Evaluation of CO2 Low Salinity Water-Alternating-Gas for enhanced oil recovery

Journal of Natural Gas Science and Engineering 35 (2016) 237e258

Contents lists available at ScienceDirect

Journal of Natural Gas Science and Engineering journal homepage: www.elsevier.com/locate/jngse

Evaluation of CO2 Low Salinity Water-Alternating-Gas for enhanced oil recovery Cuong Dang a, *, Long Nghiem a, Ngoc Nguyen b, Zhangxin Chen b, Quoc Nguyen c a

Computer Modelling Group Ltd., Canada Department of Chemical & Petroleum Engineering, University of Calgary, Canada c Department of Petroleum & Geosystems Engineering, The University of Texas at Austin, United States b

a r t i c l e i n f o

a b s t r a c t

Article history: Received 19 May 2016 Received in revised form 5 July 2016 Accepted 2 August 2016 Available online 4 August 2016

Low Salinity Waterflooding (LSW) is an emerging attractive enhanced oil recovery (EOR) method because of its oil recovery performance and relatively simple, environmentally friendly implementation, when compared with conventional high salinity waterflooding and EOR approaches. More importantly, another advantage of LSW is that it can be integrated with other EOR methods (in hybrid LSW processes), i.e. chemical or miscible gas flooding. The merits of combining LSW with CO2 injection is investigated in this paper, and a novel EOR method, Low Salinity Water Alternating CO2 (CO2 LSWAG), is proposed. CO2 LSWAG injection promotes the synergy of the mechanisms underlying these methods which further enhances oil recovery and overcomes the late production problems frequently encountered in conventional WAG. CO2 LSWAG has been evaluated in both one-dimensional and full-field scale with positive results compared with conventional high salinity WAG. To investigate the advantages of CO2 LSWAG, a comprehensive ion exchange model associated with geochemical processes has been implemented and coupled to the multi-phase multi-component flow equations in an equation-of-state compositional simulator. 1D simulation of different CO2 LSWAG schemes are first conducted to understand the combined effects of solubility of CO2 in brine, dissolution of carbonate minerals, ion exchange, and wettability alteration. CO2 LSWAG performance is then evaluated on a field scale through an integrated workflow that includes geological modeling, multi-phase multi component reservoir flow modeling and process optimization. The simulation results indicate that CO2 LSWAG has the highest oil recovery compared to conventional high salinity waterflood, high salinity WAG, continuous CO2 flooding, and low salinity waterflood. A number of geological realizations are generated to assess the geological uncertainty effect, in particular clay distribution uncertainties, on CO2 LSWAG efficiency. CO2 LSWAG injection strategies are optimized by identifying key WAG parameters. Finally, CO2 LSWAG is evaluated in the full field scale for a North Sea reservoir and the simulation results shows that CO2 LSWAG yields about 4.5% incremental OOIP compared to the conventional high salinity WAG. The proposed workflow demonstrates the synergy between CO2 WAG and LSW. Built in a robust reservoir simulator, it serves as a powerful tool for screening, design, optimization, and uncertainty assessment of the process. CO2 LSWAG is a promising EOR technique as it not only combines the benefits of CO2 injection and low salinity water floods but also promotes the synergy between these processes through the interactions between geochemical reactions associated with CO2 injection, ion exchange process, and wettability alteration. This paper demonstrates the merits of this process through modeling, optimization and uncertainty assessment. © 2016 Elsevier B.V. All rights reserved.

Keywords: Low Salinity Water-Alternating-Gas Low salinity water flooding WAG Enhanced oil recovery

1. Introduction Low Salinity Waterflooding (LSW) has received increasing * Corresponding author. E-mail address: [email protected] (C. Dang). http://dx.doi.org/10.1016/j.jngse.2016.08.018 1875-5100/© 2016 Elsevier B.V. All rights reserved.

attention in the oil industry and is currently identified as an important EOR technique as it shows more advantages than conventional chemical EOR methods in terms of chemical costs,

238

C. Dang et al. / Journal of Natural Gas Science and Engineering 35 (2016) 237e258

environmental impact, and field process implementation. The modification of the injected brine composition could improve the oil recovery factor of conventional waterflooding up to 38% (Webb et al., 2004), leading to a new concept of optimal injection brine composition for waterflooding. Other than using high salinity reservoir water, extensive laboratory experiments (Tang and Morrow, 1997; Morrow et al., 1998; Tang and Morrow, 1999a, 1999b; Zhang and Morrow, 2006; Kumar et al., 2010; Loahardjo et al., 2010) and pilot tests (McGuire et al., 2005; Lager et al., 2008; Skrettingland et al., 2010; Thyne and Gamage, 2011) have confirmed the advantages of using low salinity brine as an injected fluid on the oil recovery for both secondary and tertiary modes. Although the benefits of LSW have been realized, the mechanism for incremental oil recovery by LSW is still a topic for open discussions. Several mechanisms have been proposed during the last two decades including fines migration, wettability alteration, multi-component ionic exchange (MIE), saponification, pH modification, and electrical double layer effects. Dang et al. (2013b) provided a critical review and discussion of these mechanisms. Among the proposed hypotheses, wettability alteration towards increased water wetness during LSW is the widely accepted cause for enhanced oil recovery. The effects of low salinity brine on wettability modifications have been reported by many authors (Jadhunandan and Morrow, 1995; Tang and Morrow, 1999a; Drummond and Israelachvili, 2002, 2004; Vledder et al., 2010; Zekri et al., 2011). It has been experimentally found that the low salinity brine has a significant effect on the shape and the end points of the relative permeability curves (Webb et al., 2004; Kulkarni and Rao, 2005; Rivet, 2009; Fjelde et al., 2012), resulting in a lower water relative permeability and higher oil relative permeability. The mechanisms of wettability alteration due to ion exchange and geochemical reactions have been successfully implemented in a compositional simulator for modeling of LSW (Dang et al., 2013a). Excellent agreements between simulation results and important measurements from coreflood experiments and pilot observations were obtained with this modeling approach. LSW can be considered as secondary and tertiary recovery or can be combined with other EOR approaches such as chemical flooding (e.g., polymer or surfactant), referred as hybrid LSW in the literature. The first attempts at implementing hybrid LSW EOR was to combine LSW with low-tension surfactant flooding. This is an economically attractive hybrid EOR process since using surfactant under low salinity conditions improves surfactant solubility and reduces adsorption and retention, resulting in improved economic performance (Kozaki, 2012; Dang et al., 2014). Alagic and Skauge (2010) reported that significant incremental oil recovery was obtained for a Low Salinity Surfactant (LS-S) flood when the core was pre-flushed with low-salinity brine compared with an LS-S flood in a high-salinity environment. LS-S in aged Berea sandstone cores lowered residual oil saturations down to around 0.05. It was concluded that surfactant stayed in the aqueous phase and microemulsion was successfully formed under the low-salinity conditions, instead of moving over to the oil phase and being trapped there under increased salinity conditions. This phenomenon could well be explained by Winsor type I, II, and III phase behavior in chemical flooding. Along with low-tension surfactant flooding, polymer flooding is also an attractive EOR approach that has been widely investigated in laboratory and field scales. Based on a number of coreflood experiments, Kozaki (2012) concluded that the use of low-salinity polymer flooding has significant benefits because of considerably lower amount of required polymer for a target viscosity. Additionally, low-salinity polymer flooding can also increase oil recovery by lowering residual oil saturation and achieving faster oil recovery by wettability alteration. These observations have been confirmed by Mohammadi and Jerauld (2012)

