- Email: [email protected]
Evaluation of the EOR potential of gas and water injection in shale oil reservoirs
Journal of Unconventional Oil and Gas Resources 5 (2014) 1–9
Contents lists available at ScienceDirect
Journal of Unconventional Oil and Gas Resources journal homepage: www.elsevier.com/locate/juogr
Evaluation of the EOR potential of gas and water injection in shale oil reservoirs James J. Sheng ⇑, Ke Chen 1 Bob L. Herd Department of Petroleum Engineering, Texas Tech University, P.O. Box 43111, Lubbock, TX 79409, United States
a r t i c l e
i n f o
Article history: Received 13 September 2013 Revised 4 December 2013 Accepted 5 December 2013 Available online 21 December 2013 Keywords: Gas flooding Waterflooding Shale oil Enhanced oil recovery Unconventional oil and gas resources Simulation
a b s t r a c t With the relatively modest natural gas price, producing oil from unconventional shale reservoirs has attracted more and more interest from oil operators. Although many tremendous efforts have been made to develop shale resources, the ultimate oil recovery is still low (5–10%). Because of the important role of shale resources in the future oil and gas industry, more stimulation and production strategies are being considered and tested to find ways to improve oil production from shale reservoirs. Before applying a specific method to enhance oil recovery (EOR) in the field or conducting a research in laboratory, the simulation approach is a cheap and fast approach to evaluate its EOR potential. The analysis of simulation results will be helpful in designing laboratory experiments and field testing. In this paper we use a simulation approach to evaluate the EOR potential in shale oil reservoirs by gas flooding and waterflooding. Production behavior and oil recovery of different schemes are discussed through sensitivity studies. Simulation results of primary production, gas injection and water injection are compared. Results show that miscible gas injection has a higher potential to improve oil recovery from shale oil reservoirs than water injection. Gas injection above a minimum miscible pressure (MMP) can be fully miscible with oil, thus reducing oil viscosity significantly, in addition to the mechanism of pressure maintenance. Simulation results indicate that the oil recovery factor can be increased up to 15.1% by gas injection in a hydraulically fractured shale reservoir, whereas the oil recovery factor from the primary depletion is only 6.5%. The oil recovery from waterflooding is about 11.9% which is lower than that from gas flooding. The results indicate that miscible gas flooding could be a way to enhance oil recovery in shale oil reservoirs. Ó 2013 Elsevier Ltd. All rights reserved.
Introduction Unconventional shale reservoirs are generally produced by stimulation techniques. A horizontal well with multiple transverse fractures has proven to be an effective technique for shale gas and shale oil production. However, shale oil production faces more challenges compared with shale gas production. Even applying multi-stage hydraulic fracturing techniques, the final oil recovery factors using existing methods are only a few percent. Oil rate and reservoir pressure drop very quickly. Because most of oil remains unrecoverable even using the expensive horizontal well drilling combined with hydraulic fracturing, seeking the ways to enhance oil recovery will be the continuous efforts. In addition to drilling horizontal well and hydraulic fracturing stimulation, several EOR methods have been proposed or tested. Wan et al. (2013a,b) evaluated cyclic gas injection to improve shale oil recovery. Chen et al. (2013) studied the effect of reservoir
⇑ Corresponding author. Address: Bob L. Herd Department of Petroleum Engineering, Texas Tech University, P.O. Box 43111, Lubbock, TX 79409, United States. Tel.: +1 (806) 834 8477. E-mail address: [email protected] (J.J. Sheng). 1 Present address: PetroChina. 2213-3976/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.juogr.2013.12.001
heterogeneity on cyclic CO2 injection. Gamadi et al. (2013) did an experimental study of cyclic gas injection in shale rocks. Apparently, no gas injection has been reported to have been applied in shale oil reservoirs. Takahashi and Kovscek (2009) investigated the impact of different brine formulations in tight formations. Makhanov et al. (2012) showed that the imbibition could be a viable mechanism to transfer fluids from fracture to matrix in Horn River shales. Morsy et al. (2013a–e) studied the effects of water compositions (salinity, acid and alkali) on the EOR potential of water imbibition. Fakcharoenphol et al. (2013) pointed out that waterflooding changes the formation in situ stress due to the increase in reservoir pressure and decrease of reservoir temperature, which enhances oil recovery of shale formations by reactivating existing natural fractures and/or creating new fractures. Although waterflooding is a mature secondary recovery method for conventional reservoirs, it has not been applied in shale oil and gas reservoirs in a large commercial scale. A waterflooding pilot was started in Bakken shale in late 2006, (Wood and Milne, 2011). Because of ultra-low permeability of shale reservoirs, either waterflooding or gas flooding may have issues of injectivity and fluid transport. In this paper, we initiate a simulation study to evaluate whether gas flooding and waterflooding could have a potential
2
J.J. Sheng, K. Chen / Journal of Unconventional Oil and Gas Resources 5 (2014) 1–9
to improve oil recovery in shale oil reservoirs. The EOR potentials of gas injection and water injection are evaluated. A further sensitivity study is conducted to improve the understanding of the mechanism of gas and water injection in shale reservoirs. Before we present simulation results, the base simulation model is described next.
Base model setup Rubin (2010) used an extremely fine grid model as a reference model to simulate fracture flow. In the reference model, small cells of the actual width of fractures (assumed 0.001 ft) are used to capture the flow from the matrix to the fractures. He also showed that it is possible to accurately model flow from a fractured shale reservoir using logarithmically spaced, locally refined grids with fractures represented by 2.0-ft-wide cells and by maintaining the same conductivity (kfwf) as that of the original 0.001-ft-wide fracture. Using Robin’s approach, we can use much less cells to simulate fracture-related flow. Wan (2013) used Robin’s approach to have built a model of 200 ft long, 1000 ft wide and 200 ft thick to simulate a horizontal well and one transverse fracture (Fig. 1). 2-ft wide grid cells with 83.3 md-ft conductivity (k = 41.65 md, wf = 2 ft) were used to represent the actual fractures of 0.001-ft wide and 83.300 md. The reservoir property data used by Wan are similar to the Eagle Ford shale data (Table 1). The initial reservoir pressure is 6425 psi. The permeability is 100 nano-Darcy. The porosity is 0.06. During the primary production process, the well is controlled by bottom-hole pressure (BHP) of 2500 psi. These data and the approach are used in the base model as well. The objective of this study is to evaluate the potential of gas and water injection for improving oil production from shale oil reservoirs. Modeling a whole reservoir requires a large number of grid blocks, and it is of course time-consuming to model these complex fracture networks. Therefore, we built a small model which is 200 ft long, 1000 ft wide and 200 ft thick (Fig. 2). This small model is produced with two half-vertical wells connected with two halffractures, respectively. Thus each fracture is 1-ft wide and has a conductivity of 46.65 md-ft. We use such a simple model to simulate the flow between the two lateral hydraulic fractures of
Fig. 1. Cyclic gas injection simulation model used in Wan (2013).
Table 1 Reservoir properties used in the base model. Initial reservoir pressure Porosity of shale matrix Initial water saturation Compressibility of shale Shale matrix permeability Reservoir temperature Gas specific gravity Reservoir thickness Bubble point for oil
6425 psi 0.06 0.3 5106 psi1 0.0001 md 255 °F 0.8 200 ft 2398 psi
a horizontal well. Assume the flow between any two lateral fractures is the same, such small model can be used to simulate the flow through a part of a horizontal well. The reservoir properties data used in this model is shown in Table 1. Before conducting gas or waterflooding simulation using this small base model, we need to validate the model first. In Wan’s model, one vertical well was used to connect one lateral vertical fracture. The fracture is 2-ft wide and has a conductivity of 83.3 md-ft. In the base model, we use two vertical wells to connect two half-fractures, because we need to use one vertical well to represent an injection well and the other well to represent a production well, so that an injection-production pattern is built in the model. The area or volume of our base model is the same as that in Wan’s model. If the injector is changed to a producer in our base model, the oil production from our model should be same as that from Wan’s model. Because Wan’s model has been validated in his thesis, if the oil production performance is the same as that from Wan’s model, then our model is validated. This validation has been performed in Chen’s thesis (Chen, 2013), and it is not repeated here because of limited space allowed for this paper. In the above base model, the miscibility model proposed by Todd and Longstaff (1972) is used. Todd and Longstaff (1972) proposed a method of simulating miscible displacement performance without considering detailed compositions. They introduced a mixing parameter x, which determines the degree of mixing between the miscible fluids within a grid block. A value of zero corresponds to the immiscible displacement, whereas a value of one corresponds to complete mixing. The mixing of solvent and oil is controlled by a pressure-dependent mixing parameter, xo (Omegaos in the simulator IMEX used in this study). IMEX, the black oil simulator, is developed by Computer Modeling Group (CMG). If the individual block pressure is much lower than the minimum miscibility pressure (MMP), xo = 0.0, and gas is displacing oil immiscibly. As the block pressure increases, this mixing parameter reaches its maximum value xomax at the MMP, as shown in Fig. 3. When no better data is available, the user manual of the simulation software IMEX suggests a value in the range of 0.5–0.8 as a first approximation. xo is considered to be a function of pressure and xomax is set as 0.72 in the base model. The black oil simulator IMEX uses 4-components but three phases to simulate miscible flooding. The 4-components are water, oil, dissolved gas and injected gas (solvent). The three phases are water, oil and gas which includes dissolved gas and injected gas in the free gas state. Both the injected gas and the original dissolved gas have the same properties. The specific gravity (ratio of the gas density to the air density) is 0.8. The oil compressibility is 1105 psi1. The mixing of solvent and free gas is governed by xg (OMEGASG in IMEX), which is assumed pressure independent. xg is bounded by zero and one. Since solvent-gas has a lower mobility ratio than oil, xg is usually greater than xomax. In our case OMEGASG is set as 1.0, assuming solvent and free gas have a complete mixing. In this base simulation model, the injection is controlled by the maximum solvent injection rate for a half well of 400 Mscf/day and the maximum injection pressure of 7000 psi.
