Heat and mass transfer in MHD Williamson hybrid nanofluid flow over a stretching/shrinking cylinder with modified Buongiorno model

This study numerically analyses the chemically reactive magnetohydrodynamic Williamson hybrid nano liquid (SWCNT-MWCNT/kerosene oil) flow over a stretching/shrinking cylinder with thermal radiation and a modified Buongiorno model. The main objective of this research is to enhance thermal efficiency using diverse types of nanomaterials, namely SWCNT-MWCNT, with kerosene oil as the base fluid for the calculations. The PDEs are converted into ODEs using similarity transformations. The Bvp4c technique and a hybrid machine learning approach (MLR) are employed to solve them mathematically. The results of this investigation reveal that the shrinking cylinder has a greater influence on the parameters compared to the stretching cylinder. The Grashof number ($$\:1.5\le\:{G}_{T}\le\:5.5)$$ and magnetic impact ($$\:0.01\le\:M\le\:0.1)$$ increase the velocity profile, while the effect of the Weissenberg number ($$\:0.1\le\:We\le\:5.1)$$ is the opposite. The thermophoresis parameter ($$\:0.1\le\:Nt\le\:0.3)$$ and Brownian motion ($$\:0.01\le\:Nb\le\:0.03)$$ enhance the thermal profile. The complex multivariable relationships in the mix nanofluid system were well captured by the MLR analysis results. With the highest accuracy and generalization (R² > 0.99, MAPE < 1%), the skin friction rate model is perfect for high-precision applications. The mass transfer rate model offers balanced performance for general use, while the Nusselt number model exhibits higher errors and is less appropriate for precise tasks. All things considered, the skin friction rate model is the most dependable.