On the phase behaviour of Athabasca vacuum residue + n-decane

Phase behaviour measurement and prediction for ill-defined hydrocarbon mixtures, such as vacuum residues, are subject to significant experimental and theoretical challenges. Consequently, despite the industrial importance of such fluids, experimental data sets are sparse with respect to the composition, pressure, and temperature ranges available and computational results are typically of poor quality. There are typically large differences between measured and predicted pressure–temperature and pressure–composition phase boundaries, and the numbers and natures of phases present are misidentified in many cases. Athabasca vacuum residue + n-decane mixtures exemplify these challenges. In the present work, computed phase behaviour results are based on the Peng–Robinson Equation of State. Parameters are identified using the Marrero and Gani group contribution method (J. Marrero, R. Gani, Fluid Phase Equilibria 183–184 (2001) 183–208 [13]) applied to pseudo components identified by Sheramata (J.M. Sheremata, Ph.D. Thesis, University of Alberta, 2008 [7]). Binary interaction parameters are estimated using the PPR78 method (J.-N. Jaubert, F. Mutelet, Fluid Phase Equilibria 224 (2004) 285–304 [15]) and a predictive correlation developed by Gao et al. (G.H. Gao, J.L. Daridon, H. Saintguirons, P. Xans, and F. Montel, Fluid Phase Equilibria 74 (1992) 85–93 [18]). At temperatures above 267 °C there is quantitative agreement between the measured and the predicted phase behaviour. L1V to L1 and L1L2V to L1L2 phase transitions are identified to within 5 °C and 5 bar and L-points (L1 = L2 + V) on the LV–LLV boundary are also predicted. At temperatures less than 267 °C, the breadth of the experimental LLV region shrinks dramatically with respect to composition. This change is not picked up in the calculations and there is a mismatch between the measured and the predicted phase behaviour over the composition range 35–65 wt.% vacuum residue, where the predicted phase behaviour includes L1L2V and L1L2 phase behaviours not observed experimentally. Even with this limitation, these proof of concept computational results provide a significant advance over current practice for ill-defined hydrocarbons, in general, and provide an accurate phase behaviour model for key deasphalting and other refining processes not previously available for Athabasca vacuum residue in particular.