- Email: [email protected]
Mechanisms behind low salinity water injection in carbonate reservoirs
Fuel 121 (2014) 11–19
Contents lists available at ScienceDirect
Fuel journal homepage: www.elsevier.com/locate/fuel
Mechanisms behind low salinity water injection in carbonate reservoirs Emad Waleed Al Shalabi ⇑, Kamy Sepehrnoori, Mojdeh Delshad Petroleum and Geosystems Engineering, The University of Texas at Austin, Austin, TX, United States
h i g h l i g h t s History matching of recently published corefloods for low salinity water injection. Investigating the best way of history matching low salinity water injection effect on oil recovery. Highlighting important parameters that should be considered in modeling low salinity water injection effect in carbonates. Wettability alteration is still believed to be the main mechanism underlying the incremental oil recovery by LSWI. Oil endpoint relative permeability is more sensitive to the LSWI effect than is water endpoint relative permeability.
a r t i c l e
i n f o
Article history: Received 29 August 2013 Received in revised form 12 December 2013 Accepted 14 December 2013 Available online 28 December 2013 Keywords: Low salinity water injection Wettability alteration in carbonates History matching LSWI in carbonates IOR using LSWI
a b s t r a c t The low salinity water injection method (LSWI) has become one of the important research topics in the oil industry because of its enormous possible advantages. The objective of this paper is to investigate the mechanism behind the LSWI effect on oil recovery through data matching. The UTCHEM simulator was used to match the cycles of the injected seawater and different dilutions of the latter for two recently published coreflooding experiments. The result from the history matching revealed that the wettability alteration mechanism is believed to be the main contributor to LSWI. Based on this finding, an analytical model for oil recovery predictions can be developed. Ó 2013 Elsevier Ltd. All rights reserved.
1. Introduction Carbonate rocks account for more than half the world’s hydrocarbon proven reserves. Oil recovery from these reservoirs is a challenge due to their complex nature. The problem becomes even more complicated when wettability ranges from mixed-wet to oilwet rocks with a low permeability matrix and high fracture density as in most carbonate reservoirs. Consequently, several enhanced oil recovery techniques have been proposed to improve oil recovery of these reservoirs and overcome the high negative capillary pressure that holds oil in place. One of the suggested techniques is altering the wettability of these reservoirs from oil-wet towards more water-wet, which turns the capillary pressure positive and begins spontaneous water imbibition into the low permeability rock matrix resulting in higher oil recovery. One of the emerging improved oil recovery (IOR) techniques for wettability alteration in carbonate reservoirs is low salinity water injection. The popularity of this technique is due to its efficiency in displacing light to medium gravity crude oils, ease of injection into ⇑ Corresponding author. Tel.: +1 512 550 6279; fax: +1 512 471 9678. E-mail address: [email protected] (E.W.A. Shalabi). 0016-2361/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.fuel.2013.12.045
oil-bearing formations, water availability and affordability, and lower capital and operating costs, all of which lead to favorable economics compared to other chemical and thermal EOR methods. The only concerns with this technique are water sourcing and water disposal. This low salinity water injection EOR technique is also known in the literature as LoSal, Smart Waterflood, and Advanced Ion Management. Several studies have been done on LSWI water injection at laboratory scale and to a limited extent at fieldscale. A review of the effect of LSWI water injection on both sandstone and carbonate rocks is presented below. Extensive laboratory studies were conducted on sandstone rocks after producing 15% additional oil from the Kansas field when brine was used as injection fluid compared to fresh water [1]. The relative effectiveness of fresh and salt water on oil recovery from synthetic and natural cores containing clays was investigated by Bernard [2]. An increase in oil recovery from Berea sandstone cores by increasing the salinity up to a certain level after which recovery did not increase significantly was reported by AlMumen [3]. Both connate and invading brines have major effects on wettability and oil recovery at reservoir temperature as was reported by both Tang and Morrow [4]. Several coreflood experiments investigated the effect of low salinity brine on improving
12
E.W.A. Shalabi et al. / Fuel 121 (2014) 11–19
Nomenclature CPC EPC krl nl
parameter related to the maximum capillary pressure capillary pressure exponent phase endpoint relative permeability phase Corey’s exponent
oil recovery from Berea sandstone cores for both secondary and tertiary injection modes [5]. The incremental oil recovery from sandstone rocks was in the range of 5–20% of OOIP as reported by several studies [6–9]. The suggested mechanisms include fines migration, pH increase, multi-ion exchange (MIE), salting-in, and wettability alteration [10]. The wettability alteration process is the reason behind the low salinity effect as the decrease in salinity increases the size of the double layer between the clay and oil interface leading to organic material release [11]. Wettability alteration in sandstone rocks is related to the presence of clay minerals, oil composition, formation water with high concentrations of divalent cations (Ca2+, Mg2+), and the salinity level of the low salinity water (1000–5000 ppm) [12]. The effect of low salinity water injection on carbonate has not been thoroughly investigated in contrast to sandstone rocks because wettability alteration by low salinity water is related to the presence of clay, which is not the case in carbonate rocks. To our knowledge, no field scale pilots have been conducted so far to investigate the effect of low salinity water injection on carbonate rocks. However, some work was done at laboratory scale on the effect of low salinity water injection on oil recovery from carbonate rocks. Based on their spontaneous imbibition experiments, Hognesen et al. [13] concluded that increasing sulfate ion concentration at high temperature leads to increased oil recovery. Also, through spontaneous imbibition experiments, Webb et al. [14] investigated the effect of sulfate on oil recovery from North Sea carbonate core samples. They found that seawater has the ability to alter the wettability of the carbonate system to a more water-wet state compared to sulfate free water. Wettability alteration in North Sea chalk reservoirs in the Ekofisk field, showing the effect of adding calcium and/or magnesium ions at various temperatures, was studied by Zhang et al. [5]. In the latter study, they concluded that wettability alteration occurs if the imbibing water contains either 2+ Ca2+ and SO2 and SO2 4 or Mg 4 . The feasibility of low salinity water injection (Smart Waterflood) on carbonate rocks to improve oil recovery by using different dilutions of sea water was investigated by Yousef et al. [12]. Their coreflooding tests showed incremental oil recovery up to 18% with stepwise dilution of the sea water in the tertiary water injection mode. Incremental 15% and 20% OOIP recovery were feasible using 3 borate (BO3 3 ) and phosphate (PO4 ) as modified ions, respectively [15]. They concluded that the wettability alteration mechanism in carbonate rocks takes two forms: either dissolution by softening of the injected brine or surface charge change by modifying the injected ions. Several mechanisms describing the low salinity process have been suggested in light of this enormous body of research. However, there is no consensus on a single main mechanism for the low salinity effect. This is due to the complex nature of interaction between crude oil, brine, and rock, as well as a number of conflicting observations from experimental studies. This paper investigates the mechanism behind the LSWI effect on oil recovery through history matching of oil recovery and pressure drop data on Yousef et al.’s [12] first and second coreflooding experiments. We use the UTCHEM simulator, which is a 3D multiphase flow,
np Sl Slr
r
number of phases phase saturation phase residual saturation interfacial tension
transport, and chemical flooding simulator, developed at The University of Texas at Austin. UTCHEM is a three dimensional non isothermal finite different compositional flow simulator. It has the capability of modeling different chemical enhanced oil recovery processes such as polymer, surfactant/polymer flooding. It uses high-order numerical technique to control numerical dispersion. The method used to solve flow equations is implicit in pressure and explicit in concentration. Understanding the mechanism and matching the data enable the development of a LSWI model for oil recovery prediction. 2. Experimental data By measuring both IFT and contact angle at reservoir conditions, Yousef et al. [12] suggested that the obtained incremental oil recovery using LSWI is due to wettability alteration. They conducted two corefloods at reservoir conditions using two sets of composite carbonate cores. Carbonate reservoir cores with average porosity of 25.1% and 24.65%, and average liquid permeability of 39.6 mD and 68.3 mD were used for both first and second coreflooding experiments, respectively. In these corefloods, the cores were saturated with live reservoir oil at the irreducible water saturation, and then field seawater was injected at the reservoir temperature and pressure, followed by the injection of various seawater dilutions for the tertiary recovery. The seawater was diluted in four cycles with the dilution factors of twice, ten-times, twenty-times, and hundred-times. The experimental procedures and fluid properties, including oil and dilutions of seawater, are described elsewhere [12]. 2.1. Experimental data analysis Data digitizing, JBN Method, pressure drop data analysis, and capillary number analysis were conducted on Yousef et al.’s work as discussed in this section. 2.1.1. Data digitizing Oil recovery and pressure drop data were digitized using Engauge Digitizer software to history match the data using UTCHEM. Error bars with upper and lower bounds were used to account for the uncertainty in the data by assuming a 2% error in oil recovery data and 1 psi error in pressure drop data. These error ranges were assigned based on the accuracy of equipment used for conducting the experiments and assuming the least encountered experimental error. Thus, history matching is acceptable within the defined upper and lower error bars. 2.1.2. JBN method A set of parameters is needed to history match the experimental data using UTCHEM which includes relative permeability parameters (endpoint relative permeability and Corey’s exponents), and capillary pressure parameters (CPC and EPC). These parameters are used as input data to calculate relative permeability curves using Corey’s Model [16] and capillary pressure curves using Brooks–Corey Model [17].
