The
objective of the present study is to develop robust statistical
models for the prediction of critical diesel properties such as cloud
point, pour point, and cetane index with composition inputs such as n-Paraffins, Iso-paraffins, Naphthenes, and Aromatics (PINA)
obtained by flow modulated two-dimensional gas chromatography with
flame ionization detection (GC×GC-FID). A single gas chromatographic
measurement coupled with models to predict the key physical properties
is attractive for refiners to make quick decisions in optimizing diesel
blending. We present a partial least-squares (PLS) linear regression
statistical model that has been developed using 41 data sets of diesel
samples with different compositions, out of which 33 samples were
used for the calibration and eight samples for validation of the model.
The R
2 values obtained for cloud point,
pour point, and cetane index were 0.92, 0.93, and 0.92 with standard
deviations of 1.20, 1.50, and 0.40, respectively. The average relative
errors for predicted values of cloud point, pour point, and cetane
index are found to be 0.86, 1.02, and 0.25, respectively. The PINA
analyses of diesel and kerosene samples were carried out using flow
modulated GC×GC with flame ionization detection (FID). The technique
adapts reverse phase gas chromatography with two capillary chromatographic
columns; the columns differ in length, diameter, stationary phase,
and film thickness to get maximum peak resolution. The gravimetric
blends of high purity reference standards of paraffins, naphthenes,
and aromatic compounds (PINA) with variable carbon numbers were used
for identification and to draw the boundaries for group types. Monoaromatic
and polyaromatic content obtained for diesel and kerosene samples
by the flow modulated GC×GC method were comparable to the results
obtained by the High Performance Liquid Chromatographic (HPLC) method
as per IP 391 or ASTM D 6591. Repeatability and reproducibility of
the GC×GC analysis were performed for several samples to validate
the method. It has been found that the HPLC method for the determination
of aromatics content using a single calibration standard for each
type, such as mono-, di-, and polyaromatics, causes a small error
in the quantification in some of the samples as the refractive indices
of all the aromatic species present in the diesel and kerosene samples
vary depending on the addition of alkyl side chains; the presence
of heteroatoms such as sulfur, nitrogen, and oxygen; etc.