Figure 1 exhibits a visual foam performance test device used in this study. According to the experimental protocol, certain concentration of foaming agent solution was pre-configured, and the foam was prepared by the Waring Blender experimental method (Aleksandrov et al., 2018). The air tightness of the experimental equipment was first evaluated, and then the foam evaluation device (Model: Foam Evaluator, China MSC Group, pressure range <32 MPa and temperature range <300°C) was put into an oven thermostat (temperature range <300°C). Further, the stability of the foaming agent was determined by measuring the liquid elution half-life at a temperature of 100–300°C. Next, the oil washing efficiency of foam was calculated.

Two sandpacks with high and low permeability were, respectively, used to form a dual-sandpack model, which was used to evaluate the plugging capacity and profile control capacity of the prepared heat-resistant CO₂ foam. The experimental apparatus is shown in Figure 2.

Sandpack is a stainless tube packed with silica powder with varying particle size distributions as a model for a reservoir. The length of the sandpack was 30.0 cm and its internal diameter was 2.5 cm. The apparatus shown in Figure 2 can be used to study the foam plugging performance of a single sandpack and the foam profile control performance of dual-sandpack, respectively. The ISCO pump (Model 100DX, Teledyne Technologies, United States) with flow accuracy <0.25 μL min⁻¹ and pressure accuracy < ±0.5% was used herein. The pressure difference between the inlet and outlet of the sandpack was measured with a differential pressure sensor (model: 3210PD, China MSC Group, measuring range <50 MPa, accuracy <0.1%FS). The measuring range of the graduated cylinder was <50 ml, and the accuracy was <0.01 ml. The temperature was controlled using the oven thermostats to achieve the desired reservoir temperature.

In the experiment, the two-dimensional (2D) visualization plate model mainly composed of two tempered glass sheets and a metal frame was used. The length, width, and thickness of the tempered glass were 25, 25, and 2 cm, separately. First, two tempered glass sheets were put into metal frame and a ring was installed between them for sealing. A space of 25.0 cm × 25.0 cm × 0.6 cm was left between the two tempered glass for filling silica sand. Two mental frames with the same size cutting window were then bolted together to fix the glass, and a ring was also placed between them to prevent leakage. Some holes were present at the edge of metal frame to simulate oil well, which connected with the space between the two glasses. The apparatus used in visual oil displacement experiment is shown in Figure 3.

Silica sand containing particles with different diameters was used to pack the 2D visualization plate model to obtain the anticipated permeability and porosity. The crude oil, CO₂, and surfactant were placed inside the container. The oil was delivered to the 2D visualization plate model at a constant volumetric flow rate using a piston pump (model 100DX, with flow accuracy <0.25 μL min⁻¹ and pressure accuracy < ±0.5%). Complete the oil saturation process. Then a foam generator was connected to the model. Foam was generated by injecting the surfactant and CO₂ simultaneously through the foam generator. The produced oil and gas were separated with the separator, and they were measured using a balance (accuracy <0.001 g) and cylinder (accuracy <10 ml), separately.

(1) Evaluation of foaming ability and high temperature resistance

The commonly used foaming agents in oil fields (DHF-1/B2/XJ-1/CNS/RSB-IV) were selected, and foaming agent (0.5 wt%) solutions were prepared to measure the interfacial tension of different types of foaming agents. Then, the foaming agent was placed on a stirrer with a rotating rate of 1,000 rpmmin⁻¹, stirred for 5 min, and filled with CO₂ to obtain heat-resistant CO₂ foam. The CO₂ foam was placed in a foam evaluation device, and the initial foam volume was recorded as an indicator of foaming ability. Then the temperature of oven thermostats was set to increase from 100 to 300°C, and the time for which the volume of CO₂ foam decreased to half the initial volume was recorded as the half-life, which is an indicator of foam stability. Each sample was tested three times, and the average of the measured values was taken as the final result.

(2) Evaluation of oil washing ability

The crude oil was mixed with 120 mesh sand at a mass ratio of 1:49 and aged for 120 h in the oven thermostats at 80°C. Then the aged mixture (3 g) was added to the foaming agent (0.5 wt%) and the contents were stirred to make them react fully for 48 h, during which the contents were shaken again every 6 h. Finally, the washed liquid was taken out and dried in the oven thermostats at 90°C for 24 h to obtain the amount of washed oil, and then the oil washing efficiency was calculated according to Eq. 1 as follows: δ = m M × 100 % (1) where δ is oil washing efficiency, %; m is the oil washed out by surfactant solution, Ml; M is the total oil mixed with the silica sand, mL.

The dried 80 and 120 mesh sand were mixed uniformly in a ratio of 1:4, and filled into two sandpacks, respectively. The sandpack was monitored to find out whether the air tightness of the sandpack was qualified, and the sandpack was vacuumed after ensuring that the air tightness was qualified. Subsequently, the porosity and permeability of the sandpack were measured with saturated water. The high-permeability sandpack and the low-permeability sandpack were prepared, respectively. The final permeability gradient of the two sandpacks was 5.9. The physical parameters of dual-sandpacks are presented in Table 1.