13
E.W.A. Shalabi et al. / Fuel 121 (2014) 11–19
Due to the lack of data provided in Yousef et al. [12], a thorough analysis was performed to extract needed parameters for history matching. The Johnson, Bossler, and Naumann (JBN) method was used to extract relative permeability data on both corefloods. This method was used because both experiments were conducted in an unsteady state mode. The JBN method analysis was used to produce fractional flow curves for the two corefloods. The JBN method showed that water breakthrough occurs at high fractional flow values which indicate that lower saturations do not exist physically in the cores and hence no conclusion can be drawn about the oil relative permeability. Moreover, applying the JBN method in the tertiary mode can be tedious. Therefore, Darcy’s law is used to extract the endpoint relative permeabilities for each injection cycle assuming a steady state flow corresponding to the end of each cycle when the pressure drop stabilizes. 2.1.3. Pressure drop data analysis Three main observations can be made from the experimental pressure drop data provided in Yousef et al. [12]. First, endpoint water relative permeability values (krw ) can be obtained at the end of each cycle using stabilized pressure drop, and endpoint oil relative permeability value (kro ) can be obtained for the first cycle only using initial pressure drop as the composite cores were flooded with oil at irreducible water saturation conditions. The endpoint relative permeability values are easily calculated using Darcy’s law (Table 1). Table 1 shows that as the pressure drop decreases, the calculated endpoint relative permeability of water decreases as well. This is consistent with Darcy’s law because the viscosity of water decreases as the salinity of injected seawater decreases. This finding reflects the dominance of wettability alteration mechanism as a main contributor to LSWI over the saturation effect on endpoint relative permeability of water. Second, pressure stabilization of the first seawater cycle is a major concern because the experiments were not run for an adequate period of time at a rate of 1 ml/min to allow the pressure drop to stabilize. However, the pressure drop stabilizes at the end of the 4 ml/min injection rate. Therefore, two pressure drop values (stabilized and unstabilized) were used in the seawater cycle to match the pressure drop data. These pressure drops are taken into account by calculating the corresponding endpoint relative permeability of water, which is used as an input parameter in UTCHEM runs. It should be noted that the krw value for the first cycle reported in Table 1 corresponds to the stabilized pressure drop. Third, the critical capillary number for the composite carbonate cores used (109–108) [18] was exceeded by Yousef et al. through applying high pressure gradients in the laboratory starting with 1 ml/min followed by 2 ml/min and then 4 ml/min for each cycle. The capillary number analysis is presented in the next section. 2.1.4. Capillary number analysis The capillary number for the first cycle was calculated as follows:
Nc ¼
DP K ; L r
ð1Þ
where DP is the measured pressure drop across the composite core, L is the composite core length, K is the absolute liquid permeability, and r is the water–oil interfacial tension. Capillary number was calculated for both sets of experiments for the first cycle and exceeded the critical value for carbonate rocks. This effect was taken into account using UTCHEM to match the first seawater cycle. The trapping number (NTl) is defined in UTCHEM which takes into account the combined effects of viscous, capillary and gravity forces [19]
NTl ¼
i ~ ~ h ~U 0 ~ ~ ~ 0 q Þrh k k gð r q l l l
rll0
ð2Þ
;
~ U 0 is the where l is the displaced phase, l0 is the displacing phase, r l ~ k is the permeability, g is the gravitagradient of the flow potential, ~ tional force constant, and rll0 is the interfacial tension between the displacing and displaced phases. The dependence of Sor on the trapping number was introduced by Delshad [20]:
Slr ¼ Shigh þ lr
high Slow lr Slr for l ¼ 1; . . . ; np; 1 þ T l N sT l
ð3Þ
where l is the phase number, T l is a positive input parameter based on an experimental observation of the relation between residual saturations and trapping number, Slow and Shigh are the residual lr lr phase saturations at low and high trapping numbers, respectively. s is a parameter dependent on the pore size distribution of the rock. Similar carbonate rocks were found in the literature having about the same residual oil saturation values at the applied capillary number in the laboratory [20,21]. The composite carbonate cores used in Yousef et al.’s work [12] are similar to sample K5 in the work reported by Kamath et al. [18]. This is evident from the capillary desaturation curve (CDC) (Fig. 1) and the Sor curve as a function of the applied pressure gradient (Fig. 2). In both figures, at a similar applied pressure gradient of 50 psi/ft and capillary number of the order of 106, the resulting residual oil saturations are almost identical. Moreover, a study was conducted by Abrams [21] on an Indiana carbonate rock sample with similar rock properties and capillary desaturation curve. These CDCs were used to find suitable UTCHEM input parameters for modeling the dependence of residual oil saturation on caphigh illary numbers. This was done by using Slow of zero, s of lr of 0.35, Slr 0.8, and Tl of 33,000 (Fig. 3). Both relative permeability endpoints and exponents change because of the change in residual saturation at high trapping numbers according to Delshad et al. [22]: olow
K orl ¼ krl
þ
nl ¼ nlow þ l
0 Slow l 0 r Sl r
Slow l0 r
ohigh
Shigh l0 r
0 Slow l 0 r Sl r
high Slow l0 r Sl0 r
ðkrl
olow
0
krl Þ for l; l ¼ 1; . . . ; np;
ð4Þ
0
ðnhigh nlow l Þ for l; l ¼ 1; . . . ; np: l
ð5Þ
Table 1 Endpoint relative permeability data analysis (first coreflooding experiment). Oil viscosity Composite core length Cross sectional area Injection rate Absolute brine permeability Injection cycle First Second Third Fourth Fifth
0.691 (0.691 103) 0.5328 (0.1624) 1.22 102 (1.13 103) 1 39.6 (3.96 1014) Water viscosity cp (Pa.s) 0.272 (0.272 103) 0.242 (0.242 103) 0.232 (0.232 103) 0.212 (0.212 103) 0.193 (0.193 103)
cp (Pa.s) ft (m) ft2 (m2) cc/min mD (m2) Pressure drop psi (Pa) 6.90 (4.76 104) 6.19 (4.27 104) 6.03 (4.16 104) 6.10 (4.21 104) 5.79 (3.99 104)
Oil–water IFT
34
dynes/cm
Initial pressure drop
23 (1.59 105)
psi (Pa)
Sor 0.295 0.233 0.151 0.135 0.135
kro 0.266 – – – –
krw 0.349 0.346 0.341 0.308 0.295
14
E.W.A. Shalabi et al. / Fuel 121 (2014) 11–19
3. Simulation data This section includes a description of the simulation models used to history match both corefloods using the UTCHEM simulator. 3.1. Description of models
Fig. 1. Remaining oil saturation as a function of capillary number, white circles represent the unsteady restored state (Kamath et al. [18]).
The first coreflooding model used a 1D Cartesian grid system with dimensions of 20 1 1; width and thickness are the same (3.3677 cm) for all grid blocks; however, lengths of grid blocks vary to match the length dimensions of the four core plugs used in the first coreflooding experiment (Table 2). The heterogeneous model maintained the permeability of each core plug as shown in Fig. 4. A similar model was used for the second coreflooding experiment, with dimensions of 30 1 1; width and thickness are the same (3.3765 cm) for all grid bocks; however, lengths of grid blocks also vary to match the length dimensions of the six core plugs used in this case. Table 3 summarizes the composite core simulation models used for both corefloods. The results of only the first coreflooding experiment are presented in this paper due to space limitations.
4. Results and discussion Fig. 2. Remaining oil saturation as a function of pressure gradient, white circles represent the unsteady restored state (Kamath et al. [18]).
This section covers seawater cycle matching and methods used for the wettability alteration effect, matching for different dilutions of seawater injected cycles. The analysis for both first and second coreflooding experiments was done systematically without using the wettability alteration option in UTCHEM. 4.1. Seawater cycle match
Fig. 3. Modeled CDC curve for the first coreflooding experiment. ohigh
For the water phase, krw was assumed to be 0.4 for the first coreflooding experiment and 0.5 for the second coreflooding experiment; these values are reasonable because the endpoint relative permeability of water begins to decrease with successive olow injections due to wettability alteration by LSWI. From Eq. (4) krw o value was calculated using the known value of krw for the first cycle in Table 1. For the oil phase, endpoint relative permeability and Corey’s exponent remained constant because of the assumed constant irreducible water saturation.
Experimental results of Yousef et al.’s work concluded that substantial oil recovery by LSWI is due to a fluid-rock property (wettability alteration), not a fluid–fluid property (reduction in IFT). Moreover, the results showed that wettability alteration occurs in the second and subsequent cycles. Therefore, the first seawater cycle is matched without taking into account the wettability alteration effect. The seawater cycle was matched, taking into account that the critical capillary number was exceeded for composite carbonate cores used, as was discussed previously. The capillary pressure effect was neglected because of the high pressure gradient applied by setting the CPC parameter to zero. Hence, the data was matched by changing only Corey’s exponents, because water and oil endpoint relative permeability values are known for the seawater cycle. Reported residual oil saturations [12] of 0.295 and 0.221 were used as input values for the first and the second corefloods, respectively.
