We will now discuss the design of carbon storage. The simulation tool used for this work was a streamline-based simulator that used a three-phase (hydrocarbon, water and solid) four-component (oil, Co2, water and salt) formulation. The transport equations were solved along streamlines, giving a computationally efficient solution method that allowed the effects of fine-scale heterogeneity to be captured accurately. Details of the simulation method are given in Qi et al. (2009) while Fig. 6.4 shows a schematic of the phases and components considered. This is a relatively simple model in terms of its geochemistry and phase behavior but is sufficient to study the key processes of interest for this discussion: capillary trapping and dissolution.
The relative permeabilities used in this study are those measured on water-wet Berea sandstone by Oak (1990). To accommodate trapping and relative permeability hysteresis we apply the model developed by Spiteri et al. (2008) that is based on pore-scale modeling studies.
We will now consider how to design an injection scheme in an aquifer that traps as much Co2 as possible. in the oil industry, it is standard practice to inject gas and water together or, more commonly, in alternating slugs – since the mobility of the combination of the two phases has a lower mobility than gas alone, leading to a more stable displacement and a more efficient sweep of the reservoir (Lake, 1989). We propose the same strategy here: water and Co2 are injected together to provide a more stable displacement, forcing the Co2 into more of the formation. Brine is then injected on its own to trap the Co2. at typical reservoir conditions, Co2 is less dense and much less viscous than water and will tend to rise to the top of the formation, channeling along high-permeability pathways and bypassing most of the storage space. on the other hand, we do wish
Phases (3) Components (4)
Hydrocarbon
Aqueous
Solid
CO2 Water Salt
Oil
6.4 The simulator used to design CO2 storage considers three-phase, four-component flow and reaction. CO2 partitions in both the hydrocarbon and aqueous phases, and can precipitate as carbonate. Oil is only found in the hydrocarbon phase, water can reside in both aqueous and hydrocarbon phases (it can partially evaporate in the presence of CO2), while salt dissolved in water can precipitate if CO2 evaporates sufficient water.
locally the saturation of Co2 to be as high as possible, since this allows more to be trapped.
Table 6.1 lists the properties used in the simulations presented in this section; we use conditions representative of a saline aquifer under the north Sea. Figure 6.5 shows the corresponding mobility ratio between injected Co2 and brine and resident brine as a function of the Co2 injection fractional flow – that is, the volume of injected fluid that is CO2. in this example, while the displacement is unstable (mobility ratio greater than 1) for pure Co2 injection, an injection fractional flow of 0.85 or lower allows a stable displacement that should lead to an improved sweep of the aquifer and greater storage. also plotted is the mobility ratio between chase brine in the formation and the injected Co2–brine mixture. This displacement is always stable. The reason for this is that we consider the mobility of chase brine with residual Co2: this residual phase greatly lowers the mobility. We can now consider an injection sequence where both floods are stable – both the injection of Co2 and the use of brine to trap the Co2 are stable displacements that should allow the Co2 to penetrate and be trapped in a large fraction of the aquifer pore volume.
Figure 6.6 shows a one-dimensional analytical solution to the governing transport equations compared to a numerical solution: the good agreement helps to validate the solver and enables us to understand the sequence of fluid fronts: the injected CO2 and brine move together, followed by chase brine. The chase brine front moves much faster than the Co2 and will soon catch up, trapping all the Co2. near the well (distance 0) the chase brine has dissolved the residual Co2. Hence, there is no possibility of CO2 escaping back through the well: there is no Co2 at all nearby, while all the Co2 is trapped further away.
Table 6.1 Parameters used in the simulations
CO2 viscosity 6 ¥ 10–5 Pa s
Brine viscosity 5 ¥ 10–4 Pa s
Temperature 80 °C
Reference pressure 27 MPa
CO2 solubility (mole fraction) at 27 MPa 0.0228
Porosity 0.15
1D simulation parameters
Darcy velocity 1/15 m/day
3D simulation parameters
CO2 injection rate 1.065 ¥ 106 kg/day Chase brine injection rate 1500 m3/day
CO2 density 710 kg m–3
Brine density 1050 kg m–3
Brine density saturated with CO2 1061 kg m–3
We now investigate this design in a heterogeneous three-dimensional geological model of an aquifer. The model is based on SPE10 (Christie and Blunt, 2001), a representation of a Brent-like reservoir in the North Sea that has channels, shale and over four orders of magnitude variation in permeability. Co2 and brine are injected into one corner of the model, while brine is extracted (and re-injected) from the other corner of the model.
Figure 6.7 shows some example results where the distribution of Co2 in this three-dimensional displacement is shown. The injected Co2 tends to rise to the top of the system and channel along high-permeability streaks. However, the chase brine injection broadly follows the same flow paths, trapping most of the Co2 and leaving only a small amount of mobile Co2 at the fringes of the plume. This indicates that even when heterogeneity and gravity are taken into account, our injection strategy can lead to a significant fraction of the injected Co2 becoming a trapped phase just a few years after the end of Co2 injection.
Figure 6.8 shows a compilation of the results for different injection fractional flows. The storage efficiency is the fraction of the porespace of the aquifer
Mobility ratio between CO2/brine mixture and formation brine
Mobility ratio between chase brine and CO2/brine mixture during chase brine injection
Mobility ratio = 1.0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 fgi
10
1
0.1
0.01
Mobility ratio
6.5 The mobility ratio between an injected CO2–brine mixture and formation brine as a function of the injection fractional flow of CO2, fgi. Also plotted is the mobility ratio between chase brine (including trapped CO2) and the CO2–brine injection mixture. It is possible to choose an injection fractional flow such that both displacements are stable.
Simulation Analytical solution Mobile CO2
Trapped CO2 Dissolution
front Advancing
CO2 front Chase brine
front
0.3
0.2
0.1
0
0 400 800 1200
Distance (m) Sg
6.6 Gas (CO2) saturation saturation (Sg) as a function of distance where brine and CO2 injection with a fractional flow of 0.5 for 1000 days is followed by 50 days of brine injection alone: the injection well is at distance zero. Near the injection well all the CO2 has dissolved. Beyond this there is the chase brine front that is moving much faster than the leading CO2. After 89 days the chase water will have trapped all the CO2. Numerical and analytical solutions are shown and are in good agreement.
0.33 0.30 0.27 0.23 0.20 0.17 0.14 0.11 0.07 0.04 0.01 Mobile CO2 saturation
Trapped CO2 saturation
2280 m 3200 m
170 m
2280 m 3200 m
170 m
Z Z
X Y
X Y
6.7 Saturation distributions near the injection well for a three-dimensional simulation with an injected CO2 fractional flow of 0.85: trapped CO2
(left) and mobile CO2 (right). Twenty years of CO2 and brine injection is followed by two years of chase brine injection.
filled with CO2, while the trapping efficiency is the fraction of the injected mass that is either dissolved or residual saturation. in this example, it is possible to trap the majority of the Co2 with a relatively modest amount of chase brine injection. The storage efficiency is controlled by two competing effects. The higher the injection fractional flow, the higher the initial CO2 saturation in regions of the field where the CO2 penetrates and hence more can be trapped locally. However, the lower the fractional flow, the more favourable the mobility contrast and the Co2 sweeps more of the aquifer.
The optimum is when the injection is first stable – fgi is around 0.85.