(1) Plugging ability of heat-resistant CO₂ foam

The high-permeability sandpack was used to evaluate the plugging ability of heat-resistant CO₂ foam. First, the oil was saturated into the high-permeability sandpack at a constant rate of 3 mlmin⁻¹ at 90°C until no more water emerged from the sandpack outlet. Then, at the experiment temperature of 200°C, the 0–6 PV volume of heat-resistant CO₂ foam was injected into the sandpack. Since the goal of foam application is profile control and water plugging, the difference in water flooding seepage resistance before and after foam injection is used to characterize the plugging effect of foam. After foam slug was injected, the pressure difference between the inlet and the outlet of the sandpack during the water flooding injection was recorded as P1. Then, the parameters such as gas-liquid ratio, foam injection amount, and oil content were changed, respectively, re-inject foam slug and water flooding, the pressure difference of water flooding process after changing parameters was recorded as P2. The resistance factor (RF) of foam was defined as P2/P1. By comparing the changes in the RF, the optimal foam injection scheme was designed to achieve the best plugging effect.

(2) Heat-resistant CO₂ foam profile control ability

The prepared dual-sandpacks were recorded for porosity and saturated with oil. The water flooding experiment was carried out at 200°C and injection rate of 3.5 mlmin⁻¹, and the produced liquid at the outlet of the dual-sandpacks was collected separately. The volume of produced liquid of the low-permeability sandpack was recorded as Q₁, and that of the high-permeability sandpack was recorded as Q₂. The liquid production ratio of the dual-sandpacks was recorded as Q₂/Q₁. After the water flooding, 0–6 PV volume of heat-resistant CO₂ foam was injected into the dual-sandpacks, and the water flooding of 3 mlmin⁻¹ was carried out via continuous injection after the CO₂ foam slug injection. The produced fluid at the outlet of the dual-sandpacks was collected and recorded again as Q₁’ and Q₂’, respectively. By comparing the liquid production ratio of the dual-sandpacks before and after the addition of heat-resistant CO₂ foam, the profile control ability of the foam was verified.

The experimental steps are as follows:

(1) The 2D visualization plate model was washed with distilled water and absolute ethanol to ensure the complete removal of dirt. The pipeline was installed to simulate the injection well in the upper and lower center of the model, and to simulate the production well on the right side of the model. It was ensured that the production well outlet was higher than the model to avoid oil flow. The model was sealed as a whole and exhibited good air tightness.

Owing to its high apparent viscosity, foam bursts in oil, stabilizes in water, and exhibits high impermeability in areas with high water saturation (Ahmed et al., 2021). Moreover, the larger surface tension allows the foam to block the higher-permeability channel and promote the displacement range in the low- and medium-permeability channel. According to the research, parameters such as gas-liquid ratio, foam injection volume, foam injection rate, and oil content are important factors that affect the RF of foam (Charlier et al., 1995). Therefore, the plugging experiment of heat-resistant CO₂ foam in the high-permeability sandpack under the influencing factors was carried out, and the experimental results are presented in Figure 6 and Figure 7.

Figure 6 shows that the foam generated by the mixed gas-liquid injection exhibits higher resistance and better plugging effect. Moreover, with the increase of the foam injection volume, the RF gradually increased. However, when the foam injection volume was increased to 5.4 PV, the RF growth slowed down. With the increase in the injection rate, the RF of the foaming agent increases.

Figure 7 shows the growth of the RF under the influence of different gas-liquid ratios, revealing that when the gas-liquid injection ratio is 1:1, the plugging effect of the foam is the best.

Therefore, it is recommended to adopt the injection method with the high volume of foaming agent, the gas-liquid injection ratio was 1:1, and gas-liquid mixed injection in the oil development. After adding heat-resistant CO₂ foam, the oil recovery in the high-permeability sandpack was obtained as shown in Figure 8.

Notably, the added heat-resistant CO₂ foam improves the oil recovery of the high-permeability sandpack through the foam plugging ability. When only water flooding occurs, the oil recovery is only 80.15%, and the water saturation increases rapidly. In contrast, when heat-resistant CO₂ foam is added, the start-up pressure of heavy oil is reduced, the recovery is increased to 89.26%, and the water saturation is reduced by 6.78%. The comprehensive experimental results show that the heat-resistant CO₂ foam exhibits good plugging performance. The optimal RF is as high as 63.48, and the effect of improving oil recovery of the high-permeability sandpack is obvious.

