Desenvolvimento de método para previsão de parâmetros físico-químicos do óleo diesel comercial por cromatografia gasosa ultrarrápida - CG-UR e quimiometria
Carregando...
Data
2019-03-14
Autores
Zanão, Lídia Renata
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Resumo
O óleo diesel é o derivado de petróleo mais consumido no Brasil, principalmente no setor de transporte rodoviário, o qual predomina no país, e, por isso, garantir sua qualidade é imprescindível para o funcionamento adequado do motor e redução de emissões veiculares para a atmosfera. Os métodos oficiais de análise demandam bastante tempo e volume de amostra e, em um contexto onde a preocupação ambiental e econômica prevalecem, é interessante o desenvolvimento de metodologias que otimizem os processos, reduzindo tempo de análise e quantidade de amostra e reagentes. Motivado por essa problematização, este trabalho teve como objetivo o desenvolvimento de um método de cromatografia gasosa ultrarrápida com a máxima quantidade de picos cromatográficos possíveis e resolução global satisfatória para correlacionar as áreas destes picos com um banco de dados formado por resultados de ensaios oficiais dos parâmetros físico-químicos, a fim de criar modelos de calibração multivariada por PLS baseados nestas informações. As amostras utilizadas foram de diesel mistura B8 comercial e os parâmetros avaliados foram a massa específica, teor de enxofre, teor de biodiesel, ponto de fulgor e temperaturas de ebulição para amostras 10% destiladas (T10), 50% destiladas (T50), 85% destiladas (T85) e 90% destiladas (T90). Em seis meses, 637 amostras foram selecionadas de forma representativa por HCA e analisadas através do método cromatográfico otimizado, onde a separação cromatográfica ultrarrápida foi realizada em apenas 6,12 minutos com resolução global de 1,09 e possibilidade de identificação de toda a faixa de hidrocarbonetos presentes no diesel e dos ésteres de ácidos graxos do biodiesel. Este método foi validado e apresentou seletividade; linearidade adequada (r>0,99) sem falta de ajuste e heterocedasticidade; valores adequados segundo o MAPA de precisão intermediária e repetibilidade (menores que 5,0 e 4,8, respectivamente), de erro avaliados intradia e interdias (menores que 4,8 e 4,9, respectivamente) de limites de quantificação (52,62 μg/g) e detecção (15,95 μg/g); e robustez. 100 dessas amostras de óleo diesel (18 não conformes e 82 conformes) foram novamente selecionadas por HCA de forma a manter a representatividade do conjunto amostral disponível e facilitar a construção dos modelos PLS. Estes foram construídos utilizando as áreas dos 30 picos principais de cada amostra utilizada como matriz X e os dados dos ensaios normativos como matriz Y. Os melhores resultados foram com a utilização do pré-processamento autoescalado e o tratamento de normalização para as características temperaturas de destilação (T10, T50, T85 e T90); teor de biodiesel e teor de enxofre S10 e S500. Para a massa específica, foi utilizado o pré-processamento centrado na média e o tratamento de normalização. O modelo para o ponto de fulgor foi construído com apenas os 3 primeiros picos cromatográficos e o melhor resultado foi obtido centrando os dados na média e com transformação logarítmica. Em geral, os modelos apresentaram previsão dos parâmetros com boa correlação (rval>0,6033) e erros de previsão e padronizados um pouco acima, porém similares aos valores de repetibilidade e reprodutibilidade obtidos pelos guias normativos ASTM, EN e NBR e fornecem, com o uso de apenas 0,1 μL de diesel, a obtenção simultânea de 8 parâmetros físico-químicos de 5 amostras por hora, em contrapartida aos métodos oficiais, onde apenas a análise de 8 amostras são possíveis por dia. Isso torna o método atrativo e possível de ser utilizado como método de screening em rotinas laboratoriais, reduzindo o número de amostras que necessitam de análise por métodos oficiais e, com isso, diminuindo o tempo e volumes gastos de amostras.
