Search

Summary Broad availability of continuous pressure measurements from Permanent Downhole Gauges (PDGs) has revived interest in using Pressure Transient Analysis (PTA) to improve history matching of reservoir models. Reproducing pressure transient responses in reservoir simulations as a way of conditioning parameters and realizations of geological models, also known as geological well testing, is a rapidly growing R&D area with many applications recently reported. The main focus of this paper is integration of PTA into reservoir simulation. We present a workflow which combines time-lapse PTA, employing analytical, single-phase models, with smooth transition to numerical, multi-phase simulations. Application of this workflow is illustrated by a case study with horizontal well injection into a segment of a fractured carbonate reservoir. The value of using pressure derivative in history matching is demonstrated through the evaluation of a hypothesis of well connection to conductive faults. The analysis of pressure transients in combination with numerical simulations finally argued for absence of such connection for the analysed history period.

CO2 injection into saline aquifers causes fast pressure diffusion in large reservoir domains due to low water-rock compressibility. This gives a possibility to enlarge explored reservoir domain using pressure transient analysis (PTA). Installation of permanent downhole gauge (PDG) in the well injecting CO2 into the Tubaen formation of the Snohvit field provided pressure recording during three years of injection history with many shut-in periods. Analysis of time-lapse pressure transients from both injection and fall-off periods revealed unique pressure derivative trend indicating flow barriers. This inspired the use of analytical reservoir model to match both pressure transients and the whole injection history. Usually PTA provides input data for reservoir simulation reducing uncertainty in history matching. Another advantage of PTA models is highlighted in this study: capability of long-term injection performance forecast. The field case confirmed that a reliable model may be assembled based on interpretation of a single injection or fall-off pressure response (e.g. well test prior to injection), which provides accurate forecast. Such a well test, interpretation and forecast do not require in-depth reservoir characterization and simulation. The key point is accounting for major features of the reservoir governing injection performance such as aquifer size and flow barriers.

Water flooding of stratified reservoirs often resulted in non-uniform oil displacement and fast water breakthrough in high permeability layers. Increase in oil recovery may be achieved by improvement of oil displacement in different layers and getting simultaneous water breakthrough in production wells. A similar problem may arise when water flooding multiple reservoirs with constrains on total injection / production. Control of injection rate allocation may provide uniform oil displacement in a layered formation or multiple reservoirs. Constrains on local (per layer) and total (per formation) injection / production rates and on production period for each layer may be accounted for. In this study optimization algorithms have been developed to determine an optimal strategy enabling maximum possible oil production with minimum possible water cut from a group of non-communicating layers or reservoirs. A solution of the optimization problem was found using linear and discrete programming methods under an assumption of two-phase piston-like incompressible flow in the reservoir. An optimal strategy with maximum oil recovery is defined under constrains on local and total production rates, injection start-up time and duration of production period for each layer or reservoir. An analytical solution was also found for a partial statement of the problem, where injection rates remain constant over the whole production period, which in turn is the same for all layers. The general problem statement has flexible constrains available, while the analytical solution for the partial statement may be easily implemented and used without dimension (number of layers or reservoirs) limitations. Both solutions were coded and further tested at mechanistic reservoir models to confirm applicability and efficiency of the developed algorithms in the reservoir simulation practice. A two-dimensional cross-sectional model with non-communicating layers was set up to test efficiency of the optimization algorithms for a typical reservoir simulation problem where fluid / rock compressibility and relative permeability effects are accounted for. The reservoir simulation results have confirmed that the optimal solution remained in force and therefore the optimization algorithms may be successfully integrated in reservoir simulation workflow.