based on numerical simulation. The simulation results show that low-salinity polymer flooding gave about 5% incremental oil recovery over high-salinity polymer flooding, and a five times reduction in chemical costs per barrel of oil recovered can be realized when polymer is added to low-salinity brine. Not limited to chemical flooding, a new concept of hybrid LSW has been explored by combining LSW with CO2 Water-AlternatingCO2 (WAG) under miscible injection conditions and this concept is referred to as CO2 LSWAG Kulkarni and Rao (2005) conducted miscible and immiscible WAG with varying brine composition on Berea sandstone cores. They reported a decrease in oil recovery with decreasing the salinity of the injected water due to an increase in the solubility of CO2 in brine. The effect of injection brine salinity on CO2-WAG performance in the tertiary mode was investigated by Jiang et al. (2010) on Berea sandstone cores by changing the salinity of the injected brine up to 32,000 ppm. The coreflooding experiments were run at 60
C and at a pressure of 20% above the minimum miscibility pressure (MMP) to ensure miscible CO2 flooding. The results revealed that the WAG recovery increases with increasing the salinity of the injection brine and this was explained due to a salting out effect, as the solubility of CO2 in water decreases with increasing the salinity. This allows more CO2 available for oil displacement resulting in higher oil recovery. However, it is important to note that the sandstone core samples used in these investigations were strongly water-wet with very low clay content. This is an unfavorable condition for achieving wettability alteration, which is the key factor to obtain the benefit of LSW (Rivet, 2009; Dang et al., 2015b). Fjelde and Asen (2010) investigated wettability alteration during water flooding and CO2 flooding on reservoir rocks from the North Sea at different temperatures (50 and 130
C). The experimental work started with formation water as a first phase, followed by sea water as a second phase and finally a cycle of sea water CO2 WAG as a third phase. The results showed that wettability alteration towards more water-wet was observed after the WAG slug resulted in residual oil saturation between 3 and 5%. One of the promising investigations in this area was conducted by Zolfaghari et al. (2013). They reported, based a series of coreflood experiments under conditions favorable for LSW application, that CO2 LSWAG gave additional oil recovery of up to 18% OOIP. Interesting findings from their results are that CO2 LSWAG is also highly effective for heavy oil and the ultimate recovery by LSW is even higher than that by CO2 HSWAG. These positive results encourage the extension of LSW and CO2 LSWAG, currently limited to application in light/medium oil reservoirs, into heavy oil reservoirs. Recently, Ramanathan et al. (2015) conducted an investigation of the effect of salinity on the waterflooding as well as water alternating CO2 injection process through six coreflood experiments. They found that wettability alteration towards a more water-wet state was the cause of improved oil recovery by LSW and the solubility of CO2 due to water composition is very important on the performance of CO2 LSWAG. Fines migration was also observed during low salinity brine alternating CO2 injection process. Teklu et al. (2016) observed that the higher solubility of CO2 in low salinity water as compared to high salinity water is the main reason for the improvement in residual oil mobilization as compared to conventional WAG and higher CO2 solubility in brine can lead to stronger carbonated water in situ to alter wettability and reduce IFT and viscosity further. Al-Shalabi investigated the potential of combining LSW and carbon dioxide in carbonate reservoir using numerical simulation and they concluded that this hybrid process is promising as CO2 controls the residual oil saturation whereas LSWI boosts the oil production rate through increasing oil relative permeability by wettability alteration (Al-Shalabi et al., 2016). Despite significant growing interest in CO2 LSWAG, most of previous evaluations for this process have been done with

C. Dang et al. / Journal of Natural Gas Science and Engineering 35 (2016) 237e258

laboratory-based coreflood experiments. The ability of reservoir simulators to accurately model this process is very limited and it is desirable to have a mechanistic model that can provide better prediction and a more successful implementation of this hybrid process. This paper aims to overcome the gaps in the past evaluations of CO2 LSWAG using an advanced and comprehensive simulation approach with a mechanistic LSW model in an equation-ofstate compositional simulator. The focuses of this research are: (1) theory background and process modeling aspects; (2) 1D comparison between conventional high salinity waterflooding (HSW), LSW, high salinity WAG (HSWAG), continuous CO2 flooding, and CO2 Low Salinity Water-Alternating-Gas (CO2 LSWAG) recovery processes; (3) Evaluate the critical effects of geology on CO2 LSWAG performance; (4) propose an efficient CO2 LSWAG robust optimization method under geological uncertainties; and (5) Full fieldscale implementation of CO2 LSWAG process. 2. Concept of hybrid Low Salinity Water-Alternating-Gas LSW might have significant benefits when combined with water alternating miscible CO2 injection. While LSW is an emerging EOR method based on modification of wettability, WAG is a proven method for improving gas flooding performance by controlling gas mobility. Therefore, CO2 LSWAG injection promotes the synergy of the mechanisms underlying these methods (i.e., ion-exchange, wettability alteration, and CO2 miscible effects and mobility control) to further enhance oil recovery. CO2 LSWAG can be used in oil production in two modes:  As an effective EOR approach for green and brown oil fields, to overcome current challenges associated with LSW and CO2 WAG, by taking advantage of the synergy between them.  As an agent for improving conformance control by blocking high conductivity zones and diverting the injected fluid into unswept layers. The unfavorable mobility of pure gas flooding results in viscous fingering, which reduces volumetric sweep efficiency. WAG helps overcome this problem, reducing the large amount of gas required for EOR projects, which is particularly important in offshore oil fields. Oil production is usually delayed in conventional CO2 WAG processes. Although oil recovery is predicted to be higher for a WAG process compared with pure CO2 flooding, the economics may not be favorable because of the delayed production. CO2 LSWAG may help overcome this challenge with current CO2 WAG applications. LSW can accelerate oil production in the early stage, and CO2 WAG can help promote ion exchange and reservoir geochemical

reactions that are favorable conditions for LSW. It is expected, therefore, that CO2 LSWAG will promote the synergy of these separate process mechanisms, thereby overcoming the late production problem frequently encountered in conventional WAG. The second mode is similar to creating a “salinity shock”, where a drastic decrease in salinity gradient, with insufficient amounts of Caþþ in the injected water (i.e., lower than 1/10 of the Naþ/Caþþ ratio), can mobilize clay minerals, plug the porous media, and reduce the absolute permeability in the watered-out layers (Jones, 1964). The injected fluid is then diverted into low permeability zones, providing additional oil recovery from these regions. Until now, there is a lack of experimental evidence for concluding that LSW induces water blockage. Although this idea is promising, more detailed laboratory experiments must be carried out to confirm its effectiveness. In the literature, LSW has been evaluated both in secondary and tertiary flooding modes and CO2 LSWAG can be implemented either after waterflooding or LSW. Several prescreening conditions that are important for ensuring the highest efficiency on combining LSW and CO2 WAG from experimental work and field observations are listed in Table 1 and will be further examined in this paper:

3. Modeling aspects of hybrid Low Salinity WaterAlternating-Gas In fact, CO2 LSWAG is a hybrid recovery process that combines the effect of LSW and miscible CO2 gas flooding. Therefore, it is important to have an integrated modeling approach that simultaneously captures the fundamentals of these processes. From numerous experimental studies, wettability alteration, leading to increased water wetness due to ion exchange and geochemical reactions, has been found to be the primary LSW mechanism, whereas CO2 gas mobilizes and displaces residual oil using a multiple contact miscible process, which is the key for miscible CO2 gas flooding. To demonstrate the benefits of this hybrid EOR method, it is important to model the important phenomena in each process as well as the interactions among them when low salinity brine is combined with CO2 gas flooding. The key elements for CO2 LSWAG modeling is indicated in Fig. 1. We used Nghiem et al.'s model (2004) to simulate the CO2 flooding process, and Dang et al. model (2013a, 2015a) to simulate the LSW process. All of these proposed models have been implemented in an equation-of-state compositional simulator. The simulator is capable of accounting for complex phase behavior, geochemical reactions, petrophysical properties, and heterogeneous porousmedia properties. The key features are:

Table 1 Prescreening conditions for CO2 LSWAG. Property

Preferred condition

Reservoir

              

Crude oil Clay minerals Reservoir minerals Formation water Initial Wettability Reservoir energy Injected fluid

239

Sandstones Carbonates Must contain polar components (not effective with synthetic oil) Sandstone reservoir must contains sufficient amount of clay High Cation Exchange Capacity (CEC) and clays is preferred Calcite Dolomite Presence of divalent ions such as Caþþ and Mgþþ Presence of connate water Oil Wet or Mixed Wet Reservoir Small or ineffective in strong water wet reservoir Sufficient high pressure for achieving miscibility condition. Lower salinity concentration than formation water Must contain divalent ions Sufficient CO2 source for WAG implementation

240

C. Dang et al. / Journal of Natural Gas Science and Engineering 35 (2016) 237e258

 

0

KNa\Ca ¼

0:5

 

0

KNa\Mg ¼

 

0:5

zðNa  XÞ m Caþþ g Caþþ     0:5  þ g Naþ ½zðCa  X2 Þ m Na 0:5

 

(6) 0:5

zðNa  XÞ m Mgþþ g Mgþþ     0:5  g Naþ ½zðMg  X2 Þ m Naþ

(7)

where zðNa  XÞ , zðCa  X2 Þ , and zðMg  X2 Þare the equivalent fractions of Naþ, Caþþ and Mgþþ on the exchanger, respectively. And, m and g are the molality and activity coefficient, respectively. 4. Various intra-aqueous reactions involved in LSW and WAG processes can be modeled.

