Available online at www.sciencedirect.com Energy Procedia 37 (2013 ) 3663 3671 GHGT-11 The coupled OpenGeoSys-Eclipse simulator for simulation of CO 2 storage code comparison for fluid flow and geomechanical processes Katharina Benisch a *, Bastian Graupner b, Sebastian Bauer a a Institute of Geosciences, Kiel University, Ludewig-Meyn-Strasse 10, 24118 Kiel, Germany b Eidgenössisches Nuklearsicherheitsinspektorat ENSI, Industriestrasse 19, CH-5200 Brugg, Switzerland Abstract This paper presents a code comparison of coupled multiphase flow and geomechanical processes resulting from CO 2 injection into deep saline formations. The coupled simulator OpenGeoSys-Eclipse as well as Eclipse-Visage, GEM and OpenGeoSys are used for this purpose. Comparison of the results of the different simulators shows a strong dependence of the results on the grid discretization and the numerical methods applied. Further investigations of the one-way coupled geomechanical simulations shows that even if the multiphase flow results are nearly identical for the simulators used, the geomechanical response on the pressure build up can be different. 2013 The Authors. Published by by Elsevier Ltd. Ltd. Selection and/or peer-review under responsibility of of GHGT GHGT Keywords: coupled processes, geomechanics, CO 2 injection, code benchmarking 1. Introduction It is well known, that the technology of CO 2 capture and storage constitutes one method to reduce the atmospheric CO 2 concentration within the next decades. Especially deep, saline aquifers represent an important storage opportunity due to the large available capacity of these formations [1]. The injection of CO 2 into the subsurface will change the hydraulic, geochemical and geomechanical conditions in the storage formation and potentially in the overlying and underlying rock formations. It causes pressure increase in the storage formation as well as the surrounding reservoir depending on the boundary conditions and reservoir parameters [2,3], which will lead to changes in the stress field and therefore can cause rock deformation or even failure [4]. These processes will affect the flow regime due to porosity * Katharina Benisch. Tel.: +49-431-880-1347; fax: +49-431-880-6706. E-mail address: kb@gpi.uni-kiel.de 1876-6102 2013 The Authors. Published by Elsevier Ltd. Selection and/or peer-review under responsibility of GHGT doi:10.1016/j.egypro.2013.06.260
3664 Katharina Benisch et al. / Energy Procedia 37 ( 2013 ) 3663 3671 and permeability changes and hence also the distribution of the CO 2 phase. The deformation induced may also propagate to the land surface, and may thus be used for monitoring the CO 2 storage operation [5]. Thereby, the reservoir fluids and the reservoir rock interact, showing a coupled and non linear behavior, which can be assessed using numerical simulation codes [6]. The demand for risk assessment of geological storage of CO 2 has increased and led to an enhanced development of simulation codes capable of modeling multiphase flow and geomechanical processes (H 2 M processes) in a coupled manner [6]. An overview of suitable H 2 M codes is given e.g. by [7] and [8]. Additionally, investigations are done regarding site assessment and monitoring [9] or regarding the development of geochemical modules able to model CO 2 -brine-systems at appropriate pressures and temperatures [10]. Coupling between multiphase flow and geomechanical processes can be implemented in different ways. Basically, four forms of coupling are distinguished: fully, iterative, explicit and pseudo. Each method has advantages and disadvantages, depending on the problem type used for [11],[12]. Using iterative or fully coupled methods, both the pressure impact on the deformation process and the impact of stress changes on fluid flow are considered (two-way coupling). For the fully coupled method, the equations for fluid flow and geomechanical processes (e.g. pressure and displacement) are solved simultaneously, whereas in the iterative approach both processes are solved sequentially and iteratively in two equations systems, where coupling is achieved through coupling terms in the respective equations. Using explicit coupling, geomechanical feedback on the fluid flow is neglected and it is therefore also called one-way coupling. It can be used, if phase compressibility dominates rock compressibility [11]. Finally, coupling effects can be simplified and described by empirical correlations using pseudo coupling. The iterative coupling is the most common coupling method and has been applied to several established reservoir simulators [8, 13, 14]. In the following study, the numerical simulators OpenGeoSys, GEM, Eclipse100-Visage as well as the coupled simulator OpenGeoSys-Eclipse [15] are used for the investigation