Cracked Naphtha Coinjection in Steam-Assisted Gravity Drainage

Jun 6, 2016 - This requires large quantities of water and natural gas, resulting in sizable ... conditions is made mobile by heating and mixing solven...
0 downloads 0 Views 9MB Size
Article pubs.acs.org/EF

Cracked Naphtha Coinjection in Steam-Assisted Gravity Drainage Mohammed T. Al-Murayri,† Brij B. Maini,‡ Thomas G. Harding,*,† and Javad Oskouei‡ †

Nexen Energy ULC, 801 7th Avenue SW, Calgary, Alberta T2P 3P7, Canada Chemical and Petroleum Engineering, University of Calgary Schulich School of Engineering, Calgary, Alberta T2N 1N4, Canada



ABSTRACT: Steam-assisted gravity drainage (SAGD) is a thermal in situ recovery method for heavy oil and oil sands that has been employed to exploit the vast petroleum deposits in the Athabasca region of northern Alberta that are not amenable to surface mining. Nevertheless, in spite of its success in recovering highly viscous bitumen, SAGD remains a costly technology that requires large energy input in the form of steam for each barrel of produced oil. This requires large quantities of water and natural gas, resulting in sizable greenhouse gas (GHG) emissions and extensive postproduction water treatment. There are ongoing efforts to make SAGD more energy-efficient and environmentally sustainable by reducing steam consumption while maintaining favorable oil production rates and ultimate oil recovery. Such efforts include the coinjection of steam and solvent in a process called expanding solvent SAGD (ES-SAGD) wherein bitumen that is essentially immobile at initial reservoir conditions is made mobile by heating and mixing solvent into the oil. In this work, experiments are reported in which cracked naphtha is applied as an additive to steam for the production of Long Lake bitumen. The results conclusively demonstrate that coinjecting cracked naphtha with steam, under specific conditions, significantly increases oil production rates and reduces overall steam requirements.



INTRODUCTION This paper is based on the Ph.D. thesis of the lead author and most of the material presented here is available in that document.1 The steam-assisted gravity drainage (SAGD) process was invented by Roger Butler and his former colleagues at Imperial Oil in the late 1970s. In SAGD, high pressure (1−7 MPa) steam is continuously injected into an oil-bearing formation through a horizontal injection well that is placed close to and directly above a horizontal production well. The distance between the SAGD injection and production wells is typically around 5 m. Steam injected into the upper well rises in the formation under gravitational force and forms a steam chamber above the well. The production well in SAGD is normally situated near the bottom of the reservoir to enhance the gravitational drainage of bitumen and steam condensate, which frees up space that is then occupied by the continuously injected steam. As more oil and steam condensate drain, the steam chamber continues to grow. The idea of using solvents instead of steam, or along with it, to reduce viscosity of highly viscous oil to promote its recovery, has been studied by many researchers. Butler and Mokrys2 proposed a process called vapor extraction or VAPEX, which is essentially a solvent-based analogue of SAGD, governed mainly by mass transfer and gravity drainage. Laboratory results indicated that VAPEX was promising for field applications; however, thus far, VAPEX has not been proven to be commercially viable. The field application of VAPEX appears to have had limited success because the process is dependent on molecular diffusion which is slower and less effective in mobilizing oil than thermal diffusion. One possible approach to increase the rate of molecular diffusion is to augment it with heat. Numerous processes continue to evolve in a quest to combine the advantages of thermal and solvent-based processes. “Expanding Solvent−Steam Assisted Gravity Drainage (ES-SAGD)” is such a process that was originally envisioned by Nasr and Isaacs3 which combines the benefits of © 2016 American Chemical Society

SAGD and VAPEX. The rationale of using this name was not explained by them but may be related to increasing the vapor phase mole fraction of solvent as steam migrates toward the edge of the steam chamber and a portion of the water vapor condenses out. In ES-SAGD, steam and solvent are coinjected using typical SAGD/VAPEX well configuration to mobilize highly viscous bitumen through a synergistic mix of heat and mass transfer processes to increase oil production, while maintaining lower energy and water requirements. The interrelated effects of heat transfer, diffusion and dispersion, in addition to reservoir properties and operating conditions, play important roles in determining the appropriate solvent composition, concentration, and coinjection strategy that can be used to enhance the overall performance of SAGD. It is hypothesized that the coinjected solvent will be retained in the vapor phase and will travel with steam until it condenses at or near the boundary of the steam/vapor chamber. Once at the boundary, the condensed solvent would then dilute the highly viscous oil and, in conjunction with heat, increase its mobility and thereby provide a higher drainage rate than heating alone. As the diluted oil drains, the interface advances toward the undrained portion of the reservoir thereby exposing more of the formation to heating and dilution. Enhancing SAGD performance through solvent addition is a promising technique that warrants more attention because it can plausibly increase the range of applicability of the recovery process. Some physical processes that are associated with hybrid steam-solvent oil recovery techniques cannot be adequately captured using numerical simulators, thus encouraging the use of evaluation tools such as physical testing. Numerical simulators depend strongly on the reliability and quality of input parameters including, but not limited to, relative permeability curves, fluid Received: November 24, 2015 Revised: May 19, 2016 Published: June 6, 2016 5330