Table 2 Composite core model data (first coreflooding experiment). Parameter
Value
Comments
Number of grid blocks
20
1D (20 1 1)
Grid block sizes (DI, DJ, DK), ft (m)
X direction: 1–5, Dx is 0.0279 (0.0085) 6–10, Dx is 0.0258 (0.00786) 11–15, Dx is 0.0264 (0.00804) 16–20, Dx is 0.0265 (0.00808) Y direction: 1–1, Dy is 0.110489 (0.033677) Z direction: 1–1, Dz is 0.110489 (0.033677)
Variable grid size on a regional basis, constant grid size in the y and z directions, however, variable grid size for the x direction.
Composite core model dimensions, ft (m)
0.5328 (0.1624) 0.110489 (0.033677) 0.110489 (0.033677)
Length width thickness
E.W.A. Shalabi et al. / Fuel 121 (2014) 11–19
15
Fig. 4. Simulation model used in different runs with heterogeneous permeability (first coreflooding experiment).
Table 3 Summary of composite core model for both experiments. Property
First coreflooding
Second coreflooding
Porosity Pore volume ft3 (m3)
0.2510 1.62 103 (4.59 105) 0.1247 (0.0380) 0.5328 (0.1624) 6.50 103 (1.84 104) 1.22 102 (1.13 103) 20 1 1
0.2465 2.35 103 (6.65 105) 0.1250 (0.0381) 0.7759 (0.2365) 9.53 103 (2.70 104) 1.23 102 (1.14 103) 30 1 1
Diameter ft (m) Length ft (m) Bulk volume ft3 (m3) Cross sectional area ft2 (m2) Model dimensions (length width height)
Grid block dimensions (variable length) Width ft (m) 0.110489 (0.033677) Height ft (m) 0.110489 (0.033677)
0.110778 (0.033765) 0.110778 (0.033765)
For the first coreflooding experiment, the calculated capillary number values for the laboratory, field, and UTCHEM along with the corresponding applied pressure gradients are listed in Table 4. The field pressure gradient was considered as 0.5 psi/ft. There was no direct control on capillary number values using UTCHEM because the experiments were run in the constant injection rate mode. Nevertheless, the obtained results are close enough to those applied in the laboratory. The sets of relative permeability curves used before and after exceeding Nc are shown in Table 5 and Fig. 5. We assumed a negligible change in the irreducible water saturation. Moreover, the oil endpoint relative permeability value used in data matching is lower than the water endpoint relative
permeability, which is acceptable for a mixed-to-oil-wet carbonate core. The possibility of having a lower kro compared to krw for oilwet rocks is supported by the results of Masalmeh [23], in which Middle East carbonate rocks of the same nature were used showing that for an oil-wet rock, kro is lower than krw . For both coreflooding experiments, the first cycle data including oil recovery and pressure drop were matched. The pressure drop was matched using stabilized and unstabilized pressure drop data as discussed earlier (Fig. 6); however, the oil recovery match is only shown for the stabilized pressure case (Fig. 7). The simulated and the Buckley Leverett Theory calculated fractional flow curves were matched and they showed similar water breakthrough points (Fig. 8).
4.2. Dilutions of seawater injected cycles match Wettability alteration could be the reason for incremental oil recovery because of the injection of various dilutions of seawater from the second cycle to the fifth. We systematically simulated the wettability alteration effect by using two methods to match the experimental data for the various dilutions of seawater injection cycles. Parameters accounting for relative permeability and residual oil contributions are included in the matching, while capillary pressure contribution was ignored due to the high pressure gradient applied. For the sake of simplicity, we ignored the dependence of residual oil saturation on capillary number due to the wettability alteration effect, which changes the capillary desaturation curve at each cycle. In the first and second methods, the data was matched by tuning relative permeability parameters and taking into account the reported residual oil saturation, without using the wettability alteration option in UTCHEM. A detailed description of each method is included in this section.
16
E.W.A. Shalabi et al. / Fuel 121 (2014) 11–19
Table 4 Capillary number and pressure gradient calculations (first coreflooding experiment). Maximum delta P (Lab) Maximum delta P (UTCHEM)
30 (2.06 105) 22 (1.52 105)
Nc calculations Laboratory Field UTCHEM
1.48E06 1.32E08 7.35E07
Applied pressure gradient Laboratory Field UTCHEM
56.31 (1.27 106) 0.5 (1.13 104) 41.29 (9.34 105)
psi (Pa) psi (Pa)
psi/ft (Pa/m) psi/ft (Pa/m) psi/ft (Pa/m)
Table 5 Relative permeability parameters before and after exceeding Nc (first coreflooding experiment). Fig. 7. Oil recovery match for seawater cycle (first coreflooding experiment). First cycle matching parameters Below Nc (critical) nw no krw kro Sor Swirr
Exceeding Nc (critical) 2.19 2.00 0.34 0.266 0.35 0.1044
nw no krw kro Sor Swirr
2.00 2.00 0.349 0.266 0.295 0.1044
Fig. 8. Fractional flow match for seawater cycle (first coreflooding experiment).
Fig. 5. Relative permeability curves before and after exceeding N c (first coreflooding experiment).
inputting the laboratory reported residual oil saturation for each cycle. This method is called the Sor contribution because it shows the significance of residual oil saturation in matching the experimental data. Results showed that residual oil saturation is not the main contributor to low salinity water injection because matching the experimental data was not possible. Also, the results reflect the need to change the relative permeability parameters for oil recovery and pressure drop data matching. A better matching method is discussed in the next section. 4.2.2. Second method The second method is called the kr and Sor contributions. Here three approaches are proposed to match the experimental data: change Corey’s exponents while keeping endpoint relative permeability values constant, change endpoint relative permeability values while keeping Corey’s exponents constant, and change both endpoint relative permeability values and Corey’s exponents. Each approach is described in detail in the next section highlighting the related pros and cons.
Fig. 6. Pressure drop match for seawater cycle (first coreflooding experiment).
4.2.1. First method In this method, we used the same relative permeability parameters including endpoints and Corey’s exponents for the seawater cycle matching. The oil recovery of each cycle was matched by
4.2.2.1. First approach. In this approach, relative permeability endpoints were kept constant and Corey’s exponents for water and oil were tuned to match both sets of experimental data. This approach is proposed in light of Seccombe et al.’s work [24], where constant water relative permeability was observed at residual oil saturation for both low and high salinity water injections. The observation was made from coreflooding experiments, numerical matching of
17
E.W.A. Shalabi et al. / Fuel 121 (2014) 11–19
Fig. 9. Overall oil recovery match using second method (First approach – first coreflooding experiment).
Fig. 10. Overall pressure drop match using second method (First approach – first coreflooding experiment).
data, and the constant productivity of the wells from the SWCTTs at the Endicott field in the North Slope of Alaska. Corey’s exponents were tuned while keeping the endpoints relative permeability constant at seawater cycle values for the case when Nc critical was exceeded. The overall oil recovery obtained and the pressure drop data matching using this approach are depicted in Figs. 9 and 10, respectively. A summary of the relative permeability parameters used for both sets of experiments is presented in Table 6, showing a shift towards a more water-wet state, which is associated with higher oil recovery. This approach is not suitable for matching both sets of experimental data for the following reasons: High nw values reach 9, while the maximum typical acceptable value is about 4. Low no values reach 1.25, while the minimum typical acceptable value is about 1.5.
Fig. 11. Overall oil recovery match using second method (Second approach – first coreflooding experiment).
Fig. 12. Overall pressure drop match using second method (Second approach – first coreflooding experiment).
Water fractional flow values for the fifth cycle are higher than for the fourth cycle, which does not make sense physically. Jumps are noticed in the pressure drop data matching.
4.2.2.2. Second approach. This proposed approach to better match both sets of experimental data involves tuning the endpoint relative permeability values for water and oil, while keeping Corey’s exponents constant at seawater cycle values when Nc critical is exceeded. The water endpoint relative permeability values are taken from Table 1; however, the oil endpoint relative permeability values are tuned to match each cycle. For the first coreflood, the obtained match using this approach for overall oil recovery and pressure drop are depicted in Figs. 11 and 12. For both sets of experiments, a summary of the relative permeability parameters used in the second approach is presented in Table 7. The second approach is also not suitable because the overall oil recovery curve is not well matched especially at the starting
Table 6 Summary of relative permeability parameters for both sets of experiments (Second method – first approach). Injection cycle
First cycle Second cycle Third cycle Fourth and fifth cycles
First coreflooding experiment
Second coreflooding experiment
nw
no
nw
no
2 6 7 7.5
2 1.5 1.3 1.25
2 8 8.5 9
2 1.35 1.3 1.25
18
E.W.A. Shalabi et al. / Fuel 121 (2014) 11–19
Table 7 Summary of relative permeability parameters for both sets of experiments (Second method – second approach). Injection cycle
First coreflooding experiment
First cycle Second cycle Third cycle Fourth cycle Fifth cycle
Second coreflooding experiment
krw
kro
krw
kro
0.349 0.346 0.341 0.308 0.295
0.266 0.800 0.900 0.95 0.970
0.466 0.440 0.410 0.390 0.359
0.367 0.800 0.950 0.960 0.970
Fig. 15. Relative permeability curves using second method (Third approach – first coreflooding experiment).
