Experimental Investigation on the Characteristic Mobilization and Remaining Oil Distribution under CO 2 Huff-n-Puff of Chang 7 Continental Shale Oil

: The Chang 7 continental shale oil reservoir is tight. The recovery factor is extremely low, the remaining oil is very high, and injecting water to improve oil recovery effectiveness is too hard. Therefore, in this paper, physical simulation experiments of CO 2 huff-n-puff shale oil and NMR tests were conducted to study the cycle numbers and permeability on the recovery degree, as well as the characteristics of shale oil mobilization and the remaining oil micro distribution. The results showed that the cumulative oil recovery factors (ORFs) gradually increased in the natural logarithmic form, the single cycle ORFs decreased rapidly in exponential form with the huff-n-puff cycle number, and the biggest economic cycle numbers were between approximately 3 and 5. Furthermore, the higher the permeability, the higher the ORF, but the difference of ORF decreased between the two experimental samples with the cycles. In addition, the gap of production and recovery degree was large between the different scale pores, the ORF of macropores was 6–8 times that of micropores, and the ﬁnal remaining oil was mainly distributed in the micropores, accounting for 82.29% of the total amount; meanwhile, the macropores comprised less than 0.5%. In the process of huff-n-puff, CO 2 ﬂowed into macropores, mesopores, and smallpores under the pressure differential effect, but a small amount of CO 2 slowly diffused into micropores, resulting in the ORF of the former with more free oil being higher and the ORF of micropores with more adsorbed oil being lower. Therefore, promoting a better contact and reaction between CO 2 and shale oil of micropores is one of the key ways to effectively develop the Chang 7 continental shale oil and enhance oil recovery.


Introduction
In recent years, global oil and gas consumption has been increasing. However, the remaining reserves of conventional resources have decreased with exploration and development, resulting in a scarcity of energy. As a result, unconventional resources (such as tight/shale oil and gas) will play an extremely significant role in the petroleum industry. According to the assessment results of the US Energy Information Administration (EIA, 2019), the technical recoverable resources of shale oil in the world are about 473 × 10 8 t, and the US shale oil production reached 3.20 × 10 8 t, which accounted for 64.7% of the total oil production in 2018 [1,2]. The exploration results showed that China has abundant shale oil geological reserves of 1420 × 10 8 t, and the amount of medium and high maturity continental shale oil with good development prospects is about 200 × 10 8 t [3][4][5]. However, there are issues regarding depletion developments, such as the low production of an individual well, rapid decline, and ultra-low recovery (generally below 10%) [6,7]. Therefore, where T 2 is the total pore fluid transverse relaxation time (ms), T 2D is the diffusion relaxation time (ms), T 2B is the volume relaxation time that is determined by the self-properties of the fluid (ms), T 2S is the surface relaxation time that is related to the pore structure (ms), ρ is the transverse surface relaxivity (µm/ms), S is the pore surface area (µm 2 ), V is the pore volume (µm 3 ), D is the diffusion coefficient (cm 2 /s), γ is the nuclear gyromagnetic ratio, G is the magnetic field gradient (Guss/cm), and T E is the echo spacing of measurement sequence (ms).
In lab NMR tests, the magnetic field of sets can be nearly uniform, and the magnetic field and echo spacing of measurement sequence can be too small. In this case, the diffusion relaxation 1/T 2D can be neglected. Moreover, the volume relaxation time T 2B depends on the self-properties of fluids. For specific liquids with a low viscosity such as water and light oil, the T 2B is very long and a constant under a specific condition. Additionally, the surface relaxation time T 2S is proportional to the pore radius, and the pore radius of shale is mainly at the nano and micro scales, resulting in the surface relaxation time T 2S being extremely low in shale core. Generally, the volume relaxation time T 2B is much larger than the surface relaxation time T 2S in shale rock, which means that the 1/T 2B can also be neglected [26]. Therefore, Equation (1) can be simplified as follows [27]: where F s is the dimensionless pore shape factor and r is the pore radius (µm). Equation (2) shows that the T 2 value is directly proportional to the pore radius r. A previous study [28] found that the relationship between T 2 value and pore radius r is a power function according to a large number of experimental data statistics, as seen in the following equation: where C and n are the coefficients that can be obtained by conversion from high pressure mercury injection data.