DOI: 10.1021/acs.energyfuels.5b02773 Energy Fuels 2016, 30, 5330−5340

Article

Energy & Fuels

and outlines the procedure that was followed to produce and interpret the results. The experimental rig used in this study was similar to that used by Hosseininejad Mohebati et al.4 It has been described in detail in Al-Murayri1,8 The major pieces of equipment in the experimental setup include a confining pressure jacket, sand-pack model including both injection and production wells, positive displacement fluid pumps, nitrogen gas supply, helium/argon gas supply, a water transfer system to pressurize the model, visual separators, condensate traps, back pressure regulators, pressure transducers, two gasometers and a gas chromatography system and a control/monitoring panel as shown in Figure 10. The 2D partially scaled physical model shown in Figure 2 was fabricated using type 316 stainless steel. Its dimensions and char-

characterization, and phase behavior. Furthermore, there are inherent deficiencies in the existing numerical simulators that make capturing certain physical phenomena such as asphaltene precipitation somewhat challenging. Utilizing solvents to enhance the performance of SAGD necessitates a firm understanding of key parameters such as the most appropriate solvent types, concentrations, and operating strategies. Needless to say, these parameters can be influenced, to a great extent, by reservoir properties. In this study aspects of cracked naphtha solvent/steam coinjection are investigated using Long Lake bitumen, reservoir properties and operating conditions. Cracked naphtha is a multicomponent solvent that contains paraffinic, aromatic, and olefinic hydrocarbons and a typical composition is shown in Table 1. Furthermore, a Table 1. Typical Composition of Cracked Naphtha components

mole fraction

C4 C5 C6 C7 C8 C9 C10 C11 C12+ Total

0.0063 0.1381 0.1391 0.1741 0.1772 0.1249 0.0821 0.0553 0.1029 1.0000

Figure 2. Sand-pack model. acteristics have been described previously.8 The top and bottom plates of the model were attached to the body of the sand-box with bolts and sealed with O-rings to isolate the interior of the box from confining pressure. The model had 240 thermocouples inserted into it to allow monitoring of temperature profiles inside the sand-pack. These thermocouples were arranged in three planes perpendicular to the horizontal wells with each plane containing 20 four-point thermocouples, each of which measured the temperature at four different locations. Figures 3 and 4, with dimensions in centimeters, illustrate

WinProp-based phase diagram of cracked naphtha is presented in Figure 1. Several high pressure/high temperature SAGD and

Figure 3. Side view of thermocouple planes (dimensions in centimeters).

Figure 1. Phase envelope for cracked naphtha.

ES-SAGD experiments were conducted using a partially scaled physical model. Produced fluid samples collected during the tests and samples of the porous matrix were analyzed to assess process performance in each test. Criteria used for evaluating each experiment include the rate of vapor chamber growth, rates of oil and water production, steam-oil ratio (SOR), oil and water saturations remaining in the test cell, the degree of asphaltene precipitation if any, the viscosity and density of produced oil, and finally the amount of solvent recovered.



Figure 4. Thermocouple locations on the plane perpendicular to the horizontal wells (dimensions in centimeters).

EXPERIMENTAL DETAILS

Model and Analysis. A partially scaled 2D physical model was fabricated to conduct the SAGD and ES-SAGD experiments. Cracked naphtha was coinjected with steam at different concentrations ranging from 5 to 15% by volume. This section describes the experimental setup that was used to conduct the SAGD and ES-SAGD experiments

how the thermocouples were placed within the physical model. The thermocouples were situated at the intersections of the horizontal and vertical lines and the red arrows represent the distance between the thermocouples. Following insertion of the thermocouples, the 5331

DOI: 10.1021/acs.energyfuels.5b02773 Energy Fuels 2016, 30, 5330−5340

Article

Energy & Fuels model was filled with Ottawa sand, which was then saturated with deionized water and Long Lake oil. Thereafter, the sand-pack model was installed in a confining pressure jacket, as shown in Figure 5, which can handle pressures as high as 550 psi.

Figure 8. Preheater and steam generator. Figure 5. Confining pressure jacket. was attached to the sand-pack model to enhance thermal efficiency. The preheater is a 3/8-in. 316 type stainless steel coiled tube (1.5 m long) with a 1/8-in. diameter internal heater (1.5 kW) to heat up the water to partially saturated steam conditions. A thermocouple was placed at the outlet of the preheater to record the temperature of the partially saturated steam before it flowed into the steam generator to ensure that steam in the injection well could reach slightly superheated conditions. The outlet temperature of the steam generator was recorded using a thermocouple at the heel of the injection well. The produced fluids from the production well were collected periodically using two visual separators as shown in Figure 9. One separator was used to receive the produced fluids from the model

As illustrated in Figure 6, a horizontal production well was placed at the middle of the model and close to the bottom as an outlet for the

