Multivariate data analysis applied in the evaluation of crude oil blends

• Published in: Fuel (Guildford) 2019-03, Vol.239, p.421-428
• Main Authors: Sad, Cristina M.S., da Silva, Mayara, dos Santos, Francine D., Pereira, Laine B., Corona, Rayane R.B., Silva, Samantha R.C., Portela, Natália A., Castro, Eustáquio V.R., Filgueiras, Paulo R., Lacerda, Valdemar
• Format: Article
• Language: English
• Subjects: Blend; Cluster analysis; Crude oil; Data analysis; Data processing; Mixtures; Multivariate analysis; Outlier detection; Outliers (statistics); Physicochemical properties; Principal components analysis; Properties (attributes); Quality standards
• ISSN: 0016-2361; 1873-7153
• DOI: 10.1016/j.fuel.2018.11.045

[Display omitted] •201 crude oils of different wells contributed to the formation of two Blends.•PCA and HCA were used to monitor the production quality of blends.•Identification of outlier’s samples that may favor the quality control of blends.•Different wells in the blends composition caused a change in the oil profile. In this paper, monitoring of the physicochemical properties of crude oil blends during production stages is described. The data of the properties of crude oil blends were obtained by laboratory characterization, and then analyzed by principal component analysis (PCA), hierarchical cluster analysis (HCA), and Mahanalobis distance. Thus, the quality of the blends was monitored quickly with simple multivariate tools. The results indicate that a change in the contribution of different wells in the blends caused a change in the profile. The PCA demonstrated that in each period, the physicochemical properties in the blends contributed to verifying the spread of the data. The blends could be organized by HCA, and it was possible to identify outlier samples with different quality standards for the oil. This information is important because it allows checking the changes in the oil profile, which helps in making adjustments to improve the quality of the final product in the primary process.