Energy Convers. Mgmt WI. 36, No. 6-9, pp. 535-538, 1995 Copyright @ 1995 Elsevier Science Ltd 0196~8904(95)00061-5 Printed in Great Britain. All rights reserved 0196-8904/95 $9.50 + 0.00 UNDERGROUND STORAGE OF CO, IN AQUIFERS AND OIL RESERVOIRS T. HOLT, J.-I. JENSEN and E. LINDEBERG IKU Petroleum Research N-7034 Trondheim, Norway ABSTRACT Reservoir simulations of CO* injection into a water flooded oil reservoir show that significant amounts of oil may be recovered, and a high storage capacity of CO, is obtained also through displacement of water. Simulated storage capacities for CO, injection into an aquifer vary in the range 13-68% pore volume, depending on the prevailing displacement mechanisms. KFxwoRDs COz disposal; enhanced oil recovery; tertiary injection; aquifer storage; storage capacity INTRODUCTION CO, may be disposed underground by injection into oil and gas reservoir and aquifers. Although aquifers offer the largest storage capacity, the petroleum reservoirs may give a significant contribution The technical CO, storage potential in petroleum reservoirs, on world basis, corresponds to two thirds of the CO, produced by combustion of the reserves (Holt and Lindeberg 1992). Further, oil and gas reservoirs have a proven ceiling capacity and are usually well described, and for producing fields the existing infrastructure may be utilized. Injection of CO, into an oil reservoir increases the recovery of oil. In contrast to other methods for CO, disposal, enhanced oil recovery (EOR) by CO, injection is a storage alternative where CO, may have a positive economic value. Tertiarv g@ iniection Most known oil fields have been or are under water flooding, either by water injection and/or by natural aquifers invading the oil zone during depletion, Tertiary gas injection is gas injection after water flooding. During this process large amounts of water will be co-produced with the recovered oil. In practical operations water is often injected with the gas, either to reduce gas mobility and improve volumetric sweep, or to minimize the volume of valuable injected gas. This may lower the storage potential as the gas saturation is reduced by water. CO, is an attractive injection gas. This is demonstrated through a large number of field projects. Table 1 summarizes the results of 25 US tertiary CO, injection projects (full field and pilots, terminated and ongoing). The performance of the ongoing projects is estimates by the operators. Many of the projects are described by Brock and Bryan (1989) which gives references to the original literature needed for the calculations performed here. Additional and/or newer data are given by Beliveau (1991), Burbank (1992), Davis (1994), Flanders and DePauw (1993) Stein et al. (1989) and Wackowski and Masoner (1994). lhe average incremental recovery of these projects is 13.2% of original oil in place (OOIPj, and the average estimated disposal of CO, is 1080 Sm3/m3 stock tank oil. This agrees with data for some Canadian projects summarized by Todd and Grant (1993). 535 536 HOLT et al.: CO2 IN AQUIFERS AND OIL RESERVOIRS CO, disposed per volume of incremental oil is on average 2.3 reservoir m3/reservoir m3. This indicates a net reduction in the water saturation. The large amount of CO, stored is also due to oil recovered which is not defined as incremental. In the calculations underlying Table 1 it is assumed that all purchased CO, is retained in the reservoir , ie. insignificant Table 1 Summary of CO, field experience amounts are vented (if otherwise is not given). The figures above, and taking into account that the connate water saturation typically is 20-30% Field ncrm. oil COz utilization pore volumes (PV), show on average, that an (ZOOIP) M CFlSTB r.m3/r.m3 amount of CO, corresponding to more than 20% Dollarhide 0.7 19 2.4 of the reservoir PV is stored. 8 6.3 Fast vacum 2.1 Ford Geraldine 1.6 17 5 Aauifer storage of CQ, 11 4.5 7.1 Means 2.1 7.5 4.6 NortheastPurdy CO, disposed in aquifers may appear partly as a 0.9 6.8 2.4 Rangely free phase and partly as CO, dissolved in brine. 