Figure 6. Locations of the injection and production wells in the model. produced fluids. Two different injection ports were incorporated in the design at heights of 7.5 and 12.5 cm from the bottom. The distance between the production well and these ports is 4.5 and 9.5 cm, respectively. In order to ensure that steam flow to the sand-pack model was uniform, the size of slots in the injection well was increased toward the toe of the well as shown in Figure 7. In the production well, a

Figure 9. Visual separators. Figure 7. Slot size variation in the SAGD injection well. uniform slot size was used along the entire length. Mesh screens were installed on both wells to lessen the accumulation of fines and clays, which can potentially plug up the slots and restrict well injectivity and productivity. The temperatures of the fluids inside the injection and production wells were recorded using the thermocouples that were installed inside the wells. For both the production and injection wells, three thermocouples were implanted at the heel, middle and toe areas, respectively. These thermocouples provided direct measurements of the temperature of the injected and produced fluids. An automated data acquisition system was used to record temperature, pressure and injection rates. Furthermore, a constant-flow ISCO pump was used to inject deionized water into the in-line steam generator at the desired injection rates. Cracked naphtha was added to the deionized water stream before passing through the preheater and the steam generator. A preheater and a steam generator were placed in series during the experiments as illustrated in Figure 8 to ensure that 100% steam quality was injected. They were both buried prior to the commencement of the experiments in a Perlite insulation filled box which

Figure 10. Front view of the control panel. 5332

DOI: 10.1021/acs.energyfuels.5b02773 Energy Fuels 2016, 30, 5330−5340

Article

Energy & Fuels while the other separator was allowed to partially cool down before withdrawing the collected liquids into glass jars. The produced gases from the visual separators flowed through intermediate condensate traps and were collected using two gasometers located downstream of the separators. An Agilent (7890A) gas chromatography system was utilized to analyze the composition of the produced gases every 30 min throughout the life of the ES-SAGD experiments. The stainless steel physical model was placed in a confining pressure jacket. To reduce heat losses from the model, the annular space between the confining jacket and the physical model was filled with insulating ceramic wool. Furthermore, the top and bottom sides of the model were internally insulated with 1/4 in. polytetrafluoroethylene (PTFE) sheet. Model Scaling. The SAGD scaling procedure proposed by Butler et al.5 was used to design the sand-pack model based on the following criteria from Oskouei et al.:6 • The operating pressure of the experiment is equal to the reservoir pressure. • Thermal diffusivity of the matrix in the sand-pack model is equal to that of the reservoir. • The viscosity of the oil used to saturate the sand-pack model is equal to that of the reservoir. • ϕΔSo is the same in the sand-pack model and the reservoir, where ϕ is the porosity and ΔSo is the change in oil saturation. • Transmissibility (kH) of the model is equal to that of the reservoir, where k is the permeability and H is the sand-bed height. These requirements are needed to ensure that the gravity drainage process in physical model experiments is dimensionally similar to the field scale SAGD process.7 Ensuring that (kH)model is equal to (kH)field entails using high permeability coarse sand in order to align the scaling equation with the dimensions of the sand-pack model. For this reason, Ottawa sand (ASTM mesh 12/20) with a permeability of 555 Darcies was used to fill the sand-pack model for all experiments to ensure that (kH)model is equal to (kH)field for a reservoir with a height of 25 m and a permeability of 5 Darcies. The permeability of 12/20 mesh Ottawa sand was measured independently by preparing a cylindrical sand-pack and running a steady-state flow test at 100% saturation of water. It is assumed that the permeability of this sand is similar in the physical model to that in the cylindrical sand-pack. Such models are considered only partially scaled because some of the process parameters, such as ratio of capillary forces to gravity forces, are not scaled. Model Preparation. The sand-pack model was filled with high permeability sand while being vibrated to enhance sand homogeneity. Around 3 kg of sand was added every hour until it was impossible to add more sand into the sand-pack model. It normally took around 43.5 kg of sand to fill the sand-pack model over a period of 15 h. The model was then evacuated overnight for a period of around 12 h to remove air from the sand matrix after which deionized water was added. The bulk volume of the sand matrix was determined by filling the sand-pack model with deionized water before sand addition and was found to be around 24 080 cm3. The pore volume of the sand matrix was calculated using the amount of water which was added to saturate it, assuming water density of 1 g/cm3 for pore and bulk volume calculations. The total porosity of the sand matrix was then calculated by dividing pore volume by bulk volume. After water addition, the model was flooded with oil until residual water saturation was reached. Due to the high viscosity of the oil used, oil flooding was

