Artificial Intelligence based Assisted Computational Analysis of Micropolar Hybrid Nanofluid Flow with Non-Fourier Heat Flux and Thermal Radiation Effects

Abstract

In this study, a hybrid artificial intelligence and computational fluid dynamics (AI-CFD) method is used to study the thermal-flow behavior of micropolar CuO-Ag/water hybrid nanofluids under magnetohydrodynamic (MHD) mixed convection over a vertical porous plate with non-Fourier heat flux effects. Thermal transport phenomena are modeled accurately by the mathematical model, which includes the effects of thermal radiation, joule heating, viscous dissipation, suction/injection, heat generation/absorption, and cattaneo–christov heat flux. Similarity transformations are used to reduce the governing nonlinear partial differential equations to a dimensionless system of ordinary differential equations (ODEs) which can then be numerically solved by the Runge-Kutta fourth-order (RK4) method with the shooting technique. A data-driven machine learning framework was designed based on synthetic data generated from numerical solutions to improve the predictive capability and to minimize computational complexity. The supervised learning algorithms implemented were Linear Regression, Random Forest Regressor, Artificial Neural Networks (ANNs) using Python based TensorFlow/Keras and Scikit-learn libraries. To enhance the stability and accuracy of the model, data preprocessing and feature normalization techniques were used. The ANN model showed better performance with a maximum R² = 0.984716 for velocity prediction and R² = 0.993335 for temperature estimation, better than conventional regression models. The numerical analysis shows that raising the values of thermal radiation, Eckert number, and nanoparticle volume fraction, the temperature profiles and thermal conductivity are significantly improved, while the effect of the magnetic field is decreased the fluid velocity. Proposed AI-assisted surrogate framework is suitable for capturing nonlinear thermo-fluid interactions in an efficient manner and provides a computational efficient alternative for predicting heat transfer characteristics in advanced thermal engineering systems. The results show the promise of applying machine learning to numerical fluid dynamics for smart thermal management, energy system, and new heat transfer applications..

Keywords: Artificial intelligence; Machine learning; Computational fluid dynamics; Hybrid nanofluids; Non-Fourier heat flux; Micropolar fluid; Thermal radiation; Artificial neural networks; Magnetohydrodynamics; Heat transfer enhancement.