The residue from vacuum distillation contains the major portion of the asphaltene fraction of the crude. This residue contains high concentrations of the Conradson Carbon residue and metal components. It also contains high levels of heteroatoms such as nitrogen and sulfur. Vacuum residue cannot be used as feedstock for catalytic cracking because its high metal content leads to catalyst deactivation. Solvent deasphalting of vacuum residue produces a demetallized oil (DMO) fraction of relatively low metal content and a heavier fraction containing the rest of the metals. The main variables of this process are: vacuum residue and solvent quality, solvent/charge ratio (S/C), temperature and pressure.

A thermodynamic model to predict DMO yield and properties of the two phases is developed based on Regular Solution-Flory-Huggins theory combined with Lee-Kesler three parameter corresponding state theory to describe liquid-liquid equilibrium that results when solvent and residue are mixed. In this work, the residue is divided into four fractions: saturates, aromatics, resins and asphaltenes (SARA). Asphaltenes are further separated into fractions by using a three-parameter gamma distribution function that gives better representation of its aggregation state. Densities and solubility parameters for SARA compounds are then calculated by applying several empirical correlations based on molecular weight data. The solubility parameters of the solvents are obtained with the Lee-Kesler three parameter corresponding state theory, which takes into account the variations with temperature and pressure. The densities of the solvents are obtained from a modified Hankinson-Brobst-Thompson correlation (mHBT) that considers the pressure effect (compressed state). Binary interaction coefficients between solvent and SARA “compounds” are used to achieve a better prediction of DMO yield and properties. These coefficients are calculated using a simple relation that only involves molecular weights with a pre-multiplying parameter fitted to an experimental value of DMO yield. The resulting model can be used to predict the deasphalting process' behavior at different solvent/charge ratios, temperatures and pressures. The developed model has been tested with pilot plant data for twenty vacuum residues, which proceed from different Colombian crude oils. The model adequately describes DMO yields and its variations produced by changes in temperature, pressure and solvent/charge ratio.