Experimental study and numerical simulation of a solvent-assisted start-up for SAGD wells in heavy oil reservoirs

Journal of Petroleum Science and Engineering (xxxx) xxxx–xxxx

Contents lists available at ScienceDirect

Journal of Petroleum Science and Engineering journal homepage: www.elsevier.com/locate/petrol

Experimental study and numerical simulation of a solvent-assisted start-up for SAGD wells in heavy oil reservoirs Zhe Yuana,1, Pengcheng Liua, a b

⁎,1

, Shengfei Zhangb, Yuwei Jiaob, Xiuluan Lib

School of Energy Resources, China University of Geosciences, Beijing 100083, China Research Institute of Petroleum Exploration and Development, PetroChina, Beijing 100083, China

A R T I C L E I N F O

A BS T RAC T

Keywords: Heavy oil Start-up Solvent-assisted Well communication Solvent soak

Effective communication in the well pair during the preheating period has a significant effect on the success of the production period by using steam-assisted gravity drainage (SAGD) technology in heavy oil reservoirs. However, the time-consumption and the economic cost associated with steam generation and circulation limit the application of this technique. Because of these limitations and concerns, we have investigated the effectiveness of the solvent injection-assisted start-up period. A sand-pack physical experiment was designed to evaluate the effect of solvent on heavy oil-viscosity reduction. Different solvents were tested to examine the effects on viscosity reduction and asphaltene precipitation. Xylene was selected as asphaltene-soluble solvent for this experiment. The solvent was injected at different injection rates and diffused into the bitumen during the soaking period. Furthermore, a numerical model was constructed to simulate the performance of the preheating and early production period with or without solvent injection in the start-up process. The observations and simulation results indicate that the injected solvent diffuses to improve the sweep area to further reduce oil viscosity significantly in the soaking time. However, too long soaking time results in lower concentrations of solvent has the opposite effect on oil-viscosity reduction. The relative permeability of the oil phase increases as well. Whereas, according to numerical simulation, the solvent-assisted start-up process of SAGD shortens preheating time and reduces steam consumption. Both the production rate and the cumulative oil production improve during the early production period.

1. Introduction As the reserve and production of conventional oil decrease gradually, the oil industry has turned its attention to heavy oil and bitumen reservoirs with high initial viscosity. To enhance the mobility of oil by the strong temperature sensitive of viscosity, the steam-assisted gravity drainage (SAGD) technique is one of the major EOR (enhanced-oil recovery) processes applied to heavy oil reservoirs (Butler and Stephens, 1981; Butler, 1994). In general, before the SAGD production process, it is necessary to preheat the formation by steam circulation and to establish effective thermal hydraulic connectivity between the two horizontal wells to provide drainage channels for the production phase (Gates and Leskiw, 2010; Zhu et al., 2012). However, the warmup time is too long, which usually lasts for more than 6 months. Whereas, this process requires huge amounts of water for steam generation and circulation, resulting in high energy consumption and CO2 emissions (Mohsenzadeh et al., 2014; Ji et al., 2015). The cost of the start-up process by conventional steam cycle is an expensive and

⁎

1

economically unfeasible application, particularly in the highly fractured oilfield (Souraki, et al., 2013; Kazeem, 2014; Mohammed and Babadagli, 2014; Zendehboudi et al., 2014). Improving the efficiency and economy of the start-up process has become the focus of this research. A simple, economic method is needed to quickly warm up the formation and establish well communication so that the start-up period can be turned to the production process. To solve these problems, much research has involved variations in the operating methodology (Kaiser and Taubner, 2015; Saks et al., 2015). Parmar et al. (2009) presented a sensitivity analysis about the effects of circulation time, steam circulation rates, and pressure differences on start-up time. A similar SAGD project was conducted in East Senlac with a steam soak in the start-up process, indicating that a soak mechanism is an effective method for enhancing thermal communication between the well pairs (Boyle et al., 2003). Anderson presents a bullheading technique to decrease start-up time by leaving the injected steam deep into the formation for heat transfer (Anderson and Kennedy, 2012). However, the method likely forms a flow channel

Corresponding author. E-mail address: [email protected] (P. Liu). P. Liu and Z.Yuan contributed equally to this work (Co-first authors).

http://dx.doi.org/10.1016/j.petrol.2017.01.010 Received 21 September 2016; Received in revised form 4 January 2017; Accepted 4 January 2017 0920-4105/ © 2017 Elsevier B.V. All rights reserved.

