under magnetic fields through hybrid nanofluid: a computational fluid dynamics-artificial neural network approach

In this study, an advanced cooling method for microelectronic modules in a magnetic field is examined through a numerical and artificial intelligence approach. The research utilized a serpentine microchannel with a rectangular cross section and a laminar and steady flow, assessing various flow rates ranging from 0.01 to 0.1 mL s⁻¹. A hybrid nanofluid made of soybean oil with MgO and Al₂O₃ nanoparticles served as the coolant, tested at different mixing ratios and volume fractions of 1–4%. The impact of several parameters, including forced convection, magnetic field intensity (0–1 Tesla), module heating capacity (3–7 W), and aspect ratio of the microchannel section (AR) (0.5, 1, and 2), was explored using computational fluid dynamics. The research findings revealed that a 1 Tesla magnetic field notably increased module temperatures, with the maximum cooling efficiency observed with the application of the hybrid nanofluid, optimized convection, an AR of 0.5, and the inclusion of a magnetic insulator. This optimal setup resulted in a significant temperature reduction ranging from 5.98 to 112.88 K compared to the use of pure soybean oil without forced convection and a magnetic field strength of 1 Tesla. Furthermore, the study integrated an artificial neural network (ANN) model, which demonstrated superior accuracy and reliability in predicting the system's performance compared to genetic algorithm models. The effectiveness of ANN as a predictive tool suggests a reduction in dependency on elaborate physical experiments for evaluating the cooling performance of microelectronic modules under magnetic influence. This research proposes a potent and systematic approach for managing the thermal challenges faced by microelectronic devices in magnetic fields and deems further exploration into hybrid ANN models with meta-heuristic algorithms beneficial for enhancing solar collector modeling.