The profile control ability of foam is reflected in the following way: When both the high-permeability channels and low-permeability channels are present at the same time, the foam preferentially enters the high-permeability channels, increases the flow resistance of the high-permeability channels, and promote the subsequent injection of fluid into the low-permeability channels (Lopes et al., 2021). Consequently, the sweep degree between the high-permeability and low-permeability channels gets balanced. The results of the high-resistant CO₂ foam balancing the fluid flow rate between the high-permeability sandpack (3,975 mD) and the low-permeability sandpack (673 mD) according to the dual sandpack model are presented in Figure 9.

Figure 9 illustrates that the permeability gradient of two sandpacks was 5.9. In the range of 0–6 PV foam injection volume, the liquid production rate of high-permeability sandpack is always greater than that of low-permeability sandpack. However, under the action of heat-resistant CO₂ foam, the liquid production rate difference gradually shrinks between two sandpacks. At this time, the fluid flow rate of high-permeability sandpack is 1.81 mlmin⁻¹, and that of low-permeability sandpack is 1.47 mlmin⁻¹.

Figure 10 presents the comparative analysis of the ratio of the produced fluid of the sandpack to the total produced fluid after adding the heat-resistant CO₂ foam with different injection volume in the range of 0–6 PV.

Figure 10 demonstrates that at the beginning of the oil displacement, the high-permeability sandpack was the first to produce fluid, while the low-permeability sandpack produced almost no fluid. With the increase of the amount of foam injected, the foam gradually blocked the high-permeability channel, which led to the adjustment of the permeability profile of the two sandpacks. This resulted in the gradual increase in the liquid production of the low-permeability sandpack, and the liquid production difference gradually decreased between the two sandpacks. When the amount of foam injected reached 4.5–5.5 PV, the profile control effect of foam was found to be the best, and the difference of liquid production value of the two sandpacks was reduced to 22.64%. The experimental results show that heat-resistant CO₂ foam has good profile control ability.

The oil displacement effect of heat-resistant CO₂ foam injection was studied through the 2D visualization plate model. First, the CO₂ gas cap drive was adopted. When serious gas channeling occurred, the injection of heat-resistant CO₂ foam started plugging the channels.

Figure 11 shows that the oil recovery in the 2D visualization model is only 37.16% when the oil is displaced only by the gas cap drive. Subsequent injection of heat-resistant CO₂ foam and continued production lead to the increase in the oil recovery to 61.01%.

Figure 12 shows the variation of the remaining oil distribution with the development time. Panels (A-D) show the distribution of the remaining oil when the production time was 20 min, 40 min, 60 min, and 80 min, respectively. The black area is the remaining oil distribution area, the white area indicates the foam flooding area, and the red frame shows the as-formed high-permeability channels.

Figure 12 exhibits that when CO₂ is injected from the gas cap, gas channeling occurs in about 20 min of production. The gas first flows through the high-permeability channel to displace the oil in the channel, and the speed of oil displacement in the four channels basically remains consistent. When the production time reaches 40 min, the gas channeling becomes very serious. At this time, there is still a large amount of remaining oil in the matrix. Thus, heat-resistant CO₂ foam is injected, which preferentially increases the flow resistance in the simulated fracture and inhibits gas channeling; as a result, more CO₂ enters the low-permeability area and improves the sweep area. At the same time, after the foam bursts, the released gas enters the top of the model under the action of gravity differentiation. Moreover, the accumulated gas forms a secondary gas cap, which continues to displace the remaining oil and increases the sweep coefficient. The surfactant in the foam system can lead to significant reduction in the oil-water interfacial tension, so that the oil washing efficiency gets improved. Eventually, the remaining oil at the bottom of model stays near the production well location. When the production time is 80 min, most of the remaining oil in the matrix is ​​produced, and the oil displacement process completed.

The intermediate container is filled with certain amount of CO₂ and connected to the 2D visualization plate model to simulate the process of injecting CO₂ as a gas cap to assist heavy oil recovery, which is also an important way to realize carbon capture (Karzina et al., 2019). The results of the oil recovery affected by gas cap volume are shown in Figure 13.

Figure 13 shows that when the gas cap volume is small, the CO₂ in the gas cap remains relatively stable, and the recovery rate is low. With the increase in the gas cap volume from 0.05 to 0.1 PV, the CO₂ volume participating in the driving of heavy oil increases. The increase in gas volume makes gas channeling more likely to occur, thus forming high-permeability channels, providing oil flow channels, and increasing oil recovery. With the continuous increase in the gas cap volume to 1.5 PV, the high-permeability channel gets completely broken through, CO₂ is discharged through the channel, serious gas channeling occurs, and it hardly participates in oil displacement, resulting in a substantial decrease in oil recovery. The gas cap volume is further increased to 2.5 PV. With the increase of gas flow rate, the pressure difference between the inlet and the outlet of the 2D visualization plate model increases, providing a larger driving pressure difference, thereby increasing the oil recovery. Therefore, with the increase of the gas cap volume, the recovery factor shows a trend of first increasing, then decreasing, and then increasing, which proves that the oil recovery can be enhanced by capturing CO₂ to form a gas cap.