Please cite this article as: Yuan, Z., Journal of Petroleum Science and Engineering (2017), http://dx.doi.org/10.1016/j.petrol.2017.01.010

Journal of Petroleum Science and Engineering (xxxx) xxxx–xxxx

Z. Yuan et al.

f μo μs Sw krw krow Sl krg krog

Nomenclature SAGD Steam Assisted Gravity Drainage ES-SAGD Expanding-Solvent Steam Assisted Gravity Drainage EOR Enhanced oil recovery RF Recovery Factor CSOR Cumulative Steam-to-Oil Ratio CMG Computer Modeling Group EOS Equation of State T Temperature, °C

Fraction Oil viscosity, mPa s Solvent viscosity, mPa s Water saturation Relative permeability of Relative permeability of Liquid saturation Relative permeability of Relative permeability of

water phase in Oil–water oil phase in Oil–water gas phase in gas-liquid liquid phase in gas-liquid

precipitation. Zhang et al. (2012) tested the effects of different solvents (ligarine, diesel, and toluene) on heavy oil-viscosity reduction. They recommended that ligarine and toluene for the start-up process. Ahmadloo and Yang (2014) conducted a method by injecting xylene into a reservoir after 70 days of pre-heating. The results showed a shorter preheating time and better communication. However, solvent injected into a larger heated formation would be diluted and would impair utilization. Diesel was selected as the solvent to inject into the formation (Wu et al., 2015). However, considering the economic benefit, the cost of diesel was expensive, and the asphaltene precipitation consisted of diesel and bitumen. Thus, the precipitation would fill the pore space and prevent the liquid from flowing. In this study, a sand-pack physical experimental apparatus was designed to investigate the mechanisms of the solvent-assisted startup process of SAGD. The main objective was to examine and evaluate the effectiveness of xylene solvent on oil-viscosity reduction and the mechanism of solvent soak. Moreover, the start-up period was simulated by a homogeneous model of 46 × 40 × 57 grids of a heavy oil reservoir in Xinjiang Oilfield, China. Through this innovative technique, solvent is injected initially into the cold reservoir to reduce the viscosity of oil near the wellbore, making it easier for the steam injection that follows. Simulation included preheating and early production with and without solvent injection, which showed

for the steam and is more applicable for a thicker reservoir. These studies about start-up process were all considered as possibilities for changing the well designs or working operations. However the costs of steam consumption and heat loss were still expensive. Many attempts have been tested to investigate the efficiency of chemical solvent injection to enhance the recovery factor (RF) in both the laboratory and the oilfield pilot (Mohammadzadeh et al., 2012; Jha et al., 2013; Zhao et al., 2013; Bayat et al., 2015). To take the advantage of the mass transfer by a solvent with a light hydrocarbon component, Nasr et al. (2003) proposed solvent-assisted steam gravity drainage (ES-SAGD), hydrocarbons (alkanes, aromatic hydrocarbons) mixed with steam injected into the reservoir to reduce the amount of injected steam and the steam consumption and to improve oil recovery. As a result, experimental research and industrial operations showed significant increases in oil production compared to the SAGD process (Souraki, et al., 2013; Al-Murayri et al., 2016). The mass transfer and diffusion of solvent is the crucial mechanism. Therefore, the solvent intended to reduce oil viscosity can also be used during the start-up process. However, the amount of experimental research focusing on the start-up period remains limited. In past experimental studies, propane was known as an adaptive solvent to reduce the viscosity of heavy oil (Leyva-Gomez and Babadagli, 2013; Voskovab et al., 2016). However, its solubility depended strongly on temperature and pressure and caused strong asphaltene

Pressure transducer Value

P

Back pressure

Bypass

Back pressure value

Sand pack

Value

Xylene Piston container Water

Value

P

Pressure transducer ISCO pump Water tank Fig. 1. Experimental apparatus for solvent injection and soak.