6.5 2.0 7.5 Sacrock 17 patterns The solubility of CO, in brine decreases with Sacrock 4 patterns 3.2 1.0 9.8 increasing salinity, but is nevertheless significant S.Wasson Clearfork 10 (e.g. 50 kg/m3 in 5% brine at 200 bar and 62’C, South Welch 7.6 Enick and Klara 1990). Stored CO, may also 8 2.8 Twofreds-east 15.8 react with the solid porous medium. The amount Wertz 10 4.5 10 of CO, consumed in this manner depends on the 14 6 2.5 Garber mineralogy in question, and may be of the same Little Creek 21 12.6 4.7 order as the amount dissolved in brine (Gunter et 8.4 2.5 Maljamar 13 al. 1993). Aquifer disposal of CO, is a relatively 3 Midale 20 little studied process, and no field experience on North Coles Levee 21 the displacement efficiency of CO, is available. Port Neches 19 Van der Meer (1992) performed reservoir Quarantine Bay 1.8 0.7 14.7 simulations of CO, injection into aquifers, and Rock Creek 7.1 2.1 7 calculated a storage capacity of two percent PV. Slaugther Estate 19.6 Only CO, stored as a free gas was considered in W eeks Island 19 3.3 1.0 the study. The importance of diffusion for CO, W est Sussex 8 10.3 3.8 storage in aquifers has not been thoroughly Lick Creek 2.8 1.7 13.3 treated in the literature. 14 W ilmington TERTIARY CO2 INJECTION A description of a Norwegian continental shelf oil reservoir was used to study CO, injection after periods of water injection of variable lengths, using numerical simulations. The reservoir is a high permeable sandstone with horizontal permeabilities typically in the range 100-2000 md with an average of 340 md. The ratio between vertical and horizontal permeabilities is determined to 0.004 as an average for all blocks, and to 0.04 for the ratio of average vertical and average horizontal permeabilities. The reservoir has a dip of lo”, and consist of three distinct zones with low permeability layers in between. A description of the reservoir, positioning of the wells, fluid properties and process parameters are given elsewhere (Lindeberg and Holt 1994). The simulations were performed in a fraction of the reservoir. This fraction originally contained 64 million Sm3 oil. It is assumed that simulated production profiles can be scaled up to full reservoir size. Various process combinations were simulated using a compositional reservoir simulator. These were 25 years of CO, injection and combinations of water and gas injection starting with 5,10,20 or 25 years of water followed by 25 years of CO,. For all scenarios fluid was injected downdip at a rate corresponding to 3.8% of OOIP per year. During CO, injection some of the layers had to be shut in due to restrictions on the produced gas/oil ratio (GOR). At the end of the simulations these layers were opened without HOLT et al.: CO2 IN AQUIFERS AND OIL RESERVOIRS 537 restrictions on the GOR. Incremental oil is in the following defined as the oil produced at the end of the CO, injection minus the oil produced after 30 years of watertlooding (giving a final water cut of 90%). The results of the simulations are given in Figure 1. It is seen that all the process combinations involving CO, resulted in almost the same amount of stored CO, in the reservoir, approximately 63% hydrocarbon pore volume (HCPV). This corresponds to 66% of the injected CO,. The amounts of incremental oil produced are slightly sensitive to the process combinations. With a recovery of 47.2% HCPV after 30 years of water injection, the total recovery of oil at the end of the processes varies from 69.3 to 73.1% of OOIP. These high oil recoveries indicate a favourable reservoir for CO, injection. The 63% HCPV of CO, stored corresponds to 50% PV, which is higher man the corresponding US reservoir values. The reason for this is the high oil recoveries, and most important, that CO, is not followed by water for me process combinations simulated here. AQUIFER STORAGE OF CO2 Aquifer storage of CO, was simulated with a black oil simulator using the same reservoir as for tertiary gas injection. Pressure and temperature were set to 200 bars and 62OC.CO, was injected through one well at the top of the reservoir. The vertical permeability of one low permeability layer was increased