**4.7 Effect of presence of gas (gas reservoirs or oil reservoirs with a gas cap)**

CO2 can also be injected into depleted gas reservoirs in order to produce additional gas, this is called enhanced gas recovery (EGR). The injected CO2 increases reservoir pressure which supports gas production. As in the case of oil reservoirs or indeed aquifers the caprock failure stress must not be exceeded. Natural gas is a mixture of various components (cp. Table 4); the exact composition varies with the location of the gas fields and it is determined by the original hydrocarbon generation (Dandekar 2006).

In the reservoir, the CO2 flood front mixes with the natural gas by dispersion and diffusion. In parallel to the CO2 – gas mixing process, CO2 also equilibrates with the formation brine, similar to the mixing processes occurring in deep saline aquifers. The main advantage of CO2-EGR is profitability as in CO2-EOR, and an optimum between additional gas production and CO2 sequestration needs to be found. There are several CO2-EGR pilot units where these processes are tested, e.g. in the Lacq demonstration project in southwest France, <sup>5</sup> 10 t of CO2 will be injected and stored in a depleted gas field at a depth of 4500m (Total 2011).

A thorough study of nine natural gas fields (including sandstone and carbonate reservoirs) concludes that the main trapping mechanism over millennial timescales is dissolution trapping. At most 18% of injected CO2 is stored as a solid mineral phase (Gilfillan et al. 2009) and mineral trapping is predicted to happen only for siliclastic reservoirs.

In the case of oil reservoirs with a gas cap, the mixing thermodynamics are a combination of CO2-gas mixing, CO2 dissolution in oil and CO2 dissolution in brine. These complex

Dissolution Trapping of Carbon Dioxide in

empirical-statistical expression (equation 8).

D = diffusion coefficient c = concentration

μCO2 = CO2 viscosity μH2O = H2O viscosity

increases slightly with pressure.

and subscript 2 CO2.

where

where

t = time z = depth

Reservoir Formation Brine – A Carbon Storage Mechanism 247

A limited number of measurements of the CO2 diffusion coefficient in water DCO2-H2O at high pressure have been conducted. Renner (1988) measured DCO2-H2O for 0.25 N NaCl brine at 311 K for a pressure range 1.54-5.86 MPa and recorded DCO2-H2O values in the range 3.07- 7.35 x <sup>9</sup> 10<sup>−</sup> m2/s. More measurements at atmospheric pressure were conducted and DCO2- H2O values between 1.8 x <sup>9</sup> 10<sup>−</sup> – 8 x <sup>9</sup> 10<sup>−</sup> m2/s (Mazarei and Sandall 1980, Unver and Himmelblau 1964) were reported. Based on these datasets, Renner (1988) developed an

> 2 2 2 2 0.1584 6.911 6391 *DCO H O* μ

Renner's analysis (1988) indicated that water viscosity and CO2 viscosity were highly correlated with the diffusion coefficient, but molecular weight of CO2, molar volume of CO2, pressure or temperature were not statistically significant. However Renner states in his paper and Renner's data show that DCO2-H2O increases with an increase in pressure. Therefore it can be expected that CO2 diffusion processes under CCS conditions are faster than at atmospheric pressure conditions – which is positive news for dissolution trapping as

Hirai et al. (1997) measured DCO2-H2O via laser-induced fluorescence at 286 K and 29.4 and 39.3 MPa (DCO2-H2O = 1.3 x <sup>9</sup> 10<sup>−</sup> and 1.5 x <sup>9</sup> 10<sup>−</sup> m2/s). Their results fit perfectly with the empirical equation (9) suggested by Wilke and Chang (1955). ι is an association parameter equal to 2.26 for water. The experimental data measured by Shimizu et al. (1995) (DCO2-H2O is approximately 1.8 x <sup>9</sup> 10<sup>−</sup> m2/s at 286 K and 9-13 MPa) is however 40% larger than predicted by equation (9). Hirai's data and the Wilke-Chang equation both indicate that DCO2-H2O

> ( ) ( ) 2 2 <sup>2</sup> 2 2 0.5 8 0.6 *D M CO H O* 7.4 10 ι

> > ( )

*p c* ε

*kM T*

2 2 <sup>4</sup> <sup>5</sup>

*CO H O k k r*

*D*

( )

μ

1 12 12 ,2 ,2 2 2

2 3

<sup>−</sup> <sup>=</sup> (10)

*k k r*

More recently, Mutoru et al. (2010) developed a semi-empirical model for calculating diffusion coefficients for infinitely diluted CO2 and water mixtures (equation 10) based on 187 experimental data points. The subscript 1 denotes CO2 and the subscript 2 denotes water. However, in case of water diffusing into the CO2 phase, subscript 1 denotes water

 μ

<sup>−</sup> = ⋅ (9)

*H O <sup>T</sup>* / *CO CO <sup>V</sup>* <sup>−</sup>

it minimizes leakage risks by absorbing the mobile CO2 faster in the aqueous phase.

 μ *CO H O* −

<sup>−</sup> = (8)

<sup>∂</sup> <sup>∂</sup> <sup>=</sup> <sup>∂</sup> <sup>∂</sup> (7)

2 2 *c c <sup>D</sup> z t*

processes are topic of current research (DaVega 2011). These mixing processes result in three-phase flow in the reservoir (oil, gas and water flow as separated phases); in addition it is possible that additional phases are formed (e.g. a second immiscible oil phase or a solid asphaltene phase) which can further complicate fluid dynamics at the pore-scale and in the whole reservoir. Depending on rock surface wettabilities CO2 dissolution into brine can be slowed down, e.g. in case of a water-wet surface water covers the rock surface, and an oil layer may separate the brine from the CO2 (Piri and Blunt 2005). This oil layer then essentially acts as a barrier through which the CO2 has to pass in order to reach the brine and to be stored there safely by the dissolution trapping mechanism.


Table 4. Typical composition of natural gas (McCain 1990). Apart from methane and ethane traces of medium sized hydrocarbons can be found. In addition, natural gas can contain significant amounts of H2S, CO2 or N2 – up to 90 mol% (Firoozabadi and Cheng 2010). Such non-hydrocarbon gases usually need to be separated out of the production stream in order to achieve sellable gas quality
