Detection of discoloration in diesel fuel based on

Authors: Barbara Krakowska Ivana Stanimirova Joanna Orzel Michal Daszykowski Ireneusz Grabowski Grzegorz Zaleszczyk Miroslaw Sznajder

Publish Date: 2014/11/19

Volume: 407, Issue: 4, Pages: 1159-1170

PDF Link

Abstract

In the countries of the European Community diesel fuel samples are spiked with Solvent Yellow 124 and either Solvent Red 19 or Solvent Red 164 Their presence at a given concentration indicates the specific tax rate and determines the usage of fuel The removal of these socalled excise duty components which is known as fuel “laundering” is an illegal action that causes a substantial loss in a government’s budget The aim of our study was to prove that genuine diesel fuel samples and their counterfeit variants obtained from a simulated sorption process can be differentiated by using their gas chromatographic fingerprints that are registered with a flame ionization detector To achieve this aim a discriminant partial least squares analysis PLSDA for the genuine and counterfeit oil fingerprints after a baseline correction and the alignment of peaks was constructed and validated Uninformative variables elimination UVE variable importance in projection VIP and selectivity ratio SR which were coupled with a bootstrap procedure were adapted in PLSDA in order to limit the possibility of model overfitting Several major chemical components within the regions that are relevant to the discriminant problem were suggested as being the most influential We also found that the bootstrap variants of UVEPLSDA and SRPLSDA have excellent predictive abilities for a limited number of gas chromatographic features 14 and 16 respectively This conclusion was also supported by the unitary values that were obtained for the area under the receiver operating curve AUC independently for the model and test setsThe steadily increasing level in the consumption of petrol oil worldwide generates considerable profits for the petroleum industry and an increase in the price of petrol oil Apart from the economic factors the price of fuel is dependent on the local regulations that define the level of excise tax In general many countries apply different levels of excise duty on fuel depending on its usage For instance the diesel fuel that is used for heating purposes and in agricultural machinery in Poland has a rebated excise tax that is regulated by law in comparison to the diesel fuel that is used for regular transport In order to differentiate an oil product with respect to its usage a dye red dye Solvent Red 19 or Solvent Red 164 and a marker Solvent Yellow 124 are deliberately added 1 The presence of these specific excise duty components neither modifies the physicochemical properties of fuel nor limits its further usage The substantial financial reward that can be gained from the difference in excise tax has stimulated the illegal practice of removing the excise duty components from rebated fuel and selling it at a higher price This procedure is known as fuel “laundering” and can be done by an adsorption process using widely available materials The laundering of commercially available fuel causes a substantial loss in a government’s budget and therefore the development of analytical procedures for the detection of counterfeit diesel fuel is extremely necessaryDetection of fuel laundering specifically requires an analytical technique that is capable of revealing chemical changes in the composition of the fuel since the removal of the excise duty components does not influence its physicochemical parameters In our previous studies we developed an analytical methodology to detect any chemical changes before and after a simulated laundering process by using diesel fuel fingerprints that were obtained using fluorescence spectroscopy 2 3 It was confirmed that genuine samples can definitely be discriminated from samples after the laundering process However only limited information about the chemical composition of complex mixtures can be obtained from their fluorescence fingerprints and that is why gas chromatography coupled with flame ionization detection GCFID was investigated in this study Gas chromatography GC is one of the most popular separation techniques for studying complex petrochemical samples because the chromatograms contain comprehensive chemical information Gas chromatographic fingerprints 4 are widely used for monitoring quality and/or for identification purposes These fingerprinting techniques have also been accepted by the World Health Organization WHO for the quality assessment of herbal products 5 To the best of our knowledge up to the present GCFID has not been used for studying the laundering process The reason is that the excise duty components are not stable and degrade under a high temperature Their instability was studied and confirmed throughout our experiment when using GC with the nitrogen chemiluminescence detector sensitive to the presence of compounds containing nitrogen in their structures Furthermore the low concentration levels of excise duty components in diesel oil samples make the identification of their peaks among or