On hot water flooding strategies for thin heavy oil reservoirs

Fuel 153 (2015) 559–568

Contents lists available at ScienceDirect

Fuel journal homepage: www.elsevier.com/locate/fuel

On hot water flooding strategies for thin heavy oil reservoirs David W. Zhao, Ian D. Gates ⇑ Department of Chemical and Petroleum Engineering, Schulich School of Engineering, University of Calgary, Canada

h i g h l i g h t s  In Western Canada, 80% of heavy oil resources are in reservoirs <6 m thick.  Cold production has low recovery factor, <10%, for thin heavy oil reservoirs.  Recovery strategy of thermal processes is unclear in thin heavy oil reservoirs.  Optimization yields variable injection pressure/temperature hot water flood.  Permeability distribution controls energy-to-oil ratio and economic performance.

a r t i c l e

i n f o

Article history: Received 31 January 2015 Received in revised form 10 March 2015 Accepted 11 March 2015 Available online 21 March 2015 Keywords: Heavy oil Hot water flood Thin reservoirs Optimization Thermal efficiency Recovery process design

a b s t r a c t Cold production methods for heavy oil resources in Western Canada yield recovery factors averaging about 10% and as yet, there are no commercially successful technologies to produce oil from these reservoirs with recovery factor greater than 20%. This means that the majority of oil remains in the reservoir. The objective of this study is to determine technically and economically feasible recovery processes for thin heavy oil reservoirs by using a simulated annealing algorithm. The results reveal that high injection pressure is critical to a successful hot water flooding strategy. Also, they show from a thermal efficiency point of view that it is most efficient to adopt an injection temperature profile where the injection temperature starts high earlier in the process and ends at lower water temperature. The lower temperature injection at later stages of the recovery process partially recovers the heat stored in the reservoir matrix and therefore increases the overall heat utilization efficiency. A sensitivity analysis shows that the permeability distribution affects the performance of the hot water flooding process most significantly. The existence of a higher permeability zone in the lower part of the reservoir leads to earlier oil production and water breakthrough. High permeability was found to lead to more oil and water production in the early stage of operation and achieved the best economic performance. The low permeability case exhibited relatively low oil production volume. Although it has the lowest cumulative injected energy to oil produced ratio, poor oil production renders the operation process uneconomic. Given the volume of currently inaccessible thin heavy oil resources, the optimized strategies developed here provide important guidelines to convert these resources to producible reserves. Ó 2015 Elsevier Ltd. All rights reserved.

1. Introduction The majority of heavy oil resources, roughly 1.3 trillion barrels of oil, in the Western Canada Sedimentary Basin are found in thin reservoirs with thickness less than 6 m [1]. Due to heat losses to the overburden or understrata or both, current commercial steam-based techniques such as Steam-Assisted Gravity Drainage (SAGD) and Cyclic Steam Stimulation (CSS) are not economically feasible in thin heavy oil reservoirs (<6 m). In these processes, in thin reservoirs, the amount of steam invested in the reservoir ⇑ Corresponding author. Tel.: +1 (403) 220 5752; fax: +1 (403) 284 4852. E-mail address: [email protected] (I.D. Gates). http://dx.doi.org/10.1016/j.fuel.2015.03.024 0016-2361/Ó 2015 Elsevier Ltd. All rights reserved.

versus the oil revenues renders the processes uneconomic. In cold production (CP) processes, the only energy input is that of the pump to move the produced fluids from the reservoir to the surface; thus their energy investment is relatively small. However, the average recovery factors of cold production processes are low being equal to about 10% [1]. By encouraging sand production along with oil recovery, the Cold Heavy Oil Production with Sand (CHOPS) technique can recover as much as 15% of the OOIP [2]. In CHOPS operations, sand production creates an extensive connected wormhole network in the reservoir with zones adjacent to the network depleted of reservoir pressure [3]. In Western Canada, after primary production, in most cases, water flooding and polymer flooding have been the most widely