J.J. Sheng, K. Chen / Journal of Unconventional Oil and Gas Resources 5 (2014) 1–9
3
Fig. 2. Base reservoir model.
Fig. 4. Average reservoir pressure and oil recovery factor vs. time (Scenario G1).
Fig. 3. x vs. P.
For the production well, the flowing bottom-hole pressure is 2500 psi which is above the bubble point pressure of 2398 psi. Simulation results and analysis In this section, gas injection and waterflooding scenarios are presented and analyzed. Gas injection We first compare several gas injection scenarios. Scenario G1: 3600 days of primary production followed by 60 years of gas flooding production In this scenario, gas injection starts after 3600 days (10 years) of primary production. Figs. 4 and 5 show the results for oil recovery factor, average pressure and oil rate versus time. The reservoir pressure decreases fast from the initial reservoir pressure of 6425 psi to 3000 psi as the reservoir is mainly in depletion drive in the first 10 years’ primary production. Once gas is injected, the reservoir pressure increases from 3000 psi to about 5200 psi gradually. Because the high pressure gradient exists between the injector and pressure owing to the ultra-permeability, the pressure in the producer side is low, and the average reservoir pressure is
Fig. 5. Oil production rate vs. time (Scenario G1).
always lower than the initial reservoir pressure during the injection and production period. From the oil production rate graph (Fig. 5), the oil rate decreases from the initial rate 27.47 bbl/day to 10.26 bbl/day after 200 days of production and to 2.72 bbl/day within 5 years during the primary production. At the end of primary production period, the oil rate is 0.57 bbl/day from two half-fractures. After gas injection is started, the oil rate gradually increases to 1.3 bbl/day from one half fracture. At the end of 60 years of gas injection, 37.912 MSTB oil is produced, corresponding to an oil recovery factor of 15.12% as shown in Fig. 4.
4
J.J. Sheng, K. Chen / Journal of Unconventional Oil and Gas Resources 5 (2014) 1–9
The pressure data and oil rate data are further analyzed. At the end of primary production, the average reservoir pressure is about 3000 psi. The well BHP is 2500 psi. The pressure drawdown is about 500 psi. During the gas injection period, the average reservoir pressure is about 5200 psi. Then the pressure drawdown becomes 2700 psi which is about 5 times that at the end of primary production. During this time, the well rate from one half-fracture is 1.3 bbl/day which is about 2.3 times that rate (0.57 bbl/day) at the end of primary production. Because 1.3 bbl/day is from one half-fracture, then the rate from two half-fractures would be 2.6 bbl/day if both of the half-fractures produced. Then the rate of 2.6 bbl/day is 4.6 times the rate of 0.57 bbl/day from the two half-fractures. This rate ratio is close the pressure drawdown ratio of about 5 (the pressure drawdown during the injection period to that at the end of primary production). From the Darcy rate equation (Eq. (1)), if the pressure drawdown Dp is increased 5 times, the rate is also increased about 5 times only if the viscosity l does not change, because the other parameters in Eq. (1) (permeability k, flow area A and distance DL from the matrix to fracture) are almost unchanged.
q¼
kAðDpÞ : lðDLÞ
ð1Þ
From this analysis, it is clear that the main mechanism from gas injection is mainly pressure maintenance. Figs. 6 and 7 show the pressure variation and oil saturation distribution during the production period for Scenario G1, respectively. When starting gas injection process, the solvent is injected into the reservoir through injection well and mix with reservoir fluids, leading to oil viscosity decrease. Oil is pushed away from the injection well as shown in the saturation map. In the meantime the reservoir pressure builds up. Owing to the ultra-low permeability, fluid transport in such kind of reservoirs is much more difficult than that in conventional reservoirs. This also results in a small increase in oil rate after gas injection owing to pressure increase as just discussed. Fig. 7 also shows that the gas is only in the injector side in the first 50 years. Gas could not reach the producer. Near the producer, it is oil only. Therefore, the viscosity in Eq. (1) is the oil viscosity which is almost unchanged. The viscosity decreases owing to gas injection occurs near the injector
not the producer. As a result, the viscosity decrease mechanism owing to the gas and oil miscibility is not significant. Scenario G2: 3600 days of primary production followed by 60 years of cyclic gas flooding production For this scenario, cyclic gas injection begins after 3600 days (10 years) of primary production. Each injection cycle has 5 yearinjection and 5 year-shut in. The recovery factor at the end of 70 years is 14.42%. Scenario G3: 70 years of gas flooding production In this scenario, gas injection is started at the beginning of the development. Keep gas flooding and oil production simultaneously for 70 years which is the total production period in the previous scenarios. Because of gas injection, the reservoir pressure maintain high around 5000 psi. However, the total oil recovery is 13.48%. We have considered three production scenarios so far. In the first scenario, gas injection begins after 10 years of primary production and continues for 60 years. 74.245 MMSCF of gas is injected which corresponds to 30.7 pore volume (PV = 430.504 Mrbbls), and 37.912 MSTB of oil is produced which corresponds to 15.12% oil recovery factor. If we use a 10% discount rate, the produced oil of 37.912 MSTB is 11.893 MSTB at the present value. In the second scenario, cyclic gas injection starts after 10 years of primary production. The cyclic process has 5 years of injection and 5 years of shut in. In this process, 64.473 MMSCF (26.67 PV) of gas is used to produce about 14.42% of original oil in place which corresponds to 11.555 MTB at the present value. For the scenario 3, gas injection is implemented at the beginning of the development. It can be seen that the scenario 3 has a lower oil production in the first 10 years because only one well is producing instead of two producing wells in the other two scenarios. As shown in Table 2, the recovery factor from the scenario 3 in the first 10 years of primary production is 3.4%, compared with those in the scenarios 1 and 2 of 5.75%. The recovery factors at the end of 30, 50 and 70 years and the oil produced at the present value during the 70 years are all lower than those of the scenarios 2 and 3. Therefore, it is better to implement gas flooding after several years of natural pressure depletion. The results of three simulation scenarios show that the ultimate recovery factors are not quite different for these three different injection
Fig. 6. Reservoir pressure distribution at different times in the gas injection case (Scenario G1).
5
J.J. Sheng, K. Chen / Journal of Unconventional Oil and Gas Resources 5 (2014) 1–9
Fig. 7. Oil saturation distribution at different times in the gas injection case (Scenario G1).
Table 2 Gas flooding simulation results.
