Uso de espectroscopia no infravermelho médio e cartas de controle multivariadas no monitoramento da qualidade de biodieseis de soja e fritura residual em misturas com diesel

Abstract

Biodiesel is a renewable fuel, whose the addition of 5% biodiesel (v/v) to the diesel fuel sold in Brazil has been compulsory since 2010. This fuel can be adulterated with various less costly products and even contaminated during shipment and storage. Therefore, this study used mid-infrared spectroscopy (MIR) allied with multivariate control charts based on net analyte signal (NAS) to monitor the quality of biodiesel/diesel mixtures containing 5% biodiesel (v/v). The biodiesels in this study were produced from soybean oil and used frying oil using the methyl and ethyl routes. Three charts were developed for each type of biodiesel: the NAS chart referring to the analyte of interest (biodiesel); the chart of the interferents related to the contribution of all the other components in the sample (diesel); and the chart for the residues referring to non-systematic variations in the spectra. Based on the samples within the quality specifications, that is, the set of calibration samples (78 samples), statistical thresholds were established for each control chart. In order to validate the model, nineteen sets of validation samples were constructed (total of 1,544 samples), one set containing 31 samples within the quality specifications; one set of 39 samples containing less than 4.5% (v/v) of biodiesel; one set of 38 samples with more than 5.5% (v/v) of biodiesel; five sets of samples (118 to 120 samples by set) where the biodiesel was partially replaced (5.0% to 30.0% (v/v)) with vegetable oils (soybean, corn, sunflower, canola) and used frying oil, three sets of samples (78 to 120 samples by set) where the diesel which was partially replaced (5.0 to 30.0% (v/v) with petroleum products (used motor oil, kerosene, and gasoline), and eight sets of 640 samples which were adulterated with between 5.0 and 30.0% (v/v) of the adulterants cited the diesel and biodiesel samples. From these sets, it was seen that the model can properly monitor the samples with respect to biodiesel content of the mixtures as well as the presence of adulterants therein, since the samples within the specifications were within the established statistical limits, and the samples which were outside of specifications exceeded the limits on at least one of the three charts; of 1.662 samples analyzed in total, only two of calibration was erroneously monitored. Thus, the results obtained were satisfactory, with the correct monitoring of more than 95% of the samples for each of the constructed models.