2

Data acquisition system

Journal of Petroleum Science and Engineering (xxxx) xxxx–xxxx

Z. Yuan et al.

(Marciales and Babadagli, 2016). Because experimental results indicate that a mixture of hexane and oil leads to significant precipitation, xylene was selected for this experiment.

encouraging results. According to the oilfield pilot test, solventassisted start-up technology could enhance effective communication between the injector and the producer, shorten preheating time, and decrease the cost of steam consumption.

2.3. Experimental procedures 2. Experiments All experiments were performed at a room temperature of 20 °C. The experimental procedures are briefly described as follows:

2.1. Experimental apparatus

(1) Before the experiment, the sand pack was filled with sand, and a heated heavy oil sample was pressed into the sand pack, displacing residual water from the pore, which was then saturated with oil. The sand pack experiment for solvent- assisted start-up was conducted when the oil temperature cooled down to the atmospheric temperature. (2) The sand pack was set vertically, and the catheters were connected. The pump pressure was set to 5.0 MPa (constant), which was a little less than the caprock pressure. The xylene solvent was injected continuously into the sand pack until the mixture of oil and xylene flowed from the outlet. (3) After the mixture fluid flows from the outlet, the pressure of the inlet is no longer constant. Xylene was injected at different flow rates of 0.5, 1, 2, and 5 ml/min, respectively, into the sand pack. The pressures of the inlet and the outlet were recorded, the differential pressure was calculated, and the viscosities of the production liquid at different flow rates were measured. Then, according to Darcy's formula, the permeability of the oil phase in the sand pack was calculated. This was the first process. (4) The inlet valve and the outlet valve were closed, and the solvent soak was kept with pressure for 3 days. After soaking, the xylene injection continued in a similar manner at flow rates of 0.5, 1, 2, and 5 ml/min, respectively, into sand pack. The viscosity of the fluid produced was measured. This was the second process. (5) The solvent soak process was operated a second time for 2 days, and Steps 4 were repeated. The production fluid was collected, and the viscosity was measured. This was the third process. (6) The solvent soak process was operated third time for 2 days, and Steps 4 were repeated. This was the fourth process. (7) The weight of the production fluid was measured, and the xylene was vapored by heating. The weight of the produced oil was measured, and the solvent content of the output liquid was calculated.

Fig. 1 shows the experimental apparatus of the sand pack experiment. A piston container is used for storing the experimental solvent with a volume of 200 ml, and the maximum working pressure is 32.0 MPa. A sand pack is with a steel tube as the main part of the experiment. Solvent, injected from the inlet, flows towards the outlet. The maximum temperature is 300 °C, and the maximum working pressure is 32.0 MPa. Back pressure valves control the maximum pressure of the outlet and ensure the security of experiment. An ISCO pump displaces the fluid above the piston into the sand pack and controls the pressure and flow rate of the inlet. A data-acquisition system, with two pressure transducers, records the transient pressures of the inlet and the outlet of the sand pack as well as the transient flow rates of the pump. A scale measures the weight of the production fluid and oil to calculate the solvent concentration. 2.2. Experimental materials The experimental materials are as follows: 2.2.1. Sand pack The sand pack was made of quartz sand with grain sizes 120–160 meshes. The diameter and length of sand pack were 3.0 and 50.0 cm, respectively. The measured volume of the sand pack was 350 cm2. For best results and to prevent bubbles, sands were alternately filled with water, stirred, and compacted during the filling. The measured porosity of the sand pack was 32.0%, and the irreducible water saturation was 5.0%, both of which were consistent with an actual oil field. 2.2.2. Oil The oil sample used in this study was collected from the Xinjiang Oilfield. Oil viscosity was measured by using a Thermo Scientific HAAKE MARS III Rheometer under atmospheric pressure and temperature. The initial viscosity was approximately 1,167,243.1 mPa s, which was a super-heavy oil sample.