Q a  Keq;a ¼ 0; a ¼ 1; …; Raq

(8)

Fig. 1. Key elements for modeling CO2 LSWAG hybrid process.

where 1. Geochemical reactions are fully coupled to the multiphase multicomponent flow equations and the equations for EOS flash calculations; e.g.: for nh oil and gas components:

ji ≡

X a¼o;g;w





DTau yuia DP nþ1 þ DPcua  ~rua gDd þ

V nþ1 þV snþ1  i;aq þqi Dt



Ninþ1 Nin



X q¼g;o;w

DDuiq Dyuiq (1)

¼0; i¼1;…;nh ;

 V  nþ1 n ¼0; j¼1;…;na Nja Nja Dt

(2)

for nm mineral components

jk ≡V snþ1  k;mn

 V  nþ1 Nk  Nkn ¼ 0; Dt

# (9)

k¼1

naq is the number of species in an intra-aqueous equilibrium reaction, ak is the activity of each species in the aqueous phase, Keq is the equilibrium constant for the chemical reaction, R is the universal gas constant, T is the fluid temperature, nka is the species stoichiometric coefficient for the chemical reaction, and DG0f is the Gibbs free energy for each species.

b k 1  Qb rb ¼ A b b K



nþ1 jj ≡DTwu yujw DP nþ1  ~ruw gDd þ DDuiw Dyuiw þV snþ1 j;aq þV sj;mn

þqnþ1  j

naq   1 X nka DG0f ka RT

5. Incorporation of various mineral dissolution and precipitation reactions can affect the ion exchange process.

for na aqueous components:



" Keq;a ¼ exp 

k ¼ 1; …; nm

(3)

The superscripts n and nþ1 denote the old and new time levels, respectively. These equations are discretized in an adaptive implicit manner. The superscript u ¼ n is for explicit gridblocks and nþ1 for implicit gridblocks. The terms (V snþ1 ) and (V snþ1 ) correspond, i;aq k;mn respectively, to the intra-aqueous reaction rates and mineral dissolution/precipitation rates. 2. Multiple ion exchange and wettability alteration during the course of LSW is considered as the main mechanism of the additional oil recovery.

1 1 Naþ þ ðCa  X2 Þ4ðNa  XÞ þ Caþþ 2 2

(4)

1 1 Naþ þ ðMg  X2 Þ4ðNa  XÞ þ Mgþþ 2 2

(5)

3. The multiple ion exchanges were modeled based on chemical equilibrium between ions in the aqueous phase and clay minerals.

eq;b

! ; b ¼ 1; …; Rmn

(10)

where ^ b ¼ reactive surface area of reactant mineral b per unit bulk A volume of porous medium (m2/m3) kb ¼ rate constant of mineral reaction b (mol/m2$s). Keq,b ¼ chemical equilibrium constant of mineral dissolution/ precipitation reaction Qb ¼ activity product of mineral b dissolution reaction rb ¼ dissolution/precipitation rate per unit bulk volume of porous medium [mol/(m3$s)] Rmn ¼ number of mineral reactions 6. Multiple relative permeability sets can be used to model the alteration of wettability. The relative permeabilites of oil and water are altered by a scaled ion exchange equivalent fraction that represents the ion exchange and clay properties.

4. 1D numerical simulation of CO2 LSWAG A one-dimensional model was developed in order to simulate this process and the scaled ion exchange equivalent fraction was used as the interpolant value for wettability alteration in heterogeneity of porosity and permeability of the 1D model. The use of the scaled ion exchange equivalent fraction represents a more realistic relative permeability modification by considering both of the ion exchange and clay properties. The scaled ion-exchange equivalent fraction is defined as the equivalent fraction of Ca-X2*CEC/CECmax. Fig. 2 shows the 1D linear model for simulation of CO2 LSWAG with the main properties shown in Table 2. In this model, we considered the reversible ion exchange between Caþþ and Naþ as well as

C. Dang et al. / Journal of Natural Gas Science and Engineering 35 (2016) 237e258

241

aqueous and mineral reactions. Relative permeability curves for oil wet and water wet conditions are shown in Fig. 3. The reactions that are modeled are:

CO2 ðaqÞ þ H2 O4Hþ þ HCO 3

(11)

Hþ þ OH 4H2 O

(12)

Calcite þ Hþ 4Caþþ þ HCO 3

(13)

1 þþ 1 Ca þ Na  X4 Ca  X2 þ Naþ 2 2

(14)

Various simulation scenarios were performed to compare CO2 LSWAG with other recovery approaches such as conventional HSW, LSW, CO2 HSWAG, and pure CO2 flooding as shown in Table 3. The composition of formation water and injected brine, and the fluid property model are indicated in Tables 4 and 5, respectively. Four cycles of CO2 WAG were conducted after 0.4 injected pore volume of HSW or LSW with a WAG ratio of 1:1 (Fig. 4). First, we consider the effect of the conventional HSW (Run A) on the oil recovery compared to LSW (Run B). Fig. 5 indicates that LSW has a great advantage on oil recovery. This benefit is due to ion exchange and mineral reactions, which were discussed in detail in Dang et al. (2013a). When the high salinity brine was injected, no wettability alteration occurred since the injected brine composition is similar to the formation water composition. On the contrary, the adsorption of Caþþ during LSW altered the original mixed wetness to preferential water wetness, leading to a significant increase in the oil recovery. Although LSW has higher oil recovery than the conventional HSW, large amount of oil is still trapped in the reservoir. CO2 HSWAG (Run C) was considered to increase oil recovery. In this run,

Fig. 3. Relative permeabilities for 1D CO2 LSWAG simulation.

about 0.4 and 0.6 pore volumes of high salinity brine were injected before and after four cycles of high salinity WAG, respectively. Fig. 6 shows that oil recovery by high salinity WAG increases by 40.2% and 26.4% of the original oil in place (OOIP) compared to the HSW and LSW, respectively. The additional oil recovery comes from the effects of CO2 miscible flooding. Fig. 7 compared the oil recovery by four different recovery methods including HSW, LSW, CO2 HSWAG, and pure CO2 flooding. Although CO2 HSWAG has a higher ultimate oil recovery factor than HSW and LSW and the final oil recovery factors by CO2 HSWAG and pure CO2 flooding are relatively similar, CO2 HSWAG experiences with the problem of delayed production as indicated in the earlier discussions. It sometimes prevents the application of WAG in the field because of economic issues. However, this challenge can be overcome by using CO2 LSWAG in which the ultimate oil recovery factor is maximized and oil is produced much faster compared to CO2 HSWAG in the early stage of WAG cycles (Fig. 8). These observations are clearly presented in Figs. 9 and 10. The oil production rate continuously declines with time in HSW and LSW; however, CO2 LSWAG keeps a high oil rate that is more sustainable than single slug CO2 flooding and is higher than CO2 HSWAG. The previous results confirm the advantage of CO2 LSWAG on the oil recovery. This section aims to provide further insights into the role of geochemistry in CO2 LSWAG. Generally, geochemical reactions play an important role in CO2 LSWAG. The dissolution of Calcite can promote the wettability alteration by supplying the Caþþ source for the ion exchange process. Fig. 11 indicates the benefit of Calcite mineral dissolution on the oil recovery by CO2 LSWAG. Oil is produced faster when the reaction involving Calcite (Eq. (13)) is included. The dissolution of calcite increases the ion exchange level as shown in Fig. 12. There are four important aqueous and mineral reactions (Eqs. (11)e(14)) that are involved in this process and the injection

Fig. 2. 1D linear model of CO2 LSWAG.

Table 2 Basic reservoir properties for 1D simulation. Parameter

Value for base case

Grid blocks system Grid block sizes Horizontal permeabilities Vertical permeabilities Porosity Initial water Saturation Selectivity coefficient

20  1 x 1 Dx ¼ 3.66 m, Dy ¼ 30.48 m, Dz ¼ 15.24 m 2*2000 2*1900 2*1800 2*1700 2*1500 2*1200 2*1600 2*1850 2*2000 2*2100 Equals to horizontal permeability 2*0.265 2*0.24 2*0.22 2*0.2 2*0.16 2*0.13 2*0.19 2*0.23 2*0.26 2*0.275 0.3 0.4 at 25
C

242

C. Dang et al. / Journal of Natural Gas Science and Engineering 35 (2016) 237e258

Table 3 Injection schemes for 1D numerical Simulation.

Table 4 Formation and injected brine molality.

Caþþ Naþ

Formation

Injected brine

0.024414 0.4892

0.001892 0.01196

Table 5 Fluid property model. Oil/Gas components

Initial oil mole fraction

Aqueous components

Initial aqueous molality

CO2 CH4 C3H8 FC6 FC10 FC15 FC20

0.0001 0.2999 0.03 0.17 0.3 0.15 0.05

Hþ Caþþ Naþ Cl OH HCO3-

1.0E-7 0.0244 0.4892 0.538028 1.704E-7 0.002118

Fig. 5. Cumulative oil recovery from runs A and B.