undertaken by placing the sand-pack model and the transfer vessel inside an oven at a temperature of 60 °C. Since the density of the oil used was higher than that of water, oil flooding was done in vertically upward direction to improve sweep efficiency. In view of the highly favorable mobility ratio in the displacement of water by bitumen and additional stability provided by gravity, it can be assumed that the oil will displace water in a pistonlike manner and no fingering will take place. Oil flooding was normally done over a period of 48 h. Finally, the produced water was weighed and the corresponding volume was considered to be equal to the volume of oil in the model. The properties of the sand-pack matrix for all of the experiments are presented in Table 2. The densities of oil and water were used to calculate their volumes. The ratio of the volume of oil to the pore volume was considered to be equal to oil saturation. Similarly, the ratio of the volume of water to the pore volume was considered to be equal to water saturation. All of the experiments reported in this work were performed using dead oil. The bitumen used in these experiments was provided by Nexen and was SAGD produced oil collected from the inlet separator of their Long Lake operation. It contained significant amount of diluent which was boiled off using vacuum distillation in rotary evaporators. A large batch of such oil was prepared for these tests and the oil viscosity of the prepared oil was made equal to the viscosity of the original Long Lake oil by blending with some of the separated diluent. The prepared oil was kept in a freezer until needed for the tests. Pressurization of the Sand-Pack Model. The oil saturated sandpack model was placed inside the confining pressure jacket. Thereafter, the thermocouples in the model were connected to the automated data acquisition system and the annulus was filled with ceramic wool. Pressurization of the model was carried out by injecting nitrogen into the annulus, which also pushed deionized water from the water transfer bottle into the sand-pack model until the desired operating pressure was reached. This ensured pressure equilibrium between the inside and outside of the sand-pack model without incurring the risk of nitrogen encroachment into it. Given that the sand-pack model was originally saturated with incompressible fluids, the amount of water that was transferred into it from the water transfer bottle throughout the pressurization period was negligible. Careful attention was paid to maintaining pressure equilibrium between the inside and outside of the sand-pack model to prevent damage to model. Once pressurized, the transfer vessel was isolated from the sand-pack model and an overburden pressure of 10−20 psi was applied. Finally, either Argon or helium was used to pressurize the bias bottles of the back pressure regulators (BPR’s), which in turn led to the pressurization of the downstream production string, separators and condensate traps until the desired operating was achieved. It should be noted that, throughout the experiment, the pressure in the annulus had to be maintained such that it was slightly higher than the steam/solvent injection pressure. Depressurization of the Sand-Pack Model. The SAGD and ESSAGD experiments were terminated by following the procedure outlined below in order to prevent the model from being damaged: • Electrical power for the both the preheater and steam generator was switched off while continuing to flow water through them to cool them down.

Table 2. Properties of the Sand-Pack Matrix initial saturation (%) experiment number

mass of sand in the model (g)

1 2 3 4 5

43570 43752 43388.2 43669 43377

pore volume (g)

mass of oil in the model (g)

water

oil

porosity (%)

vol % naphtha in steam

7866 7835 7937.4 7802.9 7870.8

7287.59 7262.5 7414.47 7275.46 7155.82

9.161 9.116 8.411 8.579 10.858

90.839 90.884 91.589 91.421 89.142

32.7 32.5 33.0 32.4 32.7

0 0 10 15 5

5333

DOI: 10.1021/acs.energyfuels.5b02773 Energy Fuels 2016, 30, 5330−5340

Article

Energy & Fuels • Water flow rate was reduced gradually and was finally stopped when both the preheater and steam generator reached a safe temperature of around 100 °C or below. • While the sand-pack model was being cooled down, the domes of the back-pressure regulators were connected to the annulus through the control panel in order to prevent any damage to the sand-pack model during the cooling period. • After cooling the sand-pack model, annulus gas (nitrogen) was discharged via the purge valve while the bias bottles of the back pressure regulators were connected to the annulus. This allowed the pressure in the downstream production string, separators, and condensate traps to be reduced automatically to the same level as the pressure in the annulus while the annulus was being depressurized. • Finally, the ceramic wool in the annulus was removed, and the depressurized sand-pack model was disconnected from the injection and production lines. Fluid Sampling for the SAGD and ES-SAGD Experiments. For the SAGD experiments, the draining liquids including oil and steam condensate were collected in a pressurized visual separator before being withdrawn into glass containers. The produced oil contained significant amount of emulsified water and the produced water contained emulsified oil. These emulsions were apparently generated in situ, since the fluids did not pass through any flow restrictions from the model to the separator. The pressure in the visual separator was regulated by means of a back pressure regulator, which was located in the downstream part of the experimental setup in the gas discharge line. Make-up gas (nitrogen) was injected from a nitrogen supply bottle into the visual separator to prevent pressure reduction in the sand-pack model during the collection of the samples. Excess makeup nitrogen was vented through the back pressure regulator to maintain the pressure inside the separator and thereby in the production well within the sand-pack model. For the ES-SAGD experiments, two visual separators were used concurrently in order to collect the samples without causing any disruption as a result of the production of fluids from the sand-pack model. One separator was put off-line to withdraw the produced fluid samples into glass bottles while keeping the other separator connected to the production line. Produced gases were cooled further downstream as they passed through coiled tubes that were submerged in iced-water. Secondary traps were used to collect the condensed gases from the ES-SAGD process, if any, and the gas that did not condense was allowed to pass through the back pressure regulators, expand to atmospheric pressure for measurement by the gasometer system. A small sample of the expanded gases was sent every 30 min to the gas chromatograph system for quantitative gas analysis. The liquids collected in the visual separators and condensate traps were withdrawn into glass containers periodically throughout the life of the experiment. Produced Fluid and Sand-Pack Analysis. The produced liquid samples from the experiments included bitumen-in-water and water-inbitumen emulsions. This is similar to the field operations, where production from SAGD operations is mostly in the form of emulsions. However, the ES-SAGD experiments resulted in the production of diluted bitumen with varying amounts of solvent. Some free water separated from the emulsions when the samples were allowed to stand. The free water from the produced emulsion samples was removed and weighed before being disposed. The remaining oil-rich emulsion was then homogenized (using a high speed homogenizer) and around 50 g of it was used to measure water content using Dean−Stark Distillation. For the purpose of this study, water analysis was not considered and the impact of solvent coinjection on SAGD performance was investigated mainly based on bitumen analysis. The dehydration of bitumen by means of simple centrifugation, addition of chemical de-emulsifiers and distillation is challenging. It can potentially result in compositional changes and chemical alterations, thereby affecting the integrity of postrun fluid analysis which includes density, viscosity, asphaltene content and solvent recovery measurements. The produced bitumen samples from all of the experiments in this study were dehydrated cautiously using hygroscopic salts. The dehydrated bitumen samples