2.4. Results and discussion Fig. 3 shows the relationship between the oil viscosity and the amount of injected xylene. As shown in Fig. 3, under 1 MPa differential pressure between the inlet and the outlet, the mixture of oil and xylene was produced from outlet after 15 h. The viscosity of production fluid was 13.4 mPa s. With xylene injected continually, in the first process,

2.2.3. Solvent Before the experiment, the effects of three different solvents (hexane, xylene, and diesel) on oil-viscosity reduction were tested. Because of the high content of asphaltene in super-heavy oil, it was common for asphaltene precipitation to generate near the wellbore or the well horizontal section during the SAGD production process. It blocks the flow channel and the wellbore, which adversely affects mobility and the production of oil (Mukhametshina et al., 2015). Therefore, it was essential to choose a suitable solvent to reduce oil viscosity and prevent asphaltene precipitation. A mixture of hexane, xylene, and diesel was added to the oil sample to test the effects on viscosity reduction and precipitation. Fig. 2 shows the relationship between the content of different solvents and the viscosity of oil. From Fig. 2, it is clear that the oil viscosity reduces significantly with the increase in solvent. When the solvent content is more than 50%, oil viscosity is less than 100 mPa·s with strong mobility. The effect of hexane on viscosity reduction is best, whereas xylene somewhat better than that of diesel. Previous studies have shown that the use of alkanes rather than aromatic hydrocarbons results in greater asphaltene precipitation (Liu et al., 2015; Wu et al., 2015). At 20 °C, asphaltenes can be dissolved completely in xylene without the formation of asphalt precipitation

1000000

Viscosity(mPa·s)

100000

hexane xylene diesel

10000

1000

100

10

1

0

20

40

60

80

Solvent content(%) Fig. 2. Mixture viscosity of different solvents and oil.

3

100

Journal of Petroleum Science and Engineering (xxxx) xxxx–xxxx

Z. Yuan et al.

50

with a CMG used as a simulator, serves to investigate the solventinjection and steam-circulation processes.

oil viscosity

40

Viscosity(mPa·s)

3.1. Simulation model The single well-pair model is homogeneous in regard to porosity and permeability. The total number of blocks is 104,880, and the fundamental grid dimensions are Nx = 46; Ny = 40; and Nz = 57. The grid size along the X-direction is 10 m; in the Y-direction, 1; and in Z-direction, 1 m for the top 12 rows and 0.2 m for the rest. The key reservoir properties used in the model are presented in Table 1. Heavy-oil and xylene viscosity versus temperature are given in Table 2. Oil–water relative permeability and gas–liquid relative permeability are given in Table 3. To simulate the interaction between the oil phase and the injected solvent accurately, Peng-Robinson EOS in CMG WinProp was used to calculate the properties of the oil–solvent mixture within the reservoir volume impacted by solvent. According to the results (Fig. 2), the nonlinear interpolation of oil viscosity reduction by the injected solvent could be calculated.

30

20

10

0

0

50

100

150

200

250

300

350

400

Volume of injected solvent(mL) Fig. 3. Relationship between oil viscosity and the amount of injected xylene.

the viscosity of oil decreased to 7.0 mPa s. After 3 days of solvent soak, at the beginning of the second phase, oil viscosity increased up to 34.5 mPa s, which was later reduced to 6.4 mPa s with solvent injection. After 2 days of solvent soak (one day less than in the second process), third- and fourth-process oil viscosity decreased from 18.0 to 6.0 mPa s and from 7.9 to 6 mPa s, respectively. When the solvent is injected into the sand pack, the xylene dissolves into the heavy oil, according to the theory of like dissolving like. The increasing solvent concentration of the mixture fluid can reduce viscosity and enhance mobility significantly. During the solvent soak period, the solvent diffused to the surrounding area and improved the swept area. After which, the decreasing solvent concentration causes an increase in oil viscosity. Thus, at the beginning of the new process, oil viscosity increases significantly. The longer the soaking time, the higher the viscosity of the production mixture because the solvent dilutes after the long soaking time, which leads to increased oil viscosity. Then, by increasing the concentration of xylene solvent through injection, the viscosity of produced fluid decreases rapidly. After repeated soakings, the degree of viscosity reduction diminishes, indicating that adding more injection solvent would no longer reduce the viscosity significantly. These findings indicate that solvent has limitative reduction capabilities after at approximately 6 mPa s in this study. Fig. 4 shows the relationship between oil-phase permeability and injection rate during different processes. In the first process, oilphase permeability increases from 325 to 469 mD. After repeated solvent soakings, the oil-phase permeability increases significantly. It increases from 419 to 593 mD in the second process; in the third and fourth processes, from 411 to 604 mD and 440–627 mD, respectively. The oil-phase permeability increased with increased injection rates. The main reason is that the solvent injection rates have a significant effect on the convective mass transfer between the bitumen and solvent (Ahmadloo and Yang, 2014). With increased injection rates, the mixing rate increases, resulting in higher solvent concentration in the mixture fluid. The higher the injection rates, the lower the oil concentration in the mixture. Thus, viscosity is lower, and oil-phase permeability increases. After repeated soakings, as solvent fully dissolved into the bitumen and improve contact between the xylene and the heavy oil, the oil-phase permeability increase. Whereas, the degree of permeability increase diminishes, adding that more injection solvent would no longer improve the permeability significantly.