Diesel fuel is the most consumed petroleum derivative in Brazil, mainly in the road transport sector, which predominates in this country. In this way, guaranteeing its quality is essential for the proper functioning of the engine and the reduction of vehicular emissions to the atmosphere. The official methods of analysis of each characteristic of diesel oil quality require considerable sample time and volume, and in a context where environmental and economic concern prevail, it is interesting to develop methodologies that optimize processes, reducing analysis time, quantity sample and reagents. This work was motived by this problematization and had the objective of developing an ultrafast gas chromatographic method with a maximum amount of chromatographic peaks and a satisfactory global resolution to correlate the areas of these peaks with a database formed by results of official tests of the physicochemical parameters in order to create multivariate calibration models by PLS based on this information. Commercial diesel B8 blend were used as samples and the parameters evaluated were the specific mass, sulfur content, biodiesel content, flash point and boiling temperatures for samples 10% distilled (T10), 50% distilled (T50), 85 % distilled (T85) and 90% distilled (T90). In six months, 637 samples were representatively selected by HCA and analyzed by the optimized chromatographic method, where the ultrafast chromatographic separation was performed in only 6,12 minutes with an overall resolution of 1.09 and possibility of identification of the entire range of hydrocarbons present in diesel and the fatty acid esters of biodiesel. This method was validated and presented selectivity; linearity (r> 0.99) without lack of fit and heteroscedasticity; values in according to MAPA guide of intermediate accuracy and repeatability (less than 5.0 and 4.8, respectively), of intraday and inter-valued error (less than 4.8 and 4.9, respectively) of quantification limits (52, 62 μg/g) and detection (15.95 μg/g); and robustness. One hundred of these diesel samples (18 non-conforming and 82 conforming) were again selected by HCA in order to maintain the representativeness of the sample set available and to facilitate the construction of PLS models. These models were constructed using the areas of the 30 main peaks of each sample as matrix X and the data of the normative tests as matrix Y. The best results were with the use of self-scaling pre-processing and standardization treatment for the characteristics of distillation temperatures (T10, T50, T85 and T90); biodiesel content and sulfur content S10 and S500. For the relative denstiy, the centered pre-processing and standardization treatment were used. The model for the flash point was constructed with only the first 3 chromatographic peaks and the best result was obtained by centering the data in the mean and applying logarithmic transformation. In general, the models presented a prediction of the parameters with a good correlation (rval> 0.6033) and predicted and standard errors slightly above, but similar to the values of repeatability and reproducibility obtained by the ASTM, EN and NBR normative guides. So, using only 0.1 μL of diesel, it is possible provide simultaneously 8 physico-chemical parameters of 5 samples per hour, in contrast to the official methods, where only the analysis of 8 samples are possible per day. This makes the method attractive and possible for been used as a screening method in laboratory routines, which will reduce the number of samples that need to be analyzed by official methods and will reduce the time of total analysis of all samples.
Diesel fuel is the most consumed petroleum derivative in Brazil, mainly in the road transport sector, which predominates in this country. In this way, guaranteeing its quality is essential for the proper functioning of the engine and the reduction of vehicular emissions to the atmosphere. The official methods of analysis of each characteristic of diesel oil quality require considerable sample time and volume, and in a context where environmental and economic concern prevail, it is interesting to develop methodologies that optimize processes, reducing analysis time, quantity sample and reagents. This work was motived by this problematization and had the objective of developing an ultrafast gas chromatographic method with a maximum amount of chromatographic peaks and a satisfactory global resolution to correlate the areas of these peaks with a database formed by results of official tests of the physicochemical parameters in order to create multivariate calibration models by PLS based on this information. Commercial diesel B8 blend were used as samples and the parameters evaluated were the specific mass, sulfur content, biodiesel content, flash point and boiling temperatures for samples 10% distilled (T10), 50% distilled (T50), 85 % distilled (T85) and 90% distilled (T90). In six months, 637 samples were representatively selected by HCA and analyzed by the optimized chromatographic method, where the ultrafast chromatographic separation was performed in only 6,12 minutes with an overall resolution of 1.09 and possibility of identification of the entire range of hydrocarbons present in diesel and the fatty acid esters of biodiesel. This method was validated and presented selectivity; linearity (r> 0.99) without lack of fit and heteroscedasticity; values in according to MAPA guide of intermediate accuracy and repeatability (less than 5.0 and 4.8, respectively), of intraday and inter-valued error (less than 4.8 and 4.9, respectively) of quantification limits (52, 62 μg/g) and detection (15.95 μg/g); and robustness. One hundred of these diesel samples (18 non-conforming and 82 conforming) were again selected by HCA in order to maintain the representativeness of the sample set available and to facilitate the construction of PLS models. These models were constructed using the areas of the 30 main peaks of each sample as matrix X and the data of the normative tests as matrix Y. The best results were with the use of self-scaling pre-processing and standardization treatment for the characteristics of distillation temperatures (T10, T50, T85 and T90); biodiesel content and sulfur content S10 and S500. For the relative denstiy, the centered pre-processing and standardization treatment were used. The model for the flash point was constructed with only the first 3 chromatographic peaks and the best result was obtained by centering the data in the mean and applying logarithmic transformation. In general, the models presented a prediction of the parameters with a good correlation (rval> 0.6033) and predicted and standard errors slightly above, but similar to the values of repeatability and reproducibility obtained by the ASTM, EN and NBR normative guides. So, using only 0.1 μL of diesel, it is possible provide simultaneously 8 physico-chemical parameters of 5 samples per hour, in contrast to the official methods, where only the analysis of 8 samples are possible per day. This makes the method attractive and possible for been used as a screening method in laboratory routines, which will reduce the number of samples that need to be analyzed by official methods and will reduce the time of total analysis of all samples.
Descrição
Palavras-chave
PLS, Diesel, Cromatografia gasosa ultrarrápida, Quimiometria, Controle de qualidade