were then used to conduct density, viscosity and asphaltene content measurements. The viscosity and density of the dehydrated produced bitumen samples were measured using a Thermo Haake (VT 550) rheometer and an Anton Paar (DMA 5000) densitometer, respectively. Viscosity was measured at three temperatures and density was measured at one temperature. The method of measuring asphaltene content in the residual oil remaining in the porous medium after the tests were completed was described previously.1,8 The amount of solvent that was recovered in the liquid phase with bitumen was measured using Simulated Distillation (SimDis) while the volume of solvent that was recovered in the vapor phase was measured periodically by employing two gasometers in conjunction with chromatographic analysis of the gas. Total solvent recovery was then determined by adding the two amounts. After cooling and depressurizing the model, the porous medium was divided into 15 segments, 3 horizontal and 5 vertical, as shown in Figure 11. It is possible that some redistribution of fluids, especially

Figure 11. Sand-matrix sample locations. water, took place during the cooling period. The bitumen is less likely to redistribute because it is essentially immobile at the residual saturation in steam chamber and has high viscosity in the undrained colder portions of the model. Nonetheless, each sample was analyzed to determine its asphaltene content, along with water and residual oil saturations as previously described.8 A material balance approach was used to determine the physically separated amounts of water, bitumen, and sand. The weight of oil was determined from measured weights of the original sample, cleaned dry sand and collected water. Experimental Procedure. The experimental procedure has been previously described in detail1,8 and further information may be obtained from these references. Two SAGD experiments were conducted to ascertain the operability of the experimental setup and to establish a SAGD baseline for comparison with the subsequent cracked naphtha ES-SAGD experiments. In Experiment 1, a small leak of nitrogen occurred from the annulus into the sand-pack model approximately 100 min after the experiment started. This leakage was caused by an inadequately tightened Swagelok fitting on the east side of the sand-pack model. The impact of nitrogen inflow into the sand matrix was somewhat contained by keeping the steam injection pressure at the same level as the annulus pressure. Nitrogen leakage was not encountered during Experiment 2 that was able to proceed as planned. The materials used for the two SAGD and three cracked naphtha ES-SAGD experiments were 555 Darcy Ottawa sand, deionized water, and Long Lake bitumen. The mass of sand, pore volume, porosity, and initial oil saturation of the sand matrix for the experiments are listed in Table 2. The maximum deviations from average values were less than 1% in mass of sand, pore volume, and porosity and less than 2% in the initial oil saturation. On the basis of liquid volume, the concentration of cracked naphtha injected with steam was 10, 15, and 5% for Experiments 3, 4, and 5, respectively. Injected liquids were pumped using two constant-flow ISCO pumps. During the early times of the experiments, the injected fluid did not reach the sand-pack at saturated steam conditions because the injection rates were as low 0.2 cc/min. As a result, the small volumes of the injected fluid cooled down as they moved toward the heel of the injection well. In spite of the limited variations in the fluid injection pressures, the temperature of the injected fluid was sufficiently high to 5334

DOI: 10.1021/acs.energyfuels.5b02773 Energy Fuels 2016, 30, 5330−5340

Article

Energy & Fuels maintain superheated steam conditions beyond the start-up phase of the process. Pressure transducers were used to record injection and production well pressures. However, the pressure of the production well was monitored through the length of production line that introduced significant error due to the fluctuating pressure drop in the line. The back pressure was set at 2100 kPaa in all cases. Theoretically, in the gravity drainage process, there should be only a small pressure difference between the injection and production wells and in the case of these tests the pressure difference was kept below 20 kPa by adjusting fluid injection rates. This value was selected as an achievable practical target in view of the uncertainty of production well pressure due to pressure drop in the production flow line. Three thermocouples located along each of the horizontal wells were used to monitor temperatures. The production well temperature remained below the saturated steam temperature throughout the tests, which shows that there was no live steam production. The heel and midregion temperatures for both the injection and production wells were close but the toe temperature was relatively lower.

