Artificial neural network modeling of MHD slip-flow over a permeable stretching surface

In this work, we consider the flow of magnetohydrodynamic (MHD) fluid over a permeable surface due to continuous stretching. The stretching surface is subject to a constant magnetic field along normal direction and velocity-slip conditions. This flow is governed by nonlinear partial differential equations (PDEs) subject to associated boundary conditions. The similarity transformation technique was applied to obtain their non-dimensional form, coupled with nonlinear ordinary differential equations (ODEs). MATLAB-based program bvp5c was then used to obtain their numerical solution. Two artificial neural network models were also presented for predicting the coefficients of skin friction - f ''(0) and heat transfer rate -theta(0). The present study revealed that heat transfer rate is decreased due to increases in first- and second-order slip parameters. Results also showed that neural network models can predict thermal conductivity with high accuracy. High R squared values of 0.99 were achieved for predicting coefficients of skin friction - f ''(0) and heat transfer rate -theta(0). This shows the effectiveness of neural network models for predicting those characteristics and thus reducing the time required for numerical models for predicting MHD slip flow over a permeable stretching surface. Moreover, in comparison with the other numerical methods, the present ANN model can be applied to more complex mathematical models because it reduces the time and processing capacity required for solving the problem.

Automated summary

In this work, the behavior of magnetohydrodynamic fluid flow over a permeable surface was studied using similarity transformation technique and artificial neural network models. The study revealed a decrease in heat transfer rate with increasing first- and second-order slip parameters. The neural network models showed high accuracy in predicting skin friction coefficients and heat transfer rates, reducing the time required for numerical predictions.