Relationship between T 2 Spectrum and the Remaining Oil Content
An NMR T 2 spectrum is a relationship curve between transverse relaxation time T 2 and NMR signal amplitude. The envelope area of the curve is the total NMR signal amplitude within a certain transverse relaxation time range, which reflects the quantity of 1 H in fluid. The integral area of a T 2 spectrum curve has the following relationship with the mass of fluid in rock: where m o is the fluid mass (g), S(T 2 ) is the integral area of the T 2 spectrum curve comprising the total NMR signal amplitude, and α is the coefficient determined by experiment. The difference of integral areas under the two T 2 spectrum curves is the change of fluid mass in the specimen, as shown in Figure 1. Curves 1 and 2 are oil-bearing T 2 spectrum curves before and after the experiment, respectively. The yellow part is the integral area difference between the two curves. Then, the change of oil mass is: where mo is the fluid mass (g), S(T2) is the integral area of the T2 spectrum curve comprising the total NMR signal amplitude, and α is the coefficient determined by experiment. The difference of integral areas under the two T2 spectrum curves is the change of fluid mass in the specimen, as shown in Figure 1. Curves 1 and 2 are oil-bearing T2 spectrum curves before and after the experiment, respectively. The yellow part is the integral area difference between the two curves. Then, the change of oil mass is: Then the recovery degree η can be calculated as follows: Figure 1. NMR T2 spectrum curve of before and after the experiment. 1 denotes the oil-bearing T2 spectrum curves before the experiment, and 2 denotes the oil-bearing T2 spectrum curves after the experiment.
Therefore, after combining the relationship between rock pore radius and transverse relaxation time in Equation (3), one could analyze and calculate the distribution, producing degree, and remaining oil content of shale oil of different pore sizes.

Materials
The experimental core samples in Figure 2 were drilled in the Yanchang 7th Formation of Triassic in Ordos Basin. The continental shale of Chang 7 is a set of source rock series that are mainly composed of shale formed in the heyday of Late Triassic lake basin development. Though the two cores were cut from the same rock specimen, there was still a permeability difference due to the strong reservoir heterogeneity in the shale. Table 1 displays the properties of the core samples. Figure 1. NMR T 2 spectrum curve of before and after the experiment. 1 denotes the oil-bearing T 2 spectrum curves before the experiment, and 2 denotes the oil-bearing T 2 spectrum curves after the experiment.
Then the recovery degree η can be calculated as follows: Therefore, after combining the relationship between rock pore radius and transverse relaxation time in Equation (3), one could analyze and calculate the distribution, producing degree, and remaining oil content of shale oil of different pore sizes.

Materials
The experimental core samples in Figure 2 were drilled in the Yanchang 7th Formation of Triassic in Ordos Basin. The continental shale of Chang 7 is a set of source rock series that are mainly composed of shale formed in the heyday of Late Triassic lake basin development. Though the two cores were cut from the same rock specimen, there was still a permeability difference due to the strong reservoir heterogeneity in the shale. Table 1 displays the properties of the core samples.    The oil used in this study was n-dodecane with a density of 0.7487 g/cm 3 and a viscosity of 1.508 mPa·s at 25 • C, and the calibration curve between n-dodecane and NMR signal amplitude is shown in Figure 3. The CO 2 was from Suzhou Chenggong Co. Ltd., Suzhou, China, with a purity of 99.9%.  The oil used in this study was n-dodecane with a density of 0.7487 g/cm 3 and a viscosity of 1.508 mPa·s at 25 °C, and the calibration curve between n-dodecane and NMR signal amplitude is shown in Figure 3. The CO2 was from Suzhou Chenggong Co. Ltd., Suzhou, China, with a purity of 99.9%.

Experimental Apparatus
The main experiments carried out in this study were SEM, CO2 huff-n-puff, and NMR tests. The major apparatus for SEM included an argon ion polishing instrument (Il-ion+II697C, Gatan Co., Warrendale, PA, USA), an energy spectrometer field emission scanning electron microscope (FESEM, Quanta 450, FEI Co., Hillsboro, OR, USA), and an energy spectrometer (AMETEK, EDAX Co., Philadelphia, PA, USA). As shown in the CO2 huff-n-puff experiment schematic of Figure 4, the huff-n-puff vessel was fitted with CO2 from the cylinder before huff-n-puff. The ISCO pump (260 D, ISCO, Lincoln, NE, USA) provided power to inject CO2 from the plug-type container to the core. The hand pump provided backpressure, and the temperature control setup allowed for the maintenance of the temperature of the whole experimental system at 45 °C during the process. The NMR spectrometer was produced by Suzhou Niumag Analytical Instrument Co., Ltd., Suzhou, China. The magnetic field strength was 0.52 ± 0.05 T, and the resonance frequency

Experimental Apparatus
The main experiments carried out in this study were SEM, CO 2 huff-n-puff, and NMR tests. The major apparatus for SEM included an argon ion polishing instrument (Ilion+II697C, Gatan Co., Warrendale, PA, USA), an energy spectrometer field emission scanning electron microscope (FESEM, Quanta 450, FEI Co., Hillsboro, OR, USA), and an energy spectrometer (AMETEK, EDAX Co., Philadelphia, PA, USA). As shown in the CO 2 huff-n-puff experiment schematic of Figure 4, the huff-n-puff vessel was fitted with CO 2 from the cylinder before huff-n-puff. The ISCO pump (260 D, ISCO, Lincoln, NE, USA) provided power to inject CO 2 from the plug-type container to the core. The hand pump provided backpressure, and the temperature control setup allowed for the maintenance of the temperature of the whole experimental system at 45 • C during the process. The NMR spectrometer was produced by Suzhou Niumag Analytical Instrument Co., Ltd., Suzhou, China. The magnetic field strength was 0.52 ± 0.05 T, and the resonance frequency of the hydrogen proton was 21.3 MHz. Other devices used in this study included a core linear cutting machine, a core cleaner, an oven, a core measurement system, an analytical electronic balance, a Vernier caliper, and a gas flowmeter.

Experimental Procedures
The experimental procedures were as follows.
(1) Core preparation: we prepared the shale cores by a linear cutting machine with a diameter of 25 mm, and we cleaned the samples for 28 days. Afterwards, they with dried at 120 • C for about 3 days, and the permeability and porosity were tested by helium.
(2) SEM tests: we took the core slice from the standard sample and polished the surface.
Then, the polished sample was photographed by FESEM. (3) Rock sample saturation: we vacuumed the core samples for 72 h. Next, the specimens were saturated with n-dodecane for 48 h at 30 MPa and 45 • C, and then we put them into the container with n-dodecane for one week to completely saturate at 45 • C. Additionally, the samples were weighed before and after the processes. (4) Initial T 2 spectrum: after saturation, we measured the samples with an NMR apparatus to obtain the initial T 2 spectrum curves of the saturated oil. (5) CO 2 huff: first, we put the core into the huff-n-puff vessel, and then, as shown in Figure 4, the experimental process was connected. After we filled the huff-n-puff vessel with CO 2 from the cylinder, CO 2 in the plug-type container was injected into the huff-n-puff vessel by the ISCO pump at 15 MPa. (6) CO 2 puff: all valves were closed during the soaking period for 12 h. We opened valve #3 and controlled the back pressure to gradually drop to 0 MPa. (7) CO 2 huff-n-puff T 2 spectrum: we took out the core from the huff-n-puff vessel, and then we weighed the sample by the analytical electronic balance and measured the T 2 spectrum using the NMR apparatus. (8) We repeated steps (5)-(7) at the same injection pressure of 4 rounds for the CO 2 huff-n-puff experiments and NMR tests.
Energies 2021, 14, x FOR PEER REVIEW 6 of 18 of the hydrogen proton was 21.3 MHz. Other devices used in this study included a core linear cutting machine, a core cleaner, an oven, a core measurement system, an analytical electronic balance, a Vernier caliper, and a gas flowmeter.