560

D.W. Zhao, I.D. Gates / Fuel 153 (2015) 559–568

used techniques to raise the overall recovery factor of the reservoir [4,5]. In heavy oil reservoirs, due to the high viscosity of the oil versus that of the water, flooding processes may suffer with respect to water bypassing [4–6]. In most cases, the viscosity of the live oil ranges from 1000 to 10,000 times that of water which implies water fingering occurs. Despite this, water flooding has been actively applied in Saskatchewan and Alberta since it is technically simple to implement and has relatively low operating cost even though incremental oil recovery factors are not significantly larger than primary production. Solvent-aided thermal recovery methods have also been proposed for bitumen and heavy oil reservoirs. For example, Gates [7] examined a solvent-aided thermal recovery process for thin oil sands reservoirs by using optimization. The optimized process had lower net energy (both steam and solvent retained in the reservoir) to oil ratios compared to traditional SAGD. Solvent-only processes, such as cyclic solvent injection, have advantages in that there are no heat losses to the surrounding overburden and understrata. These methods appear to have promise for use in postCHOPS reservoirs [8,9]. Hot water flooding is a relatively low cost thermal oil recovery technique [9] since it only involves sensible heat. Compared with conventional water flooding, the use of hot water improves the mobility ratio due to a reduction of the oil phase viscosity arising from it being heated. Furthermore, heating also reduces the interfacial tension and residual oil saturation which both lead to potentially higher recovery factor. However, in hot water flooding, the heated water for injection delivers less heat to the reservoir compared to that with steam due to absence of latent heat and therefore it is less effective in reducing oil viscosity. On the other hand, for thin heavy oil reservoirs, hot water flooding has advantages over steam flooding. First, it provides larger displacement drive than steam flooding since water viscosity is much larger than that of steam [9–11]. Second, it permits the use of much higher injection pressure than steam flooding at a given temperature. Furthermore, higher-pressure injection enables greater temperatures while remaining in the hot water state. Third, due to smaller reservoir temperature, heat losses to the overburden and understrata will be substantially smaller than that encountered in steam flooding. However, less heat losses to the overburden and understrata will mean less heat delivery to the heavy oil interval. Martin et al. [10] describe the results of hot water injection into a 5–7 m thick sandstone reservoir containing oil with viscosity equal to 600 cP. They found that water injectivity and oil rates were significantly enhanced over that of cold water flooding. However, although they did not have detailed thermocouple observation wells, they concluded that 60 percent of the injected heat was lost to the overburden and understrata. Thus, there is a need to design hot water recovery processes for thin reservoirs that manage heat delivery and recovery to and within the reservoir. In the study documented here, hot water-flooding strategies are optimized by using simulated annealing, a stochastic optimization algorithm. We aimed to understand the effects of injection pressure, water temperature, as well as different reservoir conditions on the recovery process performance.

2. Models and methods 2.1. Reservoir simulation model The reservoir evaluated here has properties typical of that of a typical thin heavy oil reservoir in the Lloydminster area of Alberta, Canada described in a previous study [12]. The base case reservoir model is two-dimensional with two horizontal wells spaced 50 m apart. The thickness of the heavy oil interval is equal

to 4 m thick. The models were discretized into a regular Cartesian grid, displayed in Fig. 1, with dimensions 1 m in the cross-well direction, 1000 m in the down-well direction (into the page) and 0.4 m in the vertical direction. The length of the perforated sections of the horizontal wells in all models is equal to 1000 m. A commercial thermal reservoir simulator (CMG STARS™) was used. The commercial thermal reservoir simulator uses the finite volume approach. At the top and bottom boundaries, heat losses were permitted and were approximated by using Vinsome and Westerveld’s [14] heat loss model. At the side boundaries of the model, no flow and no heat transfer boundary conditions were applied. The reservoir simulation model and fluid properties are listed in Table 1. The relative permeability curves, listed in Table 1, are independent of temperature. The spatial distributions of oil/water saturations (average oil saturation equal to 0.65), porosity (average equal to 0.32), and base case horizontal permeability (average equal to 3650 mD) are, displayed in Fig. 1(a)–(c), respectively. The average oil saturation, porosity, and horizontal permeabilities were derived from core data taken from one of Devon Canada’s heavy oil fields located in eastern Alberta. The spatial distributions of the porosity, oil saturation and base case permeability (described below) were randomly assigned using uniform probability distributions. Given that the sand is relatively clean, the vertical-to-horizontal permeability ratio is set equal to 0.8. The initial reservoir pressure and temperature are equal to 2800 kPa and 20 °C, respectively. The solution gas-to-oil ratio at original reservoir conditions is equal to 6.17 m3/m3. To investigate the effect of permeability and its variations on the reservoir performance, five permeability cases were optimized (including the base case). These cases were chosen to span the range of reservoir characteristics that are typical in thin heavy oil reservoirs in Western Canada. Case 1: This is the base case reservoir model with permeability distribution as shown in Fig. 1(c). The average permeability is equal to 3650 mD. This case represents the expected permeability case in the study conducted here. Case 2: In this case, a permeability distribution is created with the same average permeability of Case 1 (3650 mD) but enhanced permeability at the bottom and lower permeability at the upper zone, as shown in Fig. 1(d). This vertical permeability profile would be expected in a reservoir where the sand grains were larger in size at the base of the reservoir with the finest grains at the top of the reservoir. Case 3: In this case, a permeability distribution is created with same average permeability of Cases 1 and 2, but with higher permeability at the upper zone and lower permeability at the lower part of the reservoir, as displayed in Fig. 1(e). The vertical permeability distribution of this case would be expected where the sand grains are largest at the top of the reservoir and finest at the base of the oil column. Case 4: The permeability distribution for this case, shown in Fig. 1(f), is created by scaling up the permeabilities of the gridblocks of Case 1 universally by a factor of 2. This gives rise to an average permeability of 7300 mD. This case represents the best permeability case examined here and is at the upper limit of permeabilities expect in thin heavy oil reservoirs in Western Canada. Case 5: The permeability distribution of this case, displayed in Fig. 1(g), is created by scaling down the permeabilities of the gridblocks of Case 1 universally by a factor equal to 0.6. This gives rise to an average permeability equal to 2190 mD. This case represents the worst permeability case evaluated in this study. For each of above reservoir model cases, an individual optimization of 800 runs was conducted to determine the optimum parameter set for each case. The optimization run and simulations were executed on a personal computer (3.4 GHz, dual quad core

