You are on page 1of 8

Numerical Simulation of Oil Recovery by Polymer

Injection using COMSOL
J. Wegner*1, L. Ganzer1
Clausthal University of Technology
*Corresponding author: Agricolastraße 10, 38678 Clausthal, Germany,

Abstract: In this paper we used COMSOL reservoir. Later, due to oil extraction the natural
Multiphysics to model basic physico-chemical drive energy declines.
effects relevant in polymer enhanced oil To maintain production external energy
recovery (EOR) such as non-Newtonian sources, mainly water or gas are injected into the
rheology of the displacing phase, permeability reservoir for the purpose of maintaining reservoir
reduction, adsorption and salinity effects. pressure and improving sweep efficiency. This
COMSOL's PDE interface as well as Species phase is considered as secondary recovery.
Transport in Porous Media interface was used for With increasing energy demand and high oil
solving the underlying mass accumulation, prices sophisticated tertiary recovery
transport and auxiliary equations. The validity of technologies are implemented to extract more oil
the implementation is tested by comparing the from existing hydrocarbon reservoirs. Enhanced
results obtained with COMSOL with those oil recovery consists of the injection of
obtained from the commercial simulator chemicals, heat or miscible gases which are in
ECLIPSE Blackoil [1] which is a simulator used general not normally present in the reservoir.
in the oil industry. The results show good One of the chemical tertiary technologies
agreement indicating successful implementation. which can significantly increase the oil recovery
Due to its PDE Interface and Species Transport factor is termed polymer EOR. It involves the
in Porous Media Interface COMSOL mixing of special polymers with the injection
Multiphysics provides a flexible framework to water to increase water viscosity and reduce
implement physical and chemical mechanisms water permeability during the flow through
relevant to polymer EOR. Although commercial porous media. As a result the mobility of water is
software for simulation of polymer EOR decreased, leading in a more efficient
processes exists, it lacks the multi-physics displacement process and a higher oil recovery
capabilities of COMSOL. Underlying transport factor. However, several physical and chemical
and auxiliary equations can be varied or processes accompany the flow of aqueous
extended and other physics such as temperature polymer solutions through the porous formation
or geo mechanical effects integrated. This is an resulting in loss of polymer solution viscosity,
important aspect in research. hence, in lower oil recovery. Detailed
understanding of these effects is crucial to
Keywords: Enhanced Oil Recovery, Numerical successfully design polymer EOR projects.
Simulation, Polymer Flooding, Porous Media, The goal of this work is the COMSOL
Finite Element Method implementation of a numerical model capable to
simulate the recovery of oil by means of polymer
1. Introduction EOR taking into account basic physico -
chemical effects. The model will be the basis for
Oil extraction from porous subsurface further and more detailed investigations.
reservoirs can be classified into three different
stages, namely the primary, secondary and 2. The Polymer Flood Model
tertiary recovery. The latter one is also termed
enhanced oil recovery (EOR). A two phase, four component polymer EOR
Within the primary recovery phase the model is implemented into COMSOL to simulate
natural drive energy already available in the the displacement of oil by aqueous polymer
reservoir is utilized to produce the oil. Natural solutions. The oleic phase consists of a single
energy includes rock and fluid expansion, water component oil, while the aqueous phase contains
influx, solution drive, gas cap drive and gravity the components mobile water, polymer and salt.
drainage. During the primary recovery period no Although several complex physico-chemical
further energy needs to be injected into the processes influence the flow of aqueous polymer

2. but has an impact on the viscosity of the The equation for salt transport is written as: aqueous polymer solution.3. non-Newtonian rheology and salinity phase is: effects have an impact on the aqueous phase viscosity. A mathematical relation .) The model is one-dimensional.2 Transport Equations 3. In addition to The mass conservation equation for the oelic adsorption. additional constitutive  S o  k  equations are required to close and couple the     o po   q~o .3 Cross Couplings where p .) Polymer adsorption reduces the relative required.     C p a pa   q~aC pw  a  7. (2) system of equations (1) – (5).) The oelic and aqueous phase as well as Polymer and salt are convected with the bulk the porous media are incompressible. respectively. k . For both. saturation. written as: 5.1 Flow Equations Where C . The subscripts o The viscosity of the polymer solution at “zero” and a represent the oelic and aqueous phases. For respectively.  are the pressure. shear rate depends on the salt as well as polymer ~ is the source/sink term. polymer and mobile water are fully     Cs a pa   q~aCsw (5) mixed. C pad is the mass of polymer adsorbed per unit  S a  k  mass of rock. and gravity Sa  So  1 (3) is neglected. the rock properties are homogeneous. The equation for polymer transport is permeability of the aqueous phase only.) Salt is not adsorbed to the solid surface. 2. The subscripts p . salt and polymer.1 Salinity and Concentration Effects viscosity. physical parameters such as the rock permeability change. phases in porous media can be described by the mass conservation equation for each phase.  r are the component concentration Taking into account the above assumptions the and bulk density of the rock. C pw and Csw are the polymer and the aqueous phase this is written as: salt concentration at the injection boundary.) The process is isothermal.) Multicomponent dispersion is neglected. 2.) Salt. a transport equation is 4. Therefore.     a pa   q~a (1) Due to adsorption of polymer on the rock- t  a  fluid interface. S .solutions in porous media. simultaneous flow of two immiscible fluid s represent the polymer and salt components. here we make the The fact that the void space of the porous following assumptions: media is completely filled with the oelic and aqueous phase.) Generalized Darcy’s law is applicable to  SaC p    r 1   C pad   multiphase flow. t  a  2. q concentration. t  o  2. t t (4)  k  6. effective permeability and porosity of the porous media. leads to the following relation: 1.  S aCs   k  8. Darcy velocity of the aqueous phase.  .