Fig. 13. Overall oil recovery match using second method (Third approach – first coreflooding experiment).
Fig. 16. Fractional flow curves using second method (Third approach – first coreflooding experiment).
Table 8 Summary of relative permeability parameters for both sets of experiments (Second method – third approach). Injection cycle
First coreflooding experiment
Fig. 14. Overall pressure drop match using second method (Third approach – first coreflooding experiment).
humps, and because the fractional flow curve of the fifth cycle is higher than the fourth cycle, which is not reasonable. Nevertheless, some good features of this approach are as follows:
First cycle Second cycle Third cycle Fourth cycle Fifth cycle
Second coreflooding experiment
krw
kro
nw
no
krw
kro
nw
no
0.349 0.346 0.341 0.308 0.295
0.266 0.700 0.840 0.860 0.880
2.00 2.20 2.50 2.70 2.90
2.00 1.70 1.59 1.57 1.55
0.466 0.440 0.410 0.390 0.359
0.367 0.750 0.860 0.870 0.880
2.00 2.20 2.60 2.70 2.90
2.00 1.60 1.56 1.54 1.53
Endpoint relative permeability values are reasonable. There are no jumps in the pressure drop data matching. This approach reflects the need to tune Corey’s exponents to match the humps in oil recovery curves. The approach shows that endpoint relative permeability values are more sensitive to the LSWI effect compared to Corey’s exponents, which is supported by the absence of jumps in the pressure drop data curves.
water relative permeability values used for each cycle are listed in Table 1; however, the oil endpoint relative permeability along with Corey’s exponents are adjusted to match the data. For the first coreflooding experiment, the obtained match for overall oil recovery, pressure drop data, the sets of relative permeability curves and fractional flow curves using this approach are shown in Figs. 13– 16. A summary of the relative permeability parameters used in this approach for both experimental setups is presented in Table 8. The third approach is the best and most reasonable for both oil recovery and pressure drop data matching due to the following:
4.2.2.3. Third approach. A more realistic approach to better match both sets of experimental data suggests tuning both endpoint relative permeability values and Corey’s exponents. The endpoint
Both Corey’s exponents and relative permeability endpoints were tuned. Both Corey’s exponents and relative permeability endpoint values are reasonable.
E.W.A. Shalabi et al. / Fuel 121 (2014) 11–19
Overall oil recovery curves are well matched including the humps at the start of each cycle. Pressure drop curves are well match with no jumps. Water fractional flow curve values for the fifth cycle are lower than for the fourth cycle as expected. This approach shows that oil endpoint relative permeability is more sensitive to the LSWI effect than is water endpoint relative permeability. 5. Conclusions Oil recovery and pressure drop data for both first and second coreflooding experiments of Yousef et al.’s [12] work were matched successfully using UTCHEM. The findings of this work are as follows: Wettability alteration is still believed to be the reason for the LSWI effect on oil recovery. The change in residual oil saturation due to the LSWI is dominated by wettability alteration. Altering relative permeability parameters, endpoints and Corey’s exponents, is essential for history matching. Water and oil endpoint relative permeabilities are more sensitive to the LSWI effect than are Corey’s exponents. Oil endpoint relative permeability is more sensitive to the LSWI effect than is water endpoint relative permeability. In our future work, the wettability alteration option in UTCHEM will be used for history matching to provide more insight into the low salinity water injection mechanism. Acknowledgements The authors wish to acknowledge useful discussions with G.A. Pope during the course of this work. This work was funded by Abu Dhabi National Oil Company (ADNOC). References [1] Smith KW. Brines as flooding liquids. Paper presented at seventh annual tech. meeting, Min. Ind. Expt. Sta., Penn. State College; 1942. [2] Bernard GG. Effect of floodwater salinity on recovery of oil from cores containing clays. In: SPE 1725, SPE California Regional Meeting, Los Angeles, California; 1967. [3] Al-Mumen AA. The Effect of Injected Water Salinity on Oil Recovery. MSc Thesis, King Fahad University of Petroleum and Minerals. Dhahran, Saudi Arabia; 1990. [4] Tang GQ, Morrow NR. Salinity temperature, oil composition and oil recovery by waterflooding. In: SPE 36680, SPE Reservoir Eng 1997; 12(4): 269–76.
19
[5] Zhang P, Tweheyo MT, Austad T. Wettability alteration and improved oil recovery by spontaneous imbibition of seawater into chalk: impact of the potential determining ions Ca2+, Mg2+, and SO2 4 . Colloids Surf A: Physicochem Eng Aspects 2007;301(1–3). 199-128. [6] Lager A, Webb KJ, Black CJ. Impact of brine chemistry on oil recovery. In: 14th European symposium on IOR, Cairo, Egypt; 2007. [7] Lager AK, Webb KJ, Collins IR, Richmond DM. LoSalTM enhanced oil recovery: evidence of enhanced oil recovery at the reservoir scale. In: SPE 113976, SPE/ DOE symposium on improved oil recovery, Tulsa, Oklahoma; 2008. [8] Webb KJ, Black CJJ, Edmonds IJ. Low salinity oil recovery – the role of reservoir condition core floods. In: 13th European symposium on improved oil recovery, Budapest, Hungary; 2005a. [9] Webb K, Lager A, Black C. Comparison of high/low salinity water/oil relative permeability. In: International symposium of the society of core analysts, Abu Dhabi, UAE; 2008. [10] Doust AR, Puntervold TP, Strand S, Austad, TA. Smart water as wettability modifier in carbonate and sandstone. In: 15th European symposium on improved oil recovery, Paris, France; 2009. [11] Ligthelm DJ, Gronsveld J, Hofman JP, Brussee NJ, Marcelis F, van der Linde HA. Novel waterflooding strategy by manipulation of injection brine composition. In: SPE 119835, EUROPEC/EAGE conference and Exhibition, Amsterdam, The Netherlands; 2009. [12] Yousef AA, Al-Saleh S, Al-Kaabi A, Al-Jawfi M. Laboratory investigation of novel oil recovery method for carbonate reservoirs. In: SPE 137634, Canadian unconventional resources and international petroleum conference, Calgary, Alberta, Canada; 2010. [13] Hognesen EJ, Strand S, Austad T. Waterflooding of preferential oil-wet carbonates: oil recovery related to reservoir temperature and brine composition. In: SPE 94166, SPE Europec/EAGE Annual Conference, Spain; 2005. [14] Webb KJ, Black CJJ, Tjetland G. A laboratory study investigating methods for improving oil recovery in carbonates. In: SPE 10506, International petroleum technology conference, Qatar; 2005b. [15] Gupta R, Griffin P, Hu L, Willingham TW, Cascio ML, Shyeh JJ, Harries CR. Enhanced waterflood for middle east carbonates cores – impact of injection water composition. In: SPE 142668, SPE Middle East Oil and Gas Conference, Manama, Bahrain; 2011. [16] Corey AT, Rathjens CH, Henderson JH, Wyllie MR. Three-phase relative permeability. J Petrol Technol 1956;8(11):63–5. [17] Brooks RH, Corey AT. Properties of porous media affecting fluid flow. J Irrigation Drain Div 1966;92(6):61–88. [18] Kamath J, Rober FM, Frank MN. Understanding waterflood residual oil saturation of four carbonate rock types. In: SPE 71505, SPE annual technical conference and exhibition, New Orleans, Louisiana; 2001. [19] Jin M. A study of nonaqueous phase liquid characterization and surfactant remediation. Ph.D. Dissertation, The University of Texas at Austin; 1995. [20] Delshad M. Trapping of micellar fluids in berea sandstone. Ph.D. Dissertation, The University of Texas at Austin; 1990. [21] Abrams A. The influence of fluid viscosity, interfacial tension, and flow velocity on residual oil saturation left by waterflood. SPEJ 1975;15(5):437–47. [22] Delshad M, Bhuyan D, Pope GA, Lake L. Effect of capillary number on the residual saturation of a three-phase micellar solution. In: SPE 14911, SPE enhanced oil recovery symposium, Tulsa, Oklahoma; 1986. [23] Masalmeh SK. The effect of wettability on saturation functions and impact on carbonate reservoirs in the middle east. In: SPE 78515, Abu Dhabi international petroleum exhibition and conference. Abu Dhabi, UAE; 2002. [24] Seccombe JC, Lager A, Webb K, Jerauld G, Fueg E. Improving waterflood recovery: LoSalTM EOR field evaluation. In: SPE 113480, SPE symposium on improved Oil recovery. Tulsa, Oklahoma, USA; 2008.