Cumulative oil production (70 years, present value at 10%) Cumulative gas injection (PV) Overall oil recovery (10 years) Overall oil recovery (30 years) Overall oil recovery (50 years) Overall oil recovery (70 years)
Scenario G1 – primary + gas flooding
Scenario G2 – Primary + cyclic gas injection
Scenario G3 – gas flooding only
11.893 MSTB 330.7 5.75% 8.14% 11.49% 15.12%
11.555 MSTB 26.67 5.75% 7.95% 11.05% 14.42%
7.744 MSTB 26.6 3.4% 6.68% 9.97% 13.48%
scenarios, but less solvent is injected in the scenario 2 and the ultimate oil recovery obtained from the scenario 2 is close to that of the scenario 1. In addition, this cyclic gas injection scenario is not optimized. In this case, the cyclic time is 5 years. If the cyclic time is much shorter, the recovery factor will be higher. Wan et al. (2013a) basically used a similar model. The recovery factor is 27.4% when the cyclic time is 100 days. Therefore, cyclic gas injection after some years of primary production may be a better option. However, the down time required to convert the producer into the injector was not taken into calculation here. For an analysis purpose, we simulated 70 years. 70 years may be too long in reality, we present the results at the end of 10, 30, 50 and 70 years. Waterflooding simulation In this section, we present several water injection scenarios similar to the gas injection discussed above. Scenario W1: 3600 days of primary production followed by 60 years of waterflooding production In this production scenario, water injection begins after 3600 days (10 years) of primary production. The production is driven by natural pressure depletion in the first 10 years. The reservoir pressure decreases from 6425 psi to 3000 psi in the primary production period and then gradually increases to about 4250 psi after water injection as shown in Fig. 8. Note that this average reservoir pressure of 4250 psi during the water injection period is lower than that (5200 psi) during the gas injection period in Scenario G1. This is because the water viscosity is higher than gas,
and the pressure gradient near the injection well in the water injection case is higher than that in the gas injection case, which can be convinced by comparing the pressure maps at the end of injection (Figs. 9 and 10 for gas injection and water injection, respectively). From Eq. (1), we would expect that the waterflooding performance may not be as good as gas flooding because the pressure drawdown near the producer in waterflooding is lower. Fig. 11 shows the water injection rate history and oil production rate history. The initial oil rate is 27.47 bbl/day, after 200 days of primary production it decreases to 10.26 STB/day. At the end of primary production, the oil rate is 0.57 STB/d. When water injection begins, the increase in production rate cannot be seen. Because a shale reservoir has an ultra-low permeability, the injection fluids have difficulty to transport from the injection to the producer. The response of the production well to waterflooding is poor. This also results in a low injection rate during the entire injection period except a short early time. During waterflooding, the oil rate is about 0.8 STB/day, and the oil recovery factor is 11.9% at the end. The rate histories (Fig. 11) and the comparison of pressure distribution maps (Figs. 9 and 10) together show that the main problem in waterflooding shale oil reservoirs is low water injectivity.
Scenario W2: 3600 days of primary production followed by 60 years of cyclic water production For this scenario, water injection begins after 3600 days (10 years) of primary production. In this scenario, cyclic water injection is implemented. Each injection cycle has 5 year-injection
6
J.J. Sheng, K. Chen / Journal of Unconventional Oil and Gas Resources 5 (2014) 1–9
Fig. 8. Average reservoir pressure and oil recovery factor vs. time for Scenario W1.
and 5 year-shut in. The overall recovery factor at the end of 70 years is 11.03%.
Scenario W3: 70 years of waterflooding production In this scenario, water injection is started at the beginning of the development. Keep water injection and oil production simultaneously for 70 years. When water injection starts at the beginning of the production, the initial production rate is lower than previous scenarios because only one half-fractured well is producing in the beginning, and the reservoir pressure decreases slowly from initial reservoir pressure to 4000 psi. The ultimate oil recovery factor is 11%. Three production scenarios of waterflooding have been simulated with the results presented in Table 3. In the first scenario, water injection begins after 10 years of primary production and continues for 60 years. 27.020 MSTB of water is injected which corresponds to 0.063 PV, and 29.872 MSTB of oil is produced which corresponds to 11.9% oil recovery factor. In the second scenario, cyclic water injection begins after 10 years of primary production. The cyclic injection process has 5 years of injection and 5 years of shut in. In this process, 21.883 MSTB of water (0.051 PV) is used to produce about 11.03% of original oil in place. For the scenario
Pressure (psi) 2078-12-30 100
200
File: 1inj_1prod_36 User: jsheng Date: 10/26/2013 Scale: 1:481.64478 Z/X: 1.00:1 Axis Units: ft
10,000
10,000
9,900
0
J layer: 27
0.00 0.00 0
100
35.00 10.00
70.00 feet 20.00 meters 2,490
200
Fig. 9. Pressure at the end of gas injection for Scenario G1.
Pressure (psi) 2078-12-30 100
200
File: 1inj_1prod_wa User: jsheng Date: 10/26/2013 Scale: 1:481.64478 Z/X: 1.00:1 Axis Units: ft
10,000
10,000
9,900
0
J layer: 27
0.00 0.00 0
100
35.00 10.00
70.00 feet 20.00 meters
200
Fig. 10. Pressure at the end of water injection for Scenario W1.
2,490
7
J.J. Sheng, K. Chen / Journal of Unconventional Oil and Gas Resources 5 (2014) 1–9 Table 4 Effect of half-fracture length.
Fig. 11. Oil rate & injection rate vs. time for Scenario W1.
3, water injection is implemented at the beginning of the development. We can understand that the scenario W3 has a lower oil production in the first 10 years because only one half-production well is used instead of two half-production wells in the other two scenarios. The recovery factors at the end of 30, 50 and 70 years, and the oil produced at the present value during the 70 years are all lower than those of the scenarios 2 and 3. So it is better to implement waterflooding after several years of natural pressure depletion so that some oil can be produced during early depletion. Similar to the gas injection cases, however, the down time required to convert the producer into the injector was not taken into calculation in this comparison. Further sensitivity analysis The above simulation study shows that gas flooding is a better option than waterflooding. In this section, we will further do some sensitivity analysis of gas flooding. Sensitivity analysis is a quantitative method of determining the important parameters which affect shale oil production performance. The base model is 3600 days of primary production followed by 60 years of gas injection. The parameters considered in this paper include fracture half-length, flowing bottom-hole pressure, matrix permeability and fracture spacing.
Half-fracture Length, ft
Recovery factor, %
500 365 245 125
15.1 12.6 9.1 5.5
ultimate recovery factor. The oil recovery factors at the end of 60 years of gas injection for different half-fracture lengths are shown in Table 4. The data show that the recovery factors are almost a linear function of the half-length, as shown in Fig. 12. In our model, near the fracture, it is almost linear flow. The flow area A in Eq. (1) is proportional to the half-fracture length. Therefore, if the half-fracture length is increased, A is increased proportionally. Thus the flow rate and the resulting oil recovery factor will be increased linearly with the half-fracture length.
Flowing bottom-hole pressure In the base model, we consider a system where the pressure is maintained high enough to guarantee the entire reservoir remains single phase throughout the gas flooding process. The base gas injection model is controlled by flowing bottom-hole pressure (FBHP) which was set up to 2500 psi. Now we will test the production performance under different FBHP. The flowing bottom-hole pressures selected are 1500, 1000 and 500 psi. At the end of 10 years (3600 days) of primary production, the average reservoir pressures are lowered to 2279 and 2392 psi for the flowing bottom-hole pressures of 500 and 1000 psi, respectively. As expected, with a lower flowing bottom-hole pressure, a higher rate and a higher oil recovery factor can be achieved at the end of the primary production period (at 3600 days). After that the oil rate for the lower FBHP is lower than that for the higher FBHP in the early injection period from 4000 days to 8000 days. During this period, the reservoir pressure is below the bubble point pressure. It indicates that the two-phase flow decreases the efficiency of gas flooding. Later gas injection causes the reservoir pressure to rise and become higher than the bubble point pressure.
Fracture half-length The fracture half-length used in the base model is 500 ft. Three other fracture half-lengths of 365, 245, and 125 ft are selected to compare the effect of fracture length on gas flooding production. The simulation result shows that the reservoir pressure decreases faster in the case of longer fracture half-length in primary production period. The average reservoir pressure at the end of 10 years stays lower with longer fracture half-length. A longer fracture length has a higher drainage volume of the reservoir which creates proportionately higher production rates and the gas injection process can have a better effect in maintaining the reservoir pressure which leads to a higher cumulative oil production and a higher
Fig. 12. Oil recovery factor vs. half-fracture length.
Table 3 Waterflooding simulation results.
Cumulative oil production (present value at 10%) Cumulative water injection (PV) Overall oil recovery (10 years) Overall oil recovery (30 years) Overall oil recovery (50 years) Overall oil recovery (70 years)
Scenario W1 – primary + waterflooding
Scenario W2 – primary + cyclic waterflooding
Scenario W3 – waterflooding only
11.627 MSTB 0.063 5.73% 7.59% 9.8% 11.9%
11.555 MSTB 0.051 5.73% 7.21% 9.30% 11.03%
7.578 MSTB 0.056 3.39% 6.41% 8.87% 11.05%
8
J.J. Sheng, K. Chen / Journal of Unconventional Oil and Gas Resources 5 (2014) 1–9 Table 6 Effect of matrix permeability. Matrix permeability, mD 5
510 104 2.5104 4104 5104 103
Recovery factor, % 9.1 15.1 36.1 70.5 92.1 99.3
Fig. 13. Oil rate and oil recovery factor histories for different FBHP.
Then, the miscibility is reached again in the later production time. Then the oil rate with the lower FBHP becomes higher because of a high drawdown. Thus the final oil recovery is higher with lower FBHP, as shown in Fig. 13 and Table 5. Matrix permeability The permeability value used in the base model is 1104 md (100 nano-Darcy). The other permeability values of 1103, 5104, 4104, 2.5104 and 5105 md are selected here. From the simulation results, the average reservoir pressures drop with the matrix permeability after gas injection, as shown in Fig. 14. When gas injection starts, a rapid increase in the average reservoir pressure is observed in the case of 1103 md. After gas breaks though the producer, the reservoir pressure decreases and almost all the oil has been produced. The reservoir pressures in other cases follow the same trend, with the lower permeability, the reservoir pressure is lower right after gas injection and the breakthrough time is longer. The oil recovery factors in these cases after 60 years of gas flooding are shown in Table 6, and in Fig. 15. These data show that oil recovery is very sensitive to matrix permeability. The oil recovery factor almost increases linearly with the matrix permeability.