Fig. 6. Cumulative oil recovery from runs A, B and C. Fig. 4. WAG cycling in 1D CO2 LSWAG.

scheme can play an important role in the success of CO2 LSWAG applications. A series of sensitivity analysis runs were conducted with the following observations:  The oil recovery factor tends to increase with an increase of the injected Caþþ concentration (Fig. 13).  Injected Naþ concentration must be lowered compared to the formation water to promote ion exchange and mineral dissolution, resulting in a higher oil recovery factor (Fig. 14).  HCO3- in the injected brine has detrimental effects on CO2 LSWAG performance as it may lead to the precipitation of

Calcite, and consequently a decrease of ion exchange and wettability alteration (Fig. 15).  An increase in the amount of Calcite mineral leads to an increase in the ultimate recovery factor by CO2 LSWAG (Fig. 16).

5. CO2 LSWAG integrated modeling with considerations of geology There are several key factors that affect the success of a hybrid LSW process at the field scale: (1) geology, (2) ion exchange and geochemistry, (3) relative permeability modification to reflect wettability alteration, and (4) injection scheme. Modeling and

C. Dang et al. / Journal of Natural Gas Science and Engineering 35 (2016) 237e258

243

Fig. 7. Cumulative oil recovery from runs A, B, C and E. Fig. 10. Comparative oil rate of CO2 HSWAG, CO2 LSWAG, and pure CO2 flooding.

Fig. 8. Cumulative oil from runs A, B, C, D, and E.

Fig. 11. Effect of mineral dissolution on CO2 LSWAG.

Fig. 9. Comparative oil rate of HSW, LSW, CO2 HSWAG, and CO2 LSWAG.

simulation processes must take these factors into account to achieve the best prediction of hybrid LSW performance. Since CO2 LSWAG is a process that depends critically on the reservoir geological characteristics, it is important that important geological factors be incorporated in reservoir modeling and simulation. Multiple geological realizations (facies and clay distribution) are needed for history matching, optimization and uncertainty assessment of LSW. In practice, it would be inefficient, for example,

if the geological modeling, reservoir simulation, history matching, and optimization tasks were processed separately. To overcome this, an objective of this paper is to introduce an efficient approach for effectively connecting these modeling steps in a closed-loop workflow. A comprehensive CO2 LSWAG integrated modeling approach has been developed, as shown in Fig. 17. First, the geological model is generated by considering all of the important factors, including facies, distribution and quantities of clay mineral, porosity and permeability. These geological properties are coupled with flow and physics mechanisms (e.g., ion exchange, geochemical reactions, and wettability alteration for LSW) in the reservoir simulator. The geological model and reservoir simulator are then automatically linked with a robust optimizer in a closed loop that allows fast, continuous geological model updating. As mentioned above, multiple geological realizations are required for CO2 LSWAG modeling and evaluation. By using simple scripts, the geological software can, under the command of the optimizer, geostatistically generate new geological realizations. This approach is powerful for large-scale LSW history matching when important data (well logs, core and fluid samples, and multiple relative permeabilities) is available or when robust optimization is required for capturing geological uncertainty.

244

C. Dang et al. / Journal of Natural Gas Science and Engineering 35 (2016) 237e258

Fig. 12. Comparative ion exchange with and without mineral reactions.

Fig. 13. Effect of injected Caþþ on CO2 LSWAG.

5.1. Geological and reservoir modeling A sector model of a reservoir has been constructed to demonstrate the benefits of CO2 LSWAG. It is a sandstone reservoir with nine layers from top to bottom. Three different facies: Fine-Sand (FS), Medium-Sand (MS), and Coarse-Sand (CS) were geostatistically distributed using a geological software throughout the reservoir using the facies proportions and histograms as shown in Table 6 and Fig. 18. The distributions of three

facies and clay mineral are illustrated in Fig. 19. Note that clay is distributed based on facies distribution since there is a strong correlation between the grain size and clay content, as has been reported by several authors (e.g., (Saner et al., 1996; Shahin et al., 2011). This information was incorporated in the model by assigning clay-dependent grain sizes to different facies where higher clay content results in a lowering of the mean grain-size distribution. Clay minerals have a significant influence on effective

C. Dang et al. / Journal of Natural Gas Science and Engineering 35 (2016) 237e258

245

Fig. 14. Effect of injected Naþ on CO2 LSWAG.

Fig. 15. Effect of injected HCO3- on CO2 LSWAG.

Fig. 16. Effect of mineral quantity on CO2 LSWAG.

reservoir porosity and permeability. The main types of clays are laminated, dispersed, and structural, and combinations thereof. Generally speaking, laminated clay replaces both sand grains and pore space, dispersed clay replaces only pore space, and structural clay replaces only sand grains. In this research, the effect of dispersed clay on porosity is considered since it reduces the clean sandstone porosity фmax to effective reservoir porosity фe, as follows:

Vcl ¼ фmax  фe

(15)

where Vcl is the volume of dispersed clay per bulk volume. After the porosity model is constructed based on clay distribution, the permeability distribution can be populated using the collocated cokriging technique, where porosity is treated as a variable that has a statistical relationship with permeability. Kriging is

246

C. Dang et al. / Journal of Natural Gas Science and Engineering 35 (2016) 237e258

Fig. 17. Big-loop CO2 LSWAG modeling.

Table 6 Facies proportions for CO2 LSWAG Sector Model. Facies

Proportions

Fine Sand (FS) Medium Sand (MS) Coarse Sand (CS)

0.1 0.8 0.1

a geostatistical technique for estimating properties at locations that do not have measured data. It uses a variogram model (a measure of

spatial correlation) to infer the weights given to each data point. The difference between kriging and collocated cokriging is that collocated cokriging supports the use of a second set of data, called soft data, with a lower weight than hard data at the same distance. The role of soft data is to provide additional information (typically, a trend) but is treated with less importance than hard data because soft data is considered less reliable than hard data. The purpose of this simulation approach is to generate a permeability model that reflects the continuity modeled by the permeability variogram, with the observed correlation with porosity. In the collocated

Fig. 18. Facies histogram in sector model.

Fig. 19. Facies and clay distribution in a sector model.

C. Dang et al. / Journal of Natural Gas Science and Engineering 35 (2016) 237e258

cokriging method, the correlation coefficient between hard and soft data must be specified. Fig. 20 shows the porosity and horizontal permeability distributions for the LSW base case well-placement optimization, respectively. Geological and petrophysical properties were exported from the geological software to the compositional reservoir simulator. An inverted five-spot pattern was used for LSW as shown in Fig. 21. The compositional model has seven oil and gas components, seven ions (Naþ, Caþþ, Mgþþ, Cl, Hþ, OH, CO3 ), and two minerals (calcite and magnesite). In this study, we consider multiple ion-exchange reactions involving exchanges of both calcium and magnesium ions with the clay surfaces. The geochemical reactions are modeled as shown in Eqs. (11)e(14). Two sets of relative permeability values were used to represent oil-wet and water-wet conditions in conventional high salinity and low salinity waterflooding, respectively. Since there are multiple ion exchanges in this study, the normalized equivalent-fraction ion exchange of the sodium will be used as an interpolant value for wettability alteration because it captures the dual effects of both calcium and magnesium ion adsorptions.

247

Fig. 21. 3D CO2 LSWAG sector reservoir model.

5.2. High salinity waterflooding and high salinity WAG In the first investigation, a five-spot HSW was applied for about 10 years in the sector reservoir model. Then conventional high salinity waterflooding WAG (HSWAG) was followed as a tertiary recovery method. In the other case, only HSW was applied for the rest 20 years of production. A high salinity brine term in this paper is referred to the injection brine that has the same salinity with the formation water. Fig. 22 shows the performance between two scenarios in terms of the recovery factor. In these case studies, the ultimate oil recovery by HSW after first 10 years is approximately 19.44% of the OOIP. HSW has high remaining oil saturation in the reservoir due to a high degree of reservoir heterogeneity as shown in Fig. 23. If only HSW is applied in the entire production period, the final recovery factor is only about 26.27% OOIP. A sharp decline in the production rate was observed after the first few years. By implementing CO2 HSWAG as a tertiary recovery method, it is possible to increase the final oil recovery factor to 32.37% OOIP. Thus, tertiary CO2 HSWAG provides about 6.1% incremental OOIP over HSW in a highly heterogeneous reservoir and the additional oil recovery comes from miscible effect, the key important component of the WAG process. Fig. 24 shows the effectiveness of CO2 HSWAG in decreasing remaining oil saturation in comparison with pure HSW at the end of the simulated period. One of the advantages of LSW is its potential to be applied in all recovery stages including secondary and tertiary modes (Morrow and Bukley, 2011; Dang et al., 2015b). It has been indicated that secondary LSW is more effective than conventional HSW and

Fig. 22. Oil Recovery Factor: HSW vs. HSW þ HSWAG.

tertiary LSW, in terms of timing and oil recovery. This is an important finding since LSW can be considered for the first stage of secondary recovery, instead of HSW. It provides not only better oil recovery but also reduces the cost and complexity of the additional setup and operation required for new injected-water treatment in older platforms, which is currently a challenge for tertiary recovery development in offshore and deep-water reservoirs. Regarding this important observation, this paper investigates this new potential of the secondary LSW implementation. Fig. 25 shows the performance of four different recovery methods in terms of oil recovery including: (1) conventional HSW

Fig. 20. Porosity and permeability distribution in a sector model.