3.2. Results and discussion 3.2.1. Change in the start-up process Figs. 5 and 6 show the results of the simulation parameters, including the solvent mole fraction and oil viscosity between the injector and the producer. As shown in Fig. 5, at the end of the solvent injection, the solvent is located near the wellbore only. As shown in Fig. 6, after the injection of solvent, oil viscosity near the wellbore decreased rapidly, to approximately 100.0 mPa s. As the solvent goes further into the formation after steam circulation, the oil viscosity between the well pair decreases significantly. Changes to oil viscosity and xylene mole fraction distribution at different times reflect the mechanism of the solvent-assisted start-up for SAGD. After the soak process, xylene further expands into the cross-well formation to dissolve more oil. However, too long a soak time leads to more xylene either dissolving into the heavy oil or expanding to a deeper area beyond the wellbore region, both of which have negative effects on oil-viscosity reduction. During the steam circulation process, high-temperature steam is injected into the formation between the well pair to further enhance the mass transfer rate and the formation temperature and to accelerate the oil-viscosity reduction. This process is to establish the uniform communication between wells and to avoid the steam breakthrough. Moreover, xylene is soluble in asphaltenes, which can also enhance their mobility. After steam circulation, the solvent distributes evenly within the entire interwell region, and the oil viscosity between the well pair has mostly 650

Oil phase permeability(mD)

600

3. Numerical simulation

550 500 450

first process second process third process forth process

400 350 300

0

1

2

3

4

5

Injection rate(ml/min)

To analyze the mechanism of solvent soak and its effect on the startup process fully, a conceptual homogeneous model of a single well pair,

Fig. 4. Relationship between permeability of oil phase and injection rate.

4

6

Journal of Petroleum Science and Engineering (xxxx) xxxx–xxxx

Z. Yuan et al.

profile, which is homogenous along the horizontal section, generally approaches 120 °C, indicating uniform preheating and communication. However, the necessary steam-circulation time for the solvent-assisted start-up is 140 days, whereas that of only steam injection requires nearly 30 days more. The main reason is that the oil viscosity near the wellbore is reduced, which enhances its mobility by the initially injected solvent. After which, the flow resistance is lower for steam circulation. Therefore, the rate of potential heat transfer of the hightemperature steam to the oil-saturation formation is improved. The affected zone is heated rapidly and is shortened by the preheating time, whereas less steam is needed to establish well communication.

Table 1 Key simulation parameters used in the model. Items

Values

Items

Values 2.1 × 106

Depth to reservoir top (m) 300

Rock heat capacity (J/ (m3 °C)) Initial reservoir 20 Rock conductivity (J/(m temperature (°C) day °C)) Initial oil viscosity (mPa s) 1,167,243.1 Water conductivity (J/ (m day °C)) Reference pressure (kPa) 3000 Oil conductivity (J/(m day °C)) Initial oil saturation (f) 0.75 Gas conductivity (J/(m day °C)) Porosity (f) 0.30 C6 K-value = (Kv1/P) exp(Kv4/(T+Kv5)) 2.00 Kv1 (kPa) Horizontal permeability (D) Vertical permeability (D) 1.60 Kv4 (°C) Formation compressibility 2.38 × 10−5 Kv5 (°C) (1/kPa)

