Fig. 2 compares the oil production degree after CO₂ huff-n-puff under different schemes. After the first cycle, the oil production degree in Scheme 2 is 9.89 percentage points more than that in Scheme 1, and the oil production degree in Scheme 3 is 5.60 percentage points more than that in Scheme 2. This indicates that increasing the time and area (fractures are developed in cores in Scheme 3) of contact between CO₂ and crude oil appropriately can rapidly enhance the production degree in the early stage of the huff-n-puff process. With the increase of huff-n-puff cycles, the gap between the oil production degree curves of Scheme 2 and Scheme 1 tends to be larger, indicating that increasing extraction time has a significant effect in the late stage of huff-n-puff. Moreover, the gap between the oil production degree curves of Scheme 2 and Scheme 3 increases slowly and then narrows, indicating that the fracture plays a greater role in the early stage.

Referring to the classification of pores by the International Union of Pure and Applied Chemistry (IUPAC) and combining with the characteristics of the pore structure of oil shale in the Jiyang Depression, pores are divided into micropores (<2 nm), mesopores (2-50 nm), macropores (50-1 000 nm), and bedding fractures (˃1 000 nm). Micropores and mesopores belong to nano-scale pores, while macropores and bedding fractures belong to micro-scale pores. Table 2 compares the production degree of crude oil in pores with different sizes after the same huff-n-puff cycle under different schemes. Pores in shales from the Jiyang Depression are mainly mesopores and macropores ^[23]. Data in the table 2 reveal that increasing the time of contact between CO₂ and crude oil can significantly improve the production degree of macropores. The presence of fractures connects some mesopores and macropores, greatly increasing the production degree of crude oil in these pores. As shown in Fig. 2, the oil production degree in Scheme 1 and Scheme 2 still increases apparently after the fourth cycle. It can be inferred that the active period of CO₂ injection for improving shale oil recovery is long, which is related to the micro- and nano-scale effect, kerogen, adsorbed oil and other characteristics.

The NMR T₁-T₂ spectrum can reflect the information such as the properties and mobility of fluid components. It is generally believed that a larger value of T₁/T₂ represents a heavier component of the corresponding hydrocarbon, and thus a worse mobility. The response regions of bound water and organic matter/heavy hydrocarbons on two-dimensional (2D) NMR spectra can be defined by low-temperature drying and high-temperature distillation of fresh oil shale core samples, and then the response regions of light hydrocarbons/mobile oil can be defined by oil-saturated core experiments. Fig. 3 shows the response regions of hydrocarbon and water signals in typical laminated shale cores in the Jiyang Depression, where region ① represents the signals of oil with heavy components in the cores, such as organic matter or heavy hydrocarbons, region ② represents the signals of clay bound water, and region ③ represents the signals of light hydrocarbons or mobile oil in the cores.

Cores in Scheme 3 were analyzed for microscopic production characteristics of crude oil. The core saturated with oil was put into a holder, CO₂ was injected, and the pressurized was increased step by step to 8, 12 and 15 MPa, when T₁-T₂ 2D NMR test was carried out separately. Fig. 3 shows 2D NMR spectra under different pressures. By comparing the differences in spectra, the variations of different types of fluids in the core can be determined. It can be observed from the changes in region ① that when the CO₂ pressure increases from 8 MPa to 15 MPa, the signal range and strength of fluid in region ① decrease gradually, indicating that the signals in this region gradually weaken, and the signals as a whole tend to move to the lower right, which suggests that the components in this region are changing. It is also observed that the color for 1<T₁/T₂<10 in region ③ gradually deepens, indicating that the fluid signals in this region are getting stronger, and the signal response begins to appear in the interval of 10<T₁/T₂<100. This shows that in the process of "huff", CO₂ contacts with heavy hydrocarbon components through diffusion, and gradually reduces heavy hydrocarbon components and increases light hydrocarbon components through mass transfer, with intermediate components occurred, which increases the total amount of mobile oil in the core to a certain extent. In the process of "puff", with the release of CO₂ pressure, this part of crude oil can be carried to nearby larger pores or bedding fractures, thereby improving the effect of CO₂ huff-n-puff. However, due to the scale effect, the diffusion and mass transfer of CO₂ in oil shale are slower than that in tight sandstone. Therefore, when designing the CO₂ injection scheme or implementing the on-site soaking, a proper extension of the soaking time is conducive to improving the shale oil development effect of CO₂ injection.

