Using different evolutionary algorithms and artificial neural networks to predict the rheological behavior of a new nano-lubricant containing multi-walled carbon nanotube and zinc oxide nano-powders in oil 10W40 base fluid

Abstract

This study addresses the challenge of predicting and optimizing the viscosity of nano-lubricants containing Multi-walled Carbon Nanotubes and Zinc Oxide nanopowders suspended in 10W40 base oil. Accurate viscosity control is crucial for enhancing lubrication system performance. To achieve this, an artificial neural network based on the Group Method of Data Handling was developed, integrated with eight advanced evolutionary algorithms to improve prediction accuracy and optimize viscosity under varying conditions of solid volume fraction, temperature, and shear rate. The research bridges a significant gap by combining predictive modeling with multi-objective optimization, outperforming traditional artificial neural network methods. The use of advanced evolutionary algorithms enabled precise optimization of nano-lubricant properties, while the expanded parameter space provided deeper insights into the impact of operational conditions. The framework achieved a root mean square error of 13.569 and a correlation coefficient of 0.9965, highlighting its superior accuracy. Temperature was identified as the most influential factor, with a viscosity function margin of deviation of -0.88. Further optimization using a Genetic Algorithm determined optimal conditions of 1 % solid volume fraction, 55 °C temperature, and 875.577 s⁻¹ shear rate, resulting in an optimal viscosity of 32.722 cP. This study fills a critical gap in the literature, offering a novel framework for designing high-performance nano-lubricants and significantly advancing the field of lubrication science with improved prediction and optimization methodologies for industrial applications.