248

C. Dang et al. / Journal of Natural Gas Science and Engineering 35 (2016) 237e258

Fig. 23. Remaining oil saturation after 10 years of HSW.

from the beginning to the end of simulation; (2) 10 years of HSW followed by 10 years of HSWAG; (3) LSW from the beginning to the end of simulation; (4) 10 years of LSW followed by 10 years of LSWAG. The simulation clearly indicates that the sooner the LSW process is started, the greater the benefits in term of oil recovery in comparison with the conventional HSW. Instead of using HSW at the beginning or after primary production, it is recommended to apply the LSW process so that oil could be produced much faster, yielding an additional oil recovery of 7.5% OOIP, as demonstrated in this case study. Among four recovery methods, CO2 LSWAG provided the highest oil recovery factor with an additional oil recovery of 3.5%, 5.3%, and 11.6% OOIP compared with LSW, HSW plus HSWAG, and conventional HSW, respectively. This is consistent with previous studies on delayed production, which sometimes prevents the field application of WAG for economic issues. This challenge can, however, be overcome by using LSWAG, which maximizes the ultimate oil-recovery factor and produces oil much faster than the other methods. By combining the advantages of wettability alteration and miscible CO2 WAG, this hybrid method is more effective than

Fig. 25. Comparison of Oil Recovery Factor by: HSW, HSW þ HSWAG, LSW, LSW þ LSWAG.

Fig. 24. Remaining Oil Saturation after 20 Years: HSW vs. HSW þ HSWAG.

C. Dang et al. / Journal of Natural Gas Science and Engineering 35 (2016) 237e258

LSW and conventional methods for decreasing the remaining oil saturation as shown in Fig. 26. Fig. 27 shows the comparative results of the ultimate recovery factor among CO2 LSWAG and continuous CO2 as well as the pure LSW. In consistent with 1D simulation results, CO2 LSWAG provides a better oil recovery factor with a lower remaining oil saturation than the continuous CO2 injection with the same amount of the injected gas (Fig. 28). This result indicates that CO2 LSWAG could be a promising candidate for EOR application in hostile reservoir conditions such as high temperature, high hardness, and limited CO2 sources for the continuous gas flooding.

249

As indicated in Fig. 29, clay distribution and clay content have a strong effect on the performance of CO2 LSWAG. There is a wide range of CO2 LSWAG oil-recovery factors, from 21.9% to 36% OOIP for the hundred realizations. Fig. 30 shows three representative geological distributions that are of significant difference in the ultimate oil recovery factor. Since the well locations are the same for

5.3. Role of geology in CO2 LSWAG Since the main mechanism of LSW is wettability alteration based on ion exchange between injected low salinity brine and clay surfaces, clay minerals play a very important role in LSW and hybrid LSW processes as documented in Dang et al. (2015b) study. Using the integrated modeling approach presented in the previous section, 102 geological realizations were generated to represent various clay distributions in order to effectively assess the critical effect of clay distribution and clay content in this hybrid process.

Fig. 27. Ultimate recovery factor by LSW, LSW þ continuous CO2 flooding, and LSWAG.

Fig. 26. Remaining oil saturation after 20 years: LSW þ LSWAG; LSW; HSW þ HSWAG; and HSW.

250

C. Dang et al. / Journal of Natural Gas Science and Engineering 35 (2016) 237e258

all realizations, the difference of clay distribution among the realizations has caused the difference in wettability alteration by ion exchange. The difference in permeability due to clay distribution also leads to a corresponding change in sweep efficiency, which affects the final oil-recovery factor. These three realizations represent three groups of CO2 LSWAG performance in which the ultimate oil recovery significantly decreases from realization 95 (33.6% OOIP) to realization 24 (28.7% OOIP) and to realization 19 (24.9% OOIP). The difference among these realizations was analyzed to determine the critical effects of clay minerals on CO2 LSWAG performance. In realization 19, the final oil-recovery factor is low. The main reason is that the five-spot injection pattern was located in a highclay-content region of fine-sand facies. Although the CEC in this area is very high, resulting in the strongest effect of wettability alteration, the injection water cannot flow easily through this lowporosity, low-permeability region. This results in inefficient oil displacement compared with realization 24 and realization 95, as shown in Fig. 31. This example demonstrates that tuning only the injected-brine composition without considering geological characterizations is not adequate. Theoretically, low-salinity brine contact with clay minerals should promote ion exchange and wettability alteration; however, locating an injection well in an extremely high clay, low-porosity and low-permeability zone will be detrimental to oil production and will negate the success of a CO2 LSWAG project. On the other hand, realization 95 has a higher oil-displacement efficiency in the near-injection-well region, compared with realization 19, as a result of enhanced wettability alteration in the favorable-clay-content regions, which have sufficiently high porosity and permeability for oil production. Note that although realization 24 has a higher porosity distribution inside the injection pattern, it has less amounts of clay minerals and leads to a smaller benefit of wettability alteration. Even though FS facies have high dispersed clay content, which is desirable for ion exchange and wettability alteration, it may reduce the effectiveness of LSW due to unfavorable sweep efficiency, and wells should not be placed in these regions. On the contrary, LSWAG has a better performance with favorable MS distribution and proportion. In the regions with very low clay content (e.g., coarse sand facies), which do not exhibit wettability alteration, LSWAG acts similarly to conventional high salinity WAG. Thus, well-placement optimization is very important in LSWAG implementation and must be taken into account during

Fig. 29. Effect of clay distribution and clay content in CO2 LSWAG.

the initial field development plan or infill well drilling plan for LSWAG application in secondary and tertiary recovery. 6. CO2 LSWAG well placement optimization under geological uncertainties CO2 LSWAG has been mechanistically modeled with consideration for its key mechanisms and extended to field scale simulation through an efficient big-loop modeling approach that captures the critical effect of clay mineral. This section presents new results on CO2 LSWAG optimization under geological uncertainties. Optimization of this process is important because it can help increase the ultimate recovery factor and economic profits of a CO2 LSWAG project. Based on the mechanism of wettability alteration by ion exchange between the injected low-salinity brine and clay surfaces, injected-brine composition, clay distribution and content should be considered in CO2 LSWAG optimization. From this point of view, optimization of CO2 LSWAG well-placement and injected strategy are performed simultaneously. For a given geological (facies, clay, porosity, and permeability) distribution, a well pattern should be placed in the location that optimally promotes ion exchange for wettability alteration and enhances sweep efficiency. The injected composition must be designed accurately to promote the favorable

Fig. 28. Oil saturation map after continuous CO2 flooding and CO2 LSWAG.

C. Dang et al. / Journal of Natural Gas Science and Engineering 35 (2016) 237e258

251

Fig. 30. Three comparative 3D geological realizations: Porosity distribution.

