Analysis of pressure-dependent Sutterby fluid transport in a porous channel: numerical and neural network approaches

The main focus of the current study is on the Couette and Poiseuille flows of Sutterby fluid with pressure-dependent physical effects. The one-dimensional flow mechanism is examined with pressure-dependent viscosity and pressure-dependent porous medium over an inclined channel. To deliberate the physical characteristics, four different combinations of pressure-dependent porosity with pressure-dependent viscosity are implemented. The dimensionless setup is achieved through the non-dimensional variables which is further analyzed for all considered cases of variable porosity and viscosity. A numerical illustration of the flow field of both flows corresponding to the pertinent parameters and four different cases is exhibited through graphics and tables. Some exact results of the flow velocity are also computed by considering certain conditions. This study elucidates that the extreme flow rate is examined in the middle of the channel. Neural networks trained using Levenberg Marquardt algorithm performed with fast and stable convergence to low levels of loss. This allowed avoiding overfitting as the early stopping was used, which demonstrated the effectiveness of the training process and its generalizability. Moreover, a reduction is observed in the flow field for the shear thickening flow behavior of Sutterby fluid.