Recent Developments in the CO₂-Cyclic Solvent Injection Process to Improve Oil Recovery from Poorly Cemented Heavy Oil Reservoirs: The Case of Canadian Reservoirs

Abstract

One of the limitations of Cold Heavy Oil Production with Sand (CHOPS) is the low recovery factor (5–15%). To target the remaining 85–95% heavy oil resources, several enhanced oil recovery (EOR) techniques, such as cyclic solvent injection (CSI), have been proposed. Due to its potential success in Canada and elsewhere, this paper reviews the technical and efficiency requirements of CSI EOR in post-CHOPS heavy oil reservoirs. We explain the dominant driving mechanisms of CSI with a focus on the application of CO₂ as a solvent. Limitations of current thermal and non-thermal EOR methods were compared to the CSI in thin oil reservoirs. To complete the assessment, several case studies and lessons learned were included based on the latest laboratory experiments, numerical studies, and CSI pilot/field tests. Specific to thin and shallow heavy oil reservoirs with sand production (e.g., CHOPS), the key to recover incremental oil was found to re-energize depleted reservoirs in a cyclic manner with unexpensive solvents (e.g., CO₂). Regarding the solvent use, laboratory experiences have not been conclusive about what solvent stream could improve oil recovery. To this end, successful field scale CO₂ EOR applications have been reported in several post-CHOPS reservoirs indicating that highly productive wells during primary production might also outperform during a follow up CSI process. Numerical modeling still faces challenges to properly model the main CSI driving mechanisms, including fluid–solvent interaction and the deformation of subsurface reservoirs.

1. Introduction

Heavy oil resources continue to show a significant demand in the global energy market. Despite a decline in energy consumption due to the COVID-19 pandemic, fossil fuels (i.e., oil and gas) remain dominant players, accounting for 55.9% of the global primary energy market [1]. Heavy oil production is still essential for some countries, such as Canada. The Canada Energy Regulator [2] reported that heavy oil exports have played a major role in Canada’s export participation. In 2020, the total volume of heavy oil export was estimated at 2.81 million barrels per day, with 90% of Canada’s heavy oil originating from Alberta’s heavy oil reservoirs, which amounted to 2.53 million barrels per day.

The Lloydminster area, which straddles Alberta and Saskatchewan, represents a key area in the energy framework of Canada [3]. The western region of Canada is estimated to contain a total of 5201 million m³ of heavy oil resources. Given the abundance of such resources and the unceasing energy demand, it is imperative to employ innovative techniques to recover these resources from the subsurface. Notably, approximately 80% of heavy oil reservoirs in Western Canada are less than 5 m thick [4]. For instance, in the Lloydminster area, around 95% of the oil reserves are found within sands with a maximum thickness of 10 m, highlighting the significant proportion of oil located in thin and medium-thick reservoirs (Figure 1).

According to Dusseault [6], a typical Canadian heavy oil reservoir is composed of quarzose sand zones with porosities ranging from 28% to 32%, an average connate water saturation of 25%, and permeabilities between 1 to 4 Darcies. The viscosities of heavy oil in such reservoirs vary from 500 to 50,000 cp, while the oil API gravity ranges from 10° to 16° [3]. Despite the high permeability of these unconsolidated reservoirs, the oil properties pose significant challenges for fluid production. A summary of the typical properties of Canadian CHOPS reservoirs is provided by [7,8] in

Table 1. Typical properties of Canadian CHOPS reservoirs; modified after [7,8].

Property	Value
Depth (m)	480
Net pay (m)	5
Porosity (%)	33
Permeability (Darcies)	2 to 4
Oil saturation (%)	80
Initial reservoir pressure (kPa)	2750
Reservoir temperature (°C)	20
Dead-oil viscosity (cp)	25,000
Formation compressibility (kPa−1)	5 × 10−6
Wormhole radius (m)	0.05

.

The significance of this work lies in providing a review of the key mechanisms underlying the CSI technique for post-CHOPS heavy oil reservoirs in Canada, along with field cases that offer a comprehensive understanding of the challenges involved in implementing CSI as an enhanced recovery method.

2. Recovery Techniques in Thin Unconsolidated Heavy Oil Reservoirs

Due to its high viscosity, production from heavy oil reservoirs requires additional driving mechanisms than only fluid expansion with pressure drawdown. The theoretical relationship between viscosity and flow velocity is established by the Darcy law:

$$v=-{\displaystyle \frac{k\left(\phi \right)}{\mu (T,m_s)}}{\displaystyle \frac{dP}{dx}}$$

(1)

where $v$ is the Darcy velocity, $k\left(\phi \right)$ is the permeability of the porous media as a function of the porosity, $\mu (T,m_s)$ is the fluid viscosity as a function of temperature and solvent mass, and $dP/dx$ is the pressure gradient. From (1), $v\propto {\mu }^{-1}$, which means that higher viscosity is detrimental to fluid flow. Here, fluid viscosity is basically defined as the fluid resistance to flow. A decrease in oil viscosity, an enhancement in formation permeability, or an increase in viscous forces and pressure gradient can be achieved through fluid and sand production, fluid injection, such as waterflooding, steam assisted gravity drainage, polymer flooding, in situ combustion, and solvent-based recovery techniques (e.g., VAPEX—vapor extraction), among others.