by a factor of 25. This changed the two ratios between vertical and horizontal permeabilities (see above) to 0.02 and 0.045, respectively. During the simulations the hydrostatic pressure gradient at the bottom end cross section was kept constant, and the injections were stopped when free CO, passed this plane. The relative permeability curves used are shown in Figure 2. Most of the data points are determined from core flood experiments were CO, displaced water in a 1.2 m Bentheimer sandstone at conditions corresponding to aquifer injection. Some of the data points are other IKU measured gas and water relative permeabilities of only 5 yr. my,. 2Qyr. 26 yr. waler co2 water wllie r waibr the same rock and are included based on the assumption that these permeabilities are Figure 1 Results tertiary CO, injection functions of own phase saturations only. 1.0 In the base case simulation 1.6% PV/year or 3900 ton/day CO, was injected with me well perforated through all layers. The sensitivity to various process conditions and parameters was investigated, and the results are given in Figure 3 and Table 2. In the base case simulations the reservoir was divided into 4800 grid blocks, and the Bentheimer relative permeabilities were used. Figure 2 Relative permeability curves As seen in Table 2 the CO, storage capacity depends strongly on the injection rate. At high flow rates the displacement is dominated by viscous forces and CO, flows rapidly through the most permeable paths, and the storage capacity reaches a constant lower limit. At lower rates me flood becomes dominated by gravity forces. The displacement front becomes stabilised, and after drainage of water increases the storage capacity further. The CO, storage capacity is also sensitive to permeability. Well perforations and gridding of the reservoir had less influence on the storage capacity of CO,. The CO, dissolved in brine is expressed in equivalent PV units in order to compare with me CO, stored as a free phase. The amounts CO2 dissolved are insignificant compared to the CO, stored as a free phase for the for the volumetric sweeps obtained in this study. HOLT et al.: CO2 IN AQUIFERS AND OIL RESERVOIRS 538 Table 2 Simulated CO, injection into aquifers well perforation all layers all layers all layers all layers all layers all layers top layer bot. layer 11 ,asecase 2) 13800 grid blocks 3) North Sea gas field relperm. -5 d -3 -2 -1 volume 0 1 2 InjectIon rate, log (% pore per year) Figure 3 Storage capacity vs. injection rate CONCLUSIONS Tertiary CO, injection will displace significant amounts of both oil and water giving a large storage capacity for CO,. Reservoir simulations on a real oil field resulted in 22-26% HCPV incremental oil, and more than 62% HCPV CO, was stored. The same reservoir produced 47% HCPV oil after 30 years of water flooding. CO, storage capacity of a heterogeneous dipping aquifer depends strongly on injection rate and permeabilities. An injection rate below 0.4% PV/year gave gravity stabilised displacement and high storage capacity (>30%). At rates 1.6% PV/year and higher the storage capacity became 16% PV, and was independent of rate. The storage capacity of CO, varied between 13 and 26% PV when the absolute permeability, the ratio between vertical and horizontal permeabilities and the relative permeabilities were varied at constant flow rate (1.6% PV/year). REFERENCES Beliveau, D., Payne, D.A. (1991). Paper SPE 22947. Brock, W.R., Bryan, L.A. (1989). Paper SPE 18977. Burbank, D.E. (1992). Paper SPE/DOE 24160. Davis, D.W. (1994). Paper SPE/DOE 27758. Enick, R.M., Klara, SM. (1990). _Chem.Ene.Comm.,.$!Q,23-33. Flanders, W.A., DePauw, R.M. (1993). Paper SPE 26614. Gunter, W.D., Perkins, E.H., McCann, T.J. (1993). &r .Conv.Mgmtt., &941-948. Holt, T., Lindeberg, E. (1992). EnetConv.Mgn&, 3,595-602. Lindeberg, E., Hoh, T. (1994); Paper SPE/DOE 27767. Stein, M.H., Frey, D.D., Walker, R.D., Pariani, G.J. (1989). Paper SPE 19375. Todd, M.R., Grant, G.W. (1993). ~conv.MgQ&,34.1157-1164. Van der Meer, L.G.H. (1992). J&x.Conv.Mgmt, N,611-618. Wackowski, R.K., Masoner, L.O. (1994). Paper SPE/TIOE 27755.