Experimental Procedures
The experimental procedures were as follows.
(1) Core preparation: we prepared the shale cores by a linear cutting machine with a diameter of 25 mm, and we cleaned the samples for 28 days. Afterwards, they with dried at 120 °C for about 3 days, and the permeability and porosity were tested by helium.
(2) SEM tests: we took the core slice from the standard sample and polished the surface.
Then, the polished sample was photographed by FESEM. (3) Rock sample saturation: we vacuumed the core samples for 72 h. Next, the specimens were saturated with n-dodecane for 48 h at 30 MPa and 45 °C, and then we put them into the container with n-dodecane for one week to completely saturate at 45 °C. Additionally, the samples were weighed before and after the processes. (4) Initial T2 spectrum: after saturation, we measured the samples with an NMR apparatus to obtain the initial T2 spectrum curves of the saturated oil. (5) CO2 huff: first, we put the core into the huff-n-puff vessel, and then, as shown in Figure 4, the experimental process was connected. After we filled the huff-n-puff vessel with CO2 from the cylinder, CO2 in the plug-type container was injected into the huff-n-puff vessel by the ISCO pump at 15 MPa. (6) CO2 puff: all valves were closed during the soaking period for 12 h. We opened valve #3 and controlled the back pressure to gradually drop to 0 MPa. (7) CO2 huff-n-puff T2 spectrum: we took out the core from the huff-n-puff vessel, and