2.1. Cold Production with Sand

CHOPS (Cold Heavy Oil Production with Sand) is a primary heavy oil recovery technique that involves deliberate initiation and continuous sand production to significantly increase the permeability of the target formation. In this technique, the low rock consolidation (very common in heavy oil reservoirs with unconsolidated/poorly cemented sandstones and friction angle of 15°) is used to detach the grain from the formation matrix by the action of viscous forces and effective stresses to increase oil production [6]. Some authors have noted that the sole action of viscous forces is not sufficiently large to detach the sand grains [9].

Compared to other recovery techniques, CHOPS is a relatively low-cost technology because it does not require any fluid injection (increase in heat, pressure, or add diluents). Most authors agree that CHOPS can achieve a recovery factor of 5% to 15% by the end of its primary production lifetime [10,11,12]. Field experience in the Lloydminster area also suggests an average recovery of 8% of the original oil in place for CHOPS reservoirs [3]. This provides potential to recover the remaining 90% of oil in place through follow up enhanced oil recovery techniques, such as CO₂ injection. However, the design of a successful CO₂ injection process to re-energize CHOPS reservoirs requires an understating of fluid flow and geomechanical issues that govern the behavior of deformable heavy oil reservoirs.

Table 2. Production mechanisms of CHOPS adapted from [6].

Primary Driver Category	Consequence
Gravity	Causes vertical stress from overlying rocks and impacts rock yielding and dilation
Pressure drop	Viscous force drives fluids to production well
Foamy oil	Enhances production via gas nucleation and fluid expansion dynamics
Sand production	Increases reservoir permeability and boosts extraction efficiency
Wellbore damage	Maintains flow by clearing obstructive deposits (e.g., asphaltenes, fine-grained particles, or mineral deposits).

summarizes the main recovery mechanisms in thin unconsolidated heavy oil formations; some of the major recovery mechanisms will be further described in the next section.

Foamy Oil Behavior. Due to high oil viscosity (order of 10,000–25,000 cp), the formation of dispersed gas bubbles is one of the main flow characteristics and drive forces in CHOPS [6,8]. Foamy oil drive mechanism has been extensively studied in the literature [13,14,15,16,17]. Figure 2 illustrates a sequence of images from a fluid depletion experiment in which a 2D view of the heavy oil sample, saturated with gas, was subject to fluid depletion at a rate of 300 kPa/min from 1200 kPa to 100 kPa. Figure 2 visualizes the non-equilibrium nature of the foamy oil drive; the released gas bubbles initially do not form a single gas phase and remain dispersed in the oil which causes significant fluid expansion; comparison of the total fluid column with respect to the initial oil level, depicted by the dashed lines.

Wang et al. [18] described the evolution of the foamy oil in three main stages: (i) expansion, (ii) peaking, and (iii) decay. Basically, the bubbles within the foamy oil expand during depletion and push the oil out of the reservoir. Some authors noted the dependency of foamy oil to the magnitude of the pressure depletion rate (e.g., [19]). The strength of foamy oil behavior and depletion rate in heavy oil has a direct impact on the established pore pressure during the fluid production phase. It is also worth mentioning that a lower pressure drop could lead to a more lasting foamy oil peak and declining phases.

Figure 3 shows a visual microscale confirmation of fluid flow through porous media due to the foamy oil driving mechanism [20].

Formation of High Permeability Channels. Wormhole, in the context of heavy oil and sand production, refers to a zone in which there is no grain-to-grain contact [6]; that lack of contact between grains translates into high porosity and permeability channels [21]. The formation of high permeability channels (i.e., wormhole network) has a few advantages [22].

First, since wormholes have a higher permeability, it is a preferential route for the fluid flow from the inner reservoir to reach the wellbore. If this feature is described in terms of pressure, the wormhole network can be understood as isobaric lines within the reservoir [8]. This means that the high permeability of the wormhole network creates the condition in which fluid can flow at low pressures. Figure 4 shows the inclusion of wormholes using different modeling techniques, including dual porosity models and multi-lateral wells. Pressure distribution during the depletion of a reservoir, affected by the presence of wormholes (represented through numerical multi-lateral wells) could be observed by well testing [23,24,25].

Second, the wormhole network enhances the productivity and injectivity qualities of the reservoir; in the context of fluid injection processes, such as CSI, the injection performance is strongly improved. Additionally, the creation of high-permeability channels increases the fluid access and contact area of the injected fluid within the reservoir compared to the case when a fluid is injected into an intact reservoir.