Fig. 31. Three Comparative 3D Geological Realizations: Oil Saturation after 10 years of CO2 LSWAG.

wettability alteration. In this study, a five-spot LSW injection pattern was used in a sector model to demonstrate this concept. The DECE (Designed Exploration Controlled Evolution) optimization method (Yang et al., 2007, 2009) is used for CO2 LSWAG multiple-parameters optimization, since it is a proven method for quickly finding solutions close to the true optima. DECE will efficiently explore the solution space and find the location of the injection/production pattern that maximizes ultimate oil recovery by LSW. An inverted five-spot injection pattern with four producers and one injector will be moved around the entire reservoir area to identify the best location in the reservoir for the CO2 LSWAG implementation. The objective function for the optimization process is the ultimate oil-recovery factor. In this paper, robust optimization is proposed to capture the critical effects of geology on CO2 LSWAG performance. Up to now, nominal optimization, which is based on a single geological realization has been mostly applied to solve an optimization problem in the oil and gas industry. However, as emphasized from laboratory experiments and field observations, the effectiveness of CO2 LSWAG strongly depends on geological distribution. Even if other important factors, such as rock/fluid properties, injected composition, and operating conditions, are held constant, a change in geological distribution will significantly influence CO2 LSWAG performance. Unfortunately, uncertainty is an inherent characteristic of geological models due to the general lack of observation wells and the noisy, sparse nature of seismic data, core samples, and well logs. It is not a good idea, therefore, to ignore reservoir geological uncertainty by limiting analyses to single realizations for evaluation and optimization of CO2 LSWAG process. Recent progress in computational hardware and software development allows robust optimization of complicated tasks such as reservoir modeling under conditions of geological uncertainty. The goal of this section is to present a study of CO2 LSWAG application using

advanced robust optimization (Yang et al., 2011), providing realistic prediction with reduced risk and uncertainty for the LSW process. A successful application of robust optimization for LSW wellplacement was introduced in Dang et al. (2015a); however, only well pattern location was considered as the optimized parameter in that work. For CO2 LSWAG optimization in this paper, robust optimization is performed with a larger number of the optimized parameters including injection pattern location, brine-injection composition (calcium, magnesium, and sodium concentrations), WAG parameters (water-cycle length, gas-cycle length, water displaced-volume, gas displaced-volume), and injection scheme (injection rate, BHP of production wells, and quantity of injected gas). To capture geological uncertainties, a large number of realizations honoring the geological constraints are generated. The optimization is carried out over the whole set of realizations. In essence, given N realizations, the same set of parameters are applied to each realization and the objective function for optimization is:

F¼

N 1 X F N i¼1 i

(16)

where Fi is the objective function for each realization. In this approach, each objective function evaluation involves N simulations, which could require prohibitive computational costs for large N. To reduce the computational cost of robust optimization, it is recommended to perform robust optimization on a subset of the realizations (Van Essen et al., 2006; Yang et al., 2011; Dang et al., 2015a). The same approach is used here and the uncertainty in the optimization results is evaluated. In this study, 102 realizations of clay facies and content are

252

C. Dang et al. / Journal of Natural Gas Science and Engineering 35 (2016) 237e258

generated. In those realizations, both the distribution of the facies and the clay content within each facies are varied. The first step is to rank the realizations according to the recovery factor. 102 simulation runs were performed with the initial five-spot location. The recovery factors for all those runs are plotted in Fig. 29. As carrying 102 realizations in the robust optimization process is excessive, five representative realizations were selected from those 102 realizations. The recovery factors corresponding to those five realizations are shown in Fig. 32. The five realizations are representative of the whole spectrum of recovery factors. A comparison of the mean and standard deviation of the recovery factors for all 102 realizations and the selected five realizations shows that the mean is almost identical while the standard deviation is similar, suggesting five representative realizations are sufficient for estimating reliable statistics of the entire 102 realizations. In addition to the robust optimization with five realizations, optimization with the realization corresponding to the P50 recovery was so performed. The results for those two cases are compared. In robust optimization, the DECE optimization engine was used to find the optimal location for well placement and injection strategy given the geological uncertainty. DECE provides the same new location for all realizations in each optimization step from which average oil recovery is determined. The optimization engine then tries to maximize this mean value to achieve the highest LSW performance. Fig. 33 shows the running progress of the robust optimization with five geological realizations, with each symbol in the plot representing a robust optimization. There are 300 robust optimization runs and each run consists of five simulation jobs for five realizations, a total of 1500 simulations for the study. Robust optimization has found the best location and injection parameters in terms of the final-recovery factor for the CO2 LSWAG application, with consideration for geological uncertainty in the five representative geological realizations. The original pattern was shifted approximately 70 m and 630 m in the x- and y-directions of the reservoir, respectively. To achieve the highest effect of wettability alteration in this particular study, the optimal ion concentrations for calcium, magnesium, and sodium in the injected brine are 1.013 molal, 0.1 molal, and 0.0001 molal, respectively. The average oil-recovery factor, which is the mean recovery factor of the five realizations, has increased from 30.2% in the base case to 43.5% in the best case after robust optimization. This is evidence for the effectiveness of CO2 LSWAG robust optimization, in which we have not only maximized the oil-recovery factor but also captured the effect of geological uncertainty in the five representative

Fig. 32. Recovery factor for the five representative realizations.

realizations. After the optimal solution is obtained from robust optimization, the optimal well location and injection strategy are applied to the 102 realizations, resulting in 102 predictions. Fig. 34 compares the histograms of oil-recovery factor of the 102 predictions obtained from robust optimization, P50 nominal optimization and the one from the base case. It is clear that there are significant improvements to the recovery factor after either robust optimization or nominal optimization application. In terms of uncertainty assessment, the spread of the recovery distribution is quite wide for the P50 nominal optimization, and decreases substantially with robust optimization based on five representative geological realizations. This case study demonstrates the effectiveness of robust optimization in reducing the risk associated with geological uncertainties. 7. Field-scale CO2 LSWAG implementation In this section, we extend the modeling and simulation of CO2 LSWAG to the full field scale. Although the advantages of CO2 LSWAG have been confirmed, it is necessary to quantify the benefit of this process in a larger scale. CO2 LSWAG was evaluated for a typical North Sea sandstone reservoir in a closed loop reservoir management. For this purpose, we use the Brugge field reservoir introduced by TNO (Peters et al., 2009) and populated it with geological properties including clay distribution for CO2 LSWAG assessment. The geological model was first developed using the GOCAD™ software and served as the initial input data for the CO2 LSWAG model in GEM™. Critical effects of clay mineral and important geochemistry processes like ion exchange and wettability alteration have been fully incorporated in this model. CO2 LSWAG was compared with CO2 HSWAG. Simulation sensitivity analysis and preliminary uncertainty assessments have been carried out in this study. The Brugge field consists of an E-W elongated half-dome with a large boundary fault at its northern edge and one internal fault with a modest throw at an angle of some 20
to the northern edge fault. The dimensions of the field are about 10 km  3 km. From top to bottom, the Brugge field consists of nine layers of four main formations, namely Schelde, Maas, Waal, and Schie. The Waal formation has the greatest thickness, the highest average porosity and permeability and is the major producing reservoir zone. The Schelde formation corresponds to the top two layers (layers 1 and 2) of the simulation model; The Maas formation corresponds to layers 3, 4 and 5, the Waal formation corresponds to layers 6, 7 and

Fig. 33. CO2 LSWAG robust optimization progress.

C. Dang et al. / Journal of Natural Gas Science and Engineering 35 (2016) 237e258

253

gridblocks in the x-direction, 48 gridblocks in the y-direction, and 9 gridblocks in the z-direction (Fig. 35). All nine layers in the z direction follow the geological sequence of the Brugge field. The basic rock and fluid data were provided by TNO. The initial wettability is considered as a preferential mixed wet (or weak oil-wet) based on the relative permeability categorization.

Fig. 35. 3D reservoir model of Brugge field.

Fig. 36. WAG cycling in Brugge field model. Fig. 34. Comparison of histograms resulting from 102 realizations for Base case, nominal optimization and robust optimization of CO2 LSWAG.

8, and the Schie formation corresponds to layer 9. The original high-resolution model of the Brugge field consists of 20 million gridblocks. This high resolution model is upscaled to a simulation model with 44,550 active cells. The well logs and structure of this field were used as the “hard-conditioning” data and used as input to generate a number of geological realizations. The field has been developed by 14 vertical producers and 16 vertical water injectors. Three facies are included in the petrophysics model including: (1) Fine grained sandstone e FS facies has low porosity and low permeability with high clay content; (2) Coarse grained sandstone e CS with high porosity, high permeability and low clay content; (3) Medium grained sandstone MS which is the transition facies between FS and CS facies. The geological model is created with GOCAD™ and exported to GEM. The reservoir model has 139

Fig. 37. Comparative oil recovery by CO2 HSWAG and CO2 LSWAG.

254

C. Dang et al. / Journal of Natural Gas Science and Engineering 35 (2016) 237e258

Fig. 38. Oil saturation map after CO2 HSWAG and CO2 LSWAG.

Fig. 39. Effects of WAG ratio on CO2 LSWAG.

Fig. 40. Effect of water injection period in WAG cycling on CO2 LSWAG.