D.W. Zhao, I.D. Gates / Fuel 153 (2015) 559–568

561

(a) Oil Saturation distribution (average = 0.65)

(b) Porosity distribution (average = 0.32)

(c) Base case, Case 1: horizontal permeability (mD) distribution (average = 3,650 mD)

(d) Case 2: horizontal permeability (mD) distribution – enhanced permeability at bottom and reduced permeability at top (with same overall average permeability as base case)

(e) Case 3: horizontal permeability (mD) distribution – reduced permeability at bottom and enhanced permeability at top (with same overall average permeability as base case)

(f) Case 4: horizontal permeability (mD) distribution – two times the base case permeability

(g) Case 5: horizontal permeability (mD) distribution – 0.6 times the base case permeability

Fig. 1. Distributions of the oil saturation, porosity, and horizontal permeability, scale in (c), of the reservoir models. The injection well is on the left side of the domain whereas the production well is on the right side of domain. The spacing between the wells is equal to 50 m. The dimensions of the grid blocks are equal to 0.4 m and 1 m in the vertical and horizontal directions, respectively.

with 16 GB memory). Each individual reservoir simulation took on average 2 min and 30 s to execute; given that 800 simulation runs were done each case, each optimization run took roughly 34 h to complete. 2.2. Optimization algorithm 2.2.1. The simulated annealing method In this work, a Simulated Annealing (SA) algorithm is used for operating strategy optimization as described in Gates and Chakrabarty [15]. The optimization algorithm is designed to control the thermal reservoir simulator and execute reservoir performance evaluations. Parameters for reservoir simulation are generated by the SA algorithm and then used for generating the simulation input file. Then a simulation run based on the newly generated input file is executed by the reservoir simulator. Once the simulation is complete, a computer code is called to process the reservoir simulation output data and evaluate the performance of the simulated strategy. The evaluation results are then sent back

to the optimizer to generate new parameter sets and the next iteration of the optimization algorithm starts. In the optimization procedure, the SA algorithm conducts random searches that attempt to lower the value of the cost function, i.e., the optimum value of desired reservoir operating performance. The parameters of the SA algorithm were the same as those used in previous studies [15].

2.2.2. Adjustable parameters and cost function For optimization, the adjustable parameters are the injection pressures and injection water temperature over specified time intervals, summarized in Table 2. The pressure and temperature sampled during the optimization run ensures that none of the pressure/temperature combinations are below the steam saturation line. In other words, conditions are maintained such that only subcooled water is injected into the reservoir. In total, ten pressure parameters with base value of 3000 kPa and optimization range set equal to 2000–4200 kPa, and ten water temperature

562

D.W. Zhao, I.D. Gates / Fuel 153 (2015) 559–568

Table 1 Reservoir simulation model and fluid properties.