of H 2 M processes during CO 2 injection focusing on the impact of different coupling methods and of different numerical codes on the simulation results. First results of the comparisons performed are shown in this work. 2. Method 2.1. General balance equations Modeling coupled fluid flow and geomechanical processes considers two mass balance and momentum balance equations for each, fluid and solid [7]. The general equation for balance of fluid mass for fluid phase can be written as where is the porosity, S the bulk fluid density, u the fluid velocity and Q a source term. Taking into account Darcy s law and the fact that fluid velocity has to be considered relative to the solid for deformation problems: u r = (u-u s) for u s = solid velocity results in (1)
Katharina Benisch et al. / Energy Procedia 37 ( 2013 ) 3663 3671 3665 (2) For the coupled H 2 M simulation, the balance of solid mass is considered using Biots formulation for porous media. This produces the final form of the fluid mass balance as (3) for the fluid bulk modulus K, the solid grain modulus K s k, the relative permeability k r p, the gravity vector and the solid displacement u. Solid displacement u is solved from (4) where u = and = + p with the total stress and the effective stress. Mean pressure p = S w p w + S nw p nw, where subscript w is the wetting phase and nw the non-wetting phase with S w +S nw =1. 2.2. Simulation codes OpenGeoSys The open-source scientific software OpenGeoSys (OGS) is a finite element code for simulation of thermal, hydrological and mechanical problems in porous media [16-19]. It uses an objectoriented and process-oriented approach that allows the solution of partial differential equations for different physical problems using a generic object structure [16]. This includes also multiphase flow as well as geomechanical processes. In OGS, H 2 M simulations can be solved sequentially (iterative coupling) or monolithic (fully coupling). For the following study, only the iterative method is used. Fluid flow can be solved in a pressure-pressure or in a pressure-saturation formulation, using pressure of the non wetting phase p nw and the capillary pressure p c or pressure of the wetting phase p w and saturation of the non-wetting phase S nw as primary variables, respectively. In the geomechanical process, a poroelastic model is implemented using Biot formulation for solving solid displacement. Fluid pressure is then updated from the changed stress field. Porosity and permeability are updated through constitutive laws implemented [7]. GEM is a finite difference compositional simulator developed by ComputerModellingGroup [13]. It has a geomechanical module incorporated, which uses a finite element method and can handle multiple constitutive models. One-way and two-way coupling are implemented. For the two-way coupling, fluid flow and formation deformation are coupled together in a sequential manner. The geomechanical deformation feedback is expressed in the fluid flow calculations for each time step through changing parameters in the porosity function after [20] where porosity is a function of pressure, temperature and total mean stress. Using one-way coupling, the deformation terms in the porosity function are neglected. Eclipse100-Visage The advanced and comprehensive finite element code Visage has been coupled to the reservoir simulator suite Eclipse to consider geomechanical processes in reservoir management. Fluid flow in Eclipse is calculated first and the results are then transferred from Eclipse to Visage through the
3666 Katharina Benisch et al. / Energy Procedia 37 ( 2013 ) 3663 3671 interface ECL2VIS for defined time steps. Updates of porosity and permeability can be calculated by constitutive relationships, e.g. Kozeny-Carman. For the following study, the finite difference black oil simulator Eclipse 100 is used for the fluid flow calculations. OpenGeoSys-Eclipse The conventional reservoir simulator Eclipse has been coupled with OpenGeoSys to extend the capabilities for simulation of CO 2 injection including geomechanical and geochemical processes. As OpenGeoSys is structured using a process-oriented approach, Eclipse could be implemented as alternative flow simulator within the multiphase flow process of OpenGeoSys [15]. Results from the multiphase flow simulation in Eclipse are passed at each time step to OpenGeoSys where geomechanical processes are solved. The geomechanical calculations are solves as described in the OpenGeoSys part. 2.3. Model setup The injection of CO 2 into a saline aquifer induces a significant increase of the stress field near the injection well. In the following, isothermal short-term geomechanical processes closed to the injection area within the injection formation are investigated using a synthetical