Fig. 4 compares the influences of CO₂ pre-pad and slickwater fracturing fluid on the mechanical properties of shale cores. Overall, whenever supercritical CO₂, liquid CO₂ or slickwater is used, the tensile strength of shale after immersion reduces significantly. At 48 h for slickwater, the reduction in tensile strength of shale slows down. For supercritical CO₂ and liquid CO₂, at 24 h, the tensile strength of shale almost remains unchanged. At 120 h, the average tensile strength of the core is 8.7 MPa for supercritical CO₂, 6.0 MPa for liquid CO₂, and 4.7 MPa for slickwater. Thus, supercritical CO₂, liquid CO₂ and slickwater ranks in an ascending order of influences on the tensile strength of shale cores. In general, injecting CO₂ pre-pad helps to maintain the tensile strength of reservoir rock.

By comparing the elastic modulus, Poisson's ratio, and uniaxial compressive strength of shale cores (Fig. 5), it can be seen that, at 120 h, the elastic modulus, Poisson's ratio, and peak strength of cores are similar for supercritical CO₂ and liquid CO₂, and the elastic modulus and peak strength of cores are significantly lower for slickwater than supercritical CO₂, while the Poisson's ratio is larger than that for supercritical CO₂. This indicates that the slickwater fracturing fluid can better significantly reduce the shale strength than CO₂ pre-pad. Injecting CO₂ pre-pad can effectively maintain shale brittleness and facilitate the formation of complex fracture network.

A three-dimensional (3D) fracture propagation model of shale was established using the ABAQUS finite element analysis software. The model is a cylinder with a diameter of 100 m and a height of 200 m. The model contains two parallel bedding planes, which are 30 m apart and located in the middle of the model (Fig. 6). The fracturing fluid injection point is located at the model center, and a constant-rate injection method is adopted. Meanwhile, cohesive elements with zero thickness are embedded inside the model to simulate the initiation of hydraulic fractures and bedding planes. The initial fluid saturation and porosity of the model are 100% and 20%, respectively. Considering the condition of hydrostatic excess pressure, a normal stress of 5 MPa is applied to the model surface. Pore fluid pressure at the upper, lower, or surrounding boundaries is set to 0. All external surfaces of the model are constrained by normal displacement, and rigid body displacement is not allowed on the bottom surface of the model. A displacement constraint of zero in the Z direction is applied on the bottom surface. Moreover, in order to eliminate deformation on the upper surface, Kinematic rigid displacement constraint is applied to all nodes on the upper surface. The critical damage displacement of the Cohesive element is set to 0.001 m. The basic input parameters of the model are shown in Table 3.

The CO₂ pre-pad generally remains in a supercritical state after entering into the formation, so the effects of supercritical CO₂ and slickwater on fracture initiation and propagation were simulated and compared. As shown in Fig. 7, the breakdown pressure of supercritical CO₂ is significantly lower than that of slickwater. The maximum injection pressure of supercritical CO₂ is 4.78 MPa, 36% lower than the breakdown pressure of slickwater which is 7.48 MPa. Low breakdown pressure helps to improve the complexity of fractures. The fracture width of supercritical CO₂ fracturing is significantly smaller than that of slickwater fracturing. The maximum fracture width of supercritical CO₂ fracturing is 7.2 mm, 58% smaller than that of slickwater fracturing which is 17.2 mm. The time for fracture to reach the boundary in the case of supercritical CO₂ fracturing is significantly shorter than that of slickwater fracturing. After reaching the boundary, the growth of the fracture volume slows down greatly. The induced fracture during supercritical CO₂ fracturing needs only 11.4 s to reach the model boundary, while the induced fracture during slickwater fracturing takes 25 s. The fracture in the supercritical CO₂ fracturing is narrower than that in the slickwater fracturing, but it propagates faster. Additionally, the supercritical CO₂ has strong diffusibility and zero interfacial tension, allowing it to easily filter along the bedding plane, which may induce shear failure of the bedding plane and thus increase the complexity of fractures. Overall, supercritical CO₂ is more helpful to improve the fracture complexity than slickwater, but at the same time, it can reduce the fracture conductivity and fracture length. Such adverse impacts can be overcome by subsequent injection of slickwater.

