A review of chemometrics models to predict crude oil properties from nuclear magnetic resonance and infrared spectroscopy

Abstract

Quantitative prediction of crude oil properties (e.g. API gravity, kinematic viscosity, sulfur content) can be conducted using multivariate calibration models. The growth and the popularization of chemometrics allowed the analysis of larger data sets imply in models with better performance parameters. Nuclear magnetic resonance (NMR) and infrared (IR) spectroscopy are two analytical techniques that generate a high amount of compositional information. They have been recommended to predict crude oil properties, to reduce cost, time and volume of sample used to access oil quality. The objective of this paper is to provide a general overview of the application of 1H and 13C NMR and near (NIR) and mid (MIR) infrared spectroscopy techniques associate with chemometrics applied in characterization of crude oils. Among these four acquisition methods, the most used was 1H NMR. On the other hand, 13C NMR still is the least method applied, due to the low natural isotopic abundance of 13C that adds complexity and higher cost on this technique. The key point is that NMR and IR are well established to determine several physicochemical properties of crude oil. Regarding the regression methods, partial least square (PLS) is the most used. It combines simplicity and good performance for chemical data. Meanwhile, the application of nonlinear methods to predict properties that have not yet been well modeled can be explored.