2.0 × 105 5.5 × 104 1.2 × 104 4.0 × 103

7.449 × 10

3.2.2. Predicting production rate and cumulative oil production A comparison of the oil-production rate and the cumulative oil production between well pair treated with solvent and the well pair without during the early production period is shown in Fig. 8. Fig. 8(a) shows that the production rate increases steadily and that the production rates of the solvent-assisted well pair and only steaminjection well pair are 39.71 and 29.59 m3/d, respectively, after 360 days. The CSOR of the two cases at this time is 3.14 and 3.27, respectively. Fig. 8(b) shows that the cumulative oil production with the solvent treatment is 6811 m3, nearly 44.0% higher. These findings indicate that solvent injection has a hybrid effect with high-temperature steam. Enhanced communication between the well pair is effective for developing the steam chamber. During the heat and mass transfer, the steam chamber grows rapidly and accelerates the rate of oil drainage. So the production rate with the solvent treatment is somewhat higher. The residual solvent can further extract the asphaltene to reduce residual oil saturation and dissolve potential asphaltene precipitation in the steam chamber, which can improve the oil recovery factor further. To maintain the stability of production, injecting the asphaltene-soluble solvent with steam during the production period is recommended.

6

−3120 −209

Table 2 Heavy oil and xylene viscosity versus temperature. T (°C)

μo (mPa s)

μs (mPa s)

20 40 70 100 150 200 250

1,167,243.1 64,020.8 3004.9 383.4 44.7 12 6

0.75 0.61 0.44 0.32 0.23 0.13 0.07

Table 3 Oil–water relative permeability and gas-liquid relative permeability. Sw

krw

krow

Sl

krg

krog

4. Effects in the oilfield

0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.67

0.000 0.002 0.006 0.009 0.017 0.028 0.042 0.059 0.080

1.000 0.722 0.541 0.402 0.287 0.171 0.082 0.023 0.000

0.53 0.60 0.70 0.75 0.80 0.85 0.90 0.95 0.99

0.450 0.363 0.212 0.148 0.093 0.069 0.031 0.005 0.000

0.000 0.035 0.208 0.329 0.471 0.549 0.717 0.902 1.000

To better study the effect of the use of this solvent-assisted technique, compared with the conventional method, on a SAGD start-up, a pilot test was performed in an oilfield in Xinjiang Oilfield, China. The performance of the preheating process was compared with other well pairs with no solvent injection. At the beginning of the pilot test, the designated volumes of the xylene solvent were injected into the injector and the producer, after which, the well pair was shut to soak. The solvent dissolved into the oil and diffused deep into the formation to improve contact between the xylene and the heavy oil. After several days of solvent soak, 50 m3 of xylene was injected again to expand the sweep area of xylene further. After the second solvent soak, superheated steam was injected and circulated for 3 months to reduce oil viscosity in the deep formation further and to enhance the formation temperature between the well pairs. Then, the producer turned to produce while the steam injection continued for 30 days. When the

declined to 100 mPa s. A more homogeneous xylene concentration would be helpful for the growth of the steam chamber. Fig. 7 shows the temperature profile at the end of the start-up period after different injection strategies. As shown in Fig. 7, the temperature in the inter-well region is enhanced evenly for both the solvent-assisted and the steam-only injections. The temperature

Fig. 5. Solvent fraction at different stages.

5

Journal of Petroleum Science and Engineering (xxxx) xxxx–xxxx

Z. Yuan et al.

Fig. 6. Oil viscosity at different stages.

temperature profile showed good communication, the well pair was turned to SAGD production. On the basis of the processes and effects of the pilot study, it was determined that the start-up of well pairs with solvent soak has a shorter cycle time, which lasts 130–150 days, compared with the cycle time of conventional steam circulation of more than 200 days. The volume of steam saved was approximately 4000 m3 during the preheating phase. The degree of communication before turning to SAGD production reached 85%, indicating a better effect on the formation between the well pairs. The cumulative oil production after the first 150 days showed higher production. The better performance demonstrates that xylene solvent injected for SAGD start-up is effective for improving preheating efficiency and oil production in heavy oil reservoirs.