Field trials and laboratory tests of CO2-foam have shown some success. A number of CO2-EOR field trials with either miscible or near miscible gas injection have shown that foam can lower the mobility of CO2 and improve the macroscopic sweep efficiency. However, not much work about CO2-foam flooding in fractured chalk reservoirs with low matrix permeability have been published in the literature. During CO2-foam flooding of fractured chalk media, it is expected that the injected fluids will flow in the high permeability fractures where foam will preferentially be formed. This will increase the local resistance to flow and divert more of the CO2 and CO2-foaming agent solution into the low permeable matrix.This paper presents results from laboratory experiments and simulation studies carried out to investigate the effect of mode of injection on oil recovery during CO2-foam flooding of water flooded fractured chalk models at reservoir conditions. The fractured model was created by drilling a hole through the centre of the plug and homogeneously packing it with glass beads. The CO2-foam process was conducted by injecting pre-formed foam either horizontally or vertically into the fractured models at 340 bar and 55°C. Pure CO2 and WAG processes were also carried out on similar fractured models to establish base case behaviours at the same conditions.Laboratory results show that additional oil after water flooding is recoverable by injecting pre-formed foam either horizontally or vertically (from the top) in the CO2-foam processes when compared to the injection of pure CO2 and CO2-WAG at the same effective permeability and CO2 volume. In the simulation studies carried out with CMG's compositional reservoir simulator STARS on a similar fractured model used in the laboratory experiments, the results from the core floods were quantitatively confirmed. Additional simulation carried out in large matrix blocks on the field scale showed that CO2-foam can recover additional oil at reservoir conditions. However, the adsorption of the foaming agent can make the CO2-foam process inefficient.

This paper addresses simulation aspects of the CO2 injection into the porous media. The presented results are based on the mechanistic and laboratory scale simulations. These findings would lay out the basis for the field scale CO2 injection pilot.

A streamline simulation with capillary and gravity is presented. The black oil streamline simulator [3DSL 0.25 by R.P. Batycky, SUPRI-C group, Department of Petroleum Engineering, Stanford University, 1997] was modified to incorporate capillary effects. The new simulator has been applied to several simulation cases. We discuss the influence of different factors that may impact performance prediction of water floods. The capillary forces are demonstrated to stabilize the displacement front and increase oil production even in rather viscous-dominated cases. The new approach can provide a fast and a reliable tool for the simulation of water flooding in low permeable and heterogeneous oil reservoirs as well as for reservoir screening and upscaling studies.

The oil and gas industry has accumulated significant experience in carrying out improved/enhanced oil recovery (IOR/EOR) projects. The outcome of different IOR/EOR methods applied to the fields world wide is available in open data sources as well as internal databases of oil companies. Open data typically include general information about oil field, average reservoir and fluid properties (reservoir parameters), and efficiency of an IOR/EOR method applied. Statistical analysis of such data may be applied to evaluate IOR/EOR potential of a particular field. Since such analysis requires small amount of information on a field, it is suitable for screening of large number of fields or evaluation of new discoveries or acquisitions. A new statistical approach to predict efficiency of different IOR/EOR methods for particular reservoir parameters has been developed and tested on actual filed data. The approach utilizes multi-dimensional statistical analysis based on data clustering. The K-means clustering is used to partition a filed case database into clusters based on reservoir parameters. A set of six representative parameters (porosity, permeability, depth and oil density, viscosity and temperature) has been chosen based on parameter correlation studies described in literature. Visualization of the cluster analysis results is performed via projection of six-dimensional vectors into two-dimensional space using the principal component method. IOR/EOR potential for a new field case is evaluated into two steps: (1) association of the case with a nearest cluster utilizing the discriminant analysis and (2) multi-dimensional interpolation of recovery factor for different methods within the cluster. A quality control is carried out at all stages of the statistical analysis to confirm its reliability. A list of potential methods with an estimation of recovery factor classified according to the confidence index (a measure of reliability) is outputted as a result of the analysis performed. Described algorithm was coded and tested on actual field case databases. The tests have shown good quality and reliability of the results obtained at all stages of the analysis. The testing has also revealed that reliable evaluation of IOR/EOR potential is possible for databases containing at least ten field cases with a particular method. Application of the new approach may serve as an IOR/EOR compass when potentially efficient methods have to be identified. A database containing actual field data and/or results of laboratory experiments and reservoir simulations may be used as an input for such analysis.