Table 5 Effect of flowing bottom-hole pressure. FBHP, psi
Recovery factor, %
2500 1500 1000 500
15.1 18.3 19.1 19.7
Fig. 15. Oil recovery factors vs. matrix permeability.
Higher matrix permeability means better hydraulic conductivity which corresponds to a higher oil rate and a higher cumulative oil production. Fracture spacing The distance between fractures (fracturing spacing) in the base case is 200 ft. The other two fracture spacings of 150 and 100 ft are tested to see the influence of fracture spacing on gas flooding performance. For the 150 ft spacing, the reservoir pressure decreases to lower than 3000 psi in first 10 years of primary production because of higher production rate, and then gradually increases to more than 5000 psi after starting gas injection. The oil rate decreases from the initial rate 27.32 bbl/day to 10.21 bbl/day after 200 days of production and to 1.98 bbl/day within 5 years. At the end of primary production period, the oil rate is 0.44 bbl/day. When the gas injection process is started, oil rate has a slight increasing trend. Finally, the oil rate reaches 2.13 bbl/day. 47.166 MSTB of oil is obtained which corresponds to an oil recovery factor of 25.06% (Table 7). When the fracture spacing is changed to 100 ft, the reservoir pressure is lowered to 2500 psi at the end of the first 10 years of primary production. And then it increases to around 5600 psi after gas injection. The oil rate is increased from 0.30 bbl/day to 6.6 bbl/ day after gas injection. The pressure transport and sweep efficiency in this case is much better than the other case (maps not shown here owing to limited space), corresponding to a high recovery factor which is 73.65% (Table 7). A closer fracture spacing leads to a higher cumulative oil production and higher oil production rates. Further discussion Comparing the performance of gas flooding and waterflooding presented in the preceding sections, we can see that, overall, gas
Table 7 Effect of fracture spacing.
Fig. 14. Average reservoir pressures in different matrix permeabilities.
Fracture spacing, ft
Recovery factor, %
200 150 100
15.1 25.06 73.65
J.J. Sheng, K. Chen / Journal of Unconventional Oil and Gas Resources 5 (2014) 1–9
flooding is better than waterflooding in improving oil recovery in shale oil reservoirs. Another important benefit of gas flooding is that natural gas is available in field and the price is currently low. On the contrary, water sources are limited in some areas. More importantly, as a lot of water is used in fracturing shale reservoirs, we need to try to use less water in producing oil. However, shale-water interactions are not considered in this paper. Recently, we observed that the water may help to generate microfractures or open existing microfractures (Morsy et al., 2013a–e). If so, waterflooding performance might be better than what is predicted in this paper. In the process of gas flooding, gas can be absorbed and desorbed in the shale rock, However, the adsorption and desorption have not been included in this current paper for a couple of reasons. The adsorption and desorption may primarily affect the amount of gas needed. Adsorption and desorption are complex processes, which have not been well understood. Including these processes in the current simulation will result in more uncertainty. Similar to the shale-water interaction, such work is our next research effort. Conclusions Gas and waterflooding are studied in this paper using the simulation approach. Based on this study, the following conclusions are reached. 1. Because of the ultra-low permeability of shale reservoirs, injected gas can only miscible with the oil near the fracture region. The main mechanism of gas injection is the pressure maintenance. 2. According to the sensitivity analysis, low matrix permeability is the main factor that causes low oil recovery from shale reservoirs. A close fracture spacing will have a significant effect on shale oil production. It leads to a higher initial production rate and a much better sweep efficiency for miscible gas flooding. 3. In an ultra-low porosity and ultra-low permeability shale oil reservoir, water injection through high conductivity fracture has less effect on improving oil recovery than gas injection. Unlike miscible gas injection which can reduce oil viscosity, water injection can only provide limited pressure maintenance because of high pressure loss from an injector to a producer. Ultra-low permeability results in a low productivity and low injectivity. Therefore, the performance of water injection is poor. This conclusion did not include the water-shale rock interaction which may improve waterflooding performance. 4. Comparing the simulation results of gas flooding with waterflooding, miscible gas injection has a better effect on improving oil recovery in shale reservoirs. Injected solvent can be miscible with oil, reducing oil viscosity, and lead a higher stimulated vol-
9
ume than water, in addition to pressure maintenance. Gas injection may be a potential method to improve oil production in shale oil reservoirs.
References Chen, K. 2013. Evaluation of the EOR potential by gas and waterflooding in shale oil reservoirs, Master thesis, Texas Tech University. Chen, C., Balhoff, B., Mohanty, K.K., 2013. Effect of reservoir heterogeneity on improved shale oil recovery by CO2 Huff-n-Puff, SPE 164553-MS, presented at 2013 SPE Unconventional Resources Conference, Woodlands, Texas, USA, April 10–12. Fakcharoenphol, P., Charoenwongsa, S., Kazemi, H., Wu, Y.S., 2013. T The Effect of Water-Induced Stress To Enhance Hydrocarbon Recovery in Shale Reservoirs. SPEJ 18 (5), 897–909. Gamadi, T.D., Sheng, J.J., and Soliman, M.Y., 2013. An experimental study of cyclic gas injection to improve shale oil recovery, paper SPE 166334 presented at the SPE Annual Technical Conference and Exhibition held in New Orleans, Louisiana, USA, September 30–October 2. Makhanov K., Dehghanpour H., and Kuru E. 2012. An experimental study of spontaneous imbibition in Horn River shales. SPE-162650-MS presented at SPE Canadian Unconventional Resources Conference, Calgary, Alberta, Canada, October 30–November 1. Morsy, S., Gomaa, A., Sheng, J.J., Soliman, M.Y. 2013a. Potential of improved waterflooding in acid-hydraulically-fractured shale formations, paper SPE 166403 presented at the SPE Annual Technical Conference and Exhibition held in New Orleans, Louisiana, USA, September 30–October 2. Morsy, S., Sheng, J.J., Soliman, M.Y. 2013b. Improving hydraulic fracturing of shale formations by acidizing, paper SPE 165688 presented at the SPE Eastern Regional Meeting held in Pittsburgh, Pennsylvania, USA, August 20–22. Morsy, S., Hetherington, C.J., Sheng, J.J. 2013c. Effect of low-concentration HCl on the mineralogical, mechanical, and physical properties of shale rocks, paper SPE 165689 presented at the SPE Eastern Regional Meeting held in Pittsburgh, Pennsylvania, USA, August 20–22. Morsy, S., Sheng, J.J., Hetherington, C.J., Soliman, M.Y., Ezewu, R.O. 2013d. Impact of matrix acidizing on shale formations, paper SPE 167568 presented at the Nigeria Annual International Conference and Exhibition held in Lagos, Nigeria, August 5–7. Morsy, S., Sheng, J.J., Roland O. Ezewu, 2013e. Potential of waterflooding in shale formations, paper SPE 167510 presented at the Nigeria Annual International Conference and Exhibition held in Lagos, Nigeria, August 5–7. Rubin, B. 2010. Accurate simulation of non darcy flow in stimulated fractured shale reservoirs, paper SPE 132093 presented at the SPE Western Regional Meeting, Anaheim, California, May 27–29. Takahashi S., Kovscek A.R. 2009. Spontaneous counter current imbibition and forced displacement characteristics of low permeability, siliceous rocks. SPE-121354MS presented at SPE Western Regional Meeting, San Jose, California. Todd, M.R., Longstaff, W.J., 1972. The Development, testing, and application of a numerical simulator for predicting miscible flood performance. J. Pet. Technol. 24 (7), 874–882. Wan, T., 2013. Evaluation of the EOR Potential in Shale Oil Reservoirs by Cyclic Gas Injection. Texas Tech University, Master thesis. Wan, T., Sheng, J.J., and Soliman, M.Y. 2013a. Evaluation of the EOR potential in shale oil reservoirs by cyclic gas injection, paper SPWLA-D-12-00119 presented at the SPWLA 54th Annual Logging Symposium held in New Orleans, Louisiana, June 22–26. Wan, T., Sheng, J.J., and Soliman, M.Y. 2013b. Evaluation of the EOR potential in fractured shale oil reservoirs by cyclic gas injection, paper SPE 16880 or URTeC 1611383 presented at the Unconventional Resources Technology Conference held in Denver, Colorado, USA, August 12–14, 2013. Wood, T., Milne, B. 2011. Waterflood potential could unlock billions of barrels: crescent point energy. http://www.investorvillage.com/uploads/44821/files/ CPGdundee.pdf (accessed October 27, 2013).