Table 2 List of optimization parameters.

Property

Value

Depth to reservoir top (m) Net pay (m) Porosity (dimensionless) Oil saturation (dimensionless) Solution gas-to-oil ratio (m3/m3) Horizontal permeability kh (mD) kv/kh (dimensionless) Effective rock compressibility (1/kPa) Rock heat capacity (kJ/m °C) Rock thermal conductivity (kJ/m day °C) Reference pressure (kPa) Reference depth (m) Initial reservoir temperature (°C) Dead oil viscosity (cP) 20 °C 40 °C 80 °C 160 °C 250 °C Water thermal conductivity (kJ/m day °C) Gas thermal conductivity (kJ/m day °C) Oil thermal conductivity (kJ/m day °C) Effective molecular diffusion coefficient of oil (m2/day) Effective molecular diffusion coefficient of solvent (m2/day) Methane K-value correlation in oil [13] Kv1 (kPa) K-value = (Kv1/P) exp(kv4/(T + Kv5)) Kv4 (°C) Kv5 (°C)

334 4 0.32 ± 0.02 0.65 ± 0.09 6.17 3650 ± 347 0.8 14  10 6 2600 660 2800 334 20 15,212 1884 125.4 9.66 3.09 53.5 5 11.5 4.32  10 6 4.32  10 5

Oil–water relative permeability curves

Gas–liquid relative permeability curves

Sw 0.15 0.2000 0.2500 0.3000 0.3500 0.4000 0.4500 0.5000 0.5500 0.6000 0.6500 0.7000 0.7500 0.8000 0.8500 0.9000 0.9500 1.0000 Sl 0.1500 0.2000 0.2500 0.3000 0.3500 0.4000 0.4500 0.5000 0.5500 0.6000 0.6500 0.7000 0.7500 0.8000 0.8500 0.9000 0.9500 1.0000

krw 0.0000 0.0002 0.0016 0.0055 0.0130 0.0254 0.0440 0.0698 0.1040 0.1480 0.2040 0.2710 0.3520 0.4470 0.5590 0.6870 0.8340 1.0000 krg 1.0000 0.9500 0.8400 0.7200 0.6000 0.4700 0.3500 0.2400 0.1650 0.0930 0.0750 0.0450 0.0270 0.0200 0.0100 0.0050 0.0000 0.0000

Parameter 1

Injection well pressure

0

2

Injection well pressure

7

3

Injection well pressure

13

4

Injection well pressure

19

5

Injection well pressure

25

6

Injection well pressure

31

7

Injection well pressure

37

8

Injection well pressure

43

9

Injection well pressure

49

10

Injection well pressure

59

11

Injection water temperature Injection water temperature Injection water temperature Injection water temperature Injection water temperature Injection water temperature Injection water temperature Injection water temperature Injection water temperature Injection water temperature

12 504,547 879.84 265.99 krow 0.9920 0.9790 0.9500 0.7200 0.6000 0.4700 0.3500 0.2400 0.1650 0.1100 0.0700 0.0400 0.0150 0.0000 0.0000 0.0000 0.0000 0.0000 krog 0.0000 0.0002 0.0016 0.0055 0.0130 0.0254 0.0440 0.0698 0.1040 0.1480 0.2040 0.2710 0.3520 0.4470 0.5590 0.6870 0.8340 0.9920

parameters with base value equal to 120 °C and range 20–250 °C are used to optimize the process. The cost function against which the adjustable parameters are optimized is a function of the net present value (NPV). For the simple economic model used here, the following economic factors are considered: initial capital investment (including well drilling and field equipment), operating costs, fixed costs, variable costs, water

Onset time (months)

13 14 15 16 17 18 19 20

Base value, allowed range

0

3000 kPa, 2000– 4200 kPa 3000 kPa, 2000– 4200 kPa 3000 kPa, 2000– 4200 kPa 3000 kPa, 2000– 4200 kPa 3000 kPa, 2000– 4200 kPa 3000 kPa, 2000– 4200 kPa 3000 kPa, 2000– 4200 kPa 3000 kPa, 2000– 4200 kPa 3000 kPa, 2000– 4200 kPa 3000 kPa, 2000– 4200 kPa 20–250 °C

