Search

Partition (or distribution) coefficients determine the relative equilibrium concentrations of chemical constituents (or chemical tracers) in each of the phases of a multi-phase system under dilute conditions. The various fluid phases in a reservoir have differing transport properties (e.g. varying relative permeability) and to correctly interpret the behaviour of injected chemical tracers it is essential that accurate partition coefficients are known. In the context of carbon geosequestration or long-term storage of CO2, chemical tracers will be predominantly exposed to an environment consisting of supercritical CO2 and formation water as the main fluid phases. To estimate/simulate the reservoir properties relevant to injected CO2 tagged with chemical tracers, it is therefore necessary to incorporate high pressure/temperature CO2/water partition coefficients into any model/simulation. In this paper, we present a method to determine these partition coefficients for gaseous chemical tracers using a variation of the widely used EPICS (or equilibrium partitioning in a closed system) method. With this method, only the concentration in one phase (in this case, CO2) needs to be measured. We then report these values for a series of representative chemical tracers (i.e. krypton, xenon, sulfur hexafluoride, perdeuterated methane and R134a) at pressure/temperature conditions that have been previously used at the CO2CRC's Otway CCS demonstration project in Victoria, Australia. These values were generally lower than the corresponding Henry's Law coefficients at comparable temperatures. Experiments also examined the impact of adding CH4 to the system to mimic feedstock gas at the CO2CRC Otway project and provide data pertinent to scenarios where CO2 is injected into depleted CH4 gas fields. These values are compared with Henry's Law coefficients and another recently published set of high pressure/temperature partition coefficients. With computational simulations, we have shown that these differences are potentially significant and demonstrate their impact in three typical CO2 geo-sequestration scenarios (i.e. injection into two types of aquifers and injection into a depleted reservoir with a gas cap).

Future CO 2 storage projects will require Monitoring and Verification (M&V) operations at the CO 2 storage site to understand the behavior of the CO 2 plume, including the assurance that leakage of the CO 2 has not occurred. Current surface based monitoring technologies may be unable to yield sufficient resolution or accuracy. CO2CRC, in conjunction with its Australian partners, is developing the Otway Stage 3 Project to identify and validate sub-surface monitoring techniques and configurations as a key element of a risk-based M&V program in large scale CCS projects. Subsurface monitoring approaches will be tested on a plume of CO 2 from an array of monitoring wells. Primary monitoring methods will be pressure tomography and downhole seismic, although other modalities (gravity, electromagnetic) are also being considered.

In reservoir simulation upscaling is used to obtain much faster execution times. Upscaling for CO2 storage has additional challenges such as accounting for CO2 dissolution into the surrounding formation water. Here we describe a study of upscaling applied to a 15,000 t injection at the CO2CRC Otway Research Facility. For this study 11 different coarse reservoir model geometries were created with different cell sizes. For each of the 11 geometries, 5 different methods of absolute permeability upscaling were implemented before fluid flow simulations at fine and coarse scales were performed for both single phase tracer and pure CO2 injection. The simulation computing time varied from a couple of minutes to 4.5 h for the coarse models, compared to 48 h for the fine-scale model. The results were compared for volumetric analysis, plume dimensions, pressure observed and effects on boundary conditions. In our simulation model, the different coarse model results show reliable and consistent estimation of the plume behaviour and pressure distribution. We found that flow-based permeability upscaling combined with a 0.85 factor on the solubility parameter on a 15x15x1 m grid give a good result with a substantial improvement in execution time. This study indicates the limits of coarse-scale models for CO2 storage simulations, and illustrates how these limits can be largely overcome by a careful application of absolute permeability upscaling techniques.