In the context of petroleum geomechanics, the growth of the wormhole network is governed by the yielding of the sandstone [6,27]. Due to the loss of confining stress as a result of the non-grain contact state in the cavity zone, the radial stress ${\sigma }_r$ decreases while the tangential stress ${\sigma }_{\theta }$ increases. The research in [6,22] showed a stress state diagram as a function of the radial distance into a reservoir with a cavity. The shear stress increases because the difference between the radial and tangential stresses is higher until the yielding is reached. Experimental studies have shown that under an anisotropic stress state, the wormholes tend to grow in the direction of the lower horizontal stress [28].

Those geomechanical phenomena result in an increase in porosity due to the removal of mass (i.e., sand grains) in a constant bulk volume.

Table 3. Experimental observation of an increase in porosity due to sand production.

No.	Reference	Initial Porosity [%]	Final Porosity [%]
1	[29]—Sandstone low density	36.31	85.61
2	[29]—Max. density of 1860 kg/m3	36.31	79.86
3	[29]—Experiment 2, Low density sandstone density	31.03	71.85
4	[29]—Experiment 2, max density of 1860 kg/m3	31.03	70.65
5	[29]—Experiment 3, max density of 1860 kg/m3	31.03	85.47
6	[29]—Experiment 3, max density of 1860 kg/m3	31.03	83.86
7	[30]—Silica well sorted	41.08	57.30
8	[30]—Silica poorly sorted	32.2	41.65
9	[30]—Silica well sorted 2	40.82	53.87
10	[30]—Silica no fine fraction	38.4	48.6
11	[30]—Husky field	38.2	62
12	[31] *	20.56	60
13	[32]	21.80	52.99
14	[33]	37	46
15	[34] *	32	60
15	[34] *	32	60
16	[21] ξ	32	52
	Mean	33.18	63.23
	Standard Deviation	5.69	14.01

* Used for numerical modeling; ^ξ results obtained from history matching.

shows the porosity estimations based on data and reported measurements in the literature.

A look at the CHOPS literature suggests an average initial porosity of 38.13% (standard deviation of 3.5%) for different sandstones. The final porosity has an average of 51.67% (a standard deviation of 9.2%). Similarly, some authors have noted an increase in the formation permeability due to sand production from 3 to 67 Darcies [30,35].

2.2. Waterflooding in Heavy Oil Reservoirs

Although the most common practice for heavy oil and bitumen recovery are the thermal techniques [4], the costs associated with steam generation, greenhouse gas emission, and upgrading facilities may affect the viability of thermal EOR projects. This makes viscosified waterflooding an alternative for some heavy oil reservoirs [36]. In the waterflooding process, the viscosity difference between the injected water and in situ oil often creates adverse mobility ratios with a negative impact on the oil recovery factor and sweep efficiency [36]. The mobility ratio can be described as follows:

$$M={\displaystyle \frac{{\lambda }_{displacing phase}}{{\lambda }_{displaced phase}}}={\displaystyle \frac{k_{rw}}{k_{ro}}}{\displaystyle \frac{{\mu }_o}{{\mu }_w}}$$

(2)

where $M$ is the mobility ratio, $\lambda$ is the mobility of the phase, $k_{rw,o}$ are the relative permeabilities of water or oil, and ${\mu }_{w,o}$ is the viscosity of water or oil. From Equation (2), it is apparent that if ${\mu }_o\gg {\mu }_w$ then $M\gg 1$, a condition that leads to fingering where fully water-saturated channels connect the injection and production wells [37], bypassing oil in the pore space, and ultimately leading to poor oil recovery; a further phenomenal explanation can be found in the work of [38]. Ref. [36] developed an experiment to show that an early breakthrough due to fingering can be experienced in heavy oil reservoirs.

2.3. Thermal EOR in Heavy Oil Reservoirs

The most applicable thermal EOR technique for heavy oil in Canada is steam-assisted gravity drainage (SAGD). SAGD involves the use of a pair of parallel horizontal wells for the injection of steam into the reservoir. This is intended to reduce the heavy oil viscosity by heating the oil using the latent heat of steam in the reservoir. A reduction in oil viscosity is the main reason behind all variations in thermal EOR methods, such as steam flooding, and cyclic steam stimulation (CSS). Figure 5 is a based on [4], who conducted numerical simulations to compare the efficiency of various thermal EOR techniques; they compared the thermal EOR results with cold production without sand in thin reservoirs (<10 m).