7

20–250 °C

13

20–250 °C

19

20–250 °C

25

20–250 °C

31

20–250 °C

37

20–250 °C

43

20–250 °C

49

20–250 °C

59

20–250 °C

treatment costs, and operating revenue. The following assumptions formed the basis of our evaluation: well drilling cost and other initial investment $2,500,000 (for a single well), discount rate of 10%, variable cost to be 10% of the operating revenue, heavy oil price $80.00/bbl [16], natural gas price $4.4/GJ, thermal efficiency equal to 0.75, and waste water treatment cost is $2.00/m3. The cost function (CF) is formally defined as CF = (6  106 NPV)/1  106. This indexes the value of the CF to range, in general, between 0 and 10 with lower values of the CF being more optimal. 3. Results and discussion 3.1. Injection pressure and water temperature Fig. 2 shows the optimized injection pressure and water temperature for all the optimized cases. For Case 1, the results reveal that the injection pressure remains relatively high, around 4000 kPa, throughout the majority of the operating life of the process although a lower injection pressure (2500 kPa) period exists between 1.5 and 2 years of operation. The optimized injection pressure for all the other cases generally remains high in the majority of the operating time before water breakthrough although exhibit stochastic deviations. In Case 4, the high permeability zone leads to earlier oil production compared to the other cases. The injecting pressure remains high over the first two years and shows a cyclic pattern in the later stages of operation. In Case 1, the initiate water temperature is found to be around 120 °C and then jump

D.W. Zhao, I.D. Gates / Fuel 153 (2015) 559–568

Case 1

Case 2

Case 3

Case 4

563

Case 5

Fig. 2. Comparison of injection pressure and injection water temperature profiles of the optimized strategies of Cases 1, 2, 3, 4, and 5.

to 225 °C for a period of 6 months. After this high water temperature period follows a low injecting temperature period of 1.5 years with water temperature ranging from 20 to 50 °C. The water temperature increases to 175 °C and is then further elevated to 250 °C after 3 years of operation. The 250 °C injection period persists for a year before the temperature decreases to 94 °C and then finally to 20 °C for the last 14 months of operation. From Fig. 2, one can see that there is similar pattern for the optimized injecting water temperature. The water temperature normally starts high and then gives rise to a low injection temperature period. We could call this

temperature change from high to low an injection cycle. In the 5 cases investigated here, the second cycle tends to last longer than the first cycle. In Case 5, a third cycle occurs within the six year operation life. We suggest that low injection temperature enables heat recovery from the reservoir matrix during the process which results in higher heat efficiency. During the initial period where the temperature of the injected water is relatively high, relatively high heat is injected into the reservoir and due to heat losses to the solid matrix, this results in an elevated matrix temperature. Only a small

564

D.W. Zhao, I.D. Gates / Fuel 153 (2015) 559–568

Fig. 3. Comparison of oil production rates of the optimized strategies of Cases 1, 2, 3, 4, and 5.

overall thermal efficiency and heat utilization of the recovery process. Based on the results of the optimization runs, it is suggested that a high injection pressure is critical to obtain feasible hot water flooding strategies. Essentially, high injection pressure promotes rapid fluid movement within the oil reservoir which enhances convective delivery of heat to the formation leading to a greater fraction of the heat being delivered to the oil than would be the case for low-pressure injection and low injection rate where conductive losses to the overburden and understrata would dominate heat transfer. Similar to the results for optimized SAGD operation as shown by Gates et al. [17], the optimized process promotes horizontal heat transfer over that of vertical heat transfer. In the context of hot water flooding, this is done within the constraint of hot water breakthrough to reduce direct hot water production from the reservoir. For hot water injection, the results demonstrate that it is most thermally efficient to adopt a cyclic pattern control, i.e. start at high water temperature and end at low water temperature. Multiple cycles might be beneficial depending on the reservoir condition. 3.2. Oil production rates and effects of permeability variations

Table 3 Comparison of optimized operating strategies in all the four cases in terms of cumulative oil production, cumulative water produced to oil produced ratio (cWOR), cumulative energy injected to oil ratio (cEOR), operating time and net present value (NPV). Case

Cumulative oil production (m3)

cWOR (m3/ m3)

cEOR (GJ/ m3)

1 2 3 4 5

24,366 26,400 25,655 27,319 5396

14.5 14.6 13.5 19.1 13.7

6.2 9.9 8.2 7.4 3.4

NPV* ($million) 2.8 2.9 2.9 4.7 1.6

* The blowdown performance is not considered in the NPV calculation which means the real NPV could be slightly higher than the presented values.