In order to simulate the influences of natural fractures on supercritical CO₂, a 2D finite element model of fracture propagation in shale with supercritical CO₂ fracturing was established using the ABAQUS software. A natural fracture auto-generation program was developed using Python language to randomly generate a primitive natural fracture system with a length of 2-10 m and a direction of 45° north by east. A pore pressure cohesive element with zero thickness was embedded inside the element to simulate the propagation path of induced fractures. In order to distinguish the difference in tensile strength between the matrix and the natural fractures, the cohesive element tensile strength of the matrix is set to be higher than that of natural fractures. Fig. 8 shows that supercritical CO₂ fracturing induces fractures which cannot expand further along the original direction when encountering natural fractures, but may divert to other directions, further increasing the complexity. Fig. 9 reveals that the induced fractures propagate in a clear step-like pattern in reservoirs with natural fractures. This is mainly due to the extremely low viscosity of supercritical CO₂, which makes it easier to open natural fractures. However, a certain pressure is required inside the fractures to overcome the tensile strength of the rock matrix for further propagation. Therefore, a high treatment rate should be chosen to improve fracturing efficiency when using supercritical CO₂ as fracturing fluid.

CO₂ pre-pad fracturing can efficiently replenish the reservoir energy, and also reduce the viscosity of crude oil and even produce a miscible effect, which helps to further improve the post-fracturing productivity and recovery of shale oil wells, especially in blocks with low formation pressure coefficients. CO₂ pre-pad fracturing provides a new approach and guidance for achieving economic recovery in shale oil reservoirs with normal pressure coefficients. In practices, the volume of CO₂ pre-pad determines the degree of energy replenishment and the sweep range, and its quantitative optimization is the focus of engineering design. Fracturing and reservoir simulation research was conducted using conceptual models, and an optimization method based on reservoir numerical simulation was established. Fracman and CMG simulation software were employed to achieve quantitative optimization of CO₂ pre-pad volume.

The Meyer software was used to establish a reservoir fracturing conceptual model based on a geological model of laminated shale in the target block. The model has a grid size of 305×263×115 and a total grid volume of 2.039×10¹⁰ m³. The elastic modulus of the target reservoir is set to be 26.19-40.28 GPa, and the Poisson's ratio is 0.235-0.279, with an average of 0.257. The stress of the target reservoir is ranked in a descending order as vertical stress, maximum horizontal principal stress, and minimum horizontal principal stress. The maximum horizontal principal stress is 67.41-69.8 MPa, the minimum horizontal principal stress is 57.93-58.94 MPa, the stress gradient is 2.08-2.11 MPa/100 m, and the horizontal stress difference is 8.91-9.64 MPa. The rock exhibits the characteristics of moderate elastic modulus, high Poisson's ratio, and low mechanical brittleness index. A pumping procedure of high-viscosity liquid pre-pad, rapid pumping rate increase, and gel injection was adopted to simulate the effective extension pattern of fractures in the length direction through multiple rock types, so that the combination of fracturing treatment parameters is optimized. With a comprehensive consideration to the cost of fracturing treatment, a horizontal wellbore length is set to 2 000 m, and the number of fracturing clusters is optimized to around 200. The half-length of the fracture is 180-220 m, the fracture height is 20-30 m, the conductivity is 5 μm²·cm, the single- cluster fracturing fluid volume is 300-400 m³, and the proppant volume is 25-35 m³.

A corresponding reservoir numerical model was established based on the conceptual geological model. For the target reservoir, the pressure coefficient is 1.05-1.35, the effective porosity is 3.2%-6.9%, the gas logging permeability is (17.9-81.7)×10⁻⁹ μm², the oil saturation is 48.4%-52.8%, the crude oil density is 0.860-0.864 g/cm³, and the crude oil viscosity is 18.2-32.6 mPa·s. The hydraulic fractures in horizontal wellbore are roughly coupled to the corresponding data field of the geological model through commercial software to achieve the characterization of the fracture system in the reservoir model, based on the optimized parameters such as fracture length, height, width, and conductivity. During fracturing operation, the pumping rates and injection volumes of CO₂ pre-pad and water-based fracturing fluid are simulated according to the method of the injection well. Based on the above model, the changes of energy replenishment range and pressure coefficient are investigated under the single-stage CO₂ pre-pad volumes of 25, 50, 75, 100, 150, 300, 450 and 600 t, respectively.

