DEVELOPMENT OF A MODEL SIMCA PATTERN RECOGNITION FOR CLASSIFICATION OF DIESEL FUEL

Abstract

In the quality control of the diesel fuel used test procedures formally established as reference methods by different international organizations (ASTM, ISO, IP). It is known that these methods are accurate reference though, use a measurement time and require considerable, usually of a
significant amount of sample. The application of the techniques of pattern recognition chemometrics in the oil industry has made possible the development of alternative procedures that enable to carry out the classification and quality control of various types of fuels in a fast and with the use of small amount Sample from their spectral data. The aim of our study was to develop a model SIMCE pattern recognition to detect diesel fuel samples that are outside of technical specifications, based on the use of FT-MIR. Samples of diesel were processed in the spectral range between 4000-400 cm-1 with a resolution of 4 cm-1 and 32 sweeps per point. We used a total of 45 samples, 30 for the calculation of the model and 15 for validation. The model developed SIMCE has an adequate sensitivity and selectivity, and the ability to detect diesel fuels that are out of technical specifications in a fast, economical and reliable, and therefore may be a useful tool for quality control of this product in refineries of the country.