Artificial Neural Network Modeling of Properties of Crude Fractions with its TBP and Source of Origin and Time

Abstract

The objective of present work is to inculcate the effect of the sources of crude as one of the input parameters along with volume fraction, sulfur content & specific gravity of the crude on the estimation of mean average boiling point, molecular weights by developing ANN model. It is further extended to include the effect of time element on these properties of crude for one particular source of crude. Eleven sources of crude have been selected for first part of the work & for the one particular source twenty samples at different time elements have been used. The developed ANN models are observed to be with the average accuracy of prediction within +1 % .Based on the outcome of this demonstrative work, it can be concluded that ANN has a great potential in addressing to the estimation problems related to crude properties. The novel feature of the present work is incorporation of the origin of crude & time elements along with the other properties in the ANN model developed for the prediction of important parameters like mean average boiling point & molecular weight. It is sincerely felt that the methodology adopted in the present work be extended to more comprehensive data sets.