under the peaks of major sample components difficult The larger the number of peaks the harder the separation and quantification of analytes with the GC technique is and this may provide to a failure in the characterization of excise duty components That is why the excise duty components are mainly determined with either spectroscopic or HPLCbased techniques In fact only the presence or absence of excise duty components is not indicative for a possible laundering process but the GCFID fingerprints may contain information about the overall chemical characteristics of samples before and after launderingIn general a comparative analysis of chromatographic fingerprints does not require the qualitative or quantitative evaluation of chemical components in samples but advanced chemometric techniques 6 are required On the one hand the costs of the analysis are greatly reduced because no certified reference materials/standards are required and while on the other important regions in the chromatographic fingerprints that are related to the phenomena being studied eg discrimination/classification of two or more groups of samples can be found using wellvalidated multivariate chemometric methods Once the important regions of chromatographic fingerprints are identified using a variable selection method 7 the corresponding fractions can be collected and further analyzed in detail using an orthogonal chromatographic system or a complementary analytical method A methodology that combines the fingerprint approach and chemometric analysis has gained popularity in many fields of science and technology in recent years including those such as the development of a method for the estimation of the total antioxidant capacity of green tea 8 the comparative analysis of extraction performance under different conditions 9 the analysis of secondary metabolites in citrus fruits peels 10 classification of petroleum products 11 etcIn order to investigate whether it is possible to detect diesel fuel laundering the excise duty components of a number of samples that were obtained from different suppliers in Poland were removed using an adsorption process Genuine diesel fuel samples and their counterfeit variants were analyzed using gas chromatography coupled with flame ionization detector Differences between these two groups of samples were studied using the partial least squares discriminant analysis PLSDA 12 The removal of baseline and the alignment of peaks were performed to the sample chromatographic fingerprints using penalized asymmetric least squares approach PAsLS 1 and correlation optimized warping COW 2 respectively In order to identify the key regions that are related to the chemical differences of sample groups the PLSDA approach was extended with variable selection In this study uninformative variable eliminationpartial least squares discriminant analysis UVEPLSDA 13 PLSDA combined with variable importance in projection VIP 14 and selectivity ratio SR were investigated 15 16 The effect of variable selection was monitored using a bootstrap procedure and the area under the receiver operating curve AUC and the sensitivity specificity and efficiency for the independent test set were adopted as figures of merit as wellA total of 31 samples of diesel fuel were collected from different fuel suppliers located in Poland in accordance with the sampling requirements that are specified in the PNEN ISO 31702004 norm The samples covered the majority of diesel fuel sources that are available for regular customers and crude oil used for production had a different geographical origin Poland Belarus Lithuania and The Netherlands Each investigated sample fulfilled the norm specifications and thus could be considered to be representative for a given batch of diesel fuel Prior to further analysis the samples were stored at room temperature ca 20 °CThe samples of diesel fuel were analyzed twice using a gas chromatographic system Agilent Technologies 6890N equipped with a flame ionization detector before and after the removal of the excise duty components Separation of the components of the mixture was performed using a RTX5 Restek column 60 m × 025 mm id and 025 μm film thickness with helium as carrier gas 13 ml min−1 constant flow rate gas purity 50 The following temperature program was used initial temperature 50 °C raised up to 320 °C by 3 °C per minute total analysis time 100 min Other settings of applied chromatographic method are as follows injection mode split split ratio 201 injection temperature 300 °C injection volume 01 μLEvery diesel fuel sample was subjected to a specific laboratory treatment that was aimed at removing the excise duty components the dye and marker The following laboratory procedure was applied to each genuine diesel fuel sample Ten milliliters of a sample was placed in a plastic test tube 15 ml with 2  of the adsorbent and shaken vigorously for 5 s Each test tube was shaken two to three times within a 30min period Afterwards each sample was centrifuged at 3500 rpms and the supernatant that was obtained was analyzed as is described in the “Registration of the GCFID fingerprints” section

Keywords: