Study on the production law and optimization parameters of CO2 huff ‘n’ puff for continental shale oil

In this study, nuclear magnetic resonance and computed tomography scanning were used to analyze the production law and displacement mechanism of Jimusaer continental shale oil during CO2 huff ‘n’ puff, and the optimal parameters were determined. The results indicated that CO2 huff ‘n’ puff mainly produces crude oil in pore throats with 0.1–1 μm radii, while crude oil in pore throats with radii below 0.1 μm cannot be produced. Multiple CO2 huff ‘n’ puff cycles can connect fluids in fractures with fluids in large–medium pore throats, eliminate fracture effects on the oil recovery factor, and achieve coefficient development of both fractured and unfractured shales. In the CO2 huff ‘n’ puff process, core pressure change could be divided into three stages of injection–holding–depletion, and the oil displacement mode was the piston type. The study of huff ‘n’ puff parameters revealed that huff ‘n’ puff cycles and injection pressure have a great influence on the CO2 huff ‘n’ puff efficiency, while the injection timing and soaking time imposed relatively small effects. For continental shale in the Jimusaer Sag, the optimal CO2 huff ‘n’ puff parameters are five cycles, 4‐MPa injection timing, 25‐MPa injection pressure, and 12‐h soaking time.


| INTRODUCTION
][3] In recent years, countries worldwide have achieved great efforts to solve this global problem, and China has proposed the dual carbon goals of achieving carbon peaking by 2030 and carbon neutrality by 2060. 4 There are many factors influencing carbon emissions, and one of the most important factors is fossil fuels. 56][7] Compared with other countries, China's fossil fuel use accounts for a large proportion of its energy mix, which poses a serious challenge to achieve the abovementioned dual carbon goals.Under these conditions, low-carbon technology development in the energy sector is particularly important.Currently, carbon capture, utilization, and storage technology has become an internationally recognized and effective technology for carbon emission reduction and is expected to contribute approximately one-third to the total carbon emission reduction by 2050. 3,8,93][24] Compared with other gases, CO 2 exhibits the characteristics of easy compression, low critical point, and high solubility, which enables the injected CO 2 to achieve miscibility with crude oil, extract light hydrocarbons, reduce the viscosity, increase the volume, and thus improve its movability; at the same time, CO 2 can alter the wettability and interfacial tension and enhance the flow ratio of crude oil, thus reducing the remaining oil saturation in the reservoir and achieving the purpose of EOR. 4,9,10,12,17,22,25Since its introduction, CO 2 huff 'n' puff technology has received much attention from scholars worldwide.][28][29][30][31] Compared with the marine shale occurring in North America, China mainly contains medium-to highmaturity continental shale, which is characterized by notable heterogeneity, high concentration of heavy components, high water content, and easy gas channeling, resulting in poor fluid mobility, low oil recovery, and considerable development difficulty. 32Therefore, China must still further study the production law of crude oil in continental shale and determine the optimal CO 2 huff 'n' puff parameters.At present, CO 2 huff 'n' puff has been pioneered and promoted in several major oilfields across China, such as Daqing, Xinjiang, Jilin, and Shengli, and has mainly been applied in low-permeability reservoirs. 4,16,33,34Zhu et al. 35 simulated the influence of different gas injection parameters on the production degree of tight oil in the Yanchang oilfield based on a long core displacement device 36 ; investigated the action mechanism of CO 2 in the ultralow-permeability reservoir of the Gaoshangbao oilfield by establishing a core-scale matrix-fracture seepage model.However, there remain certain problems in the study of CO 2 -EOR.On the one hand, the current research objects mostly include marine shale, and there are few reports on continental shale.On the other hand, the experimental methods are relatively simple, and they therefore fail to comprehensively characterize the oil production law in shale and determine relevant parameters.
In this paper, continental shale of the Permian Lucaogou Formation in the Jimusaer Sag of the Junggar Basin (Xinjiang, NW China) was selected as the research object, and two experimental methods, namely, nuclear magnetic resonance (NMR) and computed tomography (CT) scanning, were used to monitor each stage of huff 'n' puff experiments, analyze the oil production law of CO 2 huff 'n' puff in continental shale, visualize the huff 'n' puff efficiency, and optimize huff 'n' puff parameters.The Jimusaer Sag, an important production area of unconventional oil and gas in China, is rich in continental shale oil.However, due to technical limitations, it is mainly developed via depletion. 37According to data of the Changji Oilfield, the tight oil of the Lucaogou Formation produced through vertical wells generally follows a hyperbolic decline pattern, with an annual absolute oil production decline rate ranging from 39.7%-50.5%,averaging 46.2%, 38 which indicates that it is important to supplement the formation energy via the huff 'n' puff process.Therefore, this study is of great significance for EOR application to the Jimusaer shale and can further provide technical support for the effective development of continental shale in other regions, promote low-carbon development of oilfields, and accelerate the realization of the dual carbon goals.