model based upon the simulation setup of [21]. The magnitude and dimension of changes in the stress field are analyzed assuming linear elastic behavior. Figure 1: a) Conceptual model set up modified after Goerke et al. (2011), b) Boundary conditions used for the simulation. An axis-symmetrical model is assumed (Fig. 1a), which has a radius of 200 m and a thickness of 6 m. CO 2 is injected in the middle of the model area through a well with a diameter of 0.4 m. The model is set to a depth of 1500 m. In order to allow for a precise comparison of the codes investigated, fluid properties are set constant (Tab. 1) and porosity and permeability are homogeneously distributed. Saturation-pressure and saturationrelative permeability relationships are represented using Brook-Corey functions. A linear elastic model is assumed for the geomechanical process. Further parameters used are listed in Tab. 2. Fig. 1b) shows the boundary conditions applied for the fluid flow and geomechanical model. For the fluid flow model, the outer model boundary is set to a constant pressure condition following a hydrostatic pressure gradient. Displacement is set to zero for the upper, lower and inner boundaries. An initial total vertical stress gradient of 22621.39 Pa/m is assumed. The horizontal to vertical stress ratio is set to 0.7.
Katharina Benisch et al. / Energy Procedia 37 ( 2013 ) 3663 3671 3667 The well on the inner boundary is completed through the entire model thickness. CO 2 is injected with a rate of 1 m 3 /d for 41 days. Table 1: Fluid properties used in the simulation. Value Unit Brine density 1173 kg/m 3 Brine viscosity 1.252 10-3 Pas Brine compressibility 0.0 1/Pa CO 2 density 848 kg/m 3 CO 2 viscosity 8.1 10-5 Pas CO 2 compressibility 0.0 1/Pa Table 2: Hydrological and geomechanical parameters used in the simulation. Value Unit Permeability 3.0 10-12 m 2 Porosity 0.26 [-] Residual water saturation 0.35 [-] Residual brine saturation 0.0 [-] Entry pressure 10000 Pa Brooks-Corey Index 2.0 [-] Young s modulus 2.0 10-11 Pa Poisson s ratio 0.3 [-] Rock density 2650 kg/m 3 Biot s coefficient 1.0 [-] 3. Simulation results For the investigation of H 2 M processes, the breakthrough of the CO 2 saturation at two different locations in the model area as well as the stress along a cross section of the model area are plotted, as shown in Fig.
3668 Katharina Benisch et al. / Energy Procedia 37 ( 2013 ) 3663 3671 Figure 2: 3D model area. The red dots show the monitoring locations for CO 2 saturation and tangential stress at 20 and 60 m distance from the injection well and varying depths. 3.1. Multiphase flow simulation Prior to the coupled H 2 M simulations, the simulation codes were compared for multiphase flow without any geomechanical effects, because multiphase flow is the basis for the geomechanical interactions. Multiphase flow is sensitive to the numerical methods employed and the grid discretization used, which allows to investigate these effects in detail. For this investigation, CO 2 saturation breakthrough curves are compared at selected locations (Fig. 2), which are shown in Fig. 3. For all simulators used, a steep rise in CO 2 saturation occurs after a lag time, which represents the time the CO 2 needs to migrate from the injection well to the observation point, followed by a slower increase in saturation due to a thickening of the CO 2 phase, as more and more CO 2 is injected into the formation. It can be seen, that the results of the different simulation codes fit best near the injection well and start to diverge with increasing distance. At 20 m distance, the results of all codes show nearly identical CO 2 saturation curves except for GEM, which yield lower long term saturations. At 60 m, the results of OpenGeoSys show a steeper saturation front than those of Eclipse and GEM, but corresponds very well to the long term results from Eclipse. This difference is due to the numerical methods used in the individual codes. OpenGeoSys thus shows a less dispersive behavior than both Eclipse and GEM. The results of GEM show lower CO 2 saturations for both locations. This is due to the fact that fluid properties in GEM cannot be set constant. Thus, CO 2 density is slightly increased in the GEM simulation and therefore the density driven upward flow of CO 2 reduced. As the model thickness is relatively small, the variations in density are small and the results of the GEM simulation therefore still comparable. The results of the multiphase flow simulations show that a fine discretization of maximum 1 m in x- direction is required for an exact code comparison. Differences as a result of different numerical methods are observed mainly at the CO2 plume front (60 m and more) indicating that the plume front is most sensitive to numerical methods.