The simulation results revealed an adverse cumulative energy–oil ratio (cEOR), which is a measure of energy input required to produce a unit volume of oil. An economically reasonable cEOR is typically around 10 GJ/m³, corresponding to a cumulative steam–oil ratio (cSOR) of 4 m³/m³ [4]. The simulations showed that a high cEOR (three times larger than typical values) can be attributed to high heat loss to underburden and overburden formations in thin formations (less than 10 m thick). In a 10 m thick reservoir, heat loss was calculated to be 10%, while in a thinner reservoir of 4 m, the heat loss increased by 40%.

Considering the limitations of waterflooding and thermal techniques in thin unconsolidated heavy oil formations, the implementation of follow up recovery methods, such as Cyclic Solvent Injection (CSI), has been proposed to improve the recovery from heavy oil reservoirs. It is worth mentioning that some authors have studied the in situ upgrading of heavy crude oil using catalysts as a potential technology [39].

3. Cyclic Solvent Injection Process

The cyclic solvent injection process is intended to target the large amounts of oil remaining in the reservoir through solvent diffusion. The CSI technique is applied when either the reservoir is too thin (<10 m) for thermal EOR methods, or heavy oil is too viscous to cause an adverse mobility ratio during the flooding process. Specific to CHOPS reservoirs, solvent access to the reservoir can be through high permeability channels and taking advantage of its high contact area with the reservoir [40]. The CSI technique can also be thought of as an in situ upgrading technology because it changes the composition and properties of the heavy oil at reservoir conditions [41]. The CSI involves alternating between solvent injection, soaking, and oil production phases, using a variety of solvents, such as CH₄, CO₂, alkali metal silicide, among others [42].

During the injection phase, the wellbore is used to inject the solvent or a mixture of solvents into the heavy oil reservoir. This phase results in an increase in the pore pressure around the wellbore and a reduction in the effective stress as an immediate consequence. In field practice, the injection is applied until the initial pressure of the reservoir is reached; but in some instances, the injection can go beyond the initial reservoir pressure [5].

The injection phase is followed up by a soaking period in which the wellbore is shut in. The soaking phase is a critical step to dissipate the pore pressure and to allow the slow process of solvent diffusion to occur within the heavy oil reservoir. Solvent diffusion can permanently reduce the oil viscosity. In

Table 4. Oil properties and their change after solvent treatment. Modified after [41].

Oil/Reservoir	Viscosity [mPa·s] (Temperature)	Solvent	Viscosity After Dissolution [mPa·s]
Lloydminster	7600 (20°)	Propane	2000
Athabasca	148 (104°)	Propane	8.6
Cold lake	70,000 (20°)	Ethane	80
Athabasca bitumen	700,000(20)	Toluene	2000
Frog lake	18,600 (21.6 °C)	Butane	8
Du84, Liaohe	400,000 (60 °C)	Toluene	291.8
Gao3624, Liaohe	600 (50°)	CO2	150
Fengcheng, Xinjiang	21,584 (25.6°)	Propane	4362

, viscosity on Lloydminster’s oil varies from 7600 mPa at 20 °C to 2000 mPa due to the application of propane as a diluent (i.e., a viscosity reduction of 74%). A similar response is observed on Athabasca where bitumen’s viscosity changes from 700,000 mPa·s at 20 °C to 80–100,000 mPa·s after the application of toluene.

Once the soaking period has allowed the interaction between the injected solvent and heavy oil to occur for a sufficient time, the wellbore is opened to production. A mixture of in situ fluids (both water and oil with solvent) is produced. The production stage lasts until the oil rate becomes uneconomical, or the costs of water management become unacceptable [5].

A comparison of CSI with a continuous solvent injection is conducted in

Table 5. Differences between continuous and cyclic solvent injection; modified after [43].

	Continuous Solvent Injection	Cyclic Solvent Injection
Operation Strategy	Two types: Vapex and Lateral SVX	Huff—n-puff
Driving Mechanisms	▪ Gravity (Traditional Vapex) ▪ Constant pressure may apply for lateral SVX	▪ Gravitational forces ▪ Pressure gradient ▪ Foamy oil flow ▪ Sand influx ▪ Reduction in skin
EOR Mechanisms	▪ Viscosity reduction ▪ Asphaltene precipitation ▪ Diffusion and dispersion-based mass transfer ▪ Capillarity mixing	▪ Viscosity reduction ▪ Asphaltene precipitation ▪ Diffusion and dispersion-based mass transfer ▪ Capillary mixing ▪ Foamy oil
Rules of Wormholes	Establish communication between the injector and the producer. They may cause solvent quick breakthrough.	Increase contact area for solvent and crude oil. Help the diluted oil flow to the producer.
Main Challenges	▪ Low mass transfer rate ▪ Small gravity head (for thin-net pay reservoirs)	▪ Pressure depletion ▪ Viscosity regain
Geomechanics Implications	▪ Continuous decrease in the effective stress and tensional failure ▪ Stability of the wormholes	▪ Cyclic loading and fatigue of the rock ▪ Stability of the wormholes

.