Contents lists available at ScienceDirect
Journal of Unconventional Oil and Gas Resources journal homepage: www.elsevier.com/locate/juogr
Evaluation of the EOR potential of gas and water injection in shale oil reservoirs James J. Sheng ⇑, Ke Chen 1 Bob L. Herd Department of Petroleum Engineering, Texas Tech University, P.O. Box 43111, Lubbock, TX 79409, United States
a r t i c l e
i n f o
Article history: Received 13 September 2013 Revised 4 December 2013 Accepted 5 December 2013 Available online 21 December 2013 Keywords: Gas flooding Waterflooding Shale oil Enhanced oil recovery Unconventional oil and gas resources Simulation
a b s t r a c t With the relatively modest natural gas price, producing oil from unconventional shale reservoirs has attracted more and more interest from oil operators. Although many tremendous efforts have been made to develop shale resources, the ultimate oil recovery is still low (5–10%). Because of the important role of shale resources in the future oil and gas industry, more stimulation and production strategies are being considered and tested to find ways to improve oil production from shale reservoirs. Before applying a specific method to enhance oil recovery (EOR) in the field or conducting a research in laboratory, the simulation approach is a cheap and fast approach to evaluate its EOR potential. The analysis of simulation results will be helpful in designing laboratory experiments and field testing. In this paper we use a simulation approach to evaluate the EOR potential in shale oil reservoirs by gas flooding and waterflooding. Production behavior and oil recovery of different schemes are discussed through sensitivity studies. Simulation results of primary production, gas injection and water injection are compared. Results show that miscible gas injection has a higher potential to improve oil recovery from shale oil reservoirs than water injection. Gas injection above a minimum miscible pressure (MMP) can be fully miscible with oil, thus reducing oil viscosity significantly, in addition to the mechanism of pressure maintenance. Simulation results indicate that the oil recovery factor can be increased up to 15.1% by gas injection in a hydraulically fractured shale reservoir, whereas the oil recovery factor from the primary depletion is only 6.5%. The oil recovery from waterflooding is about 11.9% which is lower than that from gas flooding. The results indicate that miscible gas flooding could be a way to enhance oil recovery in shale oil reservoirs. Ó 2013 Elsevier Ltd. All rights reserved.
Introduction Unconventional shale reservoirs are generally produced by stimulation techniques. A horizontal well with multiple transverse fractures has proven to be an effective technique for shale gas and shale oil production. However, shale oil production faces more challenges compared with shale gas production. Even applying multi-stage hydraulic fracturing techniques, the final oil recovery factors using existing methods are only a few percent. Oil rate and reservoir pressure drop very quickly. Because most of oil remains unrecoverable even using the expensive horizontal well drilling combined with hydraulic fracturing, seeking the ways to enhance oil recovery will be the continuous efforts. In addition to drilling horizontal well and hydraulic fracturing stimulation, several EOR methods have been proposed or tested. Wan et al. (2013a,b) evaluated cyclic gas injection to improve shale oil recovery. Chen et al. (2013) studied the effect of reservoir
⇑ Corresponding author. Address: Bob L. Herd Department of Petroleum Engineering, Texas Tech University, P.O. Box 43111, Lubbock, TX 79409, United States. Tel.: +1 (806) 834 8477. E-mail address: [email protected] (J.J. Sheng). 1 Present address: PetroChina. 2213-3976/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.juogr.2013.12.001
heterogeneity on cyclic CO2 injection. Gamadi et al. (2013) did an experimental study of cyclic gas injection in shale rocks. Apparently, no gas injection has been reported to have been applied in shale oil reservoirs. Takahashi and Kovscek (2009) investigated the impact of different brine formulations in tight formations. Makhanov et al. (2012) showed that the imbibition could be a viable mechanism to transfer fluids from fracture to matrix in Horn River shales. Morsy et al. (2013a–e) studied the effects of water compositions (salinity, acid and alkali) on the EOR potential of water imbibition. Fakcharoenphol et al. (2013) pointed out that waterflooding changes the formation in situ stress due to the increase in reservoir pressure and decrease of reservoir temperature, which enhances oil recovery of shale formations by reactivating existing natural fractures and/or creating new fractures. Although waterflooding is a mature secondary recovery method for conventional reservoirs, it has not been applied in shale oil and gas reservoirs in a large commercial scale. A waterflooding pilot was started in Bakken shale in late 2006, (Wood and Milne, 2011). Because of ultra-low permeability of shale reservoirs, either waterflooding or gas flooding may have issues of injectivity and fluid transport. In this paper, we initiate a simulation study to evaluate whether gas flooding and waterflooding could have a potential
2
J.J. Sheng, K. Chen / Journal of Unconventional Oil and Gas Resources 5 (2014) 1–9
to improve oil recovery in shale oil reservoirs. The EOR potentials of gas injection and water injection are evaluated. A further sensitivity study is conducted to improve the understanding of the mechanism of gas and water injection in shale reservoirs. Before we present simulation results, the base simulation model is described next.
Base model setup Rubin (2010) used an extremely fine grid model as a reference model to simulate fracture flow. In the reference model, small cells of the actual width of fractures (assumed 0.001 ft) are used to capture the flow from the matrix to the fractures. He also showed that it is possible to accurately model flow from a fractured shale reservoir using logarithmically spaced, locally refined grids with fractures represented by 2.0-ft-wide cells and by maintaining the same conductivity (kfwf) as that of the original 0.001-ft-wide fracture. Using Robin’s approach, we can use much less cells to simulate fracture-related flow. Wan (2013) used Robin’s approach to have built a model of 200 ft long, 1000 ft wide and 200 ft thick to simulate a horizontal well and one transverse fracture (Fig. 1). 2-ft wide grid cells with 83.3 md-ft conductivity (k = 41.65 md, wf = 2 ft) were used to represent the actual fractures of 0.001-ft wide and 83.300 md. The reservoir property data used by Wan are similar to the Eagle Ford shale data (Table 1). The initial reservoir pressure is 6425 psi. The permeability is 100 nano-Darcy. The porosity is 0.06. During the primary production process, the well is controlled by bottom-hole pressure (BHP) of 2500 psi. These data and the approach are used in the base model as well. The objective of this study is to evaluate the potential of gas and water injection for improving oil production from shale oil reservoirs. Modeling a whole reservoir requires a large number of grid blocks, and it is of course time-consuming to model these complex fracture networks. Therefore, we built a small model which is 200 ft long, 1000 ft wide and 200 ft thick (Fig. 2). This small model is produced with two half-vertical wells connected with two halffractures, respectively. Thus each fracture is 1-ft wide and has a conductivity of 46.65 md-ft. We use such a simple model to simulate the flow between the two lateral hydraulic fractures of
Fig. 1. Cyclic gas injection simulation model used in Wan (2013).
Table 1 Reservoir properties used in the base model. Initial reservoir pressure Porosity of shale matrix Initial water saturation Compressibility of shale Shale matrix permeability Reservoir temperature Gas specific gravity Reservoir thickness Bubble point for oil
6425 psi 0.06 0.3 5106 psi1 0.0001 md 255 °F 0.8 200 ft 2398 psi
a horizontal well. Assume the flow between any two lateral fractures is the same, such small model can be used to simulate the flow through a part of a horizontal well. The reservoir properties data used in this model is shown in Table 1. Before conducting gas or waterflooding simulation using this small base model, we need to validate the model first. In Wan’s model, one vertical well was used to connect one lateral vertical fracture. The fracture is 2-ft wide and has a conductivity of 83.3 md-ft. In the base model, we use two vertical wells to connect two half-fractures, because we need to use one vertical well to represent an injection well and the other well to represent a production well, so that an injection-production pattern is built in the model. The area or volume of our base model is the same as that in Wan’s model. If the injector is changed to a producer in our base model, the oil production from our model should be same as that from Wan’s model. Because Wan’s model has been validated in his thesis, if the oil production performance is the same as that from Wan’s model, then our model is validated. This validation has been performed in Chen’s thesis (Chen, 2013), and it is not repeated here because of limited space allowed for this paper. In the above base model, the miscibility model proposed by Todd and Longstaff (1972) is used. Todd and Longstaff (1972) proposed a method of simulating miscible displacement performance without considering detailed compositions. They introduced a mixing parameter x, which determines the degree of mixing between the miscible fluids within a grid block. A value of zero corresponds to the immiscible displacement, whereas a value of one corresponds to complete mixing. The mixing of solvent and oil is controlled by a pressure-dependent mixing parameter, xo (Omegaos in the simulator IMEX used in this study). IMEX, the black oil simulator, is developed by Computer Modeling Group (CMG). If the individual block pressure is much lower than the minimum miscibility pressure (MMP), xo = 0.0, and gas is displacing oil immiscibly. As the block pressure increases, this mixing parameter reaches its maximum value xomax at the MMP, as shown in Fig. 3. When no better data is available, the user manual of the simulation software IMEX suggests a value in the range of 0.5–0.8 as a first approximation. xo is considered to be a function of pressure and xomax is set as 0.72 in the base model. The black oil simulator IMEX uses 4-components but three phases to simulate miscible flooding. The 4-components are water, oil, dissolved gas and injected gas (solvent). The three phases are water, oil and gas which includes dissolved gas and injected gas in the free gas state. Both the injected gas and the original dissolved gas have the same properties. The specific gravity (ratio of the gas density to the air density) is 0.8. The oil compressibility is 1105 psi1. The mixing of solvent and free gas is governed by xg (OMEGASG in IMEX), which is assumed pressure independent. xg is bounded by zero and one. Since solvent-gas has a lower mobility ratio than oil, xg is usually greater than xomax. In our case OMEGASG is set as 1.0, assuming solvent and free gas have a complete mixing. In this base simulation model, the injection is controlled by the maximum solvent injection rate for a half well of 400 Mscf/day and the maximum injection pressure of 7000 psi.