The change of energy replenishment range at different CO₂ pre-pad volumes can be calculated by counting the number of grids with pressure changes greater than or equal to 1 MPa. At the same time, the average pressure coefficient after energy replenishment can be calculated by the average pressure change amplitudes in grids with energy replenishment. Analyzed from the calculation results (Fig. 10), the range of energy replenishment and the average pressure coefficient after energy replenishment both show an increase trend as the CO₂ pre-pad volume increases. When the single-stage CO₂ pre-pad volume exceeds 100 t, the growth of the pressure coefficient tends to slow down, and the increase of the energy replenishment range tends to slow down, but at a rate smaller than the pressure coefficient. Accordingly, when the CO₂ pre-pad volume is lower than a certain amount, the increases in energy replenishment range and pressure coefficient are the dominant factors for efficiency enhancement; when the CO₂ pre-pad volume is higher than a certain amount, the influence of pressure coefficient increase gradually decreases, and the dominant factor gradually becomes the increase in energy replenishment range. As indicated by the energy replenishment range curve, the turning point of the single-stage CO₂ pre-pad volume is 100-150 t. When the CO₂ pre-pad volume exceeds the turning point, the energy replenishment effect gradually increases, but at a declined rate. Whether to further increase the CO₂ pre-pad volume needs to be determined through economic analysis.

Through reservoir numerical simulation, the 10-year cumulative oil production of single well was predicted at different single-stage CO₂ pre-pad volume (Fig. 11). When the single-stage CO₂ pre-pad volume is less than 100 t, the 10-year cumulative oil production fluctuates slightly, and basically remains at about 3.75×10⁴ m³. It is considered that, during the post-fracturing production, with the pressure decreasing, a part of the injected CO₂ is released from the crude oil, and an oil-gas-water three-phase flow zone is formed, which reduces the effective permeability of the oil phase and increases the flow resistance of the oil phase. When the CO₂ volume is small, the energy replenishing effect and flow resistance are basically balanced, and the production enhancing effect is not obvious. When the single-stage CO₂ pre-pad volume is greater than 100 t, the energy replenishing effect overrides the oil flow resistance in the three-phase flow zone, and the 10-year cumulative oil production increases gradually with the increase of CO₂ pre-pad volume. If the potential risks such as asphaltene precipitation during CO₂ injection are not considered ^[27], there is no obvious inflection point in the rising trend of 10-year cumulative oil production when the single-stage CO₂ pre-pad volume exceeds 100 t. On the other hand, through the comparative analysis of the impact of CO₂ pre-pad volume on daily oil production, it is found that the CO₂ pre-pad volume mainly affects the early daily oil production, and the impact of CO₂ pre-pad volume is basically ignored after more than one year of production. The optimized single-stage CO₂ pre-pad volume should exceed 100 t, and the economic evaluation should be performed depending on the local CO₂ price.

CO₂ pre-pad fracturing has been implemented in both shale gas wells and well groups in the Jiyang Depression, with ideal stimulation results obtained.

Well F159 is the first test well for CO₂ pre-pad fracturing in the Jiyang Depression. Its single-stage CO₂ pre-pad volume was 150 t. The well was put into production with small nozzle after backflow at 160 d of operation. When the cumulative oil production was 350 t, the tubing pressure remained at 6.7 MPa. When the cumulative oil production reached 2 000 t, the daily production rate kept stable at around 12 t, and the tubing pressure remained above 3 MPa, demonstrating a strong stable production capacity. The adjacent well F159-1 had the same operating horizon and pressure coefficient as Well F159, but it was not treated with the CO₂ pre-pad fracturing technology. For well F159-1, when the cumulative oil production exceeded 350 t, the wellhead pressure was insufficient for oil flowing; thus, mechanical pumping was adopted. When the cumulative oil production reached 1 400 t, the well was shut down due to insufficient formation energy. It can be seen that the CO₂ pre-pad fracturing is well performed in stimulation.

After the successful implementation of CO₂ pre-pad fracturing in single wells, the technology was tested in a well group FYP1 in the Boxing subsag of the Dongying Sag in the Jiyang Depression. The shales in FYP1 are buried at 3 130-3 650 m, with a maturity of 0.74%-0.83%. The main lithofacies is organic-rich laminated argilla-ceous limy shale, and the carbonate content reaches 70%. Early tests revealed the challenges in, for instance, proppant injection and fracture height extension during fracturing in limy shale due to presence of beddings, varying physical properties in horizontal sections, and other factors. Therefore, the concept of CO₂ pre-pad fracturing was fully incorporated into the fracturing design for FYP1. Eight horizontal wells in FYP1 were all treated by multi-stage subdivided volumetric fracturing. The CO₂ pre-pad volume was 100-300 t per stage, 2 500-7 000 t per well, and 4.1×10⁴ t in the well group. Other wells fractured early in this area were not treated using CO₂ pre-pad, and demonstrated the fracture pressure of 81-90 MPa and the average stimulated reservoir volume (SRV) of 39.0×10⁴ m³ per stage. In contrast, FYP1 employing CO₂ pre-pad fracturing exhibited the fracture pressure of 74-77 MPa and the average SRV of 41.2×10⁴ m³ per stage, which are 7-13 MPa lower and 2.2×10⁴ m³ higher than those of early wells. After coming into production, FYP1 achieved a peak daily oil production of single well up to 171 t. By February 2024, its cumulative oil production reached 13.7×10⁴ t.