To analyze further details about CSI and CO₂ as a solvent, the following part of this review will be split into three sections: experimental approach; field pilot projects and experiences; and modeling and numerical descriptions. It is worth noting that there have been recent developments to reduce the price of CO₂ capture and allow its use, e.g., [44,45].

3.1. Recent CSI Experimental Studies

Coskuner et al. explored how different solvents (e.g., heptane, distillate) may help to achieve a higher recovery factor [3]. The general laboratory procedure was to place saturated rock samples with oil inside a glass container, and add hot water and solvent. This experimental process was performed in multiple time steps to emulate the cyclic nature of solvent injection. Experimental data suggested recovery factors ranging from 42% to 88%; it is very likely that these high recovery factors will not be achieved in the field application of solvent injection. In these experiments, the solvent was in contact with all faces of the submerged rock sample, and the complexity of fluid flow regimes, dispersion, reservoir heterogeneity, and limited access to a large volume of solvents at reservoir conditions was not considered.

The impact of some of the characteristics of a physical model (reservoir volumes, wormholes and high-permeability channels, and their spatial location within the reservoir) was addressed by [43] using an experimental design. To capture the possible effects, ref. [43] varied the diameter of the test sample, the position of the wormholes, and the vertical and horizontal orientations of the sample. Their experimental results showed the impact of reservoir–wormhole geometry, the length and relative positions of the wormholes with respect to the reservoir limits on the final oil recovery factor. They concluded that the presence of longer, high-permeability channels and wormholes is effective in delivering solvent to access the lower-permeability portion of the reservoir; this provides increased access for the solvent to dilute heavy oil in the reservoir matrix and ultimately increases the oil recovery factor in shorter time. This means that in such geological settings, higher permeability zones and wormholes not only improve the total recovery but also increase the oil rate in the production phase [43]. This was also confirmed through a visual inspection of the sample during experiments where lower oil saturation near a high-permeability zone suggested more oil production.

Although Du et al. [43] used a broad range for their experimental study, further analysis is required to understand the impact of observed cavities in unconsolidated sandstone reservoirs, changes in in situ stresses, and heterogeneities during cyclic loading/unloading of solvent injection process. These geometries have been experimentally shown to induce different flow regimes that may impact the final recovery [29].

In addition to the impact of the reservoir discontinuities, other empirical works are available in the literature that focus on the driving mechanisms during cyclic solvent injection. In [12], the impact of parameters, such as gravity, pressure depletion rate, solvent composition, and initial oil saturation, are studied. The tests were conducted on the oil from Cold Lake formation and CH₄ was injected into the samples subject to primary production. The follow up CSI process was performed in the sand pack at low pressure (2.76 MPa); the sand pack was scanned before the primary production and after the end of each cycle of solvent injection to obtain fluid saturation profiles. The experiments showed the bubble nucleation and foamy oil behavior as the key driving mechanisms. From a linear decrease in pressure and low gas–oil ratio at the beginning of the test, the expansion of the fluid in response to the drawdown was found to mainly contribute to oil production. Once the sample reached the bubble point, the recovery factor was improved with the foamy oil flow in which the dispersed gas bubbles helped to maintain the sample pressure for an extended time. It seems that the first cycle of each test was the one with the highest impact on recovery factor because the main driving forces (especially foamy oil behavior) were strongly active during the first cycle, when no continuous gas phase was present. Similar observations have also been observed at field scale.

The composition and type of solvent stream for CSI are other highly debated topics in the literature. For EOR applications in heavy oil reservoirs, solvent type is sometimes selected based on the intended recovery mechanisms. For instance, a reservoir may benefit from activating foamy oil drive while a reduction in oil viscosity is a priority for another EOR project. As a result, a wide diversity exists on the type of solvent to meet specific EOR requirements in different regions. The cost of solvent is another issue to be considered for the success of the CSI process. In the current literature, popular gaseous solvents are methane, propane, carbon dioxide, or a combination of them.

Huerta et al. [46] proposed to take advantage of some produced gases like CO₂ and H₂S; in some fields, a stream of CO₂ and H₂S (known as acid gas) is a usual production sub-product, but they pose operational challenge in terms of management and safety, so the disposal of CO₂ and H₂S in geological formations is a more common practice than their use for EOR purposes. The researchers in [46] studied the impact of the cyclic injection of acid gas on oil recovery factor through a set of experiments where multiple tests were carried out with two cycles. They used pure CO₂ solvent (as reference), CO₂ 72%-CH₄ 28%, CH₄ 70%-C₃H₈ 30%, and CO₂ 90%-H₂S 10%. The experimental results showed that a combination of CO₂ and H₂S offers a better performance than pure CO₂ or any methane/propane mixture. Mixtures of CO₂ and propane showed the best results in terms of oil recovery. An interesting observation was that regardless of the solvent composition, all the experiments confirmed that most of the oil recovery is achieved through the first cycle. This observation has been reported in other different experimental studies and field tests.