naphtha with steam due to dilution of oil by the solvent. However, since these tests were not scaled for mass transfer, this remains to be verified by other means. It is evident from Figure 12 that the impact of solvent addition is more pronounced at the beginning of the drainage process. In fact, the slopes of cumulative produced oil versus cumulative injected steam are nearly the same after 4000 mL of steam injection. Therefore, it is apparent that solvent coinjection should be initiated early in the process. In addition to faster SAGD start up, coinjecting cracked naphtha at earlier times can be beneficial for the following reasons: • The angle of the SAGD drainage slope becomes flatter as time passes due to the lateral growth of the steam/vapor chamber. This reduces the drainage rate allowing more time for heat transfer and higher average temperature in the draining layer. The higher bitumen temperature reduces the effectiveness of viscosity reduction by the solvent. • The temperature near the production well is lower in the early period and the viscosity reduction by solvent dilution has proportionately larger impact. • Early solvent injection allows more time for the solvent to reduce the residual oil saturation by diluting the oil in the area swept by the steam/vapor chamber. It is evident from Figures 12−14 that increasing the naphtha volume fraction from 10% to 15% leads to no further improvement in the performance. This lack of additional benefits by adding more naphtha can be explained by referring to Figure 1, which shows the simulated phase behavior of cracked naphtha. At the steam temperature and pressure of 219.6 °C and 2300 kPa, pure naphtha will be in liquid state. The dew-point pressure of cracked naphtha at 219.6 °C is 282 kPa. Thus, when the partial pressure of naphtha is greater than 282 kPa, it starts condensing into the liquid phase. It is likely that at 15% naphtha in steam at 2300 kPa, where the partial pressure of naphtha in the mixture is approximately 345 kPa, a part of the added solvent was condensing out of the vapor phase before reaching the edge of the steam chamber where it could be effective in mobilizing bitumen. In comparing the performance of base case SAGD and ES-SAGD experiments, the basis for comparison chosen is that of steam only, thereby ignoring the thermal energy input provided by the solvent. Of course, there is some additional thermal energy carried by the solvent fraction of injected fluid. However, in view of the small volume fraction of solvent needed to make a large change in the performance and the fact that solvent has much smaller heat of vaporization, this is an acceptable approach. Based on energy content analysis,9 the amount of energy carried by solvent in the steam/solvent injected mixture is about 5% of the total when 15% solvent is added to steam. The role of solvent is mainly to affect the drainage rate by dilution of the oil and providing supplemental thermal energy is thought to be a minor effect. Figure 13 shows the temperature profiles for all tests in the physical model at 3 levels of produced oil (1000, 2000, and 3000 g). These temperature profiles show that the addition of cracked naphtha to steam allows similar oil production to be achieved at lower temperature. The steam requirement was lower in all cracked naphtha coinjection cases even with only 5 vol % cracked naphtha addition. Furthermore, in the cracked naphtha addition tests the temperature near the top of the vapor chamber was lower compared to the steam-only case.



RESULTS AND DISCUSSION Extensive analysis of produced fluids and samples from the porous media was conducted to evaluate the effect of cracked naphtha addition to steam on SAGD performance. Experiment 2 was used as a baseline for SAGD because, as discussed earlier, Experiment 1 experienced a leakage of nitrogen from the annulus to the sand-pack. Steam and solvent injection rates were varied throughout the experiments while attempting to maintain a constant operating pressure. Figure 12 shows cumulative solvent-free oil produced

Figure 12. Cumulative produced oil versus cumulative injected steam for the baseline SAGD and cracked naphtha ES-SAGD experiments.

plotted against cumulative steam injected that clearly shows that the cracked naphtha with steam injection produces more oil at reduced amounts of steam injected. Experiment 3 with 10 volume percent of cracked naphtha outperformed all other experiments. In SAGD, during what is normally referred to as the start-up phase, the wells are heated by circulating steam to reduce the viscosity of the bitumen between them until interwell communication is established. It can be noticed from Figure 12 that coinjecting cracked naphtha early in SAGD accelerated the start-up phase even when only 5 vol % cracked naphtha was used. The SAGD wells are normally switched from a circulation mode of operation to a SAGD mode of operation when the viscosity in the interwell region is between 600 and 1200 cP and this may be achieved sooner by coinjecting cracked 5335

DOI: 10.1021/acs.energyfuels.5b02773 Energy Fuels 2016, 30, 5330−5340

Article

Energy & Fuels

solvent additive tests were lower than their steam-only counterpart. The solvent produced in liquid and vapor phases throughout the cracked naphtha ES-SAGD experiments was carefully analyzed. As was found previously,8 solvent recovery in the vapor phase increased during pressure blowdown at the end of the tests due to flashing of the solvent. Blowdown in Experiments 3, 4, and 5 increased recovery by 5.22%, 8.4%, and 3.23%%, respectively. Total solvent recovery for the three experiments was 89.73%, 87.68%, and 72.14%, respectively. Of the total amount of recovered cracked naphtha, the fraction that was produced in the liquid phase with bitumen for Experiments 3, 4, and 5 was 79.17%, 71.05%, and 80.57%, respectively. The fraction of recovered solvent that was collected using the gasometers for Experiments 3, 4, and 5 was 20.83%, 28.95%, and 19.43%, respectively. Limited amounts if any of solvent were produced from the condensate traps during Experiments 3, 4, and 5, and they were mixed with the produced bitumen without being weighed separately. Figure 15

Figure 13. Temperature profiles of baseline SAGD and cracked naphtha ES-SAGD experiments at cumulative produced oil of 1000, 2000, and 3000 g, respectively.