Contents lists available at ScienceDirect
Fuel journal homepage: www.elsevier.com/locate/fuel
Mechanisms behind low salinity water injection in carbonate reservoirs Emad Waleed Al Shalabi ⇑, Kamy Sepehrnoori, Mojdeh Delshad Petroleum and Geosystems Engineering, The University of Texas at Austin, Austin, TX, United States
h i g h l i g h t s History matching of recently published corefloods for low salinity water injection. Investigating the best way of history matching low salinity water injection effect on oil recovery. Highlighting important parameters that should be considered in modeling low salinity water injection effect in carbonates. Wettability alteration is still believed to be the main mechanism underlying the incremental oil recovery by LSWI. Oil endpoint relative permeability is more sensitive to the LSWI effect than is water endpoint relative permeability.
a r t i c l e
i n f o
Article history: Received 29 August 2013 Received in revised form 12 December 2013 Accepted 14 December 2013 Available online 28 December 2013 Keywords: Low salinity water injection Wettability alteration in carbonates History matching LSWI in carbonates IOR using LSWI
a b s t r a c t The low salinity water injection method (LSWI) has become one of the important research topics in the oil industry because of its enormous possible advantages. The objective of this paper is to investigate the mechanism behind the LSWI effect on oil recovery through data matching. The UTCHEM simulator was used to match the cycles of the injected seawater and different dilutions of the latter for two recently published coreflooding experiments. The result from the history matching revealed that the wettability alteration mechanism is believed to be the main contributor to LSWI. Based on this finding, an analytical model for oil recovery predictions can be developed. Ó 2013 Elsevier Ltd. All rights reserved.
1. Introduction Carbonate rocks account for more than half the world’s hydrocarbon proven reserves. Oil recovery from these reservoirs is a challenge due to their complex nature. The problem becomes even more complicated when wettability ranges from mixed-wet to oilwet rocks with a low permeability matrix and high fracture density as in most carbonate reservoirs. Consequently, several enhanced oil recovery techniques have been proposed to improve oil recovery of these reservoirs and overcome the high negative capillary pressure that holds oil in place. One of the suggested techniques is altering the wettability of these reservoirs from oil-wet towards more water-wet, which turns the capillary pressure positive and begins spontaneous water imbibition into the low permeability rock matrix resulting in higher oil recovery. One of the emerging improved oil recovery (IOR) techniques for wettability alteration in carbonate reservoirs is low salinity water injection. The popularity of this technique is due to its efficiency in displacing light to medium gravity crude oils, ease of injection into ⇑ Corresponding author. Tel.: +1 512 550 6279; fax: +1 512 471 9678. E-mail address: [email protected] (E.W.A. Shalabi). 0016-2361/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.fuel.2013.12.045
oil-bearing formations, water availability and affordability, and lower capital and operating costs, all of which lead to favorable economics compared to other chemical and thermal EOR methods. The only concerns with this technique are water sourcing and water disposal. This low salinity water injection EOR technique is also known in the literature as LoSal, Smart Waterflood, and Advanced Ion Management. Several studies have been done on LSWI water injection at laboratory scale and to a limited extent at fieldscale. A review of the effect of LSWI water injection on both sandstone and carbonate rocks is presented below. Extensive laboratory studies were conducted on sandstone rocks after producing 15% additional oil from the Kansas field when brine was used as injection fluid compared to fresh water [1]. The relative effectiveness of fresh and salt water on oil recovery from synthetic and natural cores containing clays was investigated by Bernard [2]. An increase in oil recovery from Berea sandstone cores by increasing the salinity up to a certain level after which recovery did not increase significantly was reported by AlMumen [3]. Both connate and invading brines have major effects on wettability and oil recovery at reservoir temperature as was reported by both Tang and Morrow [4]. Several coreflood experiments investigated the effect of low salinity brine on improving
12
E.W.A. Shalabi et al. / Fuel 121 (2014) 11–19
Nomenclature CPC EPC krl nl
parameter related to the maximum capillary pressure capillary pressure exponent phase endpoint relative permeability phase Corey’s exponent
oil recovery from Berea sandstone cores for both secondary and tertiary injection modes [5]. The incremental oil recovery from sandstone rocks was in the range of 5–20% of OOIP as reported by several studies [6–9]. The suggested mechanisms include fines migration, pH increase, multi-ion exchange (MIE), salting-in, and wettability alteration [10]. The wettability alteration process is the reason behind the low salinity effect as the decrease in salinity increases the size of the double layer between the clay and oil interface leading to organic material release [11]. Wettability alteration in sandstone rocks is related to the presence of clay minerals, oil composition, formation water with high concentrations of divalent cations (Ca2+, Mg2+), and the salinity level of the low salinity water (1000–5000 ppm) [12]. The effect of low salinity water injection on carbonate has not been thoroughly investigated in contrast to sandstone rocks because wettability alteration by low salinity water is related to the presence of clay, which is not the case in carbonate rocks. To our knowledge, no field scale pilots have been conducted so far to investigate the effect of low salinity water injection on carbonate rocks. However, some work was done at laboratory scale on the effect of low salinity water injection on oil recovery from carbonate rocks. Based on their spontaneous imbibition experiments, Hognesen et al. [13] concluded that increasing sulfate ion concentration at high temperature leads to increased oil recovery. Also, through spontaneous imbibition experiments, Webb et al. [14] investigated the effect of sulfate on oil recovery from North Sea carbonate core samples. They found that seawater has the ability to alter the wettability of the carbonate system to a more water-wet state compared to sulfate free water. Wettability alteration in North Sea chalk reservoirs in the Ekofisk field, showing the effect of adding calcium and/or magnesium ions at various temperatures, was studied by Zhang et al. [5]. In the latter study, they concluded that wettability alteration occurs if the imbibing water contains either 2+ Ca2+ and SO2 and SO2 4 or Mg 4 . The feasibility of low salinity water injection (Smart Waterflood) on carbonate rocks to improve oil recovery by using different dilutions of sea water was investigated by Yousef et al. [12]. Their coreflooding tests showed incremental oil recovery up to 18% with stepwise dilution of the sea water in the tertiary water injection mode. Incremental 15% and 20% OOIP recovery were feasible using 3 borate (BO3 3 ) and phosphate (PO4 ) as modified ions, respectively [15]. They concluded that the wettability alteration mechanism in carbonate rocks takes two forms: either dissolution by softening of the injected brine or surface charge change by modifying the injected ions. Several mechanisms describing the low salinity process have been suggested in light of this enormous body of research. However, there is no consensus on a single main mechanism for the low salinity effect. This is due to the complex nature of interaction between crude oil, brine, and rock, as well as a number of conflicting observations from experimental studies. This paper investigates the mechanism behind the LSWI effect on oil recovery through history matching of oil recovery and pressure drop data on Yousef et al.’s [12] first and second coreflooding experiments. We use the UTCHEM simulator, which is a 3D multiphase flow,
np Sl Slr
r
number of phases phase saturation phase residual saturation interfacial tension
transport, and chemical flooding simulator, developed at The University of Texas at Austin. UTCHEM is a three dimensional non isothermal finite different compositional flow simulator. It has the capability of modeling different chemical enhanced oil recovery processes such as polymer, surfactant/polymer flooding. It uses high-order numerical technique to control numerical dispersion. The method used to solve flow equations is implicit in pressure and explicit in concentration. Understanding the mechanism and matching the data enable the development of a LSWI model for oil recovery prediction. 2. Experimental data By measuring both IFT and contact angle at reservoir conditions, Yousef et al. [12] suggested that the obtained incremental oil recovery using LSWI is due to wettability alteration. They conducted two corefloods at reservoir conditions using two sets of composite carbonate cores. Carbonate reservoir cores with average porosity of 25.1% and 24.65%, and average liquid permeability of 39.6 mD and 68.3 mD were used for both first and second coreflooding experiments, respectively. In these corefloods, the cores were saturated with live reservoir oil at the irreducible water saturation, and then field seawater was injected at the reservoir temperature and pressure, followed by the injection of various seawater dilutions for the tertiary recovery. The seawater was diluted in four cycles with the dilution factors of twice, ten-times, twenty-times, and hundred-times. The experimental procedures and fluid properties, including oil and dilutions of seawater, are described elsewhere [12]. 2.1. Experimental data analysis Data digitizing, JBN Method, pressure drop data analysis, and capillary number analysis were conducted on Yousef et al.’s work as discussed in this section. 2.1.1. Data digitizing Oil recovery and pressure drop data were digitized using Engauge Digitizer software to history match the data using UTCHEM. Error bars with upper and lower bounds were used to account for the uncertainty in the data by assuming a 2% error in oil recovery data and 1 psi error in pressure drop data. These error ranges were assigned based on the accuracy of equipment used for conducting the experiments and assuming the least encountered experimental error. Thus, history matching is acceptable within the defined upper and lower error bars. 2.1.2. JBN method A set of parameters is needed to history match the experimental data using UTCHEM which includes relative permeability parameters (endpoint relative permeability and Corey’s exponents), and capillary pressure parameters (CPC and EPC). These parameters are used as input data to calculate relative permeability curves using Corey’s Model [16] and capillary pressure curves using Brooks–Corey Model [17].
