Use of CO2 as Solvent during Steam-Over-Solvent Injection in Fractured Reservoirs (SOS-FR) Method for Heavy Oil Recovery

Abstract

Abstract Due to low efficiencies and the high cost of individual injection of steam and solvent for heavy-oil recovery, their hybrid applications have gained significant attention recently. Although numerous laboratory studies exist and there are a considerable number of field projects for sandstone environments, fractured carbonates lack technologies to drain matrix oil efficiently. An alternative methodical injection of solvent and steam was proposed and tested earlier (Babadagli and Al- Bahlani, 2008). This process applies steam initially to condition the matrix oil for succeeding solvent injection and injection of steam to retrieve solvent in the matrix and to recover additional upgraded oil. In this application, hydrocarbon solvents were tested. The present study uses CO2 as a solvent in this type of application. To clarify the physics of the process and to test the applicability of the method for different reservoir and injection conditions, we conducted a series of experiments by first injecting steam, followed by CO2 injection. In the third cycle, steam was injected again to produce upgraded oil in the matrix. The experiments were performed under static conditions (soaking sand and carbonates samples into steam or CO2 chambers) at different temperatures and pressures to determine optimal application conditions for mutual goals: heavy oil recovery and CO2 storage in the matrix. Introduction Excessive need and increasing oil prices forced the industry to focus heavily on unconventional resources. Heavy oil reserves, in particular, gained specific attention as an alternative hydrocarbon resource, yet they are still challenging cases and more research is required to ease the recovery from this type of reserve. A specific challenge is fractured and deep carbonate reservoirs containing heavy oil, where the main problem is to mobilize the heavy-oil in a tight matrix towards a high permeable fracture network. This requires the reduction of oil viscosity and interfacial tension and the best possible way to achieve this is by steam injection. Heat loss and generation costs are the main issues with thermal approaches. Although different forms of steam injection such as cyclic steam injection, steamflooding, and steam-assisted gravity drainage (SAGD) have been successfully applied in sandstones worldwide, carbonates have very limited field scale steam injection projects (Babadagli et al., 2009). In one of the rare applications, steam was injected from the crest to heat the matrix and collect the oil drained by gravity in the Qarn Alam field in Oman. Macaulay et al. (1995) reported a small primary recovery of 2% of stock-tank oil initially in place during the primary recovery, which can be raised to 20% by means of gasoil gravity drainage, which is thermally-accelerated by steam injection. This estimation is based on experimental and simulation works prior to the pilot test. Al-Shizawi et al. (1997) discussed the methods of heat monitoring in the same field for the same pilot project. Later studies reported an analysis of pilot test and further understanding of the physics of the process (Penney et al. 2005; Shahin et al. 2006; Penney et al. 2007) Matrix recoveries from fractured carbonates do not show very high amounts, typically due to low permeability, unfavorable wettability, and high viscosity. Babadagli and Al-Bemani (2007) performed an experimental analysis on Qarn Alam core samples and observed that the recovery can go up to 47% OOIP for the case with initial water saturation at 200 ºC. This recovery is expected to be lower in field conditions (27% in the Qarn Alam field), as reported by Shahin et al. (2006). Beyond this, the production rate is very slow in cases of gravity drainage, even if it is thermally accelerated. A great portion of the published work on heavy-oil recovery from fractured carbonates cover either pilot tests (only five pilots test results were documented as reported by Babadagli et al., 2009) or from numerical simulation attempts (Sedaee Sola and Rashidi 2006).