In a more detailed experimental design, ref. [40] used a 1.5 m long sand pack sample to study solvent injection and key parameters related to fluid expansion, gas dissolution, and foamy oil behavior using a combination of gaseous solvents, including methane, propane, and carbon dioxide. Soh et al. [40] observed that some gases can function as “mainly diluents” to reduce oil viscosity while other gases can act as “foamy helpers” to provide the right conditions to establish foamy oil flow. Previously, ref. [46] suggested that the mixture of methane/propane could be a good performer to increase oil recovery. However, the results from [40], specifically for the heavy oil reservoir, implied that solvents with noticeable mixing properties (e.g., propane) are not recommended to be used with the solvents that aid the formation of foamy oil (e.g., CH₄ or CO₂). This might suggest that the type and concentration of solvent is case specific to the heavy oil reservoir and that laboratory experiments are required to understand the interaction of solvent–heavy oil at reservoir pressure and temperature before conducting any CSI field test.

The physical modeling of the post-CHOPS CSI has recently found a new dimension with the use of a geotechnical centrifuge to model the drainage area of heavy oil reservoirs. Cartagena-Perez et al. [47] developed a geotechnical centrifuge cell that allows the integration of triaxial stresses with 3D printed samples that contained wormholes. The results of this work highlight the pivotal role that geomechanics has during the post-CHOPS phase of the reservoir when CSI is applied.

3.2. Recent Development in Numerical Modeling of CSI

Most of the literature on modeling the CSI process in unconsolidated heavy oil reservoirs has focused its effort on two main topics, namely, (i) the numerical modeling of solvent–heavy oil interactions that include foamy oil behavior and non-equilibrium dissolution/exsolution processes, and (ii) the numerical modeling of wormholes and high-permeability channels within the reservoir after cold production with sand.

The proper representation of the dissolution/exsolution process seems to directly impact the numerical evaluation of CSI performance in heavy oil reservoirs [48]. In general, the non-equilibrium behavior for gas dissolution into heavy oil can be described as follows:

$$SG\to SG+SL$$

(3)

where $SG$ is the solvent in gaseous phase; $SL$ is the solvent in liquid phase. Similarly, the dissolution at non-equilibrium behavior is represented as follows:

$$SG+SL\to 2SL$$

(4)

Non-equilibrium can be thought of as a delay in the gas dissolution/exsolution process, compared to equilibrium phase behavior, in which dissolved gas could release into free phase, for instance, due to pressure depletion at an instant [15,49]. The modeling can be achieved through multiple Arrhenius kinetic reaction types of modeling for the release of dissolved gas, to dispersed gas, to free gas. Ref. [48] compared the impact of non-equilibrium and instant equilibrium phase behavior on the solubility for CSI processes for a solvent mixture of CO₂ and propane (Figure 6). They reported significant differences in oil recovery, solvent recovery, and solvent-to-oil ratio. Non-equilibrium phase behavior resulted in larger bottomhole pressures and higher cumulative oil production. The observed differences were explained through the advection, diffusion, dispersion, and dissolution of the non-equilibrium mixing process.

In addition to the modeling of solvent–heavy oil interactions, it is important that the modeling tools are capable of including and history matching the presence of wormholes and high-permeability zones with field data during the cold heavy oil production phase; this step is required prior to numerically simulating the CSI process [48,50]. Different modeling approaches are presented in the literature. Rivero et al. [51] suggested the use of an effective permeability model that represents the wormholes and high-permeable zones. Rangriz Shokri et al. [52] used partial dual-permeability models, in which the matrix represents the intact reservoir, and the fractures represent the wormhole domain. Haddad et al. [53] employed dilated-zone model with wormholes and cavities represented as dilated zones around the wellbore. In a numerical study, Chang et al. [48] illustrated that the selected method to represent the reservoir and high-permeability zones affects the cumulative oil production and recovery factor, predicted for primary production and follow-up EOR scenarios.

In summary, the large-scale simulation of CSI scenarios in heavy oil reservoirs could face many other numerical challenges, including longer simulation run times, difficulty to model sand production and wormhole growth, and upscaling issues of the laboratory results when foamy oil behavior from the bulk fluid phase (e.g., PVT cell) is used to represent foamy oil flow in the porous media (e.g., [32,50,52,54]). Given the range of these modeling uncertainties, numerical simulations are still required to assess and optimize the performance of the CSI process in unconsolidated heavy oil reservoirs [26,35,55,56,57]. Optimization parameters for the CSI process include the type and concentration of solvents, injection rate and pressure, duration of injection, soaking, production phases, number of cycles, and incremental oil production per cycle. The available literature on the numerical simulation of CSI also suggests the highest oil recovery in the early cycles and a general declining trend in cumulative oil production in the subsequent cycles (Figure 7); this is consistent with previous laboratory experiments and field tests (e.g., [58]). If we assume that each cycle takes one year, after five years, the recovery would be approximately 38%, which is comparable to the performance shown for the SAGD technique in 4 years (Figure 5). It is important to note that this recovery is obtained in a reservoir where SAGD is not suitable.