13
E.W.A. Shalabi et al. / Fuel 121 (2014) 11–19
Due to the lack of data provided in Yousef et al. [12], a thorough analysis was performed to extract needed parameters for history matching. The Johnson, Bossler, and Naumann (JBN) method was used to extract relative permeability data on both corefloods. This method was used because both experiments were conducted in an unsteady state mode. The JBN method analysis was used to produce fractional flow curves for the two corefloods. The JBN method showed that water breakthrough occurs at high fractional flow values which indicate that lower saturations do not exist physically in the cores and hence no conclusion can be drawn about the oil relative permeability. Moreover, applying the JBN method in the tertiary mode can be tedious. Therefore, Darcy’s law is used to extract the endpoint relative permeabilities for each injection cycle assuming a steady state flow corresponding to the end of each cycle when the pressure drop stabilizes. 2.1.3. Pressure drop data analysis Three main observations can be made from the experimental pressure drop data provided in Yousef et al. [12]. First, endpoint water relative permeability values (krw ) can be obtained at the end of each cycle using stabilized pressure drop, and endpoint oil relative permeability value (kro ) can be obtained for the first cycle only using initial pressure drop as the composite cores were flooded with oil at irreducible water saturation conditions. The endpoint relative permeability values are easily calculated using Darcy’s law (Table 1). Table 1 shows that as the pressure drop decreases, the calculated endpoint relative permeability of water decreases as well. This is consistent with Darcy’s law because the viscosity of water decreases as the salinity of injected seawater decreases. This finding reflects the dominance of wettability alteration mechanism as a main contributor to LSWI over the saturation effect on endpoint relative permeability of water. Second, pressure stabilization of the first seawater cycle is a major concern because the experiments were not run for an adequate period of time at a rate of 1 ml/min to allow the pressure drop to stabilize. However, the pressure drop stabilizes at the end of the 4 ml/min injection rate. Therefore, two pressure drop values (stabilized and unstabilized) were used in the seawater cycle to match the pressure drop data. These pressure drops are taken into account by calculating the corresponding endpoint relative permeability of water, which is used as an input parameter in UTCHEM runs. It should be noted that the krw value for the first cycle reported in Table 1 corresponds to the stabilized pressure drop. Third, the critical capillary number for the composite carbonate cores used (109–108) [18] was exceeded by Yousef et al. through applying high pressure gradients in the laboratory starting with 1 ml/min followed by 2 ml/min and then 4 ml/min for each cycle. The capillary number analysis is presented in the next section. 2.1.4. Capillary number analysis The capillary number for the first cycle was calculated as follows:
Nc ¼
DP K ; L r
ð1Þ
where DP is the measured pressure drop across the composite core, L is the composite core length, K is the absolute liquid permeability, and r is the water–oil interfacial tension. Capillary number was calculated for both sets of experiments for the first cycle and exceeded the critical value for carbonate rocks. This effect was taken into account using UTCHEM to match the first seawater cycle. The trapping number (NTl) is defined in UTCHEM which takes into account the combined effects of viscous, capillary and gravity forces [19]
NTl ¼
i ~ ~ h ~U 0 ~ ~ ~ 0 q Þrh k k gð r q l l l
rll0
ð2Þ
;
~ U 0 is the where l is the displaced phase, l0 is the displacing phase, r l ~ k is the permeability, g is the gravitagradient of the flow potential, ~ tional force constant, and rll0 is the interfacial tension between the displacing and displaced phases. The dependence of Sor on the trapping number was introduced by Delshad [20]:
Slr ¼ Shigh þ lr
high Slow lr Slr for l ¼ 1; . . . ; np; 1 þ T l N sT l
ð3Þ
where l is the phase number, T l is a positive input parameter based on an experimental observation of the relation between residual saturations and trapping number, Slow and Shigh are the residual lr lr phase saturations at low and high trapping numbers, respectively. s is a parameter dependent on the pore size distribution of the rock. Similar carbonate rocks were found in the literature having about the same residual oil saturation values at the applied capillary number in the laboratory [20,21]. The composite carbonate cores used in Yousef et al.’s work [12] are similar to sample K5 in the work reported by Kamath et al. [18]. This is evident from the capillary desaturation curve (CDC) (Fig. 1) and the Sor curve as a function of the applied pressure gradient (Fig. 2). In both figures, at a similar applied pressure gradient of 50 psi/ft and capillary number of the order of 106, the resulting residual oil saturations are almost identical. Moreover, a study was conducted by Abrams [21] on an Indiana carbonate rock sample with similar rock properties and capillary desaturation curve. These CDCs were used to find suitable UTCHEM input parameters for modeling the dependence of residual oil saturation on caphigh illary numbers. This was done by using Slow of zero, s of lr of 0.35, Slr 0.8, and Tl of 33,000 (Fig. 3). Both relative permeability endpoints and exponents change because of the change in residual saturation at high trapping numbers according to Delshad et al. [22]: olow
K orl ¼ krl
þ
nl ¼ nlow þ l
0 Slow l 0 r Sl r
Slow l0 r
ohigh
Shigh l0 r
0 Slow l 0 r Sl r
high Slow l0 r Sl0 r
ðkrl
olow
0
krl Þ for l; l ¼ 1; . . . ; np;
ð4Þ
0
ðnhigh nlow l Þ for l; l ¼ 1; . . . ; np: l
ð5Þ
Table 1 Endpoint relative permeability data analysis (first coreflooding experiment). Oil viscosity Composite core length Cross sectional area Injection rate Absolute brine permeability Injection cycle First Second Third Fourth Fifth
0.691 (0.691 103) 0.5328 (0.1624) 1.22 102 (1.13 103) 1 39.6 (3.96 1014) Water viscosity cp (Pa.s) 0.272 (0.272 103) 0.242 (0.242 103) 0.232 (0.232 103) 0.212 (0.212 103) 0.193 (0.193 103)
cp (Pa.s) ft (m) ft2 (m2) cc/min mD (m2) Pressure drop psi (Pa) 6.90 (4.76 104) 6.19 (4.27 104) 6.03 (4.16 104) 6.10 (4.21 104) 5.79 (3.99 104)
Oil–water IFT
34
dynes/cm
Initial pressure drop
23 (1.59 105)
psi (Pa)
Sor 0.295 0.233 0.151 0.135 0.135
kro 0.266 – – – –
krw 0.349 0.346 0.341 0.308 0.295
14
E.W.A. Shalabi et al. / Fuel 121 (2014) 11–19
3. Simulation data This section includes a description of the simulation models used to history match both corefloods using the UTCHEM simulator. 3.1. Description of models
Fig. 1. Remaining oil saturation as a function of capillary number, white circles represent the unsteady restored state (Kamath et al. [18]).
The first coreflooding model used a 1D Cartesian grid system with dimensions of 20 1 1; width and thickness are the same (3.3677 cm) for all grid blocks; however, lengths of grid blocks vary to match the length dimensions of the four core plugs used in the first coreflooding experiment (Table 2). The heterogeneous model maintained the permeability of each core plug as shown in Fig. 4. A similar model was used for the second coreflooding experiment, with dimensions of 30 1 1; width and thickness are the same (3.3765 cm) for all grid bocks; however, lengths of grid blocks also vary to match the length dimensions of the six core plugs used in this case. Table 3 summarizes the composite core simulation models used for both corefloods. The results of only the first coreflooding experiment are presented in this paper due to space limitations.
4. Results and discussion Fig. 2. Remaining oil saturation as a function of pressure gradient, white circles represent the unsteady restored state (Kamath et al. [18]).
This section covers seawater cycle matching and methods used for the wettability alteration effect, matching for different dilutions of seawater injected cycles. The analysis for both first and second coreflooding experiments was done systematically without using the wettability alteration option in UTCHEM. 4.1. Seawater cycle match
Fig. 3. Modeled CDC curve for the first coreflooding experiment. ohigh
For the water phase, krw was assumed to be 0.4 for the first coreflooding experiment and 0.5 for the second coreflooding experiment; these values are reasonable because the endpoint relative permeability of water begins to decrease with successive olow injections due to wettability alteration by LSWI. From Eq. (4) krw o value was calculated using the known value of krw for the first cycle in Table 1. For the oil phase, endpoint relative permeability and Corey’s exponent remained constant because of the assumed constant irreducible water saturation.
Experimental results of Yousef et al.’s work concluded that substantial oil recovery by LSWI is due to a fluid-rock property (wettability alteration), not a fluid–fluid property (reduction in IFT). Moreover, the results showed that wettability alteration occurs in the second and subsequent cycles. Therefore, the first seawater cycle is matched without taking into account the wettability alteration effect. The seawater cycle was matched, taking into account that the critical capillary number was exceeded for composite carbonate cores used, as was discussed previously. The capillary pressure effect was neglected because of the high pressure gradient applied by setting the CPC parameter to zero. Hence, the data was matched by changing only Corey’s exponents, because water and oil endpoint relative permeability values are known for the seawater cycle. Reported residual oil saturations [12] of 0.295 and 0.221 were used as input values for the first and the second corefloods, respectively.
