Determination of Density and Viscosity of Crude Oil Samples from FTIR Data using Multivariate Regression, Variable Selection and Classification

Abstract

The use of Fourier Transform Infrared (FTIR) spectroscopy for quantification of crude oil properties was investigated using chemometric methods. Sample sets consisting of crude oil from seven different Canadian fields were analyzed. Different methods such as PLS, PCA, iPLS, and PLS-GA were used for model building and the results were compared. Evaluation of the models was conducted by determination of the coefficient of determination (R2) and cross validation error. The best results for quantification of density and viscosity were obtained by partial least squares (PLS) regression on FTIR data. Data analysis on the total sample set of 82 samples yielded a prediction error (root mean square error of cross validation) of 4.5 × 10−5 and 0.33 respectively for density and viscosity. Improvement in prediction accuracy of viscosity was obtained by using Decision tree classification on samples before applying PLS regression.