Determinação da composição de blends de petróleos utilizando FTIR-ATR e calibração multivariada

• Ano de defesa: 2011
• Autor(a) principal: Filgueiras, Paulo Roberto
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Federal do Espírito Santo; BR; Mestrado em Química; Centro de Ciências Exatas; UFES; Programa de Pós-Graduação em Química
• Palavras-chave em Português: Spectroscopy in the infra-red ray; Chemometrics; Calibration; Petroleum; Infrared Spectroscopy; Espectroscopia no infravermelho; Quimiometria; Calibração; Petróleo; Química; 54
• Link de acesso: http://repositorio.ufes.br/handle/10/6755

The exploitation of oil in various sedimentary basins gives rise to oils with variable chemical compositions in which displays great differences in their physical-chemistry properties, these individual characteristics can be maintained even after the blend. The composition of the oil is an indispensable necessity in a refinery for the adjustment of process conditions, because they define the amount of various fractions that can be obtained. In this context, methods of spectroscopy mid-infrared with attenuated total reflectance (FTIR-ATR) can be an effective alternative analytical methodologies to provide rapid, practical, not destructive sampling and easy to monitor the composition of oil. This way, this work is proposed to determine the composition of blends of oil, formed by four fields of oil producers by FTIR-ATR measurements. The modeling methodology is based on partial least squares interval (iPLS) and synergisms intervals (siPLS) assessed the errors generated in prediction of new samples. The results are promising indicating that the models best fits the oils with lower density. The absorption of water molecule in IR causes the emulsified water is interfering in a process modeling. The best results were presented by the algorithm with errors siPLS forecast of 1.5 to 1.6% and offset values around 0.99% for oils two fields with lower density. From the results, we develop a statistical model to be used in the mixing process oil in order to predict or determine the actual composition of Blend