Numerical investigation of chemical reactive MHD fluid dynamics over a porous surface with Cattaneo–Christov heat flux

A theoretical framework to investigate three-dimensional Williamson fluid flow over a bidirectional extended flat horizontal surface is proposed in this dissertation. Artificial intelligence and machine learning fields have seen tremendous growth in prominence along with the rapid advancement of related technology. This work trains a machine learning model based on artificial neural networks to handle the mathematical formulation incorporating heat source and Hall effects using the Levenberg–Marquardt approach. Additionally, the impact of activation energy on fluid concentration is incorporated into the analysis. Cattaneo-Christov double diffusion models are used to model heat transfer combined with the effects of thermal radiation. The solutions, serving as reference datasets for various scenarios, have been generated numerically using the BVP4C approach. Artificial neural networks are utilized for training, testing, and validating these numerical computations using a 70:15:15 ratio. The predictive model accuracy is evaluated using various statistical metrics, including linear regression, histograms, fitting analysis, and mean squared error evaluations, with the least error ranging between 10⁻³ and 10⁻⁴, based on individual error analysis of four parameters. The findings show that temperature rises with the M parameter, whereas velocity declines by increasing the M parameter. Concentration rises with increasing activation energy parameter and falls with decreasing Sc. The results show that artificial neural networks can provide a successful replacement for forecasts for the future, and the fluid flow structure simulated here may result in better industrial designs.