A fully coupled numerical model between geomechanics and fluid flow is still a task to accomplish. From the geomechanics perspective, extra work is needed to understand the constitutive model that may be applied to a granular material as its fabric changes with the sand production. Also, yielding criteria, such as Mohr–Coulomb, that were developed for continuous mechanics, may be limited.

3.3. Recent CSI Pilot Projects and Case Studies

In this section, field cases from the Lloydminster area, Canada, will be described with emphasis on some CHOPS wells from the following reservoirs: Nexen Plover Lake, Husky Mervin, Devon Manatokan East, and Husky Lashburn. Our focus is on the application of gaseous hydrocarbon solvents and CO₂. However, CSI projects that employed other gases, such as nitrogen, are also available in the literature [59].

Overall, the performance of CSI in an unconsolidated heavy oil reservoir is closely related to its primary production history of that reservoir; this due to the fact that large volumes of sand might have been produced with oil during primary production and that affects the creation of high permeability regions and wormholes within the reservoir. Figure 8 illustrates the typical behavior of a CHOPS well in the Lloydminster area. Initial production begins with an increase in oil and sand production rates; the reason to allow some sand production along with heavy oil is to increase the oil rate and accelerate production. With more production, sand production significantly declines; the oil is produced until the water production makes oil recovery from that particular well uneconomical [5]. During the primary production, the compaction of unconsolidated formation and other geomechanical issues are responsible for 30% of the production drive energy.

Case Study of the Nexen Plover Lake. In this field project, a propane-based CSI was applied. The wellbore experienced a high water rate and low oil production prior to the first cycle; this has made the production from this well uneconomical. The injection of propane during CSI was sufficient to increase the oil rate by six times the pre-CSI rate. It was reported that most of the injected propane was recovered from the produced reservoir fluids [5]. This retrieval of solvent added to the economic viability of the project. It is worth mentioning that one of the key challenges that determines the economic success of CSI field implementation is the high cost of solvents and the existence of the possibility to retrieve the injected solvents. During the CSI process, a reduction in oil viscosity was observed due to the use of injected propane in the reservoir [5]. A direct consequence of this oil viscosity reduction was a decrease in the water cut that resulted from an improved mobility ratio.

A description of the production record is shown in Figure 9.

Case Study of the Husky Edam. This project employed a combination of methane-propane injection. The oil recovery during the primary production is 11,000 m³, and after the first five cycles of CSI, an additional 5500 m³ oil could be produced; this translates to 50% of the initial oil recovery [5]. Each CSI cycle has resulted in a high oil rate that declined with time during the production phase [5]. The field observation suggested that the water cut decreased not only within each cycle but also showed a decreasing general trend over a total of five cycles (Figure 10). This was believed to be due to the continuous change in heavy oil properties (e.g., viscosity) caused by the solvents that positively affected the mobility ratio. Further research is required to address wettability alteration in the reservoir because of solvent use during the CSI process. Recently, there is evidence of multiple cycles of CO₂ injection (Since 2019) with a total injection of almost 5.3 × 10⁶ m³ of CO₂, showing the transition in the solvents.

Case Study of the Husky Mervin. This pilot project is of special interest, because of the CO₂ use. The first well (Mervin 05-36) injected pure CO₂ in the first two cycles (Figure 11). The primary recovery of this well was 11,700 m³; the recovery improvement during the first and the second cycles of a follow up CSI operation was over 14% and 10% of the primary recovery, respectively. From the same reservoir, another well (Mervin 15-01 Colony—Figure 12) presented a different response to the application of CSI. This well had a higher primary oil production of about 51,000 m³ (4.4 times higher than the Mervin 05-36); this probably indicates a more developed network of wormhole and high permeability channels due to sand production. The CSI process added incremental oil production, about 60% of the primary recovery in the first cycle, and 73% in the second cycle. The high recovery factor observed during the CSI cycles suggests that a productive CHOPS well may potentially perform better during CSI operation [5]. This could be associated with a larger reservoir contact area to solvents (e.g., extended wormhole network and high permeability channels due to sand influx) that had been created during primary production. A third case (Mervin 12-31) offered insights on the impact of CSI on water cut (Figure 13). The well was productive with a primary recovery of 59,000 m³, but with a high water cut of 80% [5]. The high water cut was observed in all the CSI cycles, which might be interpreted as a limited action of the injected CO₂ in gas phase on the mobility ratio.