C pad is the maximum polymer concentration adsorbed to the In order to solve the two-phase flow problem rock.7 Saturation. (9) The degree of polymer adsorption depends on  P  the type of polymer and rock. when the maximum amount of polymer is  a0   w 1  a1C p  a2 C p2  a3C 3p C sSp (6) adsorbed [1]. In this work we assume that the shear rate is 2. a3 . (11) C pad . we used the following capillary reduction factor can be expressed as [4]: pressure model [5]: 1 C pad  Rk  1  RRF  1  max (8) p c  pt  S n . Often shear thinning of such. . taking into account the apparent  (7) shear rate in porous media. which reduces the viscosity dimensionless. three additional constitutive equations need to be specified. (Equations (1) to (3)). a2 .2 Polymer Adsorption proportional to the flow velocity. but in general increases with higher polymer concentrations. Then.  1  P  1M  hence. and S p are fitting constants. where P is a viscosity multiplier.3.describing the dependence of the aqueous phase where RRF is the residual resistance factor. calculated The relation between the concentration of from equation (6) assuming no shear effect. the following relation can be defined [1]. which is expressed as [5]: phase is unaltered. Capillary Pressure and max Relative Permeability where b is a Langmuir constant. It viscosity on polymer and salt concentration is represents the decrease in rock permeability the Flory-Huggins equation [2]. that the items in the parenthesis become polymer is observed. Further the permeability In this model.4 Permeability Reduction Factor to curvature and surface tension of the interface between the phases. The surfaces the permeability to the aqueous phase is pressure difference is defined by the capillary reduced. Due 2. and polymer in the aqueous phase and the M is a shear thinning multiplier. while the permeability of the oelic pressure pc .  a   a0  .3. C pad 1  bC p max . 2.5 Non-Newtonian Rheology Where. relation may be appropriate. of the aqueous phase. Polymer adsorption causes a reduction of polymer concentration in the aqueous phase. The aqueous polymer solution behaves like a The units of the fitting constants are selected non-Newtonian fluid. a reduction in aqueous phase viscosity.3. where kw is the water permeability. For a concentration of polymer in the adsorbed state is homogeneous and high permeable reservoir this described by the Langmuir isotherm [3].3. The permeability reduction factor representing the effect of polymer adsorption on the permeability of the aqueous pc ( Sa )  po  pa (10) phase is defined as Rk  k w / k a . For low permeable and heterogeneous rocks a different approach C pad bC p will be required. 2. the pressure in the wetting As a result of polymer adsorption on the rock phase is less than in the non-wetting phase. a1 .