Katharina Benisch et al. / Energy Procedia 37 ( 2013 ) 3663 3671 3669 Figure 3: CO 2 saturation breakthrough curves at 20 m (at 1503 m depth; left) and 60 m (at 1500.5 m depth; right) distance from the injection well. Results are shown for the different simulation codes with a maximum grid increment of 1 m in the x-direction. 3.2. Coupled H 2 M simulation In the following, simulation results of the one-way coupling are shown for the simulators Eclipse- Visage, GEM and OpenGeoSys-Eclipse comparing the stress evolution due to CO 2 injection. The injection induced pressure increase leads to an increase in the stress field of the reservoir. The resulting effective stress in x-direction along a cross section at 1503 m depth after 1 day of CO 2 injection is shown in Fig. 4. Figure 4: Effective stress in x-direction along a cross section at a depth of 1503 m after 1 day simulation time. The effective stress in x-direction is increasing towards the outer boundary corresponding to an increasing displacement in x-direction. Comparing the results of the different simulators shows a good agreement for Eclipse-Visage and GEM. A smaller effective pressure increase is observed for the coupled simulation code OpenGeoSys-Eclipse, although the pressure field is calculated in both cases by Eclipse and has been verified (see Fig. 3). As the implemented pressure-stress and stress-displacement relationships in OpenGeoSys, GEM and Eclipse-Visage are identical, the reason probably is a different usage of reservoir pressure for calculation of the effective stress field.
3670 Katharina Benisch et al. / Energy Procedia 37 ( 2013 ) 3663 3671 The presented comparisons of both the multiphase flow and the geomechanical processes show that differences in the results are observable arising either from the implemented numerical methods or the included geomechanical constitutive relationships. It could also be shown that the grid discretization is crucial for a good agreement of the multiphase flow simulations, which consequently also affects the geomechanical calculations. Acknowledgment This study is funded by the German Federal Ministry of Education and Research (BMBF), EnBW Energie Baden-Württemberg AG, E.ON Energie AG, E.ON Gas Storage AG, RWE Dea AG, Vattenfall Europe Technology Research GmbH, Wintershall Holding AG and Stadtwerke Kiel AG as part of the CO 2 -MoPa joint project in the framework of the Special Program GEOTECHNOLOGIEN. The authors thank all project partners in Kiel, Leipzig and Stuttgart and colleagues of the GEOTECHNOLOGIEN program (publication number GEOTECH-1992) for their help, assistance and efficient cooperation. References [1] IEA (2008) Energy technology perspectives. IEA report, Paris [2] Benisch K, Bauer S. Short- and long-term regional pressure build-up during CO 2 injection and its applicability for site monitoring. Submitted to the Intern. J of Greenhouse Gas Tech 2012. [3] Benisch K, Bauer S. Investigation of large-scale pressure propagation and monitoring for CO 2 injection using a real site model. Proceedings ModelCARE2011. IAHS Pub. 355, 2011. [4] Rutqvist J. The Geomechanics of CO 2 Storage in Deep Sedimentary Formations. Geotechnical and Geological Engineering 2012;30(3):525-551 [5] Gemmer L, Hansen O, Iding M, Leary S, Ringrose P. Geomechanical response to CO 2 injection at Krechba, In