Table 2 Composite core model data (first coreflooding experiment). Parameter
Value
Comments
Number of grid blocks
20
1D (20 1 1)
Grid block sizes (DI, DJ, DK), ft (m)
X direction: 1–5, Dx is 0.0279 (0.0085) 6–10, Dx is 0.0258 (0.00786) 11–15, Dx is 0.0264 (0.00804) 16–20, Dx is 0.0265 (0.00808) Y direction: 1–1, Dy is 0.110489 (0.033677) Z direction: 1–1, Dz is 0.110489 (0.033677)
Variable grid size on a regional basis, constant grid size in the y and z directions, however, variable grid size for the x direction.
Composite core model dimensions, ft (m)
0.5328 (0.1624) 0.110489 (0.033677) 0.110489 (0.033677)
Length width thickness
E.W.A. Shalabi et al. / Fuel 121 (2014) 11–19
15
Fig. 4. Simulation model used in different runs with heterogeneous permeability (first coreflooding experiment).
Table 3 Summary of composite core model for both experiments. Property
First coreflooding
Second coreflooding
Porosity Pore volume ft3 (m3)
0.2510 1.62 103 (4.59 105) 0.1247 (0.0380) 0.5328 (0.1624) 6.50 103 (1.84 104) 1.22 102 (1.13 103) 20 1 1
0.2465 2.35 103 (6.65 105) 0.1250 (0.0381) 0.7759 (0.2365) 9.53 103 (2.70 104) 1.23 102 (1.14 103) 30 1 1
Diameter ft (m) Length ft (m) Bulk volume ft3 (m3) Cross sectional area ft2 (m2) Model dimensions (length width height)
Grid block dimensions (variable length) Width ft (m) 0.110489 (0.033677) Height ft (m) 0.110489 (0.033677)
0.110778 (0.033765) 0.110778 (0.033765)
For the first coreflooding experiment, the calculated capillary number values for the laboratory, field, and UTCHEM along with the corresponding applied pressure gradients are listed in Table 4. The field pressure gradient was considered as 0.5 psi/ft. There was no direct control on capillary number values using UTCHEM because the experiments were run in the constant injection rate mode. Nevertheless, the obtained results are close enough to those applied in the laboratory. The sets of relative permeability curves used before and after exceeding Nc are shown in Table 5 and Fig. 5. We assumed a negligible change in the irreducible water saturation. Moreover, the oil endpoint relative permeability value used in data matching is lower than the water endpoint relative
permeability, which is acceptable for a mixed-to-oil-wet carbonate core. The possibility of having a lower kro compared to krw for oilwet rocks is supported by the results of Masalmeh [23], in which Middle East carbonate rocks of the same nature were used showing that for an oil-wet rock, kro is lower than krw . For both coreflooding experiments, the first cycle data including oil recovery and pressure drop were matched. The pressure drop was matched using stabilized and unstabilized pressure drop data as discussed earlier (Fig. 6); however, the oil recovery match is only shown for the stabilized pressure case (Fig. 7). The simulated and the Buckley Leverett Theory calculated fractional flow curves were matched and they showed similar water breakthrough points (Fig. 8).
4.2. Dilutions of seawater injected cycles match Wettability alteration could be the reason for incremental oil recovery because of the injection of various dilutions of seawater from the second cycle to the fifth. We systematically simulated the wettability alteration effect by using two methods to match the experimental data for the various dilutions of seawater injection cycles. Parameters accounting for relative permeability and residual oil contributions are included in the matching, while capillary pressure contribution was ignored due to the high pressure gradient applied. For the sake of simplicity, we ignored the dependence of residual oil saturation on capillary number due to the wettability alteration effect, which changes the capillary desaturation curve at each cycle. In the first and second methods, the data was matched by tuning relative permeability parameters and taking into account the reported residual oil saturation, without using the wettability alteration option in UTCHEM. A detailed description of each method is included in this section.
16
E.W.A. Shalabi et al. / Fuel 121 (2014) 11–19
Table 4 Capillary number and pressure gradient calculations (first coreflooding experiment). Maximum delta P (Lab) Maximum delta P (UTCHEM)
30 (2.06 105) 22 (1.52 105)
Nc calculations Laboratory Field UTCHEM
1.48E06 1.32E08 7.35E07
Applied pressure gradient Laboratory Field UTCHEM
56.31 (1.27 106) 0.5 (1.13 104) 41.29 (9.34 105)
psi (Pa) psi (Pa)
psi/ft (Pa/m) psi/ft (Pa/m) psi/ft (Pa/m)
Table 5 Relative permeability parameters before and after exceeding Nc (first coreflooding experiment). Fig. 7. Oil recovery match for seawater cycle (first coreflooding experiment). First cycle matching parameters Below Nc (critical) nw no krw kro Sor Swirr
Exceeding Nc (critical) 2.19 2.00 0.34 0.266 0.35 0.1044
nw no krw kro Sor Swirr
2.00 2.00 0.349 0.266 0.295 0.1044
Fig. 8. Fractional flow match for seawater cycle (first coreflooding experiment).
Fig. 5. Relative permeability curves before and after exceeding N c (first coreflooding experiment).
inputting the laboratory reported residual oil saturation for each cycle. This method is called the Sor contribution because it shows the significance of residual oil saturation in matching the experimental data. Results showed that residual oil saturation is not the main contributor to low salinity water injection because matching the experimental data was not possible. Also, the results reflect the need to change the relative permeability parameters for oil recovery and pressure drop data matching. A better matching method is discussed in the next section. 4.2.2. Second method The second method is called the kr and Sor contributions. Here three approaches are proposed to match the experimental data: change Corey’s exponents while keeping endpoint relative permeability values constant, change endpoint relative permeability values while keeping Corey’s exponents constant, and change both endpoint relative permeability values and Corey’s exponents. Each approach is described in detail in the next section highlighting the related pros and cons.
Fig. 6. Pressure drop match for seawater cycle (first coreflooding experiment).
4.2.1. First method In this method, we used the same relative permeability parameters including endpoints and Corey’s exponents for the seawater cycle matching. The oil recovery of each cycle was matched by
4.2.2.1. First approach. In this approach, relative permeability endpoints were kept constant and Corey’s exponents for water and oil were tuned to match both sets of experimental data. This approach is proposed in light of Seccombe et al.’s work [24], where constant water relative permeability was observed at residual oil saturation for both low and high salinity water injections. The observation was made from coreflooding experiments, numerical matching of
17
E.W.A. Shalabi et al. / Fuel 121 (2014) 11–19
Fig. 9. Overall oil recovery match using second method (First approach – first coreflooding experiment).
Fig. 10. Overall pressure drop match using second method (First approach – first coreflooding experiment).
data, and the constant productivity of the wells from the SWCTTs at the Endicott field in the North Slope of Alaska. Corey’s exponents were tuned while keeping the endpoints relative permeability constant at seawater cycle values for the case when Nc critical was exceeded. The overall oil recovery obtained and the pressure drop data matching using this approach are depicted in Figs. 9 and 10, respectively. A summary of the relative permeability parameters used for both sets of experiments is presented in Table 6, showing a shift towards a more water-wet state, which is associated with higher oil recovery. This approach is not suitable for matching both sets of experimental data for the following reasons: High nw values reach 9, while the maximum typical acceptable value is about 4. Low no values reach 1.25, while the minimum typical acceptable value is about 1.5.
Fig. 11. Overall oil recovery match using second method (Second approach – first coreflooding experiment).
Fig. 12. Overall pressure drop match using second method (Second approach – first coreflooding experiment).
Water fractional flow values for the fifth cycle are higher than for the fourth cycle, which does not make sense physically. Jumps are noticed in the pressure drop data matching.
4.2.2.2. Second approach. This proposed approach to better match both sets of experimental data involves tuning the endpoint relative permeability values for water and oil, while keeping Corey’s exponents constant at seawater cycle values when Nc critical is exceeded. The water endpoint relative permeability values are taken from Table 1; however, the oil endpoint relative permeability values are tuned to match each cycle. For the first coreflood, the obtained match using this approach for overall oil recovery and pressure drop are depicted in Figs. 11 and 12. For both sets of experiments, a summary of the relative permeability parameters used in the second approach is presented in Table 7. The second approach is also not suitable because the overall oil recovery curve is not well matched especially at the starting
Table 6 Summary of relative permeability parameters for both sets of experiments (Second method – first approach). Injection cycle
First cycle Second cycle Third cycle Fourth and fifth cycles
First coreflooding experiment
Second coreflooding experiment
nw
no
nw
no
2 6 7 7.5
2 1.5 1.3 1.25
2 8 8.5 9
2 1.35 1.3 1.25
18
E.W.A. Shalabi et al. / Fuel 121 (2014) 11–19
Table 7 Summary of relative permeability parameters for both sets of experiments (Second method – second approach). Injection cycle
First coreflooding experiment
First cycle Second cycle Third cycle Fourth cycle Fifth cycle
Second coreflooding experiment
krw
kro
krw
kro
0.349 0.346 0.341 0.308 0.295
0.266 0.800 0.900 0.95 0.970
0.466 0.440 0.410 0.390 0.359
0.367 0.800 0.950 0.960 0.970
Fig. 15. Relative permeability curves using second method (Third approach – first coreflooding experiment).