| GEOLOGICAL SETTING
The Jimusaer Sag is located in the eastern part of the Junggar Basin, which is a secondary sag of the eastern uplift of the first-order tectonic unit of the basin, with a typical west-to-east superskid structure.From bottom to top, Carboniferous, Permian, Triassic, Jurassic, Cretaceous, Paleogene, and Neogene formations are developed in this area.Among them, the Permian Lucaogou Formation is simple in structure, with an overall west-dipping monocline that is high in the east and low in the west, belonging to a typical continental saline lake basin sedimentary system, exhibiting a three-source mixed sedimentary pattern, mainly including saline lake facies and delta facies, with a stratigraphic thickness ranging from 35 to 300 m, averaging 200 m, thinning toward the periphery.The formation is vertically divided into two sections, and the main lithology comprises shale, lithic feldspathic siltstone, doloclast sandstone, and dolomitic sandstone.The Lucaogou Formation is the main source rock of the sag and the main shale oil reservoir.
Core analysis data reveal that the Jimusaer continental shale oil reservoir is tight overall, with multiscale pore throats developed, dominated by nanoscale pore throats, and the overall permeability is less than 0.1 mD.The pore types of the Lucaogou shale mainly include inter-and intragranular dissolution pores, which are the main oil occurrence spaces, accounting for more than 85% of the total pore types.Intergranular pores are the next most abundant pores, mainly dolomite intergranular pores, while microfractures are underdeveloped, accounting for the lowest proportion (Figure 1A-D).Shale reservoirs typically exhibit millimeter-, micrometer-, and nanometer-scale pore throats.Among them, nanoscale pore throats account for more than 75% of the total reservoir space, with a suitable connectivity of pore throats and a mercury saturation higher than 80% (Figure 1E).Microscale pore throats account for approximately 20% of the total reservoir space, largely comprising intergranular dissolved pores (Figure 1F).0][41] The proportion of pore throats of different scales and the (f) intergranular pores, nanoscale pores (modified from Tan et al. 39 ).
occurrence state of crude oil determine the final recovery efficiency of different displacement media and oil displacement methods.

| Experimental materials and instruments
Materials: A full-diameter shale core retrieved from the sweet spot of the Lucaogou Formation in the Jimusaer Sag was selected, and two plunger cores with a diameter of 2.5 cm were drilled as experimental samples, one of which comprised an unfractured shale core collected from the upper sweet spot (Figure 2A).The other core was a shale core obtained after fracturing via the Brazilian fracturing experimental method, which was retrieved from the lower sweet spot (Figure 2B).Under the experimental conditions, the viscosities of crude oil and 5# white mineral oil are 352.2 and 4.1 mPa s, respectively, and the densities of the two are similar, while under the high-temperature and high-pressure conditions of the formation, the viscosity of the crude oil is only 10.4 mPa s.Therefore, to ensure the accuracy of the experimental results, this time, the crude oil and the 5# white mineral oil are mixed with a ratio of 1:50 to form a simulated oil, and this simulated oil is able to present the same properties under the experimental conditions as in the formation case.The simulated oil was used as the experimental oil in the huff 'n' puff experiments, and CO 2 with a 99.99% purity was prepared as the huff 'n' puff medium.
Instruments: A MacroMR12 NMR core analyzer (Figure 2C), Quzix 5200 constant-pressure metering pump (the injection system), Isco constant-pressure confining pump (the confining pressure control system), core gripper, intermediate container, and so forth, were employed in the experiments.The basic test parameters were as follows: the echo spacing was 0.1 ms, the waiting time was 10 s, the echo number was 2048, and the scan time was 64.

| Steps
(1) Rock sample treatment: The rock sample was placed in a constant-temperature drying oven at 105 ± 2°C after oil and salt removal through washing, 42 and the dry weight of the rock sample was measured.If the dry weight of the rock sample differed by less than 0.001 g after three measurements, it could be considered that the rock sample was completely dried.( 2) Rock sample saturation: The dried rock sample was subjected to vacuum conditions for 24 h, then pressurized to 15 MPa for saturation with the simulated oil and maintained for 12 h.After the rock sample was completely saturated, it was removed to measure the wet weight of the rock sample.The saturated oil volume is obtained from the ratio of the mass difference between the rock sample before and after saturation to the density of the simulated oil.
The rock sample fully saturated with the simulated oil was placed in an incubator for 10 h at 35°C, and the NMR response characteristics of the rock sample in the saturated state were then measured.(3) NMR measurement before huff 'n' puff application: The NMR parameters were set, and the saturated rock sample was placed in the NMR instrument for T 2 spectrum measurement.(4) Huff 'n' puff operation process: The core was placed in the core gripper, a confining pressure of 18 MPa was applied, one port of the gripper was closed, and CO 2 was injected at 15 MPa through the other port.After the injection pressure remained stable, the injection port was closed, the pressure was held, and the soaking process was simulated.After 12 h, the injection port was opened.At this time, the injection port functioned as the outlet port.The port was connected to the atmosphere.The injected gas and simulated oil flowed out of the port.A metering device was used for metering.After a certain period, if the liquid was no longer produced, it could be considered that a huff 'n' puff cycle was completed.
The oil output was recorded, and the oil recovery was calculated.( 5) NMR measurement after huff 'n' puff operation: The core was removed from the gripper, the same parameters were set as before huff 'n' puff application, the rock sample was placed in the NMR instrument for testing, and the T 2 spectrum of the sample core was obtained after huff 'n' puff application.( 6) Multiple huff 'n' puff cycles: Steps (3)-( 5) were repeated to obtain NMR experimental results after multiple huff 'n' puff cycles with CO 2 as the huff 'n' puff medium.
A schematic diagram of the oil displacement system for core samples is shown in Figure 3.

| Experimental materials and instruments
Materials: To simulate the huff 'n' puff process in the Jimusaer continental oil shale more realistically, a square core with a cross-sectional area of 4.5 cm × 4.5 cm (Figure 4A) was used as the experimental sample, and CO 2 with a 99.99% purity was prepared as the huff 'n' puff medium.
Instruments: A rock CT scanning displacement system (Figure 4B), core gripper (Figure 4C), Quzix 5200 constant-pressure metering pump, Isco constantpressure confining pump, backpressure pump, intermediate container, and so forth, were employed in the experiments.Among the components, the rock CT scanning displacement system was modified with the medical CT scanner of the China Petroleum Exploration and Development Research Institute.The scan accuracy is 0.1 mm.There are eight spherical tubes emitting Xrays.The maximum transmitting power of each spherical tube is 53.2 kW, the maximum scan voltage reaches 140 kV, and the minimum scan layer thickness is 0.625 mm.
F I G U R E 3 Schematic diagram of the oil displacement system for cores.
Since the development process of the Jimusaer continental shale first entails exploitation via natural energy depletion and then supplementation through CO 2 huff 'n' puff exploitation, to simulate the shale development process more realistically, a primary production simulation experiment was conducted before the huff 'n' puff experiment.
(2) Dry scanning of the rock sample: The core was placed in the gripper, and a confining pressure of 18 MPa was applied.The position of the gripper was adjusted to obtain the best scanning result, and the gripper was fixed.The CT scanning system was used to scan the core samples.After preheating the ball tube, the core scan area was accurately selected, and the position coordinates were recorded to ensure that the second scan was conducted at the same position.The scanning parameters were set and recorded.The bed height was 209, the scan mode was the helical 0.5 full mode, the scan area ranged from I9.250-I109.250,the number of images was 81, the scan interval was set to 1.25, 3.75, and 0.75:1, the tube voltage reached 120 kV, the tube current was 150 mA, the scan field was 9.6 cm, and the scan time was 14.3 s.This scan was recorded as the dry scan model.(3) Core saturation: A vacuum pump was connected with the gripper, and the rock sample was vacuumed via a connecting pipeline for 24 h.Then, the rock sample was saturated with simulated oil.As live oil contains dissolved gas, direct saturation could lead to the release of dissolved gas.Therefore, first, gripper B port was opened to pressurize the rock sample saturated with dead oil to 15 MPa, and the saturated oil volume was recorded after the pressure stabilized.Then, the dead oil was replaced with live oil, the ports were switched, and live oil was injected at a constant pressure.The pressure of the backpressure pump was set to 5 MPa, and gripper port A was opened for stable single-phase seepage.The pressure of the backpressure pump was gradually increased to 15 MPa, and the pressure was increased in steps of 1 MPa at 2-h intervals to ensure that all dead oil was replaced by live oil to the greatest extent.The final state entailed an injection pressure of 15 MPa, the backpressure pump was pressurized to 15 MPa, the two ports of the core gripper were closed after liquid was no longer produced from the outlet, and the gripper was maintained for 48 h.(4) Wet core scanning: After bulb preheating, CT scanning of the oil-saturated core was conducted according to the dry scanning parameter settings, and this scan was recorded as the wet scan model.(5) Primary production simulation: Gripper A port was opened, the outlet pressure of the backpressure pump was set to 5 MPa, and the liquid output was measured at the same time.When liquid production from the outlet port ceased, the oil output at the failure stage was recorded, and the recovery degree at the failure stage was calculated.(6) Huff 'n' puff simulation: CO 2 was injected via gripper A port under a pressure of 15 MPa, port A was closed after the injection pressure remained stable, the injection volume was recorded, the pressure was maintained, and the soaking process was simulated.
After soaking the sample for 12 h, the bulb was preheated, core CT scanning was performed according to the same parameter settings, and this scan was recorded as the intermediate model after the first huff 'n' puff cycle.After scanning, the A port was opened, the outlet pressure of the backpressure pump was set to 5 MPa, the liquid output was measured at the same time, the oil output at the huff 'n' puff stage was recorded, and the recovery degree A schematic diagram of the oil displacement system for core samples is shown in Figure 3.

| Parameter optimization
On the basis of the actual development process of the Jimusaer shale, combined with the huff 'n' puff mechanism and production law determined via laboratory experiments, four parameters greatly impacting the huff 'n' puff process were selected, and corresponding experiments were designed to determine the optimal parameters.The four huff 'n' puff parameters included the injection timing, injection pressure, soaking time, and number of huff 'n' puff cycles.The injection timing is expressed in terms of pressure values, when the pressure depletion from 15 MPa to a certain pressure, then begins to inject the huff 'n' puff media, the certain pressure value is referred to as the injection timing in this paper.The oil exchange ratio is the ratio of the volume of produced oil to the volume of injected gas, which is an important indicator to evaluate the oil displacement effect of huff 'n' puff media.In the experiment, the oil recovery and gas injection amounts were measured, and the oil recovery factor and oil exchange ratio were finally calculated.These two indicators could be used to evaluate the CO 2 huff 'n' puff efficiency.The equations are as follows.
where RF is the oil recovery factor, %; V o is the volume of oil produced, mL; and V s is the volume of oil in the initial saturated state of the core, mL.
where ER is the oil exchange ratio, %; V o is the volume of oil produced, mL; and V g is the volume of gas injected, mL.
The injection timing to start injection after depletion was set to 9, 7, and 5 MPa, the injection pressure was selected as 18, 15, and 12 MPa, the soaking time was set to 12, 6, and 2 h, and the number of huff 'n' puff cycles was determined as the number of cycles needed until the recovery ratio basically remained unchanged (Figure 5).The experimental process followed the principle of controlling the process variables.When studying the influence of a given huff 'n' puff parameter on the recovery degree, we adjusted the value of this parameter and maintained the remaining parameters unchanged.The multiple huff 'n' puff cycles results obtained by a F I G U R E 5 Flowchart of parameter optimization design.
7-MPa injection timing, 15-MPa injection pressure, and 12-h soaking time are considered as the control group, which is the first to be completed.The oil recovery factor and oil exchange ratio obtained under parameter variation were compared with those obtained in the control group to determine the optimal CO 2 huff 'n' puff parameters.

| Oil production law of the Jimusaer continental shale
The obtained NMR T 2 spectrum revealed that the cores of the Lucaogou Formation basically exhibited a unimodal distribution, which changed into a bimodal distribution in the presence of artificial fractures (Figure 6).The change in pore structure led to a difference in the T 2 spectrum, which in turn affected the efficiency of the CO 2 huff 'n' puff process.According to the NMR response mechanism and capillary pressure theory, there exists a certain conversion relationship between the T 2 spectrum and capillary pressure curve.Wells with NMR and mercury injection data were selected, and a relationship between the NMR T 2 relaxation time and capillary pressure was proposed and established.
From the theory of capillary pressure, the relationship between capillary pressure and pore-throat radius is shown in Equation (3).
where δ is the fluid interfacial tension, dyn/cm; θ is the wetting contact angle, °; r c is the capillary radius, μm.
For the air-mercury system, δ = 49.44 N/cm 2 , θ = 140°, and substituting into Equation (3) and omitting the negative sign, we have Equation (4).When the surface relaxation mechanism plays a major role, the T 2 distribution can be utilized to evaluate the pore-throat size and its distribution.The surface relaxation equation is Equation (5).
where ρ 2 is the rock transverse surface relaxation rate, μm ms −1 , and F s is the geometry factor, which is equal to 3 for spherical pores and 2 for columnar pore throats.
Combining Equations ( 4) and ( 5) leads to Equation (6). where By finding the conversion factor C, the relationship between the NMR T 2 relaxation time and the capillary pressure P c can be established, and then the pore-throat radius corresponding to different T 2 relaxation times can be obtained through Equation ( 5) (Table 1).
Combined with the NMR T 2 spectra (Figure 6), the pore size distribution of the Jimusaer continental shale corresponded to a relaxation time ranging from 2 to F I G U R E 6 NMR results and oil recovery factor curves of the core samples after CO 2 huff 'n' puff: (A) unfractured shale and (B) fractured shale.NMR, nuclear magnetic resonance.1000 ms, which basically covered the low-velocity nonlinear flow to pseudolinear flow interval, with a few stagnant flow states.The main pore-throat distribution corresponded to a T 2 relaxation time ranging from 10 to 100 ms, which corresponds to a radius ranging from 0.1 to 0.5 μm, indicating that the fluid flow pattern in the Jimusaer continental shale reservoir mainly comprised low-velocity nonlinear flow, and the pore-throat scale mainly occurred within the nanoscale to micron range.
In regard to unfractured shale, during CO 2 huff 'n' puff, the NMR curve with a T 2 relaxation time ranging from 10 to 200 ms greatly changed, that is, CO 2 huff 'n' puff mainly produced crude oil in microscale pore throats with a radius ranging from 0.1 to 1 μm, while crude oil in nanoscale pore throats with a radius smaller than 0.1 μm was basically not produced (Figure 6A).In the first huff 'n' puff cycle, CO 2 injection resulted in the production of a large amount of crude oil in large-medium pore throats; in the second huff 'n' puff cycle, under CO 2 injection, crude oil in large-medium pore throats was still produced, but the oil recovery factor in large pore throats declined, mainly because the crude oil occurring in medium pore throats flowed into large pore throats, resulting in some residual oil remaining in the large pore throats after huff 'n' puff.After three huff 'n' puff cycles, the final oil recovery factor reached 37.6%, indicating that CO 2 huff 'n' puff could effectively enhance the production degree of shale oil.Considering continental shale oil, overall, the CO 2 huff 'n' puff process mainly produced crude oil in large-medium pore throats, while crude oil in small pore throats was basically not produced, and the huff 'n' puff efficiency declined with increasing number of huff 'n' puff cycles.
In regard to fractured shale, CO 2 huff 'n' puff mainly produced crude oil in fractures, while crude oil in large and medium pore throats was less notably produced (Figure 6B).The main reason for this phenomenon is that the presence of fractures could lead to changes in the main seepage path of gas, and serious gas channeling occurred in the huff 'n' puff process.When CO 2 is injected into fractured shale, the huff 'n' puff medium is more likely to enter fractures, functioning as relatively large-scale seepage channels with a relatively low seepage resistance, resulting in CO 2 huff 'n' puff mainly producing crude oil in fractures.With an increasing number of huff 'n' puff cycles, the continuous change in the reservoir pressure field could lead to increasing connectivity between fractures and adjacent large to medium pore throats, forming new seepage channels.In the subsequent huff 'n' puff cycle, part of the injected CO 2 could enter the shale pore throats through these newly formed seepage channels, producing part of the crude oil occurring in the large-medium pore throats.By comparing the experimental results between the unfractured and fractured shale cores, it could be found that under the conditions of multiple huff 'n' puff cycles, the ultimate oil recovery factor of fractured shale reached 37.2%, which is only 0.4% lower than that of unfractured shale.The effect of the CO 2 huff 'n' puff process on the EOR degree for both types of shales was basically the same.Therefore, considering the Jimusaer continental shale, multiple CO 2 huff 'n' puff cycles could reduce the impact of fractures on the final recovery and achieve coefficient development of both fractured and unfractured shales.

| CO 2 huff 'n' puff mechanism of the Jimusaer continental shale
The analysis of the pressure values at both gripper ports in the huff 'n' puff process indicated that the pressure changes during each huff 'n' puff cycle could be mainly divided into three stages of injection-holding-depletion (Figure 7), and the pressure changes at the different stages indicated different seepage characteristics of the huff 'n' puff medium and pore crude oil in the shale samples.