fraction of the injected heat is produced with the produced fluids. Due to the small thickness of the reservoir pay zone, a significant fraction of the heat is lost to the overburden and understrata. After the hot water injection period, subsequent water injection at lower temperature enables heat recovery from the reservoir matrix, that is, heat is transferred from reservoir rock to water and mobile oil. Furthermore, since the injection temperature is lower than that of the overburden and understrata, heat recovery also occurs from these zones to the reservoir thus improving the

Fig. 3 shows the oil production rates for Cases 1–5. The peak oil production rates are found to range from 20 to 25 m3/day for Cases 1–4. In Case 5, the maximum oil rate seldom exceeds 5 m3/day. The results show that despite the same average permeability value, the distribution of the permeability within the pay zone impacts oil production. In Case 2, a higher permeability zone is located at the bottom zone of the reservoir. This results in earlier oil production than that of Case 3, the case where a higher permeability zone is located at the upper part of the reservoir. The higher permeability at the lower part of the reservoir causes faster hot water frontal advance in the lower part of the reservoir. This enhances heat transfer (tends to migrate upwards rather than downwards) to the oil above the higher permeability zone at the base of the reservoir. Furthermore, the accelerated water front speed leads to more oil displacement and production. As listed in Table 3, within the same operating time of 6 years, Case 2 produced 3% more oil than Case 3. On the other hand, 11% more water is produced in the optimized Case 2, which is caused by the higher permeability of the lower region of pay zone. Case 3 produces 5% more oil than Case 1 but used 40% more heat injection over the total 6 years of operation. The higher permeability interval at the upper part of reservoir contributes to larger heat losses to the overburden.

(a) after 12 months

(b) after 36 months

(c) after 60 months

Fig. 4. Oil saturation profiles of optimized Case 1.

D.W. Zhao, I.D. Gates / Fuel 153 (2015) 559–568

565

Case 1

Case 2

Case 3

Case 4

Case 5

Fig. 5. Oil saturation distributions after 4 years of operation for Cases 1, 2, 3, 4, and 5.

In addition to the effects of the spatial permeability distribution, the absolute average permeability value also impacts oil production. As shown in Fig. 3, the highest permeability case, Case 4, results in the highest oil production of all cases in the shortest time. On the other hand, the lowest permeability case, Case 5, has the lowest cumulative oil production of all cases, only 5396 m3 versus 24,366 m3 for Case 1. It should be pointed out that higher permeability also leads to higher water injection and consequent production. Fig. 4 shows the oil saturation distributions after 12, 36, and 60 months of operations for the optimized Case 1. The conformance zone created by hot water flooding is relatively high due to the thinness of the pay zone. The water front advances faster in the lower part of the reservoir with evidence of water fingering. Fig. 5 shows the oil saturation distributions of all optimized cases after 4 years of operation. In Case 2, as shown in Fig. 5b, due to higher permeability at the lower part of the reservoir, the water front moves much faster in the lower part and breaks through at an early time which lead to overall higher water cut. In Case 3, as shown in Fig. 5c, the advance of the water front is relatively uniform in the pay zone. In Case 4, the high permeability is found to result in lowest oil saturation after 4 years of operation (Fig. 5d). However, in Case 5 (Fig. 5e), due to the low permeability, the water front moves at a relatively slow pace which resulted in the lowest oil production. 3.3. Water injection rates and water production The water injection rates in all the optimized cases are shown in Fig. 6. For Cases 1–3, the initial water injection rates are generally low in the early stages of oil production but ramp up as the operation continues. Since the injection temperature drops as the operations progress, at the later stage of hot water flooding, water breakthrough does not cause substantial heat losses since lower

Fig. 6. Water injection rates of the optimized strategies of Cases 1, 2, 3, 4, and 5.

temperature water is injected. In Case 5, due to the low injectivity determined by the low permeability, the water injection rates are low and thus the oil production rate is relatively low. Fig. 7 shows the water cut of all of the cases studied here. The water cuts are generally larger than 80%. At the later stages, water cuts rise to above 95%. In Case 4 where reservoir has the largest permeability, the water cut rises to 99% by the end of the 6 years of operation. 3.4. Temperature distributions, cumulative energy injected to oil ratio (cEOR), and net present value Fig. 8 presents the spatial distributions of the temperature after 12, 36, and 60 months of operation in the optimized Case 1. Figs. 9–

566

D.W. Zhao, I.D. Gates / Fuel 153 (2015) 559–568

Fig. 7. Water cut of the optimized strategies of Cases 1, 2, 3, 4, and 5.