(1) through (3). and the saturation equation is defined as:   Q (21) A . respectively.where  . 2  3 k  Sn  k  k rw The fractional flow formulation has shown ka   good results over other formulations using Rk Rk (13) COMSOL [6]. Here. pt are the Brooks-Corey coefficient S a     u a  q~a . S a t  0  Sra Equation (1) through (14) define the coupled flow and transport system through a number of If the total rate of the aqueous phase is denoted cross couplings. In the following the by Qa . the Darcy flux at the inlet boundary is implementation of the system into COMSOL is presented and initial and boundary conditions are given by: specified. COMSOL Implementation p t  0  p0 . The ka o relative permeabilities of water krw and the oelic  k k  u   a  o p (18) phase kro are calculated using the Brooks-Corey   a o  . 2  Polymer and salt are convected with the k o  k  1  S n  1  S n   k k  2   ro velocity of the aqueous phase ua . correlation [5]. Influx boundary conditions are specified for the    u  q~a  q~o (15) polymer and salt concentrations: n   S a C p  u a C p   a C pw . dpc pt  S n   (14) dS w  1  S ra  S ro  Cp  C p 0 . (12) which are defined as: 1  S ra  S ro   u  ko ka  ua    p (17) S ra and S ro are the residual and immobile k o  a   o k a   a k o  c 1 saturations of the aqueous and oelic phase. The following initial conditions need to  1  be specified. Equation (20) implies that only given by: the aqueous phase is injected into the domain. The pressure equation is sectional area. Implementation into COMSOL requires a   Q reformulation of the two-phase flow equations n  ua  a (20) A . a fractional flow formulation is considered which transfers the  two-phase flow equations into one pressure and n is the normal vector and A the cross- one saturation equation. where p is the global pressure. Cs t  0  Cs 0 t 0 (19) 3. The cross couplings calculated analytically: defined in equations (6) – (14) are implemented as well. (16) and the capillary entry pressure and Sn is the t normalized water saturation defined as [5]:   where ua and u are the velocities of the S a  S ra aqueous phase and total velocity. which is used   as input velocity field for the Species Transport The capillary pressure derivative can be in Porous Media Interface. Sn  .

Residual Resistance Factor. simulation model was set-up.1 Injection Scheme and Results Length. Time difference simulator ECLIPSE Blackoil by Qa C pw Csw [days] Schlumberger [1]. The results obtained with COMSOL are compared to the Injection Salt Polymer results obtained using the commercial finite. Width. k m² 1e-12 Residual Oil Saturation. o mPa∙s 10 A p t  p0 (24) Capillary Entry Pressure. 0.  Q Porosity. a generic Table 2: Injection Scheme. t days 1 phase viscosity.6   Constant. Sp .3 n   S a C s  u a C s   a C sw . 2 n   S a C p  u a C p   0   (25) Constant. 0.1000 10 5 0 Parameter Unit Value 4. Rate. The input data for the simulation models are listed in Table 1. -0. a 3 . S ra Water Viscosity.200 10 5 1 Table 1: Physical parameters used in simulation. Conc.15 where  denotes artificial diffusion.  . here. 100 . Between 100 and 200 days a . h m 10 concentration of 5 kg/m³ is injected into the reservoir (Figure 1). kg/m³ 10 in the equations presented in the chapter polymer Cs 0 flood Pa 1e5 Brooks Corey Coefficient. b max for salt and polymer which will be discussed Maximum adsorption. 4. At the outlet boundary.2 RRF 4. (22) A Permeability. w m 10 During the first 100 days only water at a salt Height. 200 . Conc. As the polymer slug is Element Size COMSOL. 0. w mPa∙s 1   Q n  ua  a (23) Oil Viscosity.24 n   S a C s  u a C s   0 (26) Rock density. l m 300 The injection scheme is presented in Table 2. the salt concentration is Element Size ECLIPSE. C pad kg/kg 8. the following conditions are Residual/ Initial aqueous . x m 1 injected after 100 days.  r kg/m³ 2650 In COMSOL artificial diffusion  is used to Langmuir adsorption m³/kg 1 stabilize the solution of the convection equation constant. a 2 . Artificial [m³/day] [kg/m³] [kg/m³] diffusion is added to COMSOL only and will be 0 .0.100 10 5 0 discussed later. x m 0. Sro . a1 ..2 assigned: phase Saturation. Therefore. diffusion is explicitly included in the equations while it is not included Initial Salt Concentration.06 lower than the initial concentration (Figure 3). Model Test and Simulation Artificial Diffusion. - 1.  ..4e-5 later. which is favorable to maintain a higher aqueous Time Step.  m/s 5e-7 In order to validate the implementation of the polymer flood model presented here.