Case Study of the Husky Lashburn. This project also involved the cyclic injection of CO₂ (Figure 14). In this reservoir, the behavior of each cycle seems to decline with a high initial oil rate and low water cut. The water cut during CSI cycles was high, which again suggests that CO₂ injection in gas phase may not help improve the mobility ratio. Field operational data indicated that 40% of the injected CO₂ remained in the reservoir [5]; the increasing trend of stored CO₂ eventually helped to transition from CO₂ EOR to permanent CO₂ storage.

Case Study of the Dee Valley. This project has at least six wellbores (05-10-049-22W3, 06-10-049-22W3, 10-09-049-22W3, 14-09-049-22W3, 15-09-049-22W3, 16-09-049-22W3) where CO₂-CSI has been applied, taking advantage of the proximity of the wellbores (Figure 15). All the wells were subject to the cyclic injection of CO₂ at the beginning of 2015 with approximately three or four cycles until the end of 2018 or beginning of 2019.

The wells display a characteristic pattern observed in CHOPS reservoirs during primary production, marked by a decline in oil rate and an increase in water cut, necessitating the implementation of a CSI program. CO₂ injection in one of these wells commenced in 2015, resulting in an equal water cut compared to the end of primary production but also an increased oil rate (Figure 16). Over the next 3 years, three cycles of CO₂ injection were carried out until 2024, leading to peaks in oil production equivalent to some of the highest levels during primary production, and progressively increasing the water cut with subsequent cycles, which again challenge the possibility of a favorable mobility ratio for the oil with the injection of CO₂.

Those field observations, next to the numerical and experimental ones, may result in some driving mechanisms of the production in post-CHOPS CO₂-CSI reservoirs that are summarized in

Table 6. Driving mechanism during the CSI in post-CHOPS reservoirs.

Mechanism	Explanation	References
Larger wormhole network or cavities	The wellbores with well-developed wormholes or cavity areas have a larger contact area for the solvent to act. It has been observed in the field experience that wells with good performance during CHOPS have higher production during the CSI.	[5]
Re-energization	Due to the injection of solvent, the increase in the pore pressure derived in a re-energization of the reservoir. This means a higher reservoir pressure around the wellbore to facilitate the production.	[35]
Foamy oil	Experimental approaches have shown that the foamy oil promoted by the CO2 as a discontinuous phase may play a key role in the recovery due to the expansion of it during the drawdown.	[60]
Increase in the compaction	Physical modeling of post-CHOPS reservoirs concludes that geomechanical driving mechanics are key by influencing the flow behavior and pressure distribution in CO2-CSI processes.	[29,47,61]
Capillarity mixing	During the production stage, the zone around the wormholes reduces its saturation of oil. During the injection and soaking periods, and due to the capillary forces and wettability of the formation, there is a redistribution of saturations, allowing the oil to migrate from the far to the near zone surrounding the wormholes or cavities.	[43]
Reduction in oil viscosity	Field experiences have shown that this mechanism is limited (Case Dee Valley or Husky Mervin), and the impact of the CO2 in the mobility ratio is not always dominant.	[5]

.

4. Conclusions and Remarks

Looking at the recent development of the CSI process in thin unconsolidated heavy oil reservoirs, it appears that the popularity of cyclic solvent injection (as opposed to VAPEX and continuous solvent injection) has increased as a follow up EOR technique after cold production. Other IOR/EOR methods, including waterflooding, gas flooding, and thermals (e.g., SAGD, CSS) can face efficiency issues with fingering, sweep efficiency, and significant heat loss; this again makes CSI a feasible EOR alternative to consider for thin heavy oil reservoirs. Regarding the solvent type and concentration, laboratory experiments are not conclusive to achieve high recovery factors; however, most of the solvents either act to reduce heavy oil viscosity or to increase the strength of the foamy oil behavior (non-equilibrium nucleation and stability of the dispersed bubbles). Of interest, the injection of CO₂ in gas phase has been employed at field scale in different reservoirs with some success to improve recovery factor. Numerical simulations still need to overcome modeling challenges to properly address the complex interplay of solvent–heavy oil reactions, foamy oil flow, sand production, and stress-deformations during loading/unloading cycles of CSI. Progress has been made to integrate kinetic reactions to model non-equilibrium phase behavior. The lessons learned from the field implementation of CSI indicated that highly productive wells during primary cold production are probably good performers during follow-up CSI operations, which confirms the important role of creating high-permeability channels and a wormhole network so solvents can access more intact portions of reservoirs. A common observation from numerical simulations, laboratory experiments, and field tests is that the first CSI cycles are more effective to improve oil recovery; the additional recovery after each cycle decreases, meaning that field development strategies should be focused on the first cycles of CSI. This paradigm has been challenged by some field experiences, demanding further research about the hierarchy in the driving mechanisms of the production with CSI in post-CHOPS reservoirs.