As the results from previous simulation indicate, the secondary LSW followed by CO2 LSWAG has the highest oil recovery factor. This injection scheme is, therefore, applied to the Brugge field. LSW is implemented for the first eight years, and then is followed by either CO2 LSWAG or CO2 HSWAG. For the base case of CO2 LSWAG and CO2 HSWAG, the WAG ratio is 1:1, the sizes of the alternate slugs is 90 days (about 0.7% HCPV) and the total slug sizes of CO2 is approximately 13% HCPV at the end of the stage (Fig. 36). All of the

WAG processes are conducted under miscible conditions. Fig. 37 shows that CO2 LSWAG yields about 4.5% incremental OOIP compared to CO2 HSWAG. By combining of the advantages of wettability alteration and miscible CO2 WAG, this hybrid method is more effective than the conventional CO2 HSWAG on decreasing of the remaining oil saturation for different formations in the Brugge field as shown in Fig. 38.

C. Dang et al. / Journal of Natural Gas Science and Engineering 35 (2016) 237e258

255

Fig. 41. Summary of the cumulative oil recovery from fifteen geological realizations.

Fig. 42. Effect of the clay distributions on the CO2 LSWAG recovery factor.

Besides the injected brine composition which was discussed in the 1D model, this hybrid method can also be optimized by controlling of the WAG ratio and other injection scheme with the following important factors:  The WAG ratio has a large effect on the ultimate oil recovery and the WAG ratio of 1:2 gave the highest oil recovery in this

particular field (Fig. 39). The WAG ratio can be varied for different reservoirs depending on the geological characterization, formation water and oil properties, and the source of CO2. It is important to note that the solubility of CO2 in the brine is higher when the injection brine salinity is lower and an amount of CO2 will be lost in the aqueous phase. It thus needs to consider a make-up CO2 for achieving the highest oil recovery factor.

256

C. Dang et al. / Journal of Natural Gas Science and Engineering 35 (2016) 237e258

 The longer CO2 LSWAG cycling is applied, the higher the benefit.  The shorter the water injection period in each WAG cycle is, the better the oil recovery is (Fig. 40). Clay minerals distribution plays an important role in the CO2 LSWAG performance since this process strongly depends on the ion exchange and wettability alteration. Clay minerals can be geostasitically distributed in the geological model and calibrated with

well logs data. However, the distribution clay is an uncertain parameter, which will be studied below. Fifteen geological realizations with different facies and clay mapping have been generated from the base case. Figs. 41 and 42 indicate the results of the uncertainty assessment of clay distribution on CO2 LSWAG performance. The difference on the ultimate recovery factor between the best (realization # 31) and the worst (realization # 24) is approximately 2% OOIP.

Fig. 43. Comparative oil saturation map of the geological realizations 24 and 31.

C. Dang et al. / Journal of Natural Gas Science and Engineering 35 (2016) 237e258

257

Fig. 44. Porosity distribution on the geological realizations 24 and 31.

Fig. 43 shows the comparative oil saturation at different injection periods after the secondary LSW, middle and final stages of CO2 LSWAG implementation. It indicates that the geological realization 31 has a faster and higher oil production than the one in realization 24. One of the main reasons is the distribution of the clay mineral as represented by the porosity distribution in Fig. 44. Similar to CO2 HSWAG, this process is also sensitive to the reservoir heterogeneity. A high degree of reservoir heterogeneity may lead to low recovery. It is observed that the MS facies with average porosity and permeability and sufficient clay content has the highest benefit on promoting the CO2 WAG as shown by realization 31. Another interesting observation is that there is only a slightly difference on the oil saturation map between these two realizations after the secondary LSW, but CO2 WAG has shown a great advantage on enhancing oil recovery afterward. It is a good demonstration of the improvements with CO2 LSWAG compared to the pure LSW.

Geology plays a critical role on CO2 LSWAG and robust optimization is a promising approach to reduce the risks associated with geological uncertainties and maximize the ultimate oil recovery factor. Acknowledgments The authors thank Dr. Vijay Shrivastava, Dr. Heng Li, and Dr. Chaodong Yang of Computer Modelling Group Ltd. for their valuable comments and assistance in this research. The authors also thank Paradigm for providing the GoCad™ software for this research. The compositional simulator GEM™ and the optimization software CMOST™ from Computer Modelling Group Ltd. were used in this paper. The authors also wish to thank TNO for providing the access to Brugge Field data. Nomenclature

8. Conclusions This paper presents a comprehensive evaluation of CO2 LSWAG from a one-dimensional heterogeneous model into full field simulation. It shows that CO2 LSWAG is a promising EOR technique as it not only combines the benefits of gas and low salinity water floods but also promotes the synergy between these processes through the interactions between geochemical reactions associated with CO2 injection, ion exchange process, and wettability alteration. CO2 LSWAG overcomes the late production problem frequently encountered in the conventional WAG. CO2 LSWAG provides an incremental oil recovery of 4.5e9% OOIP compared to CO2 HSWAG. The success of CO2 LSWAG depends on: (1) type and quantity of clay; (2) initial reservoir wettability condition; (3) reservoir heterogeneity; (4) reservoir minerals such as calcite and dolomite; (5) composition of formation water and injected brine; (6) reservoir pressure and temperature for achieving CO2 miscible condition; (7) WAG parameters. Additionally, the secondary LSW followed by CO2 LSWAG has the highest oil recovery factor compared to the other injection schemes. Field scale simulation indicates that The WAG ratio has a large effect on the ultimate oil recovery and the WAG ratio of 1:2 gave the highest oil recovery in this particular field. The longer CO2 LSWAG cycling is applied, the higher the benefit. It is also observed that recovery factor increases with the shorter the water injection period in each WAG cycle.

ai b A

b

CEC CECcal CECmax

Ea F g G K K0 kb Keq mi na nh Ni P qi Qb rb R Raq Rmn T

Activity of component i Reactive surface area of mineral b Cation exchange capacity Calculated cation exchange capacity Maximum cation exchange capacity Activation energy Objective function Gravity acceleration Gibbs energy Permeability Selectivity coefficient Reaction rate constant of mineral reaction b Chemical equilibrium constant Molality of component i Number of aqueous components Number of hydrocarbon components Moles of component i Pressure Injection/production rate of component i Activity product Reaction rate of mineral b Universal gas constant Number of reactions in aqueous phase Number of mineral reactions Temperature

258

Tq V Vcl yiq

D

4 4max 4e

gi n ka

~ r

s z j

C. Dang et al. / Journal of Natural Gas Science and Engineering 35 (2016) 237e258

Molar transmissibility of phase q (q ¼ o, g, w) Gridblock bulk volume Clay volume Mole fraction of component I in phase q (q ¼ o, g, w) Difference operator Porosity Clean sandstone porosity Effective porosity Activity coefficient of component i Stoichiometric coefficient of a species for the Chemical reaction Mass density Reaction rate Ion exchange equivalent fraction Constitutive equation