Pore Characteristics and Radius Distribution
The SEM images of Chang 7 shale core samples are shown in Figure 5. It could be observed that the rock was tight shale, and microfractures with different lengths and widths, as well as a large number of micropores, were developed. Some larger pores were filled with clastic particles among the minerals, such as pyrite and illite. Hence many micropores were formed. As a whole, massive micro-smallpores and fractures developed, and relatively few large-scale pores were distributed in the rock.
filled with clastic particles among the minerals, such as pyrite and illite. Hence many micropores were formed. As a whole, massive micro-smallpores and fractures developed, and relatively few large-scale pores were distributed in the rock.  Figure 6 shows the T2 spectrum curves of initial saturated oil in cores. The curve of core #Y1 has three wave crests, but the transition between the middle and right crest is not apparent in the other one. The crest decreases from left to right, which indicates that the core pore distribution was strongly heterogeneous. The crest on the left means micropores, the middle one means mesopores, the trough between the two means smallpores, and the crest on the right represents macropores or fractures. In the T2 spectrum of core #Y2, there is a little transition between the crests that represents mesopores and macropores. Therefore, after combining this spectrum with Equation (3), the pore radius distribution could be shown in Figure 7. In this study, the shale pores could be divided into four types, as shown in Table 2, according to Lai et al. and Nie et al. [29,30], and their volume ratios are shown in Table 3. All the pore volume ratios in core #Y1 were slightly  Figure 6 shows the T 2 spectrum curves of initial saturated oil in cores. The curve of core #Y1 has three wave crests, but the transition between the middle and right crest is not apparent in the other one. The crest decreases from left to right, which indicates that the core pore distribution was strongly heterogeneous. The crest on the left means micropores, the middle one means mesopores, the trough between the two means smallpores, and the crest on the right represents macropores or fractures. In the T 2 spectrum of core #Y2, there is a little transition between the crests that represents mesopores and macropores. Therefore, after combining this spectrum with Equation (3), the pore radius distribution could be shown in Figure 7. In this study, the shale pores could be divided into four types, as shown in Table 2, according to Lai et al. and Nie et al. [29,30], and their volume ratios are shown in Table 3. All the pore volume ratios in core #Y1 were slightly larger than those in core #Y2, except for micropores. This reflects the relationship between permeability and pore size to a certain degree, such that the higher the proportion of pores with a larger radius, the greater the permeability. larger than those in core #Y2, except for micropores. This reflects the relationship between permeability and pore size to a certain degree, such that the higher the proportion of pores with a larger radius, the greater the permeability.     larger than those in core #Y2, except for micropores. This reflects the relationship between permeability and pore size to a certain degree, such that the higher the proportion of pores with a larger radius, the greater the permeability.     T 2 < 1 r < 0.020 micropore 1 ≤ T 2 < 10 0.020 ≤ r < 0.25 smallpore 10 ≤ T 2 < 100 0.25 ≤ r < 2.0 mesopore T 2 ≥ 100 r ≥ 2.0 macropore  Figure 8 depicts the measured cumulative oil recovery factor (ORF) versus cycle numbers at a pressure of 15 MPa. The results showed that the effect was positive for continental shale oil produced by CO 2 huff-n-puff. After five cycles, the cumulative ORFs of cores #Y1 and #Y2 reached 40.34% and 37.66%, respectively. In the first three cycles, the cumulative ORFs significantly increased, and the oil production accounted for 89.80% and 89.53%, respectively, but slowly increased in the succeeding two cycles. Additionally, the cumulative ORFs gradually increased in the natural logarithmic form with cycle number. Correspondingly, as shown in Figure 9, the single cycle ORFs rapidly decreased in the exponential form with cycle number, and the recoveries were only 1.37% and 1.29% for cores #Y1 and #Y2, respectively, at the last cycle. This indicates that the interaction between CO 2 and shale oil gradually reduced until it was ineffective for EOR. Thus, based on the analysis of the experimental results, efficient production required about three-to-five cycles of CO 2 huff-n-puff for shale oil reservoirs, which was consistent with the research of W. Pu and B. Wei [31].  Figure 8 depicts the measured cumulative oil recovery factor (ORF) ver numbers at a pressure of 15 MPa. The results showed that the effect was positiv tinental shale oil produced by CO2 huff-n-puff. After five cycles, the cumulativ cores #Y1 and #Y2 reached 40.34% and 37.66%, respectively. In the first three c cumulative ORFs significantly increased, and the oil production accounted for 89 89.53%, respectively, but slowly increased in the succeeding two cycles. Additio cumulative ORFs gradually increased in the natural logarithmic form with cycle Correspondingly, as shown in Figure 9, the single cycle ORFs rapidly decreas exponential form with cycle number, and the recoveries were only 1.37% and cores #Y1 and #Y2, respectively, at the last cycle. This indicates that the inter tween CO2 and shale oil gradually reduced until it was ineffective for EOR. Th on the analysis of the experimental results, efficient production required abou five cycles of CO2 huff-n-puff for shale oil reservoirs, which was consistent w search of W. Pu and B. Wei [31].    Figure 8 depicts the measured cumulative oil recovery factor (ORF) ver numbers at a pressure of 15 MPa. The results showed that the effect was positiv tinental shale oil produced by CO2 huff-n-puff. After five cycles, the cumulativ cores #Y1 and #Y2 reached 40.34% and 37.66%, respectively. In the first three c cumulative ORFs significantly increased, and the oil production accounted for 89 89.53%, respectively, but slowly increased in the succeeding two cycles. Additio cumulative ORFs gradually increased in the natural logarithmic form with cycle Correspondingly, as shown in Figure 9, the single cycle ORFs rapidly decreas exponential form with cycle number, and the recoveries were only 1.37% and cores #Y1 and #Y2, respectively, at the last cycle. This indicates that the intera tween CO2 and shale oil gradually reduced until it was ineffective for EOR. Th on the analysis of the experimental results, efficient production required about five cycles of CO2 huff-n-puff for shale oil reservoirs, which was consistent wi search of W. Pu and B. Wei [31].

Effect of Permeability
The permeabilities of experimental cores Y1 and Y2 were 0.2911 × 10 −3 µm 2 and 0.0169 × 10 −3 µm 2 , respectively, as seen in Table 1. A series of experiments were carried out to investigate the effect of permeability on the oil recovery factor during the CO 2 huff-n-puff process under 15 MPa of pressure, as shown in Figures 8 and 9. The single cycle and cumulative ORF of core #Y1 was larger than those of core #Y2. However, as shown in Figure 10, the differences of single cycle ORFs sharply decreased between the specimens with different permeabilities after two cycles, and it was only 0.07% in the fifth cycle. That was because the permeability was proportional to the pore radius of the reservoir. Under the same external conditions, the larger the permeability and the corresponding average pore radius, the smaller capillary force and seepage resistance; then, the saturation of the movable fluid increased in the pore and the recovery factor of CO 2 huff-n-puff was higher. The remaining oil decreased with the permeability increases. Generally, fracturing and acidizing could enlarge the sizes of pores, fractures, and permeability for shale oil reservoirs, thus creating a foundation to improve the recovery by CO 2 huff-n-puff.