As a result, the contact area between the steam/vapor chamber and the top of the bitumen pay within the sand-pack was bigger in the SAGD case relative to all the cracked naphtha ES-SAGD cases, which resulted in more heat loss, thereby making SAGD less energy efficient. It can be inferred from these observations that coinjecting cracked naphtha with steam can improve the economics by promoting oil drainage at lower steam-oil ratios. Energy efficiency of all cases dropped with time due to increasing contact area between the hot vapor chamber and the annulus continued to increase that in turn increased the heat losses. The role of heat loss to the annulus in the physical model is similar to heat loss to the overburden in the field; both increase as the steam chamber becomes larger. Figures 14 show comparison of cSOR as a function of time and cumulative injected steam. It is evident that the cSOR values for all

Figure 15. Solvent recovery versus time for the cracked naphtha ESSAGD experiments.

shows the variation in solvent recovery for the cracked naphtha ES-SAGD experiments with respect to time. It should be noted that the duration of these experiments was not the same. The amount of water produced during the blowdown phase in Experiments 2, 3, 4, and 5 was approximately 345, 486, 493, and 678 g, respectively. Injected steam and produced water remained in balance throughout the life of all experiments, thereby resulting in minimal fluid losses as illustrated in

Figure 14. Cumulative steam-oil ratio as a function of time and cumulative injected steam for the baseline SAGD and cracked naphtha ES-SAGD experiments. 5336

DOI: 10.1021/acs.energyfuels.5b02773 Energy Fuels 2016, 30, 5330−5340

Article

Energy & Fuels

Figure 16. Cumulative injected steam and cumulative produced water for the baseline SAGD and cracked naphtha ES-SAGD experiments.

Figure 17. Residual oil saturation within the depleted sand matrix for the baseline SAGD and cracked naphtha ES-SAGD experiments.

the residual oil saturation measurement, there was no clear-cut effect of solvent volume fraction on the residual saturation in the well-drained zone. Variations in residual oil saturation are thought to be related to the extent of the steam/vapor chamber growth since residual oil saturation was lower in regions that were effectively swept by the steam/vapor chamber and much higher at the lower corners of the sand matrix. Figure 18 shows that the water saturation profiles have significant asymmetry from left to right. This most likely is an artifact caused by the process of removing the model from the pressure vessel using a

Figure 16. The minor difference between cumulative injected steam and cumulative produced water resulted from the water retained in the sand matrix along with the water needed to pressurize the sand-pack model which was around 290 g for all tests. Fluids were extracted from each of the 15 sand-pack samples taken in order to measure water and residual oil saturation as well as asphaltene content. Fluid saturation contour maps are shown in Figures 17 and 18, respectively. Data used to generate these figures were published previously.1 Within the accuracy of 5337

DOI: 10.1021/acs.energyfuels.5b02773 Energy Fuels 2016, 30, 5330−5340

Article

Energy & Fuels

Figure 18. Water saturation within the depleted sand matrix for the baseline SAGD and cracked naphtha ES-SAGD experiments.

Figure 19. Asphaltene content within the depleted sand matrix for the baseline SAGD and cracked naphtha ES-SAGD experiments.

respectively. Figure 19 shows that asphaltene content was somewhat higher near the injection well, probably due to the relatively higher temperatures in this region, which could have resulted in the evaporation of light ends from the residual oil and possibly due to the initial exposure to solvent that occurs at low temperatures. Laboratory analysis of oil extracted from different parts of the model showed only small differences in asphaltene content resulting from cracked naphtha coinjection

crane that hangs the model on its side. Apparently, there is some redistribution of water in the depleted zone during this operation. Asphaltene content concentration maps within the depleted sand matrix were generated for each test as shown in Figure 19. The data used to make these contour maps may be found in the thesis by Al-Murayri.1 Original oil asphaltene content measured for Experiments 3, 4, and 5 was 22.5%, 22.0%, and 22.5%, 5338

DOI: 10.1021/acs.energyfuels.5b02773 Energy Fuels 2016, 30, 5330−5340

Article

Energy & Fuels

manner to that of the density. Both density and viscosity of produced oil from Experiment 4 were lower than those of the other experiments because of the higher amount of coinjected cracked naphtha. Although it is apparent that ES-SAGD can substantially reduce the steam requirements for in situ recovery of bitumen, its application in the field requires careful consideration of the cost versus benefits of adding the solvents. The solvent is more expensive than the produced bitumen and the economic performance of ES-SAGD is very sensitive to the fraction of injected solvent recovered with the produced fluids. In the physical model experiments reported here a closed system has been used so the injected solvent either stays in the model or is produced with the other fluids. In other words, in the model, it cannot migrate away from the intended location. This is not the case in the field, where the solvent can indeed migrate outside the target pattern if mobile water is present. Minimizing the loss of solvent by migration to outside of the target pattern area remains a challenge that may require modifications to the solvent injection strategy.