Salah, Algeria. First Break 2012;30(2):79-84. [6] Rutqvist J, Boergesson L, Chijimatsu M, Kobayashi A, Jing L, Nguyen TS et al. Thermohydrodynamics of partially saturated geological media: governing equations and formulation of four finite element models. Intern J of Rock Mech and Mining Sci 2001;38:105-127. [7] Hou Z, Gou Y, Taron J, Gorke UJ, Kolditz O. Thermo-hydro-mechanical modeling of carbon dioxide injection for enhanced gas-recovery (CO 2 -EGR): a benchmarking study for code comparison. Environm Earth Sci 2012;67:549-561. [8] Rutqvist J. Status of the TOUGH-FLAC simulator and recent applications related to coupled fluid flow and crustal deformations. Comp & Geosci 2011;37(6):739-750. [9] Bauer S, Class H, Ebert M, Feeser V, Goetze H, Holzheid A et al. Modeling, parameterization and evaluation of monitoring methods for CO 2 storage in deep saline formations: The CO 2 -MoPa project. Environ. Earth Sci. 2012, published online, DOI: 10.1007/s12665-012-1707-y. [10] Li D, Graupner BJ, Bauer S. A method for calculating the liquid density for the CO 2 H 2 O NaCl system under CO 2 storage condition, Energy Procedia 2011;4:3817-3824, ISSN 1876-6102, 10.1016/j.egypro.2011.02.317. [11] Tran D, Nghiem L, Buchanan L. Improved Iterative Coupling of Geomechanics With Reservoir Simulation. SPE Res Simul Sym (SPE 93244), Houston, 31 Jan.-2 Feb. 2005. [12] Longuemare, P., et al. Geomechanics in Reservoir Simulation: Overview of Coupling Methods and Field Case Study. Oil & Gas Science and Technology Rev. IFP. 2002; 57(5):471-483. [13] Computer Modeling Group. User s Guide GEM, Advanced Compositional and GHG Reservoir Simulator, 2011. [14] Khan S, Han H, Ansari SA, Khosravi N. An integrated geomechanics workflow for Caprock-integrity analysis of a potential carbon storage. In: SPE international conference on CO2 capture, storage, and utilization, SPE 139477-MS 2011. [15] Graupner BJ, Li D, Bauer S. The coupled simulator ECLIPSE OpenGeoSys for the simulation of CO 2 storage in saline formations, Energy Procedia 2011;4:3794-3800, ISSN 1876-6102, 10.1016/j.egypro.2011.02.314. [16] O. Kolditz, S. Bauer, A process-oriented approach to computing multi-field problems in porous media, J. Hydroinform. 2004;6:225-244. DOI 10.1007/3-540-26737-9_17.
Katharina Benisch et al. / Energy Procedia 37 ( 2013 ) 3663 3671 3671 [17] Wang W, Kolditz O. Object-oriented finite element analysis of thermo-hydro-mechanical (THM) problems in porous media, International Journal for Numerical Methods in Engineering 2007;69:162-201. 201. DOI: 10.1002/nme.1770. [18] Kolditz, O., Delfs, J.-O., Kalbacher, T. et al. OpenGeoSys: An open source initiative for numerical simulation of THMC processes in porous media, Environ. Earth Sci. 2012, DOI: 10.1007/s12665-012-1546-x. [19] Kolditz O, Goerke UJ, Shao H, Wang W. Thermo-Hydro-Mechanical-Chemical Processes in Porous Media: Benchmarks and Examples in Lecture Notes in Computational Science and Engineering, Springer, 2012. [20] Tran D, Settari A, Nghiem,L. New Iterative Coupling Between a Reservoir Simulator and a Geomechanics Module. SPE 78192, SPEJ 2004;9(3):362-369. [21] Goerke UJ, Park CH, Wang W, Singh AK, Kolditz O. Numerical Simulation of Multiphase Hydromechanical Processes Induced by CO2 Injection into Deep Saline Aquifers. Oil Gas Sci. Technol. 2011;66(1):105-118, DOI: 10.2516/ogst/2010032