06 1003 RWE Dea AG and Wintershall Holding GmbH ECLIPSE 0 1 950 for financing and supporting this research. Due to Table 2 lists the impact on breakthrough time for adsorption of the polymer to the rock the size of the different cases indicating differences. Principles of Polymer Chemistry. 0 0. the element size and artificial aqueous phase (Figure 2).1. Table 3: Polymer breakthrough time (BT).J. both methods to find solutions to the 1. This 5. ExxonMobil Production Deutschland COMSOL 5e-7 1 1005 GmbH. 7. After smoothing as presented in Figure 6. In order to show this effect in ECLIPSE another 3. . polymer EOR can be modeled in COMSOL. but the front is smoothed. model with larger element size ( x  1m) but Factors influencing polyacrylamide adsorption in with the same input data (Table 1) is created. Taking into account the same input presented results show good agreement data given by Table 1 and 2 and shear thinning compared to the reservoir simulator ECLIPSE by as described by equation (9) the results Schlumberger indicating a successful previously presented change. This is an important aspect in research. Technical convection equation have problems and one has Description 2011. or geo-mechanical effects integrated. Conclusions and opinions stated in this paper are In case of the finite difference method used those of the authors only. water at a salt concentration of 5 kg/m³. Underlying transport and auxiliary equations can be easily varied or diffusion (  ) the finite element method used extended and other physics such as temperature here is unstable and oscillations occur. the implementation. J. Lakatos-Szabo. relevant to polymer EOR. ECLIPSE© Blackoil Simulator. Polymer breakthrough with COMSOL and Basic physico-chemical effects relevant to ECLIPSE occurs after 1005 and 1003 days.polymer concentration of 1 kg/m³ is added to the COMSOL. Conclusion results in a reduced efficiency of the process. Although commercial Sharp fronts need to be resolved numerically software for simulation of polymer EOR in purely convective transport of chemicals such processes exists. The size of the oil bank is reduced compared to the latter case. the polymer slug decreases (Figure 2) and aqueous solution viscosity is reduced. Without adding artificial capabilities of COMSOL. P. Schlumberger. 6. Discussion 4. the Transport in Porous Media Interface COMSOL saturation equation (Equation 16) is parabolic in Multiphysics provides a flexible framework to nature. The model will be the basis for efficiency of the process is further reduced as further and more detailed investigations. J. presented in Figure 5. References size used to discretize the domain in space. This results in the diffusion is kept constant. The ECLIPSE results formation of an additional oil bank which is using the coarse elements sizes shows increased displaced to the production side (Figure 4). For porous media and their effect on flow behavior... Then. Flory. to accept either oscillations or smearing of the 2. it lacks the multi-physics as polymer and salt. I.. in ECLIPSE no oscillations occur. GDF SUEZ E&P Deutschland GmbH. No artificial diffusion is required in implement physical and chemical mechanisms COMSOL to avoid non-monotone solutions.. Toth. Acknowledgements Element Simulator  Size BT [m/s] [days] The authors of this paper would like to thank [m] DGMK. front. Hence.2 Artificial Diffusion Due to its PDE Interface and Species As the capillary pressure is non-zero. Smoothing depends on the element 8. The same 200 days the polymer slug is displaced with can be shown for COMSOL by increasing  . The respectivly. Lakatos.. Cornell University Press (1953).

Hydraulic Properties of Porous Media. R. (2008). COMSOL COMSOL no Shear Thinning 1 Eclipse 0.Paper presented at the Symposium on Surface 11 COMSOL Phenomena in EOR. excerpt 5 from the proceedings of the COMSOL 4 Conference Hannover.5 0. 6 6.3 0. 0 50 100 150 200 250 300 Length (m) Figure 3: Salt concentration after 10 and 100 days. Colorado State University.2 0 50 100 150 200 250 300 0 Length (m) 0 50 100 150 200 250 300 Length (m) Figure 1: Water saturation after 10 and 100 days.. and Corey. New scaling criteria for 10 polymer emulsion and foam flooding Salt Concentration (kg/m³) 9 experiments.65 t=200 days 0.4 0.6 0. Colorado State 7 University Hydrology Paper.2 0. (1989). Figure 4: Water saturation after 200 and 300 days. Figure 5: Water saturation after 200 days with and without shear thinning .1 0. Brooks. Islam.. Bj.I.8 0.4 0 50 100 150 200 250 300 0 20 40 60 80 Length (m) Length (m) Figure 2: Polymer concentration after 300 and 800 days.4 0.45 0 0. 8 t=100 days 5. M..3 0. Appendix A.55 0. J. T. Figures 1 COMSOL COMSOL Eclipse 0.8 COMSOL with Shear Thinning 0.8 0.5 t=200 days 0. (1964). Farouq.2 0.6 t=300 days t=800 days 0. Stockholm. Technol. 3.7 Water Saturation (-) 0.7 Water Saturation (-) 0. 28(4).6 t=300 days 0.7 Water Saturation (-) 0. Can.. 79– 87 (1989). Aker. Comparing Equations for Two-Phase Fluid Flow in Porous Media.6 0. A. S.8 0.75 Polymer Concentration (kg/m³) 0. Petrol.4 t=10 days t=100 days 0. 9..5 0. t=10 days Eclipse 4. A.9 Eclipse 0.

=5e-7 1 Eclipse. .6 t=300 days t=800 days 0.4 0.2 0 0 50 100 150 200 250 300 Length (m) Figure 6: Polymer concentration after 300 and 800 days. COMSOL. x=1.8 0. x=1 Polymer Concentration (kg/m³) 0.