Deep Learning Approach on Numerical Analysis of Time‐Based Flow Over a Decelerating Rotating Disk With Newtonian Heating

ABSTRACT The present study investigates an unsteady incompressible flow beyond a rotating stretchable disk. The surface undergoes deceleration over time. The flow model governed by the three‐dimensional Navier–Stokes equations is reduced to a dimensionalized form of self‐similar differential equations. To seek a feasible solution to the mathematical model, we implemented a built‐in Runge–Kutta–Fehlberg package. Numerical values of drag force, torque, and Nusselt number at the disk surface are computed for the underlying parameters. Artificial Neural Networks (ANNs) nowadays are an efficient tool that provides novel results to complex problems, which are very complicated to solve with existing conventional procedures. More specifically, a statistical visualization tool and deep learning techniques are used to accurately assess and estimate drag force, torque, and Nusselt number against the physical quantities. The magnitude of shear and tangential stresses is enhanced for slow deceleration, while for larger decelerating rotating disks, these stresses are first reduced and then enhanced in magnitude against increased stretching parameters. The heat transfer rate at the disk surface is reduced through conjugate parameters; however, such a rate is enhanced against enlarged values of the Prandtl number. The ANN model demonstrates remarkable accuracy in capturing both linear and nonlinear trends across all parameter interactions, validating its robustness in predicting complex relationships in fluid mechanics.