Accurate prediction of solubility of hydrogen in heavy oil fractions

Abstract

The use of heavy oil is increasing due to the exhaustion of the light crude oil wells. Meanwhile, refineries try to intensify their capacity and process the heaviest fractions of petroleum to turn them into a more valuable form. To design and operate hydro-processes, it is of great importance to acquire information about the hydrogen solubility in heavy oils. However, there are some difficulties in the hydrogen solubility measurements and modeling. Consequently, prediction of hydrogen solubility in heavy oils in an accurate and easy way is essential. In this study, reliable model namely Adaptive Neuro Fuzzy Inference System (ANFIS) was developed for prediction of hydrogen solubility in heavy oil fractions based on experimental data gathered from literature. The accuracy of the proposed model was checked using various statistical parameters and graphical representations. The model exhibit overall R2 and AARD% values of 0.99242 and 3.4 which indicate accuracy and robustness of developed model. Result show that the developed model presents accurate results by reproducing the experimental values with an acceptable error.