Dynamic modeling of geological carbon storage in an oil reservoir, Bredasdorp Basin, South Africa

Geological carbon storage provides an efficient technology for the large-scale reduction of atmospheric carbon, and the drive for net-zero emissions may necessitate the future usage of oil reservoirs for CO2 projects (without oil production), hence, dynamic modeling of an oil reservoir for CO2 storage in the Bredasdorp basin, South Africa, was therefore conducted. Injection into the reservoir was for 20 years (2030–2050), and 100 years (2050–2150) to study the CO2–brine–oil interactions, with sensitivities carried out on reservoir boundary conditions. The closed boundary scenario experienced pressure buildup with a target injection rate of 0.5 Mt/year, and a cutback on injection rate progressively until 2050 to not exceed the fracture pressure of the reservoir. The CO2 plume migration was not rapid due to the reduced volume of CO2 injected and the confining pressure. The system was gravity dominated, and gravity stability was not attained at the end of the simulation as fluid interfaces were not yet flat. The open boundary reservoir did not experience a pressure buildup because all boundaries were open, the target injection rate was achieved, and it was a viscous-dominated system. In both cases, the dissolution of CO2 in oil and brine was active, and there was a growing increase of CO2 fraction dissolved in water and oil, a decline in gaseous mobile CO2 phase between 2050 and 2150, and active trapping mechanisms were structural trapping, dissolution in oil and water, and residual trapping. The study showed that boundary condition was very crucial to the success of the project, with direct impacts on injection rate and pressure. This pioneering study has opened a vista on the injection of CO2 into an oil reservoir, and CO2–brine–oil interactions, with sensitivities carried out on reservoir boundary conditions in a closed and an open hydrocarbon system in South Africa.

Within the South African context, studies have shown that South Africa has the potential of storing ~ 150 Gt of CO 2 , with the offshore basins showing greater prospects [35][36][37][38][39][40] , but it is unlikely that commercially available CCS will be in place in the country before 2030 due to the present challenges of doubts in policies around climate action, delayed pilot/test underground carbon injection projects, low-priced carbon rates, dearth of commercial or profitable case studies, and the acceptability of the public to carbon storage projects in regions hosting potential storage sites [41][42][43][44][45][46][47] .South Africa fulfils its energy needs principally from fossil fuels, with 77% of the nation's total electricity generation from coal-fired power plants and its power sector the 9th biggest emitter among power sectors in the world (~ 218 Mt of CO 2 ) (Fig. 1), though these are low emissions in comparison to developed nations, cumulative emissions from the next nine African nations are below it 48 .Preliminary studies have been carried out to identify and estimate the capacities of potential geological CO 2 reservoirs in South Africa 39 , with particular focus on the Zululand basin 49 , Algoa 36,50 , and the Bredasdorp basin [51][52][53] , with a pilot CO 2 injection test scheduled for 2023 54,55 .These continued inquests into underground carbon storage are a pointer that it will play a big part in the decarbonization of South Africa.
In the bid to attain climate neutrality, Denmark, currently the largest oil producer in the European Union (EU), is set to completely phase out the production of hydrocarbon, new oil and gas extraction permits have been cancelled and when the calendar reads 2050, the hydrocarbon valves will be shut off for good.Presently, subsidies are being provided to stimulate large-scale CCS projects 56,57 .This means more searchlights will be beamed on oil reservoirs (without oil recovery) to make them viable CO 2 sinks.Oil reservoirs present many advantages for CO 2 storage, such as the presence of below-surface and on-the-ground installations and equipment that can be tailored to CO 2 injection and storage (usually with some modifications), the presence of quality seal and establishment of caprock integrity which have held oil in place through geologic time, and availability of geological, hydrogeological, geophysical and engineering data for characterization of the reservoir and other petroleum system elements, among others 5,[58][59][60][61][62][63][64] .
Therefore, this research aims to consider an oil reservoir in the offshore Bredasdorp basin, South Africa, for CO 2 storage, with no enhanced oil recovery (EOR).The knowledge gained from the study will also be useful for reservoirs with pockets of unrecoverable oil due to economic, technical, commercial, or logistics reasons, and bypassed oil/unswept zones.

