Pixel-Based Chemometric Analysis of Pre-Salt Crude Oils: Advancing GC×GC-TOFMS for Reservoir Characterization

Abstract

The chemical composition of crude oil provides clues about its origins, well dynamics, and reservoir performance. Target petroleum analysis using comprehensive two-dimensional gas chromatography coupled with time-of-flight mass spectrometry (GC×GC-TOFMS) has been widely used for individual identification and group-type analysis. However, untargeted analysis is sometimes faster and more effective, especially with large, complex GC×GC-TOFMS data sets. The pixel-based preprocessing approach to treating GC×GC-TOFMS data facilitates fast, easy exploration of regions or compounds in two-dimensional chromatograms, which are important for comparing complex matrices using chemometric tools. In this context, to help reservoir geochemistry researchers mitigate exploration and development risks, we investigated subtle differences in the light hydrocarbon compositions of crude oil. Fifty Brazilian crude oil samples from the Búzios presalt reservoir in the Santos Basin were evaluated. Total ion chromatograms were used to construct concatenated matrices, which were aligned, normalized, and subjected to multivariate statistical analysis. Unsupervised principal component analysis indicated minor differences between the samples, which may correspond to differences in the presalt geological formations. Supervised orthogonal partial least-squares discriminant analysis determined whether each sample came from the Barra Velha or Itapema formation; the Barra Velha formation contains lighter hydrocarbons than the Itapema formation. Additional exploratory analyses indicated slight differences among the oil samples, demonstrating that light hydrocarbons can be investigated at the molecular level using GC×GC-TOFMS high-throughput data. Pixel-based chemometrics thus proves to be a rapid, innovative approach to assisting fluid distribution in a petroleum reservoir and a faster alternative to current methodologies for processing and analyzing GC×GC-TOFMS data.