Prediction of viscosity index and pour point in ester lubricants using quantitative structure-property relationship (QSPR)

Since ancient times, lubricants have been applied in different fields of technology and, from the very beginning, there has been a wide interest on improving some of their physical-chemical properties. The turning point in the development of lubricants came in the twentieth century, when modern synthetic ester base fluids were realized. In fact, with respect to “natural” lubricants (fats and mineral oils), they can be modified in order to optimize some specific technological properties; in particular, it is desirable that they present high viscosity index and a low pour point. Nevertheless, it is not always straightforward to accustom these parameters, and different theoretical studies have been pursued on this regard. Above all, valid tools to investigate these type of problems are the Quantitative Structure-Properties/Activity relationship studies (QSPA/QSPR). Starting from this considerations, the aim of the present paper is to investigate, by means of QSPR models, whether it can be possible to individuate or design ester base lubricants with some peculiar technological specificities. In particular, a QSPR analysis has been conducted in order to predict viscosity index and pour point on 41 ester lubricants by means of partial least squares combined with Leardi's genetic algorithms. The present study has provided satisfying results from the prediction point of view, and it has led to interesting conclusions from the interpretation viewpoint. In fact, it has highlighted that, the viscosity index and, to a lesser extent, the pour point, are highly correlated to the geometry, the molecular connectivity and the spatial autocorrelation of the investigated substances. © 2018 Elsevier B.V.