Multivariate statistics for summarizing diesel feeds for flammability attributes using comprehensive two‐dimensional gas chromatography

Abstract

Comprehensive 2D gas chromatography has been utilized for analyzing complex mixtures of hydrocarbons of diesel feeds. Here, we evaluated 19 diesel feeds for their paraffinic, naphthenic, and aromatic group compositions dictating their flammability properties. Compositional ranges of feeds were as follows: paraffins: 9.6–57.8%, naphthenes: 7.9–38.5%, and aromatics: 10.5–82.3%. Diesel's flammability performance is estimated by thermodynamic conditions and rates of radical formation of hydrocarbon type in actual engine condition, limiting cetane number. However, limitations are overcome by understanding the relative compositional variations of feeds by simple ranking of feeds based on C15‐16 compositions. Due to the multidimensional variability of feeds, a principal component analysis was adopted later for its distinguishing capability. Paraffinic, naphthenic, and aromatic group's principal component analysis clustered up feeds based on the higher concentration of individual hydrocarbon group. We explored hierarchical cluster analysis to organize feeds into classes of mixed C9 to C26 paraffin's composition in the diesel range. Further, for discriminating C15–C16 enriched and depleted feeds in total paraffin composition, a row dendrogram with heat map was drawn. The above multivariate methods have led to a fair distinction of nonadditive feed compositions influencing flammability properties by radical formation rate.

Recent Advances in GC×GC and Chemometrics to Address Emerging Challenges in Nontargeted Analysis

“…33 Machakanur et al used PCA to group diesel feedstocks by their hydrocarbon composition, followed by HCA to further investigate the compositional similarities of the fuels studies. 34 Also, it is common to use both unsupervised and supervised methods in the same study. Zou et al used PCA and HCA to investigate the clustering of dodecene products by catalyst type followed by downstream supervised tools.…”

“…The most intense peaks in a loadings plot are those that drive separation in the PC scores and can be further investigated as putative marker compounds. Other important unsupervised analysis tools include hierarchical clustering analysis (HCA) − and the recently introduced variance rank initiated-unsupervised sample indexing (VRI-USI) methods. , …”

“…Facanali et al used peak table-based PCA and HCA to investigate how seasonal changes affected the profile of Varronia curassavica essential oil . Machakanur et al used PCA to group diesel feedstocks by their hydrocarbon composition, followed by HCA to further investigate the compositional similarities of the fuels studies . Also, it is common to use both unsupervised and supervised methods in the same study.…”