Effect of Permeability
The permeabilities of experimental cores Y1 and Y2 were 0.2911 × 10 −3 μm 2 × 10 −3 μm 2 , respectively, as seen in Table 1. A series of experiments were car investigate the effect of permeability on the oil recovery factor during the CO2 process under 15 MPa of pressure, as shown in Figures 8 and 9. The single cy mulative ORF of core #Y1 was larger than those of core #Y2. However, as show 10, the differences of single cycle ORFs sharply decreased between the speci different permeabilities after two cycles, and it was only 0.07% in the fifth cycle because the permeability was proportional to the pore radius of the reservoir. same external conditions, the larger the permeability and the corresponding av radius, the smaller capillary force and seepage resistance; then, the saturation o able fluid increased in the pore and the recovery factor of CO2 huff-n-puff was h remaining oil decreased with the permeability increases. Generally, fracturing ing could enlarge the sizes of pores, fractures, and permeability for shale oil thus creating a foundation to improve the recovery by CO2 huff-n-puff.

Microscopic Process with NMR Tests
Though the recovery factors can be calculated from the experimental data CO2 huff-n-puff cycles, it is difficult to quantitatively determine the oil prod remaining oil in different scale pores in a core. However, to investigate the NM trum curves shown in Figure 11, we obtained the characteristics of ORF, the rem distribution in different-scale pores, and the pores' contributions to EOR.

Microscopic Process with NMR Tests
Though the recovery factors can be calculated from the experimental data of different CO 2 huff-n-puff cycles, it is difficult to quantitatively determine the oil production and remaining oil in different scale pores in a core. However, to investigate the NMR T 2 spectrum curves shown in Figure 11, we obtained the characteristics of ORF, the remaining oil distribution in different-scale pores, and the pores' contributions to EOR.

Characteristics of Shale Oil Mobilization
As shown in Figure 11a,b, compared with the initial saturated oil, there are three obvious wave crests in the NMR T 2 spectrum curve at the end of each CO 2 injection cycle. This confirms that the target reservoir had a strong heterogeneity and once again mainly consisted of four scales of pores. The distribution curves of the T 2 spectrum obtained from the CO 2 huff-n-puff experiment gradually shift downward to the left with the cycle numbers. The curve with the T 2 larger than 10 ms more significantly shifts, indicating that the crude oil flowed out first in the mesopores and macropores, which was consistent with a study of Ma et al. [32].
In the first cycle, the oil production was greatest in the macropores. The percentages of the recovery factor were 36.95% and 39.75% for experimental samples. Following was mesopores, accounting for 33.53% and 31.85%, and then 24.11% and 16.21% in the smallpores. Finally, the percentages were at the minimum in the micropores, accounting for only 7.10% and 10.51%. However, with the increase of the injection cycles, the oil production proportion of the macropores continuously decreased and more sharply reduced after the third cycle. Additionally, the oil recovery factors of mesopores also obviously decreased after the third cycle but were generally stable in the smallpores. The percentage gradually increased in the micropores, especially more obviously after the third cycle. After five cycles, the average cumulative ORFs of different pores in the two samples were 97.26%, 70.94%, 64.71%, and 13.89%. The oil production percentages in different pores are shown in Table 4. Therefore, with the increase of the cycle numbers, most oil in macropores and mesopores had flowed out, and the oil in smallpores and micropores became the main oil-producing area after approximately three-to-five cycles of CO 2 huff-n-puff.

Characteristics of Remaining Oil Distribution
After quantitatively calculating the NMR T 2 spectