The Peng-Robinson equation of state (EoS) was adapted using group contribution methods to model asphaltene precipitation from solutions of toluene and an n-alkane and from n-alkane diluted bitumens. A liquid-liquid equilibrium was assumed between a primary liquid phase and a second dense asphaltene phase. Bitumen was characterized in terms of solubility fractions: saturates, aromatics, resins, and asphaltenes. Critical properties of the saturates, aromatics, and resins were determined that fit their measured densities and compared well with existing critical property correlations. The saturate and aromatic critical properties were also tuned to fit asphaltene precipitation data from solutions of the saturate and toluene or the aromatic and heptane. Asphaltenes were divided into fractions of different molar masses using a gamma distribution function. EoS parameters for asphaltenes were determined that fit the measured densities, fit precipitation data for mixtures of asphaltenes, toluene, and heptane, and compared well with existing critical property correlations. The model successfully fitted and predicted the onset and amount of precipitation over a broad range of compositions, temperatures from 0 to 100 °C, and pressures up to 7 MPa. The model results were within the error of the measurements except for high dilutions with n-pentane.

The saturation pressure and solubility of propane in Athabasca bitumen, as well as the liquid phase densities and viscosities, were measured for temperatures from 10 to 50 °C. Equilibration proved challenging for this fluid mixture and required some experimental modifications that are discussed. Only liquid and liquid-vapour phase regions were observed at propane contents below 20 wt%. A second liquid phase appeared to have formed at higher propane contents. The saturation pressures, where only a single dense phase formed, ranged from 600 to 1,600 kPa and these were fitted with a modification to Raoult's law. Viscosities less than 210 mPa.s were obtained at a propane content of 15.6 wt%. All of the viscosity data of the liquid phase were predicted from the propane and bitumen viscosities using the Lobe mixing rule. Introduction The worldwide original oil-in-place (OOIP) of heavy oil and bitumen is estimated to be approximately 6 trillion barrels. A major part of these resources are in Canada (~36%) and Venezuela (~27%)(1). In Canada, steam-based methods are often employed to improve heavy oil recovery. However, the industry is starting to seek alternatives to these methods because they are energy intensive and are drawing heavily on the available water supply. Solvent-based recovery methods are a potential alternative capable of providing high recovery factors without substantial water requirements(2, 3). One option is the vapour extraction method (VAPEX), which is a solvent-based analogue of the Steam-Assisted Gravity Drainage (SAGD) process(4–7). VAPEX is implemented with a pair of horizontal wells: a production well at the bottom of the reservoir and a solvent injection well located directly above the production well(5), as shown in Figure 1. The vapourized solvent is injected through the injection well and a chamber of solvent vapour forms around the well. At the walls of the chamber, the solvent diffuses into a surface layer of the heavy oil and dramatically reduces its viscosity. The diluted oil layer is then mobile enough to drain down, under the influence of gravity, into the production well. VAPEX performance depends on the viscosity and density of the liquid phase that forms at the edge of the solvent chamber. In order to design and optimize VAPEX and other solvent-based processes, it is critical to be able to determine the diffusivity of the solvent in the heavy oil, identify the phases that form in the solvent and heavy oil mixtures at various temperatures and pressures, and determine the density and viscosity of the liquid phase. Other solvent-based processes (steam and solvent injection for heavy oil recovery and solvent extraction of oil sands) require similar data. Most research on VAPEX has focused on physical model experiments with light alkane solvents; particularly mixtures of methane and propane(3). However, mixtures of carbon dioxide and propane may be a more viable option. Currently, carbon dioxide is expensive, but costs are expected to decrease if environmental incentives to sequester carbon dioxide are introduced. Carbon dioxide may also be a better VAPEX solvent than methane because it is more soluble in heavy oil and reduces the viscosity more(8).

The solubility of pure carbon dioxide in Athabasca bitumen was measured and compared with the literature data. Multiple liquid phases were observed at carbon dioxide contents above approximately 12 wt%. A correlation based on Henry's law was found to fit the saturation pressures at carbon dioxide contents below 12 wt%. The saturation pressure and solubility of carbon dioxide and propane in Athabasca bitumen, as well as the liquid phase densities and viscosities, were measured for three ternary mixtures at temperatures from 10 to 25 °C. Two liquid phases (carbon dioxide-rich and bitumen-rich) were observed at 13 wt% carbon dioxide and 19 wt% propane. Only liquid and vapour-liquid regions were observed for the other two mixtures (13.5 wt% propane and 11.0 wt% carbon dioxide; 24.0 wt% propane and 6.2 wt% carbon dioxide). The saturation pressures for the latter mixtures were predicted using the correlation for the carbon dioxide partial pressure and a previously developed correlation for the propane partial pressure. The mixture viscosities were predicted with the Lobe mixing rule. Introduction In Part I of this work(1), mixtures of carbon dioxide and propane were identified as a potential solvent for the VAPEX process. At typical heavy oil reservoir conditions (pressure of ~1.2 MPa and temperature of ~10 °C), propane and butane have sufficient solubility to reduce the oil viscosity to a level where gravity drainage can occur in an economic time scale. However, propane and butane are expensive solvents and the success of the process depends on how much solvent can be recovered. As well, the VAPEX process operates below the saturation pressure of the solvent and, therefore, propane and butane cannot be used at higher reservoir pressures where they exist only in the liquid phase. Methane can be added to achieve the desired pressures(2). However, carbon dioxide may also be a better VAPEX solvent than methane because it is more soluble in heavy oil and significantly reduces the viscosity(3). Mixtures of carbon dioxide and propane may achieve the desired reduction in viscosity while minimizing the required propane volumes. Hence, there is an incentive to evaluate mixtures of carbon dioxide and propane as a VAPEX solvent. VAPEX performance depends on the viscosity and density of the liquid phase that forms at the edge of the vapour chamber. In order to design and optimize VAPEX and other solvent-based processes, it is critical to be able to determine the diffusivity of the solvent in the heavy oil, identify the phases that form in the solvent and heavy oil mixtures at various temperatures and pressures, and determine the density and viscosity of the liquid phase. Other solvent-based processes (steam and solvent injection for heavy oil recovery and solvent extraction of oil sands) require similar data. In Part I of this work(1), saturation pressures and liquid phase densities and viscosities were measured for propane and Athabasca bitumen. There are also considerable data in the literature for mixtures of carbon dioxide and crude oils. Simon and Graue(4) measured the solubility, swelling and viscosity of mixtures of carbon dioxide and nine different oils.

No abstract

Employing CO 2 as the non-condensable gas in the Vapex process is an attractive option that could provide environmental benefits of CO 2 sequestration along with improved Vapex performance. Mixtures of CO 2 and a hydrocarbon such as propane allow the solvent to be tailored to different reservoir conditions. To test potential solvent mixtures, the phase behavior and physical properties and physical model experiments are required. We have previously reported on the phase behavior, viscosity and density of the CO 2 -propane-Athabasca Bitumen systems (Badamchizadeh et al., 2008a,b). These results confirmed the ability of carbon dioxide and propane mixtures to sufficiently reduce Athabasca bitumen viscosity and were used to design the solvent compositions utilized in the physical model tests reported here.The experimental approach used in these tests was to use a fixed composition of the CO 2 and propane mixture as the Vapex solvent. The objective of this work was to evaluate the performance of this solvent in recovering the Athabasca bitumen. The experiments were carried out at room temperature in a physical model. In-line measurements of the density and viscosity of the produced oil were used to gain further insight into the mechanisms involved in the process.