Multi-objective optimization of surface-drag phenomena in coaxial disk systems using machine learning assisted Taguchi based grey relation analysis

Purpose Integrating Taguchi optimization with grey relation analysis minimizes variability and identifies dominant flow parameters, including shear rate, thermal gradient and mass-transfer characteristics, thereby improving the boundary-layer characteristics. The present work explores the comparative impact of uniform and horizontal magnetic fields on viscous fluid flow between coaxial rotating disks. The study aims to examine the impact of Fourier’s law, thermal radiation, pollutant concentration, slip effects and variable thermal conductivity on flow behavior. Design/methodology/approach The modeled equations are converted into ordinary differential equations (ODEs) by applying the appropriate similarity transformations. The resulting ODEs are then numerically solved using the shooting technique and the Runge–Kutta–Fehlberg–4th-5th order method. Additionally, a multilayer perceptron based Levenberg–Marquardt algorithm is employed to optimize the skin friction coefficient. Furthermore, Taguchi design with analysis of variance and orthogonal arrays was used to assess the control factors for optimizing surface-drag and improving flow behavior. The present study uses grey relation analysis to obtain appropriate parameter settings for the multi-objective optimization. Findings According to the grey relation, the Reynolds number (62.92%) has a significant effect on the surface drag, followed by the wall slip parameter (20.20%) and the magnetic parameter (9.84%). Originality/value The grey relation analysis identifies the optimal configuration of parameter values as {M1, ζ4, Re4}, representing the best configurations to optimize the drag forces and enhance the fluid flow phenomena.