As part of the CO2CRC Otway Residual Saturation and Dissolution Test, a series of field tests were conducted at their project site in Victoria, Australia, with the primary goal of developing and assessing methods for quantifying residual CO 2 saturation in a saline aquifer. This paper reports the outcome of one of these tests, a single-well “push–pull” tracer test that uses novel reactive esters (i.e., propylene glycol diacetate, triacetin and tripropionin). For this tracer test, the ester is injected (pushed) into the reservoir where residual saturation has been established using CO 2 , and maintained by pushing with CO 2 saturated water (to prevent changes in saturation due to CO 2 dissolution). The ester is partially hydrolysed by the formation water to yield multiple compounds (i.e., the corresponding alcohol and acid generated from the ester). During water production (pull) from the same well, these compounds will partition differentially between the residual supercritical carbon dioxide phase and water phase, leading to chromatographic separation. By modelling the concentration profiles of these tracers in production water samples and using the experimentally determined partition coefficients, we generate two consistent residual saturation estimates using two separate modelling techniques, one a simple finite difference simulation of the tracer velocity field and the other a standard multiphase simulation code.

Most polymers used in EOR exhibit shear thinning behavior. At least theoretically, shear thickening will improve sweep while shear thinning (pseudoplasticity) will impair it through exacerbating the velocity contrast and/or inducing instability. Despite this, the effect of pseudoplasticity on sweep has not been studied in detail. An in-house streamline simulator has been extended to handle polymer flooding with Newtonian and non-Newtonian behavior. Three main modifications were implemented. First, the polymer mass balance was solved along the streamlines. Second, a polymer multiplier was defined to account for polymers' viscosifying and thinning effects. Finally, an iterative approach was implemented to solve the pressure field. This is needed since the pressure depends on the aqueous phase viscosity, which for non-Newtonian fluids depends on shear stress, and hence the pressure itself. The simulator was then used to investigate pseudoplasticity effects on sweep and recovery in various reservoir models. Two cases were run. The first had a stable connate bank; hence, thinning does not induce instability and the only effect is velocity exacerbation. The second had an unstable connate bank; hence, thinning can induce instability. The results of this work prove the importance of taking polymers' non-Newtonian behavior into account for the successful design and evaluation of polymer flooding projects. This is because pseudoplasticity will impair sweep, which can deteriorate the whole economic picture of a polymer flood. Even if instability is not induced, more pore volumes will be needed, more water will be produced, and in light of a limiting water cut, less oil will be recovered; in other words, higher operations costs, higher processing costs, but less profit. Simulations carried in a 2D heterogeneous model suggest that for unconditionally stable flooding, the injection requirement will double from 2 to 4 pore volumes. In addition, in light of an 80% limiting water cut, a thinning fluid would decrease recovery by 2% of OOIP compared to a Newtonian fluid. Furthermore, if thinning does induce instability, shear thinning will reduce ultimate recovery. Simulations carried in a 2D heterogeneous model suggest that a pseudoplastic fluid would decrease recovery by 5% of OOIP compared to a Newtonian fluid.

We implement a novel up-winding scheme for the mobility calculation using the computed velocities in an adaptive finite element (FE) unstructured-mesh reservoir simulator. In the finite-element finite-volume (FEFV) numerical method, the pressure and transport equations are decoupled. The pressure is calculated using finite elements, and the saturation is calculated using finite volumes. Each element is shared between several control volumes -- three for triangles (2D) and four for tetrahedral (3D). Consequently, the saturations used in calculating the mobilities hence updating pressure - are unclear. Some researchers use the average value between the elemental control volumes, or the integration points of the finite elements. For two-dimensional radial flow, this does not produce accurate saturations profiles when compared to the Buckley-Leverett reference solution. In this paper, we present a new formulation to calculate the FE mobility. We use the velocity vector, which is piece-wise constant in first order elements, to find the upstream saturation—where the tail of velocity vector intersects an element. This novel approach produces more accurate saturation profiles than previous methods even with higher order methods. Then, we present some benchmark simulation results where we model vertical spontaneous imbibition driven by capillarity and gravity disequilibrium between a fracture network at the bottom of the simulation domain and the matrix. The results compare favourably with semi-analytical treatments of this problem and experimental measurements. The method presented better models multi-phase displacements in complex reservoirs using FEFV. It can, also, be easily implemented in current FEFV based simulators.