Methodology
With the aid of the equation-of-state CMG-GEM compositional simulator (2022.30v),dynamic simulation was conducted for CO 2 injectivity and storage in the oil reservoir, with a focus on CO 2 plume migration, active trapping mechanisms, and supercritical CO 2 -oil-brine interactions.The reservoir used for this study is a clastic type with shales, siltstones, and sandstone units, and it has a permeability range of 3-560 mD and average effective porosity of 14% 51 .For CO 2 solubility in the aqueous phase, Henry's law was employed [65][66][67][68][69] , with the formation and fluid values, as well as the composition of the oil used for the study presented in Tables 1, 2.
Two rock types were delineated in the reservoir based on permeability zones 70 , namely high (rock type 1) and low (rock type 2) permeability zones, with permeability cut-off set at 150 mD as the reservoir had prevailing permeability values of 100-560 mD 51 (Fig. 2).Given three-phase flow in the reservoir (oil, CO 2 , and water) and at connate water saturation, for both zones, there were liquid-gas relative permeability values 71 , and water-oil relative permeability values (generated from Corey correlation) [72][73][74][75][76] (Fig. 3).
Dynamic simulations were carried out to simulate the injection of 0.5 Mt/year of CO 2 (762,000 m 3 /day) in the oil reservoir for 20 years (2030-2050) with one injection well, and a further simulation period of 100 years (2050-2150) to study the CO 2 -oil-brine interactions.Two scenarios were considered based on the boundary conditions of the reservoir, namely (i) open system (all boundaries open); and (ii) closed system (all boundaries closed).www.nature.com/scientificreports/

Injection and pressure
The maximum allowable pressure (well bottom-hole pressure, BHP) in the reservoir was set at 30,000 kPa, equivalent to 90% of reservoir lithostatic pressure 77,78 .With the target gas rate set at 0.5 Mt/year (762,000 m 3 /day) and the inability of pressure to dissipate in the system due to closed reservoir boundaries, the plot of BHP against time (Fig. 4) showed a steadily rising gas rate until it reached the maximum rate allowable (653,649 m 3 /day) to not exceed the maximum pressure, then the injection/gas rate was cut back to ensure caprock was not damaged.The average reservoir pressure rose from 25,500 kPa at the onset of injection in 2030, peaked at 30,000 kPa in 2050 and this pressure was maintained from 2050 to 2150.BHP and Well block pressure dropped to 29,741 kPa in 2050 to 2150, as the pressure due to injection of CO 2 into the reservoir had ceased.
In any underground carbon storage project, a target amount of CO 2 is proposed for injection into the ground, usually accompanied by pressure buildup in the system.The attainment of this desired injection rate is dependent on the injectivity of the reservoir, as an injection rate that is too high can cause the downhole pressure to exceed the fracture pressure, inducing reservoir and seal fracturing, or reactivation of existing sealing faults, all leading to environmental concerns 64 .A typical project that has had to deal with this scenario is the CO 2 injection into the Tubåen Formation at Snøhvit, where a considerable rate of pressure at the early onset of the project led to cutting down of injection rates, and subsequent abandonment of injection into the formation when the pressure increase considerably approached the fracture pressure [79][80][81] .
CO 2 plume migration and active flow regime CO 2 plume did not migrate rapidly away from the wellbore due to confining pressure, and reduction in the volume of CO 2 injected into the reservoir with the inability to attain the intended constant injection rate of 0.5 Mt/ year (Fig. 5a,c).Supercritical CO 2 accumulated at the top of the reservoir due to buoyancy in 2150, rather than moving rapidly laterally (Fig. 5b,d) 82 .The system was gravity dominated, as counter-current movement ensued in the vertical route owing to the gravity segregation of CO 2 , water, and oil, and the CO 2 invasion into the reservoir being stable due to gravity stability and gravity exceeding viscous forces 83 , it is also pertinent to note that low injection rate is accompanied with less lateral spreading of gas 84 .Though the gravity segregation was reduced because of relative permeability [85][86][87][88] , with increasing residence time, the system will be fully gravity stable evidenced by fluid interfaces being flat, as interfaces are usually destabilized by viscous forces 83 .There was structural trapping under the caprock, and the occurrence of non-reservoir rocks also provided local compartmentalization in the reservoir, aiding the reduction of the rapid migration of the plume.

