Hydrogen solubility in heavy oil systems: Experiments and modeling

Abstract

Hydrogen solubility measurements in heavy oils are required in order to develop accurate process models. Nevertheless, these solubility measurements are challenging at elevated temperatures and pressures and the amount of data points is scarce in the literature. This paper presents measured hydrogen solubilities in heavy oil systems at a temperature range from 498 to 598K and a pressure range from 2 to 11MPa. The experiments were conducted with a continuous flow apparatus. One of the well-characterized heavy oil systems was a hydrocracked vacuum gas oil and the second system consisted of a modified vacuum residue from Urals crude and toluene. The modified vacuum residue and toluene mixtures were prepared gravimetrically (mass fractions of vacuum residue: 0.25, 0.34 and 0.50). The experiments demonstrated that increasing the partial pressure of hydrogen and temperature increased the hydrogen solubility. Another finding was that the amount of toluene in the system had great impact to the hydrogen solubility. Four modeling approaches were compared based on their predictions on the hydrogen solubility in heavy oil systems measured in this work and four heavy oils found from the literature. The chosen models were PC-SAFT, Peng–Robinson, a simple correlation based on the corresponding theory and a method based on the Scatchard–Hildebrand theory. PC-SAFT with applied a heavy oil characterization method and the correlation based on the corresponding theory were found to predict the hydrogen solubility equally well and accurately. The benefit of using PC-SAFT instead of the simple correlation is that with PC-SAFT, phase behavior of multicomponent systems can be predicted and other properties, such as densities, can be obtained simultaneously. Peng–Robinson with a single carbon number characterization method overestimated the hydrogen solubility in the studied heavy oils and the method based on the Scatchard–Hildebrand theory could model the hydrogen solubility well after parameter regression.