(1) Injection stage: The most obvious feature of this stage is that the pressure at the gripper port rapidly increased.When the huff 'n' puff medium was injected into gripper port A, it quickly entered the shale core at the gripper from the pipeline under a high-pressure gradient.The pressure at port A rapidly rose.Upon entering the core, the flow rate decreased, and the injected gas slowly diffused throughout the core.After a certain period of seepage, the pressure was transmitted to the port of the core gripper.As a result, the pressure increase at port B lagged behind that at port A.
(2) The pressure-holding stage is also denoted as the soaking stage: When the injection pump pressure remained constant, the A port was closed, and pressure holding was initiated.At this time, the injected gas medium had not yet fully diffused in the core to yield a stable state, resulting in the pressure at port A exceeding that at port B. With gradual diffusion of CO 2 throughout the core, the gas medium finally stabilized across all pore throats, and the pressure at both ports reached equilibrium and remained stable until the end of the soaking time.
(3) Depletion stage: After maintaining the pressure for a set time, port A was opened for the depletion process.The pressure at port A rapidly declined to the initial set value.At this time, the crude oil in the shale core gradually flowed toward port A and was produced under the effect of the pressure gradient, and the pressure at port B slowly declined to the initial setting value, with a certain time lag over port A.
This occurred because crude oil is less mobile than gas in tight shale, and the time needed to reach a stable pressure at port B is thus slightly longer than the injection time needed to reach the set pressure, but both are much shorter than the soaking time.
Given the same sample, multiple cycle huff 'n' puff cycles could alter the pressure field distribution characteristics of the gas medium within the core, which in turn could affect the resultant fluid diffusion paths and seepage channels to achieve the purpose of EOR.In regard to the Jimusaer continental shale core, under the same experimental conditions, after nine huff 'n' puff cycles, the oil recovery factor gradually increased at first and then remained basically unchanged.During the first five cycles, with increasing number of huff 'n' puff cycles, the oil recovery factor significantly increased, with an overall increase of 62.66%.The oil recovery factor during the last four cycles changed more slowly with increasing number of huff 'n' puff cycles, increasing only 5.12% (Table 2).After nine huff 'n' puff cycles, the final recovery ratio reached 73.57%, indicating that the CO 2 huff 'n' puff process could provide a favorable application effect on EOR.
The CT scanning technique was used to visualize the crude oil distribution in pore throats under the different CO 2 huff 'n' puff cycles and to clarify the production law of fluid in the shale pore throats.In the CT scanning oil saturation image of the core (Figure 8), the red part represents the crude oil, and the blue-green part indicate the gas medium.In this experiment, the CO 2 huff 'n' puff process was applied in saturated cores.Therefore, the reduction in oil saturation during huff 'n' puff reflects the enhancement in the oil recovery factor.The CT scanning results after seven huff 'n' puff cycles revealed that the oil recovery process of shale basically followed the pistontype oil displacement mode.During the first huff 'n' puff cycle, the crude oil occurring in the pore throats near the port was mainly produced, and the oil recovery factor of the core increased by 24.9%.During the second huff 'n' puff cycle, the crude oil in the pore throats within 1/3 of the volume was largely produced, and the oil recovery factor of the core increased by 41.2%.The third and fourth huff 'n' puff cycles could produce crude oil in the pore throats within 1/2 of the volume, and the oil Oil production (mL) recovery factor increased by 55.7% and 63.4%, respectively.The fifth and sixth huff 'n' puff cycles could produce crude oil in the pore throats within 2/3 of the volume, and the oil recovery factor increased by 68.4% and 71.2%, respectively.With increasing number of huff 'n' puff cycles, the volume of movable crude oil changed very little, the enhancement in the oil recovery factor also remained basically unchanged, and the huff 'n' puff efficiency tended to remain stable.CO 2 huff 'n' puff could significantly enhance the shale oil recovery, and the oil displacement mechanism is closely related to the physicochemical properties of shale oil.When the huff 'n' puff medium enters the shale, the CO 2 dissolved in crude oil generates the effect of expanding the volume and increasing the formation energy, which could alter the occurrence state of crude oil in the pore throats and the fluid pressure gradient field distribution, which could facilitate effective seepage of crude oil in the micropore throats, thus increasing the crude oil production degree.4][45][46] With increasing pressure, CO 2 and crude oil were produced as a condensate gas phase rich in CO 2 , forming a dissolved gas drive mechanism and finally achieving the development purpose of notable EOR.The pressure in this experiment remained below the CO 2 miscibility pressure, and the effect of CO 2 miscibility on the recovery was therefore not considered.If CO 2 huff 'n' puff development were conducted under reservoir conditions, due to the rise in pressure and temperature, the huff 'n' puff efficiency would be better under the influence of miscibility.Multiple factors affect the oil displacement efficiency of the huff 'n' puff process, not only the nature of the injected gas but also external injection parameters, mainly including the number of huff 'n' puff cycles, injection timing, injection pressure, and soaking time.To achieve the purpose of economic and effective EOR applications, it is necessary to optimize various injection parameters to guide efficient oilfield development.