J.J. Sheng, K. Chen / Journal of Unconventional Oil and Gas Resources 5 (2014) 1–9
3
Fig. 2. Base reservoir model.
Fig. 4. Average reservoir pressure and oil recovery factor vs. time (Scenario G1).
Fig. 3. x vs. P.
For the production well, the flowing bottom-hole pressure is 2500 psi which is above the bubble point pressure of 2398 psi. Simulation results and analysis In this section, gas injection and waterflooding scenarios are presented and analyzed. Gas injection We first compare several gas injection scenarios. Scenario G1: 3600 days of primary production followed by 60 years of gas flooding production In this scenario, gas injection starts after 3600 days (10 years) of primary production. Figs. 4 and 5 show the results for oil recovery factor, average pressure and oil rate versus time. The reservoir pressure decreases fast from the initial reservoir pressure of 6425 psi to 3000 psi as the reservoir is mainly in depletion drive in the first 10 years’ primary production. Once gas is injected, the reservoir pressure increases from 3000 psi to about 5200 psi gradually. Because the high pressure gradient exists between the injector and pressure owing to the ultra-permeability, the pressure in the producer side is low, and the average reservoir pressure is
Fig. 5. Oil production rate vs. time (Scenario G1).
always lower than the initial reservoir pressure during the injection and production period. From the oil production rate graph (Fig. 5), the oil rate decreases from the initial rate 27.47 bbl/day to 10.26 bbl/day after 200 days of production and to 2.72 bbl/day within 5 years during the primary production. At the end of primary production period, the oil rate is 0.57 bbl/day from two half-fractures. After gas injection is started, the oil rate gradually increases to 1.3 bbl/day from one half fracture. At the end of 60 years of gas injection, 37.912 MSTB oil is produced, corresponding to an oil recovery factor of 15.12% as shown in Fig. 4.
4
J.J. Sheng, K. Chen / Journal of Unconventional Oil and Gas Resources 5 (2014) 1–9
The pressure data and oil rate data are further analyzed. At the end of primary production, the average reservoir pressure is about 3000 psi. The well BHP is 2500 psi. The pressure drawdown is about 500 psi. During the gas injection period, the average reservoir pressure is about 5200 psi. Then the pressure drawdown becomes 2700 psi which is about 5 times that at the end of primary production. During this time, the well rate from one half-fracture is 1.3 bbl/day which is about 2.3 times that rate (0.57 bbl/day) at the end of primary production. Because 1.3 bbl/day is from one half-fracture, then the rate from two half-fractures would be 2.6 bbl/day if both of the half-fractures produced. Then the rate of 2.6 bbl/day is 4.6 times the rate of 0.57 bbl/day from the two half-fractures. This rate ratio is close the pressure drawdown ratio of about 5 (the pressure drawdown during the injection period to that at the end of primary production). From the Darcy rate equation (Eq. (1)), if the pressure drawdown Dp is increased 5 times, the rate is also increased about 5 times only if the viscosity l does not change, because the other parameters in Eq. (1) (permeability k, flow area A and distance DL from the matrix to fracture) are almost unchanged.
q¼
kAðDpÞ : lðDLÞ
ð1Þ
From this analysis, it is clear that the main mechanism from gas injection is mainly pressure maintenance. Figs. 6 and 7 show the pressure variation and oil saturation distribution during the production period for Scenario G1, respectively. When starting gas injection process, the solvent is injected into the reservoir through injection well and mix with reservoir fluids, leading to oil viscosity decrease. Oil is pushed away from the injection well as shown in the saturation map. In the meantime the reservoir pressure builds up. Owing to the ultra-low permeability, fluid transport in such kind of reservoirs is much more difficult than that in conventional reservoirs. This also results in a small increase in oil rate after gas injection owing to pressure increase as just discussed. Fig. 7 also shows that the gas is only in the injector side in the first 50 years. Gas could not reach the producer. Near the producer, it is oil only. Therefore, the viscosity in Eq. (1) is the oil viscosity which is almost unchanged. The viscosity decreases owing to gas injection occurs near the injector
not the producer. As a result, the viscosity decrease mechanism owing to the gas and oil miscibility is not significant. Scenario G2: 3600 days of primary production followed by 60 years of cyclic gas flooding production For this scenario, cyclic gas injection begins after 3600 days (10 years) of primary production. Each injection cycle has 5 yearinjection and 5 year-shut in. The recovery factor at the end of 70 years is 14.42%. Scenario G3: 70 years of gas flooding production In this scenario, gas injection is started at the beginning of the development. Keep gas flooding and oil production simultaneously for 70 years which is the total production period in the previous scenarios. Because of gas injection, the reservoir pressure maintain high around 5000 psi. However, the total oil recovery is 13.48%. We have considered three production scenarios so far. In the first scenario, gas injection begins after 10 years of primary production and continues for 60 years. 74.245 MMSCF of gas is injected which corresponds to 30.7 pore volume (PV = 430.504 Mrbbls), and 37.912 MSTB of oil is produced which corresponds to 15.12% oil recovery factor. If we use a 10% discount rate, the produced oil of 37.912 MSTB is 11.893 MSTB at the present value. In the second scenario, cyclic gas injection starts after 10 years of primary production. The cyclic process has 5 years of injection and 5 years of shut in. In this process, 64.473 MMSCF (26.67 PV) of gas is used to produce about 14.42% of original oil in place which corresponds to 11.555 MTB at the present value. For the scenario 3, gas injection is implemented at the beginning of the development. It can be seen that the scenario 3 has a lower oil production in the first 10 years because only one well is producing instead of two producing wells in the other two scenarios. As shown in Table 2, the recovery factor from the scenario 3 in the first 10 years of primary production is 3.4%, compared with those in the scenarios 1 and 2 of 5.75%. The recovery factors at the end of 30, 50 and 70 years and the oil produced at the present value during the 70 years are all lower than those of the scenarios 2 and 3. Therefore, it is better to implement gas flooding after several years of natural pressure depletion. The results of three simulation scenarios show that the ultimate recovery factors are not quite different for these three different injection
Fig. 6. Reservoir pressure distribution at different times in the gas injection case (Scenario G1).
5
J.J. Sheng, K. Chen / Journal of Unconventional Oil and Gas Resources 5 (2014) 1–9
Fig. 7. Oil saturation distribution at different times in the gas injection case (Scenario G1).
Table 2 Gas flooding simulation results.
Cumulative oil production (70 years, present value at 10%) Cumulative gas injection (PV) Overall oil recovery (10 years) Overall oil recovery (30 years) Overall oil recovery (50 years) Overall oil recovery (70 years)
Scenario G1 – primary + gas flooding
Scenario G2 – Primary + cyclic gas injection
Scenario G3 – gas flooding only
11.893 MSTB 330.7 5.75% 8.14% 11.49% 15.12%
11.555 MSTB 26.67 5.75% 7.95% 11.05% 14.42%
7.744 MSTB 26.6 3.4% 6.68% 9.97% 13.48%
scenarios, but less solvent is injected in the scenario 2 and the ultimate oil recovery obtained from the scenario 2 is close to that of the scenario 1. In addition, this cyclic gas injection scenario is not optimized. In this case, the cyclic time is 5 years. If the cyclic time is much shorter, the recovery factor will be higher. Wan et al. (2013a) basically used a similar model. The recovery factor is 27.4% when the cyclic time is 100 days. Therefore, cyclic gas injection after some years of primary production may be a better option. However, the down time required to convert the producer into the injector was not taken into calculation here. For an analysis purpose, we simulated 70 years. 70 years may be too long in reality, we present the results at the end of 10, 30, 50 and 70 years. Waterflooding simulation In this section, we present several water injection scenarios similar to the gas injection discussed above. Scenario W1: 3600 days of primary production followed by 60 years of waterflooding production In this production scenario, water injection begins after 3600 days (10 years) of primary production. The production is driven by natural pressure depletion in the first 10 years. The reservoir pressure decreases from 6425 psi to 3000 psi in the primary production period and then gradually increases to about 4250 psi after water injection as shown in Fig. 8. Note that this average reservoir pressure of 4250 psi during the water injection period is lower than that (5200 psi) during the gas injection period in Scenario G1. This is because the water viscosity is higher than gas,
and the pressure gradient near the injection well in the water injection case is higher than that in the gas injection case, which can be convinced by comparing the pressure maps at the end of injection (Figs. 9 and 10 for gas injection and water injection, respectively). From Eq. (1), we would expect that the waterflooding performance may not be as good as gas flooding because the pressure drawdown near the producer in waterflooding is lower. Fig. 11 shows the water injection rate history and oil production rate history. The initial oil rate is 27.47 bbl/day, after 200 days of primary production it decreases to 10.26 STB/day. At the end of primary production, the oil rate is 0.57 STB/d. When water injection begins, the increase in production rate cannot be seen. Because a shale reservoir has an ultra-low permeability, the injection fluids have difficulty to transport from the injection to the producer. The response of the production well to waterflooding is poor. This also results in a low injection rate during the entire injection period except a short early time. During waterflooding, the oil rate is about 0.8 STB/day, and the oil recovery factor is 11.9% at the end. The rate histories (Fig. 11) and the comparison of pressure distribution maps (Figs. 9 and 10) together show that the main problem in waterflooding shale oil reservoirs is low water injectivity.
Scenario W2: 3600 days of primary production followed by 60 years of cyclic water production For this scenario, water injection begins after 3600 days (10 years) of primary production. In this scenario, cyclic water injection is implemented. Each injection cycle has 5 year-injection
6
J.J. Sheng, K. Chen / Journal of Unconventional Oil and Gas Resources 5 (2014) 1–9
Fig. 8. Average reservoir pressure and oil recovery factor vs. time for Scenario W1.
and 5 year-shut in. The overall recovery factor at the end of 70 years is 11.03%.
Scenario W3: 70 years of waterflooding production In this scenario, water injection is started at the beginning of the development. Keep water injection and oil production simultaneously for 70 years. When water injection starts at the beginning of the production, the initial production rate is lower than previous scenarios because only one half-fractured well is producing in the beginning, and the reservoir pressure decreases slowly from initial reservoir pressure to 4000 psi. The ultimate oil recovery factor is 11%. Three production scenarios of waterflooding have been simulated with the results presented in Table 3. In the first scenario, water injection begins after 10 years of primary production and continues for 60 years. 27.020 MSTB of water is injected which corresponds to 0.063 PV, and 29.872 MSTB of oil is produced which corresponds to 11.9% oil recovery factor. In the second scenario, cyclic water injection begins after 10 years of primary production. The cyclic injection process has 5 years of injection and 5 years of shut in. In this process, 21.883 MSTB of water (0.051 PV) is used to produce about 11.03% of original oil in place. For the scenario
Pressure (psi) 2078-12-30 100
200
File: 1inj_1prod_36 User: jsheng Date: 10/26/2013 Scale: 1:481.64478 Z/X: 1.00:1 Axis Units: ft
10,000
10,000
9,900
0
J layer: 27
0.00 0.00 0
100
35.00 10.00
70.00 feet 20.00 meters 2,490
200
Fig. 9. Pressure at the end of gas injection for Scenario G1.
Pressure (psi) 2078-12-30 100
200
File: 1inj_1prod_wa User: jsheng Date: 10/26/2013 Scale: 1:481.64478 Z/X: 1.00:1 Axis Units: ft
10,000
10,000
9,900
0
J layer: 27
0.00 0.00 0
100
35.00 10.00
70.00 feet 20.00 meters
200
Fig. 10. Pressure at the end of water injection for Scenario W1.
2,490
7
J.J. Sheng, K. Chen / Journal of Unconventional Oil and Gas Resources 5 (2014) 1–9 Table 4 Effect of half-fracture length.
Fig. 11. Oil rate & injection rate vs. time for Scenario W1.
3, water injection is implemented at the beginning of the development. We can understand that the scenario W3 has a lower oil production in the first 10 years because only one half-production well is used instead of two half-production wells in the other two scenarios. The recovery factors at the end of 30, 50 and 70 years, and the oil produced at the present value during the 70 years are all lower than those of the scenarios 2 and 3. So it is better to implement waterflooding after several years of natural pressure depletion so that some oil can be produced during early depletion. Similar to the gas injection cases, however, the down time required to convert the producer into the injector was not taken into calculation in this comparison. Further sensitivity analysis The above simulation study shows that gas flooding is a better option than waterflooding. In this section, we will further do some sensitivity analysis of gas flooding. Sensitivity analysis is a quantitative method of determining the important parameters which affect shale oil production performance. The base model is 3600 days of primary production followed by 60 years of gas injection. The parameters considered in this paper include fracture half-length, flowing bottom-hole pressure, matrix permeability and fracture spacing.
Half-fracture Length, ft
Recovery factor, %
500 365 245 125
15.1 12.6 9.1 5.5
ultimate recovery factor. The oil recovery factors at the end of 60 years of gas injection for different half-fracture lengths are shown in Table 4. The data show that the recovery factors are almost a linear function of the half-length, as shown in Fig. 12. In our model, near the fracture, it is almost linear flow. The flow area A in Eq. (1) is proportional to the half-fracture length. Therefore, if the half-fracture length is increased, A is increased proportionally. Thus the flow rate and the resulting oil recovery factor will be increased linearly with the half-fracture length.
Flowing bottom-hole pressure In the base model, we consider a system where the pressure is maintained high enough to guarantee the entire reservoir remains single phase throughout the gas flooding process. The base gas injection model is controlled by flowing bottom-hole pressure (FBHP) which was set up to 2500 psi. Now we will test the production performance under different FBHP. The flowing bottom-hole pressures selected are 1500, 1000 and 500 psi. At the end of 10 years (3600 days) of primary production, the average reservoir pressures are lowered to 2279 and 2392 psi for the flowing bottom-hole pressures of 500 and 1000 psi, respectively. As expected, with a lower flowing bottom-hole pressure, a higher rate and a higher oil recovery factor can be achieved at the end of the primary production period (at 3600 days). After that the oil rate for the lower FBHP is lower than that for the higher FBHP in the early injection period from 4000 days to 8000 days. During this period, the reservoir pressure is below the bubble point pressure. It indicates that the two-phase flow decreases the efficiency of gas flooding. Later gas injection causes the reservoir pressure to rise and become higher than the bubble point pressure.
Fracture half-length The fracture half-length used in the base model is 500 ft. Three other fracture half-lengths of 365, 245, and 125 ft are selected to compare the effect of fracture length on gas flooding production. The simulation result shows that the reservoir pressure decreases faster in the case of longer fracture half-length in primary production period. The average reservoir pressure at the end of 10 years stays lower with longer fracture half-length. A longer fracture length has a higher drainage volume of the reservoir which creates proportionately higher production rates and the gas injection process can have a better effect in maintaining the reservoir pressure which leads to a higher cumulative oil production and a higher
Fig. 12. Oil recovery factor vs. half-fracture length.
Table 3 Waterflooding simulation results.
Cumulative oil production (present value at 10%) Cumulative water injection (PV) Overall oil recovery (10 years) Overall oil recovery (30 years) Overall oil recovery (50 years) Overall oil recovery (70 years)
Scenario W1 – primary + waterflooding
Scenario W2 – primary + cyclic waterflooding
Scenario W3 – waterflooding only
11.627 MSTB 0.063 5.73% 7.59% 9.8% 11.9%
11.555 MSTB 0.051 5.73% 7.21% 9.30% 11.03%
7.578 MSTB 0.056 3.39% 6.41% 8.87% 11.05%
8
J.J. Sheng, K. Chen / Journal of Unconventional Oil and Gas Resources 5 (2014) 1–9 Table 6 Effect of matrix permeability. Matrix permeability, mD 5
510 104 2.5104 4104 5104 103
Recovery factor, % 9.1 15.1 36.1 70.5 92.1 99.3
Fig. 13. Oil rate and oil recovery factor histories for different FBHP.
Then, the miscibility is reached again in the later production time. Then the oil rate with the lower FBHP becomes higher because of a high drawdown. Thus the final oil recovery is higher with lower FBHP, as shown in Fig. 13 and Table 5. Matrix permeability The permeability value used in the base model is 1104 md (100 nano-Darcy). The other permeability values of 1103, 5104, 4104, 2.5104 and 5105 md are selected here. From the simulation results, the average reservoir pressures drop with the matrix permeability after gas injection, as shown in Fig. 14. When gas injection starts, a rapid increase in the average reservoir pressure is observed in the case of 1103 md. After gas breaks though the producer, the reservoir pressure decreases and almost all the oil has been produced. The reservoir pressures in other cases follow the same trend, with the lower permeability, the reservoir pressure is lower right after gas injection and the breakthrough time is longer. The oil recovery factors in these cases after 60 years of gas flooding are shown in Table 6, and in Fig. 15. These data show that oil recovery is very sensitive to matrix permeability. The oil recovery factor almost increases linearly with the matrix permeability.