12 present the temperature distributions after 12, 36, and 60 months in optimized Cases 2–5. In Case 1, the reservoir temperature peaks at about 100 °C by the end of high temperature water injection period (at the end of 4 years of operation). Due to the use of cold water for injection, the temperature of the flooded zone starts to decrease and declines to about 50 °C. In Cases 2–4, the

maximum reservoir temperatures during hot waterflooding were found to be in the range between 107 and 120 °C whereas the final temperature of the flooded zone was between 50 and 75 °C. In Case 5, due to low permeability and therefore low injectivity, the average reservoir temperature never exceeded 30 °C. In Cases 1–4, the overall reservoir temperature profile versus time reflects heat recovery from reservoir matrix sequestered there during hot water injection and recovered during colder water injection. The cumulative energy injected (as sensible heat in the injected water) to produced oil ratio (cEOR, expressed as GJ injected energy per m3 of oil produced) versus time for all the cases is displayed in Fig. 14 with results at the end of the six years of operation listed in Table 3. The cEOR generally starts high due to heat losses and initial low oil production rate. As the oil rate increases, the cEOR decreases. By the end of the high oil production rate period, the cEOR increases until cold-water injection is started which then recovers heat previously stored in the reservoir matrix. In Case 1, the resulting cEOR is equal to 6.2 GJ/m3, being the lowest value excluding Case 5. In Case 2, the existence of the high permeability layer in the lower part of reservoir results in relatively early water break through and therefore greater energy injection, more heat losses to overburden, a higher overall reservoir temperature (Fig. 13), and the highest cEOR equal to 9.9 GJ/m3. Case 5 achieved the lowest cEOR but also resulted in the lowest production rate and recovered oil volume and therefore had a negative net present

(a) After 12 months of operation

(b) After 36 months of operation

(c) After 60 months of operation

Fig. 8. Temperature (°C) distributions of optimized Case 1.

(a) After 12 months of operation

(b) After 36 months of operation

(c) After 60 months of operation

Fig. 9. Temperature (°C) distributions of optimized Case 2.

D.W. Zhao, I.D. Gates / Fuel 153 (2015) 559–568

567

(a) After 12 months of operation

(b) After 36 months of operation

(c) After 60 months of operation

Fig. 10. Temperature (°C) distributions of optimized Case 3.

(a) After 12 months of operation

(b) After 36 months of operation

(c) After 60 months of operation

Fig. 11. Temperature (°C) distributions of optimized Case 4.

(a) After 12 months of operation

(b) After 36 months of operation

(c) After 60 months of operation

Fig. 12. Temperature (°C) distributions of optimized Case 5.

value (NPV). This result suggests that heat losses were reduced in the low permeability case but oil production suffers resulting in an uneconomic process. Of the five cases studied, the resulting overall cEOR after six years of operation is under 10 GJ/m3, which indicates

relatively good heat utilization efficiency. The calculated reveals that hot water flooding, with the economic inputs here, can be economic in thin (<6 m) heavy oil reservoirs the base case (Case 1) properties and that high and

NPV used with low

568

D.W. Zhao, I.D. Gates / Fuel 153 (2015) 559–568

Fig. 13. Average reservoir temperature as function of operating time in optimized Cases 1, 2, 3, 4, and 5.

efficiency of the process. Multiple cycles of high/low temperature water injection might be beneficial depending on the reservoir condition.  The permeability distribution is found to affect the performance of the hot water flooding process. The existence of higher permeability zone at the lower part of the reservoir leads to earlier oil production and water breakthrough. The higher injectivity and water production also caused higher cEOR. The performance of Case 3, which has higher permeability zone at upper part of the reservoir, is comparable to that of the Case 1 but it used 40% more heat injection.  The absolute overall permeability of the reservoir impacts performance significantly. Case 4 produced the largest amount of oil and water in the early stage of operation. Although Case 4’s produced water-to-oil is also substantially higher than the other cases, it achieved the best economic performance. The low permeability of Case 5 led to slow oil production. Although it has the lowest cEOR, the poor oil production made the operation process uneconomic.

