fax 01-972-952-9435. AbstractThe need for CO 2 emissions reduction at a large scale globally implies that CO 2 injection into the subsurface be undertaken in a greater variety of geological environments that has been the case previously. Often when the storage reservoirs are saline aquifers, exploration data for proposed injection sites are extremely sparse. The special behaviour of CO 2 -water/brine systems (mutual solubility and chemical reactivity) adds complex processes, such as dry-out, salting-out, chemical reactions to the dynamic model. Simulation in these situations is one of few means of assessing an injection site and testing various scenarios. The accurate description of physics and chemistry in numerical simulation tools is fundamental for understanding processes, as well as designing appropriate injection or mitigation strategies.We present simulations of CO 2 injection into saline aquifers with a fully compositional code that has been expanded and enhanced to include specific phenomena, such as dryingout and salting-out. The examples illustrate the importance of pre-injection studies, as the wrong injection strategy may severely impact injectivity, putting the project in jeopardy.

Two methods for calculation of three-phase compressible flow in a porous media using streamlines are presented. For simplicity, gravity and capillary effects are neglected. Introduction Various aspects of streamline computations have been reported in a number of recent papers. A review of the technique was given by King et.al.[7]. As many other authors [5][9][13], he emphasis on incompressible two-phase flow. Lately, however there have been a few contributions to the field addressing compressible flow [3][10]. In this paper, we consider three-phase compressible flow. The reason why compressibility has been neglected by most authors in the field, is that it represent a strong coupling between the pressure and the saturation equations. In a streamline method, the pressure is calculated first and defines the streamlines. The saturations are then propagated along those streamlines. To obtain a stable solution using an explicit finite difference method (FDM), there is a strong limitation on the time step length. Thus, it is crucial for the efficiency of the streamline method to establish a time stepping sequence where only a small number of pressure updates is needed compared to the number of time steps required by the saturation solver. While this has been done successfully for years for incompressible flow [4], it is still a challenge to obtain both accuracy and efficiency for compressible flow. We present two different approaches to handle the couplings between pressure and saturation. First, we describe a sequential IMPES type method, where we do additional steps for reducing the mass discrepancy error. Then an implicit method for both pressure and saturation along the streamlines is presented. The results are compared to the solutions obtained using an existing black-oil simulator. The Governing Equations Consider three-phase compressible flow in a porous media. For simplicity, the effects of gravity are neglected. Gravity can be accounted for using operator splitting [5] and introducing solvers for three-phase flow along gravity lines. This work is in progress, but will not be the issue of this paper. Also, capillary forces are neglected, since physical effects transverse to streamlines complicates the streamline approach. We will study a black-oil model, with three phases and three components. We will allow gas to dissolve in the oil phase while we assume the water and gas phase to consist of only water and gas respectively. The component conservation equations now read [11] ∂∂t(ϕSwbw)​∇⋅(bwfwv→t)=qw∂∂t(ϕSobo)​∇⋅(bofov→t)=qo∂∂t(ϕ(Sgbg​RsSobo))​∇⋅(bgfgv→t​Rsbofov→t)=qg(1) for water, oil and gas respectively. By summing up the component conservation equations, we obtain the pressure equation ct∂P∂t​∇⋅v→t=Q−b→⋅∇P(2) where v→t=λt∇P(3) is the Darcy velocity. Refer to the nomenclature for an explanation on notation.

Effective geological storage of CO2 can be accomplished through a number of trapping mechanisms. Physical trapping is achieved through either CO2 being trapped under a structural closure or CO2 made immobile in the pore space, as residual saturation, by capillary action. Geochemical trapping, which might be regarded as a more secure mode of storage, is achieved through dissolution of CO2 in formation water and precipitation of carbonates. The dissolution rate depends on surface contact and is generally enhanced by greater CO2 plume movement. During site selection, a potential injection well location is commonly evaluated with respect to the proximity to potential leakage features. This paper investigates requirements for separation distance between CO2 injection location and potential leakage features in highly permeable steeply dipping brine reservoir settings. Reservoir models are simulated with a compositional code and sensitivity analyses performed with variations in reservoir permeability, hysteresis effects, and formation dip. Trapping mechanisms, over a timescale of several centuries, are illustrated as key indicators for containment and storage performance. Study results suggest that the amount of CO2 trapped by dissolution and residual saturation is enhanced by a dynamically flowing plume. The simulation results demonstrate that the separation distance requirement typically envisaged in a worst-case reservoir geometry setting is commonly overly conservative, representing opportunity for further optimisation. Numerical simulation is useful in addressing the complex reality of flow dynamics such as hysteresis in footprint prediction. Understanding CO2 plume migration scenarios relative to potential leakage risks, under various key reservoir key properties, is essential in storage containment and capacity assessments for storage site selection and development.