From 2020 to 2023, 57 horizontal shale oil wells in the Jiyang Depression were treated with CO₂ pre-pad fracturing, with 32.5×10⁴ t CO₂ injected totally. Particularly, 21 wells achieved a peak daily oil production of over 100 t, and 19 wells witnessed the cumulative oil production surpassing 5×10⁴ t. It is proved that the CO₂ pre-pad fracturing technology allows for effective stimulation of shale oil reservoirs in the Jiyang Depression, which are characterized by deep burial, high formation pressure, high lime content, high fracture pressure, and difficulty in vertical fracture propagation. In detail, CO₂ functions for lime dissolution to expand the pore space, modification to increase fractures, dissolution to reduce the viscosity, swelling to replenish energy, and diffusion to displace oil, which will effectively reduce the fracture pressure, increase the SRV and enhance the fracture connectivity, ultimately improving crude oil fluidity and formation energy.

The component model was used to describe the fluid phase behaviors, and the embedded discrete fracture model was constructed to characterize the fracture network created by multi-stage fracturing. The numerical simulation model is 1 700 m×370 m×43 m in size, with 170×37×1 grids. The model contains 30 equally-spaced induced fractures (Fig. 12), as well as microfractures randomly generated. The shale oil reservoir has the matrix porosity of 5%, the induced fracture half-length of 150 m, the induced fracture permeability of 10 μm², and the microfracture permeability of 500×10⁻³ μm². It was assumed that CO₂ huff-n-puff is initiated at the 8^th year of elastic development, with a huff-n-puff cycle of 1 year, an injection time of 60 d, an injection rate of 200 t/d, and a soaking time of 30 d.

Based on the model above, and given the formation pressure of 45 MPa and C₁₁₊ content of 25% in shale oil, three models were designed with different matrix permeability (0.001×10⁻³, 0.005×10⁻³ and 0.02×10⁻³ μm²) to simulate the development effects of elastic development and CO₂ huff-n-puff. Fig. 13 compares the pressure field distribution under the two development modes. It can be seen that the reservoir pressure after CO₂ huff-n-puff is significantly higher than that after elastic development, indicating a remarkable effect of CO₂ injection for replenishing the formation energy. Such effect is varying at different matrix permeabilities of shale reservoirs. The lower the matrix permeability, the higher the reservoir pressure and the more the remaining oil at the later stage of development. As to EOR effect, the recovery is increased by 1.05, 1.42, and 1.55 percentage points, respectively, for the models in a descending order of matrix permeability. This suggests that the lower the matrix permeability, the greater the improvement in recovery.

Based on a model with a matrix permeability of 0.02×10⁻³ μm² and a C₁₁₊ content of 25%, the formation pressure was set at 30, 35, 40, and 45 MPa respectively to simulate the ultimate recovery under the modes of elastic development and CO₂ huff-n-puff. As shown in Fig. 14, for shale oil reservoir with low pressure, the production degree is low under the elastic development mode, and the increase in recovery is greater under the CO₂ huff-n-puff mode. Essentially, the higher formation pressure corresponds to larger elastically recoverable reserves, leading to a greater degree of production in the elastic development stage and leaving less recoverable reserves. Therefore, the extent of improvement in recovery in the CO₂ huff-n-puff stage is limited.

Based on a model with a matrix permeability of 0.02× 10⁻³ μm² and a formation pressure of 45 MPa, the C₁₁₊ content was set at 25%, 30%, and 35%, respectively, to simulate the ultimate recovery under the modes of elastic development and CO₂ huff-n-puff. As shown in Fig. 15, crude oil with a higher C₁₁₊ content exhibits a lower recovery and larger remaining recoverable reserves under the mode of elastic development, due to relatively high viscosity and high resistance to oil-phase flow. After CO₂ huff-n-puff is implemented, the mobile crude oil reserves are significantly increased through viscosity reduction, volume expansion, extraction, and miscibility, ultimately resulting in a substantial increase in the recovery.

