Análise de modelos de predição de viscosidade para óleos lubrificantes

Abstract

Models that estimate the resulting viscosity of blending fluids have been studied for centuries and were the object of study in this work. To work with the combination of fluids, mineral base oils, synthetic base oils, a viscosity index improving additive and fourteen lubricating oils were used, seven of which were of mineral origin and the other seven were of synthetic origin. The method used was divided into two stages, the first of them was the addition of the additive to the lubricants, in seven concentrations that varied between 0.5% and 9.5% m/m. After that, the density, dynamic and kinematic viscosity were measured, all of them at 100 °C, and these experimental values were compared with the theoretical values calculated for thirteen different viscosity blending models. Then, the mean absolute error, the root mean square error and the mean absolute percentage error were calculated to verify the model with the best performance. The second stage is similar to the first, but this time, instead of adding the additive, synthetic base oils were added to synthetic lubricants and mineral base oils were added to mineral lubricants. After carrying out measurements and calculations, it was found that there was no model with better performance than another when adding base oils to lubricants, due to the high deviation values found. However, the Arrhenius, Power law and Grunberg-Nissan models were the best in predicting viscosity when viscosity index improving additives were added to lubricating oils.