Desarrollo de un modelo SIMCA para la clasificación de kerosinas mediante el empleo de la espectroscopía infrarroja

Abstract

Recebido em 16/10/07; aceito em 29/11/07; publicado na web em 26/8/08DEVELOPMENT OF A SIMCA MODEL FOR CLASSIFICATION OF KEROSENE BY INFRARED SPECTROSCOPY. In the petroleum refining industry, the use of crude from several origins is frequent. This leads to a product of variable chemical composition during refining, hindering quality control. Therefore, it is important to develop classification models that help to better characterize those products. The objective of this study is to develop a SIMCA recognition pattern to classify kerosene using infrared spectroscopy data. The model permits to differentiate two kerosene groups with different chemical compositions, which was corroborated by mass spectrometry. Keywords: SIMCA model; kerosene classification; infrared spectroscopy. INTRODUCCIÓNEn el control de la calidad de los destilados medios del petróleo se emplean procedimientos de ensayo que han sido oficialmente establecidos como procedimientos de referencia por diferentes organizaciones internacionales; como por ejemplo; la American Society for Testing Materials (ASTM), la ISO (Internacional Standard Organization) y el IP (The Institute of Petroleum). Es conocido que estos métodos de referencia aunque son exactos emplean un tiempo de medición considerable y requieren, por lo general, de apreciable cantidad de muestra.Los métodos quimiométricos han tenido gran aplicación en la industria de refinación del petróleo; por ejemplo, los métodos de calibración multivariada para el desarrollo de procedimientos alternativos para la predicción simultánea de propiedades físico químicas en derivados como gasolinas, kerosinas, turbocombustibles y diesel. 1-9 La exactitud de estos procedimientos alternativos es comparable con la exactitud de los métodos de referencia; presentan además las ventajas de utilizar pequeñas cantidades de muestras y de que el tiempo invertido en el análisis es pequeño. Por otra parte los méto-dos quimiométricos de clasificación y reconocimiento de patrones han sido ampliamente utilizados en la detección de adulteraciones de combustibles, así como en el control de calidad de los procesos productivos. [10][11][12][13] En nuestro laboratorio se han desarrollado modelos de regresión multivariados para la predicción simultánea de propiedades físico químicas de kerosinas mediante el empleo de la espectroscopía infrarroja de rango medio (FT-MIR) y el método de mínimos cuadrados parciales (PLS). El establecimiento de estas metodologías analíticas alternativas no culmina con el desarrollo y posterior validación del modelo; es necesario establecer un procedimiento de control de su validez con respecto al tiempo, principalmente en la industria de refinación del petróleo, donde ocurren con frecuencia variaciones de la materia prima y de los propios procesos productivos, que originan cambios en la composición química de los productos terminados; y por ende, problemas con la predicción de los modelos. 14-16 Para minimizar estas dificultades relacionadas con la detección de situaciones anómalas originadas por mues...

A Morphological and Morphometric Dental Analysis as a Forensic Tool to Identify the Iberian Wolf (Canis Lupus Signatus)

“…For the multivariate analysis, a principal components analysis (PCA) was used to explore the extent to which variables studied were related, and a partial least squares discriminant analysis (PLS-DA) was used to classify the data by sex (males and females) and to identify the critical variables that allow such differentiation. The multivariate analysis was performed with Soft Independent Modelling of Class Analogies (SIMCA) (Sartorius Stedim Biotech, Göttingen, Germany) [ 19 ] and seven cross-validation groups were used to consider similar observations in the same group, validating the calculated latent variables. The data were centred and scaled (Unit Variance, UV), and the software was set to calculate the boundaries with 95 % probability [ 19 ].…”

“…The multivariate analysis was performed with Soft Independent Modelling of Class Analogies (SIMCA) (Sartorius Stedim Biotech, Göttingen, Germany) [ 19 ] and seven cross-validation groups were used to consider similar observations in the same group, validating the calculated latent variables. The data were centred and scaled (Unit Variance, UV), and the software was set to calculate the boundaries with 95 % probability [ 19 ]. Only variables that presented statistically significant differences between the sexes ( p < 0.05) were considered for PCA and PLS-DA analysis.…”

Introducing ATR-FTIR Spectroscopy through Analysis of Acetaminophen Drugs: Practical Lessons for Interdisciplinary and Progressive Learning for Undergraduate Students

“…The data was previously centered and scaled (unit vector), and the software was set to calculate the boundaries with 95% probability to counterweigh for any magnitude unbalance and/or variance that might exist. 12 This procedure allowed eliminating any weight contribution due to the variables or observations magnitude. For eluding any model overfitting, the final number of components was based on the autofitting cross-validation setting as suggested by the principal component analysis (PCA) module in SIMCA.…”

“…The SIMCA 15 software (Sartorius Stedim Biotech, Göttingen, Germany) was also used because this multivariate analysis software is very powerful and rather easy to use while tuning the different plotting options for understanding the samples’ behavior. The data was previously centered and scaled (unit vector), and the software was set to calculate the boundaries with 95% probability to counterweigh for any magnitude unbalance and/or variance that might exist . This procedure allowed eliminating any weight contribution due to the variables or observations magnitude.…”

Morphometric Analysis of Bite Mark Patterns Caused by Domestic Dogs ( Canis lupus familiaris ) Using Dental Wax Registers

“…Furthermore, principal components analysis (PCA) and orthogonal projection to latent structures and discriminant analysis (OPLS-DA) was used to interpret to what extent a breed and/or sex differs from each other, and to identify the critical variables that allow such differentiation. The multivariate analysis was performed with SIMCA (Sartorius Stedim Biotech, France) (Dago Morales et al, 2008). The data was previously centered and scaled (Pareto), and the software was set to calculate the boundaries with 95 % probability (Dago Morales et al) to compensate for any magnitude unbalance and/or variance that could exist.…”