Experimental evaluation and development of predictive models for rheological behavior of aqueous Fe3O4 ferrofluid in the presence of an external magnetic field by introducing a novel grid optimization based-Kernel ridge regression supported by sensitivity analysis

In the present study, experiments are performed to determine the changes in the viscosity of water-Fe3O4 magnetic nanofluid (MNF) with shear rate, nanoparticle concentration and magnetic field (MF) induction. It was observed that as the shear rate elevates, the MNF viscosity first diminishes and then remains almost constant. Besides, the viscosity elevated with the application of the MF and its induction and also with increasing the concentration of nanoparticles. As another novelty of this research, a novel kernel based machine learning scheme namely, grid optimization based-kernel ridge regression (Grid-KRR) model was developed to accurate prediction of viscosity of water-Fe3O4 MNF based on volume fraction of nanoparticles, shear rate, and magnitude of external MF as input features. Besides, the Random forest (RF) and Gene expression programming (GEP) models were examined for validating the Grid-KRR model. The performance criteria demonstrated that the Grid-KRR outperformed the RF. (C) 2021 Elsevier B.V. All rights reserved.

Automated summary

Experiments in the study focus on investigating the impact of shear rate, nanoparticle concentration, and magnetic field induction on the viscosity of water-Fe3O4 magnetic nanofluid (MNF). Results show a complex relationship between these factors and viscosity. A novel machine learning model, Grid-KRR, is developed for accurately predicting viscosity based on input features such as nanoparticle volume fraction, shear rate, and magnitude of external magnetic field. Performance evaluation indicates that the Grid-KRR model outperforms other models like Random Forest and Gene expression programming.