TiO2-Based Nanoemulsion Lubrication for Hot Rolling of High-Tensile Structural Steel: Experimental Analysis and Machine Learning-Based Roll Force Prediction

Abstract

The present study focuses on the formulation of a nano-additive emulsion by incorporating titanium dioxide nanoparticles into roll bite lubricants. The goal is to improve rolling force performance, enhance surface finish, and minimize oxide scale formation during hot-rolling of E410 steel. Experiments were conducted on a rolling mill using water, emulsion, and TiO2-based nanoemulsions at varying concentrations. Different machine learning algorithms such as Artificial Neural Network, Genetic Algorithm optimized ANN, Support Vector Regression, Genetic Algorithm optimized SVR and Backpropagation Neural Network were developed and evaluated to predict rolling force. Comparative analysis indicated that hybrid optimization-based models had better predictive ability, showing applicability of hybrid models to reliable prediction of rolling force. The lowest surface roughness obtained was 2.735 μm. This shows a decrease of 18.7% compared to the standard emulsion and 53.7% compared to water lubrication. The 0.1 wt.% nanoemulsion offers the best lubrication, and the rolling force during the first pass dropped to 33.59 tons, which is much lower than observed with the standard emulsion. The results show that TiO2-based nanoemulsions are an effective and environmentally friendly lubrication strategy, and machine learning can be an effective tool for process prediction and optimization in hot rolling.