with steam. The maximum fraction of asphaltene within the depleted sand matrix for Experiments 3, 4, and 5 was 24.6%, 24.5%, and 23.5%, respectively. This level of asphaltene is only marginally higher than the original oil and is within the range of uncertainty of these measurements. Coinjection of cracked naphtha with steam resulted in reduced density and viscosity of oil produced. These changes in oil properties facilitated the dehydration of the produced emulsion samples from the steam/solvent coinjection experiments. Maximum difference in asphaltene fraction observed between original and produced oil for Experiments 2, 3, 4, and 5 was determined to be 0.9%, 3.0%, 3.5%, and 3.5%, respectively. Figure 20 presents a plot of asphaltene fraction in



CONCLUSIONS The conclusions drawn from this study may be summarized as follows: • ES-SAGD using cracked naphtha results increases oil production rates and lowers steam−oil ratios relative to SAGD. • Incremental oil recovery in ES-SAGD is expected to be a natural consequence of reduced SOR since it allows SAGD to run economically for longer periods of time. • Cracked naphtha coinjection with steam is able to accelerate the startup phase of the SAGD process by enhancing oil viscosity reduction in the interwell region. • The benefits of cracked naphtha coinjection with steam become less pronounced as the steam/vapor chamber matures. • Cracked naphtha coinjection with steam facilitates the separation of water from oil in the produced emulsions • In the concentrations used, cracked naphtha coinjection with steam does not cause significant levels of asphaltene precipitation in the sand.

Figure 20. Asphaltene content in produced oil versus time for Experiments 2, 3, 4, and 5.

produced oil versus time for Experiments 2, 3, 4, and 5. Some of the reduction in asphaltene content that is evident is attributable to the presence of solvent in the produced oil. High levels of asphaltene precipitation were not observed and this is explained by the composition of the cracked naphtha that includes aromatic and relatively heavy molecular weight solvents. If lighter hydrocarbon solvents such as propane or butane are used, it is expected that asphaltene precipitation would be higher. Figure 21 shoes the density of produced oil following removal of water plotted against time. It is evident that there was very little change in the oil density throughout Experiment 2; however, for all of the cracked naphtha ES-SAGD experiments, oil density generally decreased as the experiments proceeded. The produced oil viscosity fluctuated in a similar



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected].

Figure 21. Density and viscosity of produced oil from the baseline SAGD and cracked naphtha ES-SAGD experiments. 5339

DOI: 10.1021/acs.energyfuels.5b02773 Energy Fuels 2016, 30, 5330−5340

Article

Energy & Fuels Notes

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS Support of this work by Nexen Energy ULC is gratefully acknowledged. REFERENCES

(1) Al-Murayri, M. T. Experimental Investigation of ES-SAGD Using Multicomponent Solvents. Ph.D. Dissertation, University of Calgary, December 2012. (2) Butler, R. M.; Mokrys, I. J. A New Process (VAPEX) for Recovering Heavy Oils Using Hot Water and Hydrocarbon Vapour. J. Can. Pet. Technol. 1991, 30 (1), 97−106. (3) Nasr, T. N.; Isaacs, E. E. Process for Enhancing Hydrocarbon Mobility Using a Steam Additive. US Patent No. 6,230,814, 2001. (4) Hosseininejad Mohebati, M.; Maini, B. B.; Harding, T. G. Experimental Investigation of the Effect of Hexane on SAGD Performance at Different Operating Pressures. SPE Heavy Oil Conference, Calgary, Alberta, Canada, June 12−14, 2012; SPE 158498. (5) Butler, R. M.; McNab, G. S.; Lo, H. Y. Theoretical Studies on the Gravity Drainage of Heavy Oil During In-Situ Steam Heating. Can. J. Chem. Eng. 1981, 59 (4), 455. (6) Oskouei, S. J. P.; Maini, B.; Moore, R. G.; Mehta, S. A. Effect of Initial Water Saturation on the Thermal Efficiency of the SteamAssisted Gravity-Drainage Process. J. Can. Pet. Technol. 2012, 51 (5), 351−361. (7) Butler, R. M. Thermal Recovery of Oil and Bitumen; GravDrain Inc.: Calgary, Alberta, 1998; pp 297−299. (8) Al-Murayri, M. T.; Maini, B. B.; Harding, T. G.; Oskouei, J. Multicomponent Solvent Co-injection with Steam in Heavy and ExtraHeavy Oil Reservoirs. Energy Fuels 2016, 30, 13. (9) Hassanzadeh, H.; Harding, T. Analysis of Conductive Heat Transfer During In-situ Electrical Heating of Oil Sands. Fuel 2016, 178, 290−299.

5340

DOI: 10.1021/acs.energyfuels.5b02773 Energy Fuels 2016, 30, 5330−5340