Modeling Base Oil Properties using NMR Spectroscopy and Neural Networks

Abstract

A method to correlate average base oil structure with physical properties using an approach that combines NMR spectroscopy with an artificial neural network to develop a model is described. 13C NMR spectroscopy is used to characterize the average structure of the oil. The structural information derived from NMR is then modeled against the experimentally observed physical properties of an initial database of oils using a neural network. A number of properties have been modeled in this fashion, such as viscosity index, Noack volatility, pour point, an aniline point. This type of approach has led to an improvement in the relatively unexplored area of predicting base oil performance.