Acknowledgements Acknowledgement is extended to the Petroleum Technology Research Centre (PTRC) for their financial support and the University of Calgary for providing financial and logistical support as well as Computer Modelling Group for the use of its thermal reservoir simulator, STARSTM. References

Fig. 14. Cumulative energy injected to oil ratio (cEOR) of optimized Cases 1, 2, 3, 4, and 5.

permeability zones at the top or bottom of the reservoir realize similar NPV providing the overall permeability is similar. The results show that Case 4 achieved the best economic outcome of the cases studied here – this is a result of its enhanced permeability. 4. Conclusions In the present work, stochastic optimization was conducted to determine the optimum injecting pressure and injecting water temperature strategies in thin heavy oil reservoir in five cases. The key results are as follows.  A high injecting pressure is critical to a success hot water flooding strategy. In the present optimized cases, the injection pressures remain high during the operating process although deviations present. This promotes larger horizontal heat transfer (convective) than vertical heat losses (vertical losses adversely impact process performance due to heat losses to non-productive overburden and understrata).  For water injection, the results suggest that starting with high temperature injection to lower temperature injection later on provides opportunities to recover heat from the reservoir and overburden and understrata thus improving the thermal

[1] Adams DM. Experiences with waterflooding Lloydminster heavy-oil reservoirs. J Can Petrol Technol 1982;34:1643–50. [2] Pan Y, Chen Z, Sun J, Bao X, Xiao L, Wang R. Research progress of modelling on cold heavy oil production with sand. In: Paper SPE 133587 presented at the SPE western regional meeting, Anaheim, California, USA; 27–29 May 2010. [3] Istchenko C, Gates ID. Well-wormhole model of cold heavy oil production with sand. SPE J 2014;19(2):260–9. [4] Miller KA. State of the art of Western Canadian heavy oil water flood technology. In: Paper 2005–251 presented at the petroleum society’s 6th Canadian international conference (56th annual technical meeting), Calgary, Alberta, Canada; June 7–9 2005. [5] Asghari K, Nakutnyym P. Experimental results of polymer flooding of heavy oil reservoirs. In: Paper 2008-189 presented at the Canadian international petroleum conference/SPE gas technology symposium 2008 joint conference (the petroleum society’s 59th annual technical meeting). Calgary, Alberta, Canada; 17–19 June 2008. [6] Gao C. Advances of polymer flood in heavy oil recovery. In: Paper SPE 150384 presented at the SPE heavy oil conference and exhibition, Kuwait City, Kuwait; 12–14 December 2011. [7] Gates ID. Solvent-aided steam-assisted gravity drainage in thin oil sand reservoirs. J Petrol Sci Eng 2010;74:138–46. [8] Ivory J, Chang J, Coates R, Forshner K. Investigation of cyclic solvent injection process for heavy oil recovery. J Can Petrol Technol 2010;48:22–33. [9] Diaz-Munoz J, Farouq Ali SM. Simulation of cyclic hot water stimulation of heavy oil wells. SPE 5668-MS; 1975. [10] Martin WL, Dew JN, Powers ML, Steves HB. Results of a Teriary hot waterflood in a thin sand reservoir. J Can Petrol Technol 1967;243:739–50. [11] Gates ID. Basic reservoir engineering. 1st ed. Kendall Hunt; 2011. ISBN: 978-14652-3684-5. [12] Zhao W, Wang J, Gates ID. Thermal recovery strategies for thin heavy oil reservoirs. Fuel 2014;117:431–41. [13] Computer Modelling Group Ltd. STARSTM User’s Manual; 2012. [14] Vinsome PKW, Westerveld JD. A simple method for predicting cap and base rock heat losses in thermal reservoir simulators. J Can Pet Technol 1980;19(3):87–90. [15] Gates ID, Chakrabarty N. Design of the steam and solvent injection strategy in expanding solvent steam-assisted gravity drainage. J Can Petrol Technol 2008;47:12–20. [16] Teare M, Burrowes A, Baturin-Pollock C, Rokosh D, Evans C, Gigantelli P, Marsh R, Ashrafi B, Tamblyn C, Ito S, Willwerth A, Yemane M, Fong J, Kirsch M, Crowfoot C. Alberta’s energy reserves 2011 and supply/demand outlook 2012– 2021; June 2012. [17] Gates ID, Kenny J, Hernandez-Hdez IL, Bunio GL. Steam injection strategy and energetics of steam-assisted gravity drainage. SPE Reserv Eval Eng 2007;10:19–34.