Since 2019, field test of CO₂ huff-n-puff has been con-ducted in four slant wells in two types of shale oil reservoirs in Shengli Oilfield, China: the depression-stable shale oil reservoirs in the Niuzhuang Depression, and the complex fault-block shale oil reservoirs in the Boxing Depression.

Well GX26 was the first for CO₂ huff-n-puff test in the Niuzhuang Depression. The well encountered the 3th zone in the pure upper submember to the pure lower submember of the fourth member of Shahejie Formation of Paleogene (Es₄^cs-3-Es₄^cx). Shales in this interval are 51 m thick, with a mid-depth of 3 380 m, and the maturity of 0.64%. During the two-stage fracturing, 320 t CO₂ and 2 382.7 m³ fracturing fluid were injected in the entire wellbore, and 65.4 m³ proppant was injected. During the flowing stage, the peak daily fluid production was 46 m³, the peak daily oil production was 27.5 t, and the peak water cut was 40.2%. After the cumulative fluid production reached 1 784 m³, oil production was implemented by pumping instead of flowing. Production continued with low fluid rate and low water cut for a long time, corresponding to an average daily fluid production of 3.3 m³, an average daily oil production of 2.3 t, and an average water cut of 30%. When the huff-and-puff test was conducted, the cumulative fluid production already reached 5 954 m³. Assuming that the formation pressure was restored to the post-fracturing level (pressure coefficient 2.0), the CO₂ injection was designed with a huff-n-puff volume of 2 700 t and a rate of 100 t/d. After CO₂ injection, 50 m³ water was injected to prevent CO₂ spillover. After the operation, the actual CO₂ injection time and volume were 26 d and 2 724 t, respectively. Upon the injection, the formation pressure coefficient increased from 1.4 to 2.1, effectively replenishing the formation energy. After the well was put into production, the initial wellhead tubing pressure was 34 MPa. After stable flowing, the CO₂ huff-and-puff effectively facilitated the enhanced oil recovery and the water cut reduction, with the daily fluid production increased to over 6 t, and the water cut dropped to 2% and remained stable. During the plateau period, the daily oil increment reached 6 t. Moreover, ground sampling showed significant changes in fluid properties. The crude oil density decreased from 0.893 1 g/cm³ to 0.840 2 g/cm³, and the viscosity of the crude oil at 50 °C decreased from 28.7 mPa·s to 7.37 mPa·s. After the injection of CO₂, the crude oil fluidity improved significantly.

Well F159-1 was the first for CO₂ huff-and-puff test in the Boxing Depression. This well encountered Es₄^cs, where the shales are 51 m thick with a mid-depth of 3 185 m. During the two-stage fracturing, 600 t CO₂ and 23 131.9 m³ fracturing fluid were injected in the entire wellbore, and 137.9 m³ proppant was injected. When the cumulative fluid production was 2 417 m³ and the cumulative oil production was 1 012 t in the flowing stage, the wellhead tubing pressure was only 0.8 MPa. It was decided to conduct the CO₂ huff-and-puff test before the well stopped flowing. The designed CO₂ huff-and-puff volume was 4 000 t, with an injection rate of 100 t/d, followed by 200 m³ water injected to prevent CO₂ spillover. It was calculated that the formation pressure coefficient could be restored to around 1.8. During the test, the pump pressure was 26.73 MPa and the daily injection rate was 80 t in the early stage of CO₂ injection. At 10 d, the pump pressure increased to 29.1 MPa, and the daily injection rate dropped to only 9 t. Due to the limitation of casing pressure, a total of 881 t CO₂ was injected. To improve the energy replenishment effect, an additional water slug of 2 016 m³ was injected, and the well was shut in for 133 d. When the wellhead tubing pressure slowly decreased from 25.5 MPa to 20.29 MPa, fluid drainage began. The daily oil production steadily increased from 1.0 t before the CO₂ injection to 4.8 t, and then gradually decreased. Before the huff-and-puff operation, the water cut of the well was 65%. After soaking and putting into production, the water cut quickly rose to around 98%. As the oil production continued to increase, the water cut gradually decreased to around 45% after approximately half a month, indicating an effective extension of the flowing period (Fig. 16). The increase in production was very significant.