CO 2 dissolution in oil and brine
Displacement of oil by injected CO 2 is active in the reservoir, though only 71% of the oil can be displaced (Fig. 6a), the heavier hydrocarbon components (29%) are not dissolved into the gas phase.Oil mass density (Fig. 6b) ranged from 650 kg/m 3 in the CO 2 uninvaded zones to 699 kg/m 3 in the zones invaded by CO 2 , with water mass density (Fig. 6c) also increasing to 1031 kg/m 3 in blocks that CO 2 had invaded.This indicates active dissolution of CO 2 in the reservoir fluids, as formation fluids density increases progressively as they become enriched with CO 2 89,90 .www.nature.com/scientificreports/

Active trapping mechanism
Active CO 2 trapping mechanisms were structural trapping under the caprock (supercritical mobile), dissolution and residual trapping (Fig. 7a,c).At the end of injection in 2050, 74.9 billion moles (3.3 Mt) of CO 2 had been successfully injected into the reservoir, 63.7% was in the gaseous mobile phase, 27.7% had dissolved in oil, 4% was trapped residually and 4.5% had dissolved in brine (Fig. 7b,c).In 2150, 4.8% had dissolved in water, 35.9% dissolved in oil, 3.6% was trapped residually and 55.5% was in the gaseous mobile phase.
There was a growing increase in CO 2 fraction dissolved in oil and water and a decline in the residually trapped and gaseous mobile CO 2 phase.With further residence time, which is a core element of the CO 2 storage process [91][92][93] , this trend should continue.

Injection and pressure
With all boundaries open, there was pressure dissipation in the system.In Fig. 8, the average reservoir pressure was constant at 25,395 kPa from 2030 to 2150, maximum well bottom-hole pressure recorded was 31,000 kPa in 2050, which was below the lithostatic pressure of the formation (33,440 kPa).The pressure dropped to 25,290 kPa after cessation of injection, which was maintained till 2150.Well block pressure rose from 26,000 kPa in 2030 to 26,297 kPa in 2050 and dropped to 25,290 kPa from 2050 to 2150.Pressure dissipation in the system because of open boundaries also ensured the proposed gas injection rate of 762,000 m 3 /day (0.5 Mt/year) was fully attained, with no direct risks of caprock and reservoir fracture, or storage capacity destruction of the rock 94 .

CO 2 plume migration and active flow regime
There was gradual and active lateral migration of injected CO 2 away from the injection well (Fig. 9a-d), as there was no confining pressure in the reservoir.The system was viscous-dominated as increased injection rates were accompanied by increasing viscous forces 95 , though gravity was also active, as shown by up-dip movement (Fig. 9c,d) and vertical segregation of the CO 2 plume from the bottom of the reservoir due to buoyancy, with supercritical CO 2 accumulating and moving directly below the seal but did not get to the reservoir flanks (Fig. 9b,d).

CO 2 dissolution in oil and brine
Oil displacement by injected CO 2 was active in the reservoir, though only 70.5% of the oil could be displaced (Fig. 10a), the heavier hydrocarbon components (29.5%) are not dissolved into the gas phase.Oil mass density (Fig. 10b) ranged from < 640 kg/m 3 in the CO 2 uninvaded zones to 695 kg/m 3 in the zones invaded by CO 2 , with water mass density (Fig. 10c) also increasing from 1021 kg/m 3 in uninvaded zones to 1030 kg/m 3 in blocks that CO 2 had invaded.This is indicative of the active dissolution of CO 2 in the reservoir fluids, as formation fluids density increases progressively as they become CO 2 enriched 89,90 .www.nature.com/scientificreports/

