Obtenção e classificação de biodiesel proveniente de óleos vegetais empregando espectroscopia NIR associada à análise multivariada.

Abstract

Currently mankind lives in almost exclusive dependence on petroleum, a non renewable energy source, highlighting the need for an exponential increase in the world supply of renewable energy sources to meet the shortage of fossil fuels.At this juncture it is increasingly growing interest in biofuel production, various technological routes and obtaining a wide range of raw materials able to be harnessed, among them the biofuel, which makes it imperative to identify quality parameter for performance of this fuel. In this context, this paper proposed to evaluate predictive variables for classifying quality parameters of methyl biodiesel from vegetable oils, originated from different oilseeds (cotton, canola, sunflower and soy) and expiration date (2008, 2009, 2010, 2011, 2012 and 2013) aided by NIR spectroscopy and chemometrics.Construction of classification models 42 samples of cottonseed oil, 28 with expiry date and valid 14 were used. Classes of vegetable oils (canola, sunflower and soy) consisted of 30 samples with expiry date and valid 15 respectively. The biodiesel was obtained from synthetic oils, cotton, canola, sunflower and soybeans using methyl NaOH solution 16.7% (w / v) at a temperature of 50 ° C simultaneously with vegetable oils. Each sample of biodiesel synthesized was formed by three repetitions.The quality parameters of biodiesel used in the specification of biodiesel were composed by the acid number and viscosity. Transflectance measurements were performed in the visible infrared (VIS) 400 to 800 nm region and in the near infrared (NIR) from 1100 to 2500 nm with 0.5 nm resolution region.Data were subjected to analysis of pattern recognition using PCA (Principal Component Analysis), SIMCA (Independent Analysis by Class Analogy) and multivariate calibration method PLS (Partial Least Squares Regression) and MLR (Independent Flexible Modeling by Analogy Class).The model (PCA) applied to VIS spectra of different classes of biodiesel from vegetable oils was effective to observe the trend of class formation among these samples, both in relation to the origin of vegetable oil as the expiration date of the raw material used.The SIMCA model was able to classify more accurately the samples in the visible significance of allowing 10% to be used for identification of samples of biodiesel region.The tools used for classification of multivariate data VIS region 400-800 nm in quality control of biodiesel derived from vegetable oils are suitable to observe the promising trend of classification of samples related to acidity parameter, occurring in the NIR region 1100 2500 nm for the efficient performance of the viscosity parameter.Spectroscopy in the visible and near infrared region associated with multivariate calibration PLS and MLR that meets the requirements for determining the quality of biodiesel with respect to the class of vegetable oil and the duration of non-destructively.