Modeling of chemically reactive fluid dynamics with thermal effect and energy source through a magnetized medium

This work investigates the radiative reactive flow of Williamson fluid over a Riga plate, considering thermophoresis, heat generation, variable chemical reactions, and material viscosity. The main intention is to compute fluid velocity, temperature, and concentration under these conditions. It surpasses conventional methods in terms of efficiency and precision by employing artificial neural networks. The validity of the results has been established utilizing visual comparisons and numerical simulations, which confirm the robustness and reliability of the proposed scheme. The fluid dynamics problem is resolved by implementing a systematic design methodology incorporating training, testing, and validation. The neural network architecture is refined to acquire knowledge of patterns, and the model's performance and ability to generalize across a range of scenarios are evaluated. As a litmus test, validation against a reference dataset is implemented. The intricate interaction of flow model parameters is elucidated through visual representations, which offer profound insights into the behavior of the fluid system. The range includes the absolute error values attained throughout the testing, validation, and training phases as 10-03-10-08. The mean-squared error values for Cases 1–4, namely horizontal velocity, vertical velocity, temperature, and concentration, all fall within the interval 10-09-10-10. Maximum gradient values are observed in the range of 10-08, whereas error histograms are observed between -1.4e-05 to -8.1e-06.