Active trapping mechanism
Active CO 2 trapping mechanisms were structural trapping under the caprock (supercritical mobile), dissolution in oil and brine, and residual trapping (Fig. 11a,c).At the end of injection in 2050, 236 billion moles (10.4 Mt) of CO 2 had been successfully injected into the reservoir, 67.9% was in the gaseous mobile phase, 22.8% had dissolved in oil, 4.3% was trapped residually and 4.3% had dissolved in brine (Fig. 11b,c).In 2150, 4.7% had dissolved in water, 29.2% dissolved in oil, 4.2% was trapped residually and 61.7% was in the gaseous mobile phase.
There was a gradual decline in the gaseous mobile CO 2 phase and an increase in the CO 2 fraction dissolved in oil and water.This leaning should continue upon further residence time, and a marked increase in CO 2 dissolution in water occasioned by an influx of brine into the reservoir.

Conclusion
In this study, an oil reservoir in the offshore lying Bredasdorp basin, South Africa, has been considered for CO 2 storage, without enhanced oil recovery (EOR).The closed boundary scenario experienced a pressure buildup with a target injection rate of 0.5 Mt/year, and therefore a cutback on injection rate progressively until 2050 to ensure the reservoir and overlying seal were not damaged.Migration of the CO 2 plume was not rapid, due to the reduced volume of CO 2 that was injected and confining state of the reservoir, the system was gravity dominated but did not attain gravity stability at the end of the simulation.There was a growing increase of CO 2 fraction dissolved in water and oil and a decline in the gaseous mobile CO 2 phase between 2050 and 2150.In 2150, 4.8% had dissolved in water, 35.9% dissolved in oil, 3.6% was trapped residually and 55.5% was in the gaseous mobile phase.The open boundary state experienced no pressure buildup in the reservoir and the target injection rate of 0.5 Mt/year was achieved, and 10.4 Mt of CO 2 had been successfully injected into the reservoir.CO 2 plume  www.nature.com/scientificreports/migrated up-dip without getting to the reservoir flanks, it was a viscous-dominated system attended with gravity movement and segregation.With an increase in the density of formation fluids, the dissolution of CO 2 in brine and oil was active, active trapping mechanisms were structural trapping, dissolution in oil and water and residual trapping.There was a decline in the gaseous mobile CO 2 phase and an increase in CO 2 fraction dissolved in oil and water between 2050 and 2150.With further residence time, fractions of CO 2 dissolved in the oil and brine phases would increase, as well as residually trapped fractions, with the CO 2 gaseous mobile phase experiencing a continuous decline.Therefore, this study showed that boundary condition was key to the success of the project, as it impacts injection rate and pressure.

Figure 1 .
Figure 1.Core point sources of carbon dioxide emissions in South Africa are in the Free State, Gauteng, and Mpumalanga provinces hosting the coal mines and most of the coal-fired power plants.Emissions from oil refineries, gas-to-liquid, and coal-to-liquid firms are also covered here.Adapted after 48 .

Figure 2 .Figure 3 .
Figure 2. EW cross-sectional view of permeability distribution in the reservoir.

Figure 4 .Figure 5 .
Figure 4. Cross-plot of average reservoir pressure, well bottom-hole pressure, well block pressure and gas rate.

Figure 7 .
Figure 7. (a) CO 2 trapping mechanism, (b) fraction of CO 2 sequestered, (c) summary of volumes of CO 2 injected, and in different trapping mechanisms.

Figure 8 .
Figure 8. Cross-plot of gas rate, average reservoir pressure, well block pressure and well bottom-hole pressure.

Figure 9 .
Figure 9. (a) Gas saturation shown from the J plane after 20 years (at 2050), (b) gas saturation in 2150 shown from the J plane, (c) gas saturation shown from the I plane after 20 years (at 2050), and (d) gas saturation in 2150 shown from the I plane.

Figure 11 .
Figure 11.(a) CO 2 trapping mechanism, (b) fraction of CO 2 sequestered, (c) summary of volumes of CO 2 injected, and in different trapping mechanisms

Table 1 .
Reservoir and fluid values.