Fig. 13. Overall oil recovery match using second method (Third approach – first coreflooding experiment).
Fig. 16. Fractional flow curves using second method (Third approach – first coreflooding experiment).
Table 8 Summary of relative permeability parameters for both sets of experiments (Second method – third approach). Injection cycle
First coreflooding experiment
Fig. 14. Overall pressure drop match using second method (Third approach – first coreflooding experiment).
humps, and because the fractional flow curve of the fifth cycle is higher than the fourth cycle, which is not reasonable. Nevertheless, some good features of this approach are as follows:
First cycle Second cycle Third cycle Fourth cycle Fifth cycle
Second coreflooding experiment
krw
kro
nw
no
krw
kro
nw
no
0.349 0.346 0.341 0.308 0.295
0.266 0.700 0.840 0.860 0.880
2.00 2.20 2.50 2.70 2.90
2.00 1.70 1.59 1.57 1.55
0.466 0.440 0.410 0.390 0.359
0.367 0.750 0.860 0.870 0.880
2.00 2.20 2.60 2.70 2.90
2.00 1.60 1.56 1.54 1.53
Endpoint relative permeability values are reasonable. There are no jumps in the pressure drop data matching. This approach reflects the need to tune Corey’s exponents to match the humps in oil recovery curves. The approach shows that endpoint relative permeability values are more sensitive to the LSWI effect compared to Corey’s exponents, which is supported by the absence of jumps in the pressure drop data curves.
water relative permeability values used for each cycle are listed in Table 1; however, the oil endpoint relative permeability along with Corey’s exponents are adjusted to match the data. For the first coreflooding experiment, the obtained match for overall oil recovery, pressure drop data, the sets of relative permeability curves and fractional flow curves using this approach are shown in Figs. 13– 16. A summary of the relative permeability parameters used in this approach for both experimental setups is presented in Table 8. The third approach is the best and most reasonable for both oil recovery and pressure drop data matching due to the following:
4.2.2.3. Third approach. A more realistic approach to better match both sets of experimental data suggests tuning both endpoint relative permeability values and Corey’s exponents. The endpoint
Both Corey’s exponents and relative permeability endpoints were tuned. Both Corey’s exponents and relative permeability endpoint values are reasonable.
E.W.A. Shalabi et al. / Fuel 121 (2014) 11–19
Overall oil recovery curves are well matched including the humps at the start of each cycle. Pressure drop curves are well match with no jumps. Water fractional flow curve values for the fifth cycle are lower than for the fourth cycle as expected. This approach shows that oil endpoint relative permeability is more sensitive to the LSWI effect than is water endpoint relative permeability. 5. Conclusions Oil recovery and pressure drop data for both first and second coreflooding experiments of Yousef et al.’s [12] work were matched successfully using UTCHEM. The findings of this work are as follows: Wettability alteration is still believed to be the reason for the LSWI effect on oil recovery. The change in residual oil saturation due to the LSWI is dominated by wettability alteration. Altering relative permeability parameters, endpoints and Corey’s exponents, is essential for history matching. Water and oil endpoint relative permeabilities are more sensitive to the LSWI effect than are Corey’s exponents. Oil endpoint relative permeability is more sensitive to the LSWI effect than is water endpoint relative permeability. In our future work, the wettability alteration option in UTCHEM will be used for history matching to provide more insight into the low salinity water injection mechanism. Acknowledgements The authors wish to acknowledge useful discussions with G.A. Pope during the course of this work. This work was funded by Abu Dhabi National Oil Company (ADNOC). References [1] Smith KW. Brines as flooding liquids. Paper presented at seventh annual tech. meeting, Min. Ind. Expt. Sta., Penn. State College; 1942. [2] Bernard GG. Effect of floodwater salinity on recovery of oil from cores containing clays. In: SPE 1725, SPE California Regional Meeting, Los Angeles, California; 1967. [3] Al-Mumen AA. The Effect of Injected Water Salinity on Oil Recovery. MSc Thesis, King Fahad University of Petroleum and Minerals. Dhahran, Saudi Arabia; 1990. [4] Tang GQ, Morrow NR. Salinity temperature, oil composition and oil recovery by waterflooding. In: SPE 36680, SPE Reservoir Eng 1997; 12(4): 269–76.
19
[5] Zhang P, Tweheyo MT, Austad T. Wettability alteration and improved oil recovery by spontaneous imbibition of seawater into chalk: impact of the potential determining ions Ca2+, Mg2+, and SO2 4 . Colloids Surf A: Physicochem Eng Aspects 2007;301(1–3). 199-128. [6] Lager A, Webb KJ, Black CJ. Impact of brine chemistry on oil recovery. In: 14th European symposium on IOR, Cairo, Egypt; 2007. [7] Lager AK, Webb KJ, Collins IR, Richmond DM. LoSalTM enhanced oil recovery: evidence of enhanced oil recovery at the reservoir scale. In: SPE 113976, SPE/ DOE symposium on improved oil recovery, Tulsa, Oklahoma; 2008. [8] Webb KJ, Black CJJ, Edmonds IJ. Low salinity oil recovery – the role of reservoir condition core floods. In: 13th European symposium on improved oil recovery, Budapest, Hungary; 2005a. [9] Webb K, Lager A, Black C. Comparison of high/low salinity water/oil relative permeability. In: International symposium of the society of core analysts, Abu Dhabi, UAE; 2008. [10] Doust AR, Puntervold TP, Strand S, Austad, TA. Smart water as wettability modifier in carbonate and sandstone. In: 15th European symposium on improved oil recovery, Paris, France; 2009. [11] Ligthelm DJ, Gronsveld J, Hofman JP, Brussee NJ, Marcelis F, van der Linde HA. Novel waterflooding strategy by manipulation of injection brine composition. In: SPE 119835, EUROPEC/EAGE conference and Exhibition, Amsterdam, The Netherlands; 2009. [12] Yousef AA, Al-Saleh S, Al-Kaabi A, Al-Jawfi M. Laboratory investigation of novel oil recovery method for carbonate reservoirs. In: SPE 137634, Canadian unconventional resources and international petroleum conference, Calgary, Alberta, Canada; 2010. [13] Hognesen EJ, Strand S, Austad T. Waterflooding of preferential oil-wet carbonates: oil recovery related to reservoir temperature and brine composition. In: SPE 94166, SPE Europec/EAGE Annual Conference, Spain; 2005. [14] Webb KJ, Black CJJ, Tjetland G. A laboratory study investigating methods for improving oil recovery in carbonates. In: SPE 10506, International petroleum technology conference, Qatar; 2005b. [15] Gupta R, Griffin P, Hu L, Willingham TW, Cascio ML, Shyeh JJ, Harries CR. Enhanced waterflood for middle east carbonates cores – impact of injection water composition. In: SPE 142668, SPE Middle East Oil and Gas Conference, Manama, Bahrain; 2011. [16] Corey AT, Rathjens CH, Henderson JH, Wyllie MR. Three-phase relative permeability. J Petrol Technol 1956;8(11):63–5. [17] Brooks RH, Corey AT. Properties of porous media affecting fluid flow. J Irrigation Drain Div 1966;92(6):61–88. [18] Kamath J, Rober FM, Frank MN. Understanding waterflood residual oil saturation of four carbonate rock types. In: SPE 71505, SPE annual technical conference and exhibition, New Orleans, Louisiana; 2001. [19] Jin M. A study of nonaqueous phase liquid characterization and surfactant remediation. Ph.D. Dissertation, The University of Texas at Austin; 1995. [20] Delshad M. Trapping of micellar fluids in berea sandstone. Ph.D. Dissertation, The University of Texas at Austin; 1990. [21] Abrams A. The influence of fluid viscosity, interfacial tension, and flow velocity on residual oil saturation left by waterflood. SPEJ 1975;15(5):437–47. [22] Delshad M, Bhuyan D, Pope GA, Lake L. Effect of capillary number on the residual saturation of a three-phase micellar solution. In: SPE 14911, SPE enhanced oil recovery symposium, Tulsa, Oklahoma; 1986. [23] Masalmeh SK. The effect of wettability on saturation functions and impact on carbonate reservoirs in the middle east. In: SPE 78515, Abu Dhabi international petroleum exhibition and conference. Abu Dhabi, UAE; 2002. [24] Seccombe JC, Lager A, Webb K, Jerauld G, Fueg E. Improving waterflood recovery: LoSalTM EOR field evaluation. In: SPE 113480, SPE symposium on improved Oil recovery. Tulsa, Oklahoma, USA; 2008.