Determination of crude oil physicochemical properties by high-temperature gas chromatography associated with multivariate calibration. (15th May 2018)

• Record Type: Journal Article
• Authors: Rodrigues, Érica V.A.; Silva, Samantha R.C.; Romão, Wanderson; Castro, Eustáquio V.R.; Filgueiras, Paulo R.
• Is Part Of: Fuel. Volume 220(2018)
• Page Start: 389
• Page End: 395
• Publication Date: 2018-05-15
• Subjects: Crude oil -- HTGC -- PLS -- Multivariate regression; Fuel -- Periodicals; Coal -- Periodicals; Coal; Fuel; Periodicals; 662.6
• DOI: 10.1016/j.fuel.2018.01.139
• Languages: English
• ISSNs: 0016-2361

Abstract

Graphical abstract: Highlights: Physicochemical properties of crude oils were estimated by HTGC technique. PLS regression was applied in the modelling of physicochemical properties. The developed methods were applied in crude oil samples from wells in production. Abstract: Determining the physicochemical properties of petroleum is important for rapid decision-making during the production process. True boiling point (TBP) curve is the most important parameter in the petroleum characterization. However, the TBP can be estimated by high temperature gas chromatography (HTGC). In this paper, the HTGC technique associated with PLS regression was used to estimate API gravity, kinematic viscosity, pour point, carbon residue, saturated and aromatic content in crude oil. We use 98 samples with API gravity ranging from 11.4 to 54.0. Afterwards the developed methods were applied in nine samples from a field of production of the Brazilian coast. PLS model for API gravity, carbon residue, saturates and aromatics contents show root mean square error of prediction (RMSEP) set of 1.7, 0.83 wt%, 6.76 wt% and 4.05 wt% respectively. These models were applied to the nine samples and presented an exact equivalent to the models developed. The models for logarithm of kinematic viscosity and pour point show RMSEP of 0.31 and 12 °C respectively. Furthermore, tests applied in all models for evaluating the presence of systematic and trend errors indicate that there are no significant evidences of the presence of these types of error in the residues, at significance level of 5%.