References Alagic, E., Skauge, A., 2010. Combined low salinity brine injection and surfactant flooding in mixed-wet sandstone cores. Energy Fuels 24, 3551e3559. Al-Shalabi, E.W., Sepehrnoori, K., Pope, G., 2016. Numerical modeling of combined low salinity water and carbon dioxide in carbonate cores. J. Petrol. Sci. Eng. 137, 157e171. Dang, T.Q.C., Chen, Z., Ngoc, T.B.N., Bae, W., 2014. Numerical simulation of surfactant adsorption in micellar polymer flooding. J. Energy Sources Part A 36 (16), 1774e1785. Dang, T.Q.C., Nghiem, L.X., Chen, Z., Nguyen, Q.P., 2013a. Modeling low salinity waterflooding: ion exchange. In: Geochemistry and Wettability Alteration. Paper SPE 166447 Presented at the 2013 SPE Annual Technical Conference and Exhibition, New Orleans, LA, USA, 30 September e 2 October. Dang, T.Q.C., Nghiem, L.X., Chen, Z., Nguyen, T.B.N., 2013b. State-of-the art smart low salinity water flooding for enhanced oil recovery. In: Paper SPE 165903 Presented at the 2013 SPE Asia Pacific Oil & Gas Conference and Exhibition, Jakarta, Indonesia, 22e24 October. Dang, T.Q.C., Nghiem, L.X., Nguyen, T.B.N., Chen, Z., 2015b. New insights into critical role of geology in modeling and optimization of low salinity waterflooding. In: Paper SPE 174294 Presented at the 2015 SPE EUROPEC/EAGE, Madrid, Spain, 1e4 June. Dang, T.Q.C., Nghiem, L.X., Nguyen, T.B.N., Chen, Z., Nguyen, Q.P., 2015a. Modeling and optimization of low salinity waterflooding. In: Paper SPE 173194 Presented at the 2015 SPE Reservoir Simulation Symposium, Houston, TX, USA, 23e25 February. Drummond, C., Israelachvili, J., 2002. Surface forces and wettability. J. Petrol. Sci. Eng. 33, 123e133. Drummond, C., Israelachvili, J., 2004. Fundamental studies of crude oil-surface water interactions and its relationship to reservoir wettability. J. Petrol. Sci. Eng. 45, 61e68. Fjelde, I., Asen, S.M., 2010. Wettability alteration during water flooding and carbon dioxide flooding of reservoir chalk rocks. In: Paper SPE 130992, SPE EUROPEC/ EAGE Annual Conference and Exhibition, Barcelona, Spain. Fjelde, I., Asen, S.V., Omekeh, A., 2012. Low salinity water flooding experiments and interpretation by simulations. In: Paper SPE 154142 Presented at the Eighteenth SPE Improved Oil Recovery Symposium, Tulsa, OK, USA, April 14e18. Jadhunandan, P.P., Morrow, N.R., 1995. Effect of wettability on waterflood recovery for crude-oil/brine/rock systems. SPE Reserv. Eng. 10 (1), 40e46. Jiang, H., Nuryaningsih, L., Adidharma, H., 2010. The effect of salinity of injection brine on WAG performance in tertiary miscible carbon dioxide flooding: experimental study. In: Paper SPE 132369 presented at the Western Regional Meeting, Anaheim, CA, USA, 27e29 May. Jones, F.O., 1964. Influence of chemical composition of water on clay blocking of permeability. JPT 16 (4), 441e446. Kozaki, C., 2012. Efficiency of Low Salinity Polymer Flooding in Sandstone Cores. University of Texas at Austin. Master Thesis. Kulkarni, M.M., Rao, D.N., 2005. Experimental investigation of miscible and immiscible WAG process performance. J. Petrol. Sci. Eng. 48, 1e20. Kumar, M., Fogden, A., Morrow, N.R., Buckley, J.S., 2010. Mechanisms of improved oil recovery from sandstone by low salinity flooding. In: Paper SCA 2010-25 Presented at the 24th International Symposium of Core Analysts, Halifax, Canada, 4-7 October. Lager, A., Webb, K.J., Collins, I.R., Richmond, D.M., 2008. LoSalTM enhanced oil recovery: evidence of enhanced oil recovery at the reservoir scale. In: Paper SPE 113976 Presented at the SPE/DOE Improved Oil Recovery Symposium, Tulsa, 19e23 April. Loahardjo, N., Xie, X., Morrow, N.R., 2010. Oil recovery by sequential water flooding of mixed-wet sandstone and limestone. Energy & Fuels 24 (9), 5073e5080. McGuire, P.L., Chatam, J.R., Paskvan, F.K., Sommer, D.M., Carini, F.H., 2005. Low salinity oil recovery: an exciting new EOR opportunity for Alaska's north slope.

In: Paper SPE 93903 Presented at the SPE Western Regional Meeting, Irvine, CA, USA, 30 March-1 April. Mohammadi, H., Jerauld, G.H., 2012. Mechanistic modeling of the benefit of combining polymer with low salinity water for enhanced oil recovery. In: Paper SPE 153161 Presented at the Eighteenth SPE IOR Symposium, Tulsa, OK, USA, 14e18 April 2012. Morrow, N.R., Bukley, J.S., 2011. Improved oil recovery by low-salinity waterflooding. JPT 106e112. Distinguished Author Series. Morrow, N.R., Tang, G.Q., Valat, M., Xie, X., 1998. Prospects of improved oil recovery related to wettability and brine composition. J. Pet. Eng. 20 (3e4), 267e276. Nghiem, L.X., Sammon, P., Grabenstetter, J., Ohkuma, H., 2004. Modeling CO2 storage in aquifers with fully-coupled geochemical EOS compositional simulator. In: Paper SPE 89474 Presented at the SPE Fourteenth Symposium on Improved Oil Recovery, Tulsa, OK, USA, April 17e21. Peters, L., Arts, R., Brouwer, G., Geel, C., 2009. Results of the Brugge benchmark study for flooding optimisation and history matching. In: Paper SPE 119094 Presented at the SPE Reservoir Simulation Symposium, the Woodlands, Texas, February 2e4. Ramanathan, R., Shehata, M., Nasr-El-Din, A.H., 2015. Water alternating CO2 injection process e does modifying the salinity of injected brine improve oil recovery?. In: Paper OTC-26253 presented at the Offshore Technology Conference Brasil, Rio de Janeiro, Brazil, 27e29 October. Rivet, S.M., 2009. Coreflooding Oil Displacements with Low Salinity Brine. Master’s Thesis. University of Texas at Austin. Saner, S., Cagatay, M.N., Sanounah, A.M., 1996. Relationships between shale content and grain-size parameters in the safaniya sandstone reservoir, NE Saudi Arabia. J. Petrol. Geol. 19, 305e320. Shahin, A., Tatham, R., Stoffa, P., Spikes, K., 2011. Optimal dynamic rock-fluid physics Template validated by petroelastic reservoir modeling. Geophysics 76 (6), O45eO58. Skrettingland, K., Holt, T., Tweheyo, M.T., Skjevrak, 2010. Snorre low salinity water injection e core flooding experiments and single well field pilot. In: Paper SPE 129877 Presented at the SPE Improved Oil Recovery Symposium, Tulsa, OK, USA, 24e28 April. Tang, G.Q., Morrow, N.R., 1999b. Oil recovery by water flooding and imbibition e invading brine cation valency and salinity. In: Paper SCA-9911 Presented at the 1999 International Symposium of the Society of Core Analysis, Golden, Colorado, USA, 1e4 August. Tang, G.Q., Morrow, N.R., 1997. Salinity, temperature, oil composition and oil recovery by water flooding. SPE Reserv. Eng. 12 (4), 269e276. Tang, G.Q., Morrow, N.R., 1999a. Influence of brine composition and fines migration on crude oil/brine/rock interactions and oil recovery. J. Petrol. Sci. Eng. 24 (2e4), 99e111. Teklu, T.W., Alameri, W., Graves, R.M., Kazemi, H., AlSumaiti, A.M., 2016. Low salinity water-altertaning-CO2 EOR. Petrol. Sci. Eng. 142, 101e118. Thyne, G., Gamage, P., 2011. Evaluation of the effect of low salinity water flooding for 26 fields in Wyoming. In: Paper SPE 147410 Presented at the SPE Annual Technical Conference and Exhibition, Denver, Colorado, USA, 30 October-2 November. Van Essen, G.M., Zandvliet, M.J., Van den Hof, P.M.J., Bosgra, O.H., Jansen, J.D., 2006. Robust waterflooding optimization of multiple geological scenarios. In: Paper SPE 102913 Presented at the SPE Annual Technical Conference and Exhibition, San Antonio, TX, USA, 24e27 September. Vledder, P., Foncesca, J.C., Wells, T., Gonzalez, I., Ligthelm, D., 2010. Low salinity water flooding: proof of wettability alteration on a field wide scale. In: Paper SPE 129564 Presented at the SPE Improved Oil Recovery Symposium, Tulsa, OK, USA, 24e28 April. Webb, K.J., Black, C.J.J., Al-Ajeel, 2004. Low salinity oil recovery e log-inject-log. In: Paper SPE 89379 Presented at the SPE/DOE Symposium on Improved Oil Recovery, Tulsa, 17e21 April. Yang, C., Card, C., Nghiem, L., 2009. Economic optimization and uncertainty assessment of commercial SAGD operations. J. Can. Petrol. Technol. 48 (9), 33e40. Yang, C., Card, C., Nghiem, L., 2011. Robust optimization of SAGD operations under geological uncertainties. In: Paper SPE 141676 Presented at the SPE Reservoir Simulation Symposium, the Woodlands, Texas, USA, 21e23 February. Yang, C., Nghiem, L., Card, C., 2007. Reservoir model uncertainty quantification through computer assisted history matching. In: Paper SPE 109825 Presented at the SPE Annual Technical Conference an Exhibition, Anaheim, California, USA, 11e14 November. Zekri, A.Y., Nasr, M.S., Al-Arabai, Z., 2011. Effect of LoSal on wettability and oil recovery of carbonate and sandstone formation. In: Paper SPE 14131 Presented at the International Petroleum Technology Conference, Bangkok, Thailand, 7e9 February. Zhang, Y., Morrow, N.R., 2006. Comparison of secondary and tertiary recovery with change in injection brine composition for crude oil/sandstone combinations. In: Paper SPE 99757 Presented at the SPE/DOE Symposium on Improved Oil Recovery, Tulsa, 22e26 April. Zolfaghari, H., Zebarjadi, A., Shahrokhi, O., Ghazanfari, M.H., 2013. An experimental study of CO2 low salinity WAG injection in sandstone heady oil reservoirs. Inranian J. Oil Gas Sci. Technol. 2 (3), 37e47.