In this study, two indicators, namely, the oil recovery factor and oil exchange ratio, were used to evaluate the effects of different parameters on the oil displacement efficiency.The changes in the curves depicted in Figure 9 indicated that under the conditions of the different injection timings, injection pressures, and soaking times, with an increasing number of CO 2 huff 'n' puff cycles, the oil recovery factor first increased and then tended to remain stable, while the reservoir oil exchange ratio exhibited the opposite behavior, revealing a changing trend of first decreasing and then stabilizing.The inflection points of both curves occurred at five huff 'n' puff cycles.
Under the conditions of the different injection timings and soaking times, the huff 'n' puff efficiency gradually stabilized with an increasing number of huff 'n' puff cycles, and the final oil recovery factor was similar, which indicates that the injection timing and soaking time imposed a relatively limited influence on the final CO 2 huff 'n' puff efficiency (Figure 8B,D, respectively), while the final oil recovery factor corresponding to the different CO 2 huff 'n' puff cycles and injection pressure greatly varied, indicating that both quantities notably influenced the CO 2 huff 'n' puff efficiency (Figure 8A,C, respectively).The oil exchange ratio only indicated a sharp decrease at first with an increasing number of huff 'n' puff cycles and then tended to remain stable.The other three parameters slightly influenced this index.The oil exchange ratio under the different injection timings, injection pressures, and soaking times tended to remain consistent after five huff 'n' puff cycles, with the difference not exceeding 1%.After nine huff 'n' puff cycles, the difference in oil exchange ratio did not exceed 0.2%.
The obtained CT scanning oil saturation images of the cores could visualize the experimental results and more clearly and realistically reflect the influence of each parameter on the CO 2 huff 'n' puff efficiency.As shown in Figure 8, with an increasing number of CO 2 huff 'n' puff cycles, the oil recovery factor gradually increased.The oil displacement efficiency after the first five cycles was notable, the oil saturation rapidly decreased, and the oil recovery factor rapidly increased.After the fifth cycle, the oil saturation change was slight, the oil displacement efficiency worsened, and the oil exchange ratio rapidly decreased.The oil exchange ratio after the fifth cycle reached only 3.5%.Comparing the CT scanning oil saturation images after nine huff 'n' puff cycles under the different injection timings, injection pressures, and soaking times (Figure 10), it could be observed that the higher the injection pressure was, the higher the final oil displacement efficiency.With every 3 MPa increase in the injection pressure, the final oil recovery factor increased by approximately 12%.Under an injection pressure of 18 MPa, the oil recovery factor increased by 85.6%.Under the condition of an injection pressure higher than the saturation pressure, the final huff 'n' puff efficiency under the three injection timings basically remained the same, but there were still slight differences.The later the injection timing was, the lower the associated depletion pressure and the higher the oil recovery factor.The final oil recovery factor increased by approximately 0.2% for every 2 MPa reduction in the depletion pressure.Under an injection timing of 5 MPa, the oil recovery factor increased by 73.8%.The longer the soaking time, the fewer huff 'n' puff cycles were required to achieve the best effect.The shale oil recovery factor first increased with increasing soaking time and then gradually stabilized.After 12 h of soaking, the shale oil recovery factor increased by 74.6%.
To further clarify the influence of the different parameters on the CO 2 huff 'n' puff efficiency and determine the optimal displacement parameters, based on nine huff 'n' puff experiments, we continued to increase the number of huff 'n' puff cycles and analyzed the recovery change rule.Considering the cost, five cycles still represented the optimal number of huff 'n' puff cycles (Figure 11A).Maintaining the other experimental conditions unchanged, two additional huff 'n' puff experiments were performed with injection timings of 11 and 13 MPa, and predictions were obtained.The results demonstrated that the oil recovery factor gradually increased with decreasing injection timing on the premise that the pressure exceeds the saturation pressure of crude oil.Therefore, the pressure associated with the optimal injection timing for the Jimusaer continental shale is slightly higher than the saturation pressure, with a value of approximately 4 MPa (Figure 11B).In addition, maintaining the other experimental conditions unchanged, we performed additional experiments under injection pressures of 9, 21, and 24 MPa, and we obtained predictions.The results revealed that the oil recovery factor increased with increasing injection pressure.When the injection pressure reached 25 MPa, the recovery degree basically remained unchanged.Therefore, the optimal injection pressure in the CO 2 huff 'n' puff process is 25 MPa (Figure 11C).The oil recovery factor first increased with increasing soaking time, but after soaking for approximately 12 h, when the soaking time was further increased, the recovery ratio basically remained unchanged.Therefore, the optimal soaking time is 12 h (Figure 11D).In summary, considering the continental shale oil of the Lucaogou Formation in the Jimusaer Sag, the optimal number of CO 2 huff 'n' puff cycles for EOR is five cycles, the pressure associated with the optimal injection timing is slightly higher than the saturation pressure, with a value of approximately 4 MPa, the optimal injection pressure is 25 MPa, and the optimal soaking time is 12 h.