Table 5 Effect of flowing bottom-hole pressure. FBHP, psi
Recovery factor, %
2500 1500 1000 500
15.1 18.3 19.1 19.7
Fig. 15. Oil recovery factors vs. matrix permeability.
Higher matrix permeability means better hydraulic conductivity which corresponds to a higher oil rate and a higher cumulative oil production. Fracture spacing The distance between fractures (fracturing spacing) in the base case is 200 ft. The other two fracture spacings of 150 and 100 ft are tested to see the influence of fracture spacing on gas flooding performance. For the 150 ft spacing, the reservoir pressure decreases to lower than 3000 psi in first 10 years of primary production because of higher production rate, and then gradually increases to more than 5000 psi after starting gas injection. The oil rate decreases from the initial rate 27.32 bbl/day to 10.21 bbl/day after 200 days of production and to 1.98 bbl/day within 5 years. At the end of primary production period, the oil rate is 0.44 bbl/day. When the gas injection process is started, oil rate has a slight increasing trend. Finally, the oil rate reaches 2.13 bbl/day. 47.166 MSTB of oil is obtained which corresponds to an oil recovery factor of 25.06% (Table 7). When the fracture spacing is changed to 100 ft, the reservoir pressure is lowered to 2500 psi at the end of the first 10 years of primary production. And then it increases to around 5600 psi after gas injection. The oil rate is increased from 0.30 bbl/day to 6.6 bbl/ day after gas injection. The pressure transport and sweep efficiency in this case is much better than the other case (maps not shown here owing to limited space), corresponding to a high recovery factor which is 73.65% (Table 7). A closer fracture spacing leads to a higher cumulative oil production and higher oil production rates. Further discussion Comparing the performance of gas flooding and waterflooding presented in the preceding sections, we can see that, overall, gas
Table 7 Effect of fracture spacing.
Fig. 14. Average reservoir pressures in different matrix permeabilities.
Fracture spacing, ft
Recovery factor, %
200 150 100
15.1 25.06 73.65
J.J. Sheng, K. Chen / Journal of Unconventional Oil and Gas Resources 5 (2014) 1–9
flooding is better than waterflooding in improving oil recovery in shale oil reservoirs. Another important benefit of gas flooding is that natural gas is available in field and the price is currently low. On the contrary, water sources are limited in some areas. More importantly, as a lot of water is used in fracturing shale reservoirs, we need to try to use less water in producing oil. However, shale-water interactions are not considered in this paper. Recently, we observed that the water may help to generate microfractures or open existing microfractures (Morsy et al., 2013a–e). If so, waterflooding performance might be better than what is predicted in this paper. In the process of gas flooding, gas can be absorbed and desorbed in the shale rock, However, the adsorption and desorption have not been included in this current paper for a couple of reasons. The adsorption and desorption may primarily affect the amount of gas needed. Adsorption and desorption are complex processes, which have not been well understood. Including these processes in the current simulation will result in more uncertainty. Similar to the shale-water interaction, such work is our next research effort. Conclusions Gas and waterflooding are studied in this paper using the simulation approach. Based on this study, the following conclusions are reached. 1. Because of the ultra-low permeability of shale reservoirs, injected gas can only miscible with the oil near the fracture region. The main mechanism of gas injection is the pressure maintenance. 2. According to the sensitivity analysis, low matrix permeability is the main factor that causes low oil recovery from shale reservoirs. A close fracture spacing will have a significant effect on shale oil production. It leads to a higher initial production rate and a much better sweep efficiency for miscible gas flooding. 3. In an ultra-low porosity and ultra-low permeability shale oil reservoir, water injection through high conductivity fracture has less effect on improving oil recovery than gas injection. Unlike miscible gas injection which can reduce oil viscosity, water injection can only provide limited pressure maintenance because of high pressure loss from an injector to a producer. Ultra-low permeability results in a low productivity and low injectivity. Therefore, the performance of water injection is poor. This conclusion did not include the water-shale rock interaction which may improve waterflooding performance. 4. Comparing the simulation results of gas flooding with waterflooding, miscible gas injection has a better effect on improving oil recovery in shale reservoirs. Injected solvent can be miscible with oil, reducing oil viscosity, and lead a higher stimulated vol-
9
ume than water, in addition to pressure maintenance. Gas injection may be a potential method to improve oil production in shale oil reservoirs.
References Chen, K. 2013. Evaluation of the EOR potential by gas and waterflooding in shale oil reservoirs, Master thesis, Texas Tech University. Chen, C., Balhoff, B., Mohanty, K.K., 2013. Effect of reservoir heterogeneity on improved shale oil recovery by CO2 Huff-n-Puff, SPE 164553-MS, presented at 2013 SPE Unconventional Resources Conference, Woodlands, Texas, USA, April 10–12. Fakcharoenphol, P., Charoenwongsa, S., Kazemi, H., Wu, Y.S., 2013. T The Effect of Water-Induced Stress To Enhance Hydrocarbon Recovery in Shale Reservoirs. SPEJ 18 (5), 897–909. Gamadi, T.D., Sheng, J.J., and Soliman, M.Y., 2013. An experimental study of cyclic gas injection to improve shale oil recovery, paper SPE 166334 presented at the SPE Annual Technical Conference and Exhibition held in New Orleans, Louisiana, USA, September 30–October 2. Makhanov K., Dehghanpour H., and Kuru E. 2012. An experimental study of spontaneous imbibition in Horn River shales. SPE-162650-MS presented at SPE Canadian Unconventional Resources Conference, Calgary, Alberta, Canada, October 30–November 1. Morsy, S., Gomaa, A., Sheng, J.J., Soliman, M.Y. 2013a. Potential of improved waterflooding in acid-hydraulically-fractured shale formations, paper SPE 166403 presented at the SPE Annual Technical Conference and Exhibition held in New Orleans, Louisiana, USA, September 30–October 2. Morsy, S., Sheng, J.J., Soliman, M.Y. 2013b. Improving hydraulic fracturing of shale formations by acidizing, paper SPE 165688 presented at the SPE Eastern Regional Meeting held in Pittsburgh, Pennsylvania, USA, August 20–22. Morsy, S., Hetherington, C.J., Sheng, J.J. 2013c. Effect of low-concentration HCl on the mineralogical, mechanical, and physical properties of shale rocks, paper SPE 165689 presented at the SPE Eastern Regional Meeting held in Pittsburgh, Pennsylvania, USA, August 20–22. Morsy, S., Sheng, J.J., Hetherington, C.J., Soliman, M.Y., Ezewu, R.O. 2013d. Impact of matrix acidizing on shale formations, paper SPE 167568 presented at the Nigeria Annual International Conference and Exhibition held in Lagos, Nigeria, August 5–7. Morsy, S., Sheng, J.J., Roland O. Ezewu, 2013e. Potential of waterflooding in shale formations, paper SPE 167510 presented at the Nigeria Annual International Conference and Exhibition held in Lagos, Nigeria, August 5–7. Rubin, B. 2010. Accurate simulation of non darcy flow in stimulated fractured shale reservoirs, paper SPE 132093 presented at the SPE Western Regional Meeting, Anaheim, California, May 27–29. Takahashi S., Kovscek A.R. 2009. Spontaneous counter current imbibition and forced displacement characteristics of low permeability, siliceous rocks. SPE-121354MS presented at SPE Western Regional Meeting, San Jose, California. Todd, M.R., Longstaff, W.J., 1972. The Development, testing, and application of a numerical simulator for predicting miscible flood performance. J. Pet. Technol. 24 (7), 874–882. Wan, T., 2013. Evaluation of the EOR Potential in Shale Oil Reservoirs by Cyclic Gas Injection. Texas Tech University, Master thesis. Wan, T., Sheng, J.J., and Soliman, M.Y. 2013a. Evaluation of the EOR potential in shale oil reservoirs by cyclic gas injection, paper SPWLA-D-12-00119 presented at the SPWLA 54th Annual Logging Symposium held in New Orleans, Louisiana, June 22–26. Wan, T., Sheng, J.J., and Soliman, M.Y. 2013b. Evaluation of the EOR potential in fractured shale oil reservoirs by cyclic gas injection, paper SPE 16880 or URTeC 1611383 presented at the Unconventional Resources Technology Conference held in Denver, Colorado, USA, August 12–14, 2013. Wood, T., Milne, B. 2011. Waterflood potential could unlock billions of barrels: crescent point energy. http://www.investorvillage.com/uploads/44821/files/ CPGdundee.pdf (accessed October 27, 2013).