Physics-informed deep learning modeling of magnetohydrodynamic-driven electroosmotic thermal transport in tangent hyperbolic fluids within a microchannel

The present study investigates the problems associated with the thermal transport analysis of magnetohydrodynamic-induced electroosmotic flow through a tangent–hyperbolic fluid-saturated porous medium. The solution for the electrical potential distribution in the electric double layer is given in closed form by the linearized Poisson–Boltzmann equation, which precisely characterizes the electrostatic field produced near the charged walls. Additionally, an approximate solution of the governing equations is generated by incorporating the boundary conditions into the loss function and optimizing the unknown parameters of a feed-forward deep neural network, which is defined as a series of linear transformations followed by a nonlinear activation function of a certain number of layers. According to the findings of the research, the increase in the electroosmotic parameter corresponds to a smaller Debye length that intensifies the electric force and leads to an increase in entropy generation. A higher Eckert number signifies a greater transformation of mechanical energy dissipation into the internal energy of the fluid. Consequently, the fluid's temperature rises due to the heat generated from the mechanical energy dissipation. Moreover, the magnetic field still dominates flow in highly permeable media, reducing velocity and producing plug-like behavior. Proper control of heat flow helps manage biochemical reactions and separate chemicals in lab-on-a-chip devices, making medical tests faster and more reliable.