3D thermally laminated MHD non-Newtonian nanofluids across a stretched sheet: intelligent computing paradigm

The primary subject of this article is the study of the viscous flow of nanofluids consisting of copper-methanol and water in the presence of a three-dimensional stretched sheet, which is subjected to magnetohydrodynamic effects (3D-MHD-NF) by employing artificial recurrent neural networks that are optimized using a Bayesian regularization technique (ARNN-BR). The viscosity effect is recognized to be dependent on temperature, with methanol and water being used as the base fluid. The presented model is employed in the manipulation and creation of surfaces within the field of nanotechnology. Its applications include stretching, shrinking, wrapping, and painting devices. The Adams method was employed to generate a dataset for the 3D-MHD-NF model for four scenarios by varying the Hartmann number (H), volume fraction of nanoparticle (\(\varphi\)), and viscosity parameter (α). The ARNN-BR technique employed a random selection of data 70% for training, 20% for testing, and 10% for validity. It has been found that boundary layer becomes thinner as the volume percentage of nanoparticle increases. Additionally, it is observed that augmentation in the viscosity parameter results in a proportional rise in temperature. Moreover, it is observed that increment in the variables H, φ, and α have an impact on the velocity boundary thickness in both the x- and y-directions. The newly introduced ARNN-BR technique's dependability, stability, and convergence were assessed using a fitness measure based on mean squares errors, histogram drawings, regression, input-error cross-correlation, and autocorrelation analysis for each scenario of the 3D-MHD-NF model.