50

Oil production rate(m3/d)

40

5. Conclusions

30

20

solvent steam

10

0

0

50

100

150

200

250

300

350

400

Production days(d)

On the basis of the present study, the following conclusions can be drawn:

(a) Relationship between production days and production rate

(1) Compared to with that of hexane and diesel, the effect of xylene on oil-viscosity reduction is significant, and because of an absence of asphaltene precipitation, fluid flow is not hindered. Also, the mobility of heavy oil is enhanced by the mixture of xylene. (2) Solvent injection at higher rates can enhance the mass transfer and the rate of mixing between the injected solvent and the heavy oil. After soaking, the injected solvent diffuses into the bitumen and can greatly improve the relative oil-phase permeability. However, if the soak time is too long, the solvent will diffuse deeper into the formation, which has the opposite effect on oil-viscosity reduction. (3) The main goal of the solvent assisted start-up is to preliminarily reduce the viscosity of the oil near the wellbore by solvent injection instead of by steam. After which, the sweep area is improved by high-temperature steam circulation to reduce the oil viscosity in the formation further and to easily form a steam chamber. These novel results indicate that solvent injected into an initial reservoir can effectively shorten the start-up process and decrease the cost of steam consumption.

Cumulative oil production(m3)

10000

solvent steam

8000

6000

4000

2000

0

0

50

100

150

200

250

300

350

400

Production days(d)

(b) Relationship between production days and cumulative oil production Fig. 8. Production performance of the two cases.

Fig. 7. Temperature profile after the start-up period.

6

Journal of Petroleum Science and Engineering (xxxx) xxxx–xxxx

Z. Yuan et al.

bitumen deposit. J. Pet. Sci. Eng. 116, 28–35. Leyva-Gomez, H., Babadagli, T., 2013. Numerical simulation of heavy-oil/bitumen recovery by solvent injection at elevated temperatures. J. Pet. Sci. Eng. 110, 199–209. Liu, P., Li, W., Shen, D., 2015. Experimental study and pilot test of urea-and urea-andfoam-assisted steam flooding in heavy oil reservoirs. J. Pet. Sci. Eng. 135, 291–298. Marciales, A., Babadagli, T., 2016. Selection of optimal solvent type for high-temperature solvent applications in heavy-oil and bitumen recovery. Energy Fuel 30, (2563–273). Mohammadzadeh, O., Rezaei, N., Chatzis, I., 2012. Production characteristics of the steam-assisted gravity drainage (SAGD) and solvent-aided SAGD (SA-SAGD) processes using a 2-D macroscale physical model. Energy Fuel 26 (4), 4346–4365. Mohammed, M., Babadagli, T., 2014. Alteration of Matrix Wettability during Alternate Injection of Hot-water/solvent Into Heavy-oil Containing Fractured Reservoirs. SPE170034, Alberta, Canada, June 10–12. Mohsenzadeh, A., Al-Wahaibi, Y., Jibril, A., Al-Hajri, R., Shuwa, S., 2014. The novel use of deep eutectic solvents for enhancing heavy oil recovery. J. Pet. Sci. Eng. 130, 6–15. Mukhametshina, A., Kar, T., Hascakir, B., 2015. Asphaltene precipitation during bitumen extraction with expanding-solvent steam-assisted gravity drainage: effects on porescale displacement. SPE J., 380–392. Nasr, T.N., Beaulieu, G., Golbeck, H., Heck, G., 2003. Novel expanding solvent-SAGD process "ES-SAGD". J. Can. Pet. Technol. 42 (1), 13–16. Parmar, G., Zhao, L., Graham, J., 2009. Start-up of SAGD wells: history match, wellbore design and operation. J. Can. Pet. Technol. 48 (1), 42–48. Saks, D., Kyanpour, M., Onamade, O., 2015. Evaluation of Thermal Efficiency of the Preheat Period in the SAGD Process for Different Completion Methods. SPE 174450, Alberta, Canada, 9–11 June. Souraki, S., Ashrafi, M., Torsaete, O., 2013. A comparative field-scale simulation study on feasibility of SAGD and ES-SAGD processes in naturally fractured bitumen reservoirs. Energy Environ. Res. 3 (1), 49–63. Voskovab, D., Zaydullinbd, R., Luciac, A., 2016. Heavy oil recovery efficiency using SAGD, SAGD with propane co-injection and strip-SAGD. Comput. Chem. Eng. 88, 115–125. Wu, Y.B., Li, X.L., Zhao, R.,Li, J., Liu, X.Q., Zhou, Y., Du, X., 2015. Case Study of Solventassisted Start-up in Fengcheng SAGD project. SPE 174440, Alberta, Canada, 9–11 June. Zendehboudi, S., Rajabzadeh, A.R., Bahadori, A., Chatzis, L., Dusseault, M.B., Elkamel, A., Lohi, A., Fowler, M., 2014. Connectionist model to estimate performance of steam-assisted gravity drainage in fractured and unfractured petroleum reservoirs: enhanced oil recovery implications. Ind. Eng. Chem. Res. 53 (4), 1645–1662. Zhang, J., Li, C.N., Zeng, F.H., Chen, P., Zhang, Z., Zhou, C.H., Zhang, W.L., 2012. To Start up Extra-heavy Oil or Bitumen Well by Solvents after Drilling. SPE 158341, Alberta, Canada, 12–14 June. Zhao, D.W., Wang, J., Gates, I.D., 2013. Optimized solvent-aided steam-flooding strategy for recovery of thin heavy oil reservoirs. Fuel 112, 50–59. Zhu, L., Zeng, F., Zhao, G., Duong, A., 2012. Using transient temperature analysis to evaluate steam circulation in SAGD start-up processes. J. Pet. Sci. Eng. 100, 131–145.

