Near infrared spectroscopy employment in the petroleum feed stream characterization of an atmospheric distillation process. / Aplicação da espectroscopia de infravermelho próximo na caracterização da carga de petróleo para o processo de destilação atmosférica.

AUTOR(ES)
DATA DE PUBLICAÇÃO

2006

RESUMO

The knowledge of the petroleum composition and its products is an indispensable tool in a refinery for the process conditions settlement. The Near Infrared Spectroscopy (NIR) presents-itself as a helpful tool that attends to this need, because of its high potential and applications. In this context, it was developed the experimental methodology for NIR spectroscopy using two kinds of sensors: Transmission (Transmission system) and Attenuated Total Reflectance (ATR system). There were analyzed 63 petroleum samples by transmission system and 41 samples by the ATR system. The API degree of the samples ranges from 29,8 to 35,6. In the first step of this work, NIR spectra and SIMDIS chromatography data (Simulated Distillation) of the petroleum samples were obtained. From these data some samples were eliminated due to the presence of inconsistent points. In the second step, the analysis of the spectra OFF-LINE was carried out. In both systems, by choosing the region of 5600 to 6000 cm-1 as the spectral range which corresponds to the NIR overtone region for the hydrocarbons. Moreover, repeatability tests for the spectra were carried out and tests about the influence of the optical fibers on them. A pre spectrum treatment was carried out in order to become their information more sensible to the differences. Besides, the ON-LINE spectrum analysis by a closed LOOP were developed for the transmission system, with oil diesel and petroleum, in order to test the effect of the temperature and flow rate variations on the spectra. In this step, tests of repeatability at different temperatures were also carried out. In the following step, it was developed the Neural Network (RN) model. The entry variables of the data input are the random absorption spectra represented in 20 wavelengths with pre processing and the petroleum API degree. The net output variables are the weight percentages of the SIMDIS.

ASSUNTO(S)

caracterização de petróleo atr spectroscopy espectroscopia infravermelha infrared spectroscopy redes neurais simulated distillation - simdis destilação simulada - simdis petroleum characteristic neural networks espectroscopia atr

ACESSO AO ARTIGO

Documentos Relacionados