| CONCLUSION
The main conclusions of this study are as follows: (1) In regard to unfractured shale, CO 2 huff 'n' puff mainly produces crude oil in large-medium pore throats with radii ranging from 0.1 to 1 μm, while crude oil in pore throats with radii smaller than 0.1 μm is basically not produced.Regarding fractured shale, due to severe gas channeling, the first several huff 'n' puff cycles mainly produce crude oil in fractures.However, with an increasing number of huff 'n' puff cycles, the changes in the internal pressure field lead to an increased connectivity between the fractures and large-medium pore throats, thus achieving coefficient development of both unfractured and fractured shales with essentially similar final recoveries.process, the oil displacement mode in the shale cores is basically piston-type displacement.(3) The number of huff 'n' puff cycles and injection pressure greatly impact the CO 2 huff 'n' puff efficiency, while the injection timing and soaking time slightly influence the CO 2 huff 'n' puff efficiency.Considering the continental shale of the Lucaogou Formation in the Jimusaer Sag, five huff 'n' puff cycles, an injection timing of 4 MPa, an injection pressure of 25 MPa, and a soaking time of 12 h are the optimal CO 2 huff 'n' puff parameters, which provide the best effect on the EOR.
The optimal parameters provided in this work can be directly applied to the field oil test in Jimusaer Sag.On the basis of the production law of CO 2 huff 'n' puff on crude oil, combined with the field oil test result, the optimal plan can be formulated for the oilfield to facilitate its efficient development.

F I G U R E 7
Pressure changes at ports A and B during huff 'n' puff.T A B L E 2 Results for the different CO 2 huff 'n' puff cycles.

F
I G U R E 8 CT saturation image after the different CO 2 huff 'n' puff cycles.CT, computed tomography; EOR, enhanced oil recovery.

F
I G U R E 9 Influence of the different parameters on the oil recovery factor and oil exchange ratio: (A) influence of cycle number on the oil recovery factor and oil exchange ratio; (B) influence of injection timing on the oil recovery factor and oil exchange ratio; (C) influence of injection pressure on the oil recovery factor and oil exchange ratio; (D) influence of soaking time on the oil recovery factor and oil exchange ratio.

( 2 )
The pressure changes in the CO 2 huff 'n' puff process can mainly be divided into three stages of injection-holding-depletion.The pressure changes at the different stages indicate various seepage characteristics of the huff 'n' puff medium and crude oil.When CO 2 enters shale, due to its characteristics of volume expansion, energy enhancement, extraction, and viscosity reduction, in the huff 'n' puff F I G U R E 11 Relationship between the different parameters and oil recovery factor: (A) relationship between cycle number and oil recovery factor; (B) relationship between injection timing and oil recovery factor; (C) relationship between injection pressure and oil recovery factor; (D) relationship between soaking time and oil recovery factor.
the first huff 'n' puff cycle was calculated.Finally, the core was scanned after the first huff 'n' puff cycle, and this scan was recorded as the intermediate model after the first huff 'n' puff cycle.(7)Multiplehuff 'n' puff cycles: The conditions were maintained unchanged, and Step (6) was repeated several times to obtain CT experimental results after multiple huff 'n' puff cycles with CO 2 as the huff 'n' puff medium.(8) CT image processing: The dry scan model, the wet scan model, and the intermediate model in a certain state were imported into specific CT image processing software, area setting parameters were selected, and a CT scan saturation image of the current state was obtained.The imported dry and wet scan models were maintained unchanged, and only the intermediate model was changed to obtain a CT scan saturation image of the entire multicycle CO 2 huff 'n' puff process.
Experimental materials and CT scanning equipment: (A) core sample; (B) CT scan instrument; (C) core gripper.CT, computed tomography.after