(4) Simulations of a well pair with different injection strategies showed that less time is needed to obtain the same temperature profile than when only steam injection is used. The consumption and cost of steam injection during the preheating period can be reduced to achieve greater economic benefits. During the early production period, the amount of cumulative oil produced by solvent injection is slightly higher than that with no solvent injection. Acknowledgements This research is financially supported by the National Oil & Gas Key Special Project of China for 2016ZX05012-002, 2016ZX05016-006 and 2016ZX05015-002 the contribution of which and the permission for publication are gratefully acknowledged. References Ahmadloo, F., Yang, P., 2014. Solvent-Assisted Start-up of SAGD Wells in Long Lake project. SPE 170052, Alberta, Canada. 10–12 June. Al-Murayri, M.T., Maini, B.B., Harding, T.H., Oskouei, J., 2016. Multicomponent solvent co-injection with steam in heavy and extra-heavy oil reservoirs. Energy Fuel 30 (4), 2604–2616. Anderson, M.T.I., Kennedy, D.B., 2012. SAGD Startup: Leaving the Heat in the Reservoir. SPE 157918, Alberta, Canada, 12–14 June. Bayat, A.E., Junin, R., Kharrat, R., Shamshirband, S., Akib, S., Buang, Z., 2015. Optimization of solvent composition and injection rate in vapour extraction process. J. Pet. Sci. Eng. 128, 33–43. Boyle, T.B., Gittins, S.D., Chakrabarty, C., 2003. The evolution of SAGD technology at East Senlac. J. Can. Pet. Technol. 42 (1), 58–61. Butler, R.M., 1994. Steam-assisted gravity drainage: concept, development, performance and future. J. Can. Pet. Technol. 33 (2), 44–50. Butler, R.M., Stephens, D.J., 1981. The gravity drainage of steam heated to parallel horizontal wells. J. Can. Pet. Technol. 20 (2), 90–96. Gates, I.D., Leskiw, C., 2010. Impact of steam trap control on performance of steam assisted gravity drainage. J. Pet. Sci. Eng. 75 (1), 215–222. Jha, R.K., Kumar, M., Benson, I., Hanzlik, E., 2013. New insights into steam/solventcoinjection-process mechanism. SPE J. 18 (5), 867–877. Ji, D.Q., Dong, M.Z., Chen, Z.X., 2015. Analysis of steam-solvent-bitumn phase behavior and solvent mass transfer for improving the performance of the ES-SAGD process. J. Pet. Sci. Eng. 133, 826–837. Kaiser, T.M.V., Taubner S., 2015. Factors Contributing to SAGD Start-up Performance. SPE 178451, Alberta, Canada, 23–25 November. Kazeem, A.L., 2014. Economics of steam-assisted gravity drainage for the Nigerian

7