document preview

Abstract
Sequestration of carbon dioxide in geological formations has drawn increasing consideration as a potential method to reduce the level of CO2 in the atmosphere, and therefore mitigate climate change. In particular, saline aquifers can potentially provide a large storage volume world-wide. It is essential to assess the risk involved in storing CO2 in the subsurface, and simulations of CO2 injection play an important role. Detailed simulations using a compositional simulator, which solves the equation of state for the fluids and calculates the partitioning of fluids between phases, is time consuming. It is therefore advantageous to use a simpler method for simulation, such as a modification of a black-oil simulator (designed for use in the oil industry), where fluid properties are input using look-up tables.

In this study, we have tested the accuracy of flow simulations of CO2 storage in saline aquifers using a black-oil simulator (BOS) compared with a compositional simulator (CS). A range of models was investigated: 2D, 3D and radial models, horizontal and tilted, and homogeneous and heterogeneous. On the whole the results compared well, although accuracy of the BOS depended on the type of grid used, being less accurate for radial models, where discretisation effects were evident.

In agreement with other studies, we found that the black-oil simulations were, on average, a factor of four faster than compositional simulations.

Introduction
CO2 capture and storage has attracted much attention in recent years as a method to reduce the greenhouse gas emissions to the atmosphere and thus meeting the requirements of the Kyoto protocol (December, 1997). Mechanisms to trap large amounts of CO2 include dissolution into the oceans, sorption by vegetation and geological sequestration. Several geological settings are envisaged as potentials storage sites which can be accomplished by the aid of oil and gas reservoirs, either reservoirs in production giving enhanced oil recovery, or abandoned reservoirs, non mineable coal seems, and deep saline aquifers. It should be noted that in all cases the CO2 should be stored as a supercritical fluid. The critical point of CO2 is 31.1°C and 73.9 bars, above which CO2 has a high density, like a liquid, but it still acts as a gas (IPCC, 2005). This is achieved by storing the CO2 at a depth of more than 800 meters. This study focused on CO2 storage in saline aquifers. It is important to be aware of the factors affecting CO2 migration in the saline aquifers, such as advection, buoyancy and thermal effects and the influence of reservoir heterogeneity (e.g. Ukaegbu et al, 2009). CO2 is retained in situ through four basic trapping mechanisms: stratigraphic and structural, solubility, residual trapping at the pore scale and mineral trapping.

Compositional simulation was originally developed for modeling enhanced oil recovery processes, such as miscible gas injection. More recently, compositional simulation packages have been extended to model CO2 injection into saline aquifers, including the mutual solubility of CO2 and brine and the increase of the density of brine with dissolved CO2 (e.g. CO2STORE module, Schlumberger, 2010). However, assuming the CO2 is pure and is always in a supercritical state, the simulation of CO2 storage is a simpler procedure than miscible gas injection. In this case, fully compositional simulation is unnecessary, and the PVT properties can be input using pre-calculated tables. One method of doing this is to adapt black-oil simulators for CO2 storage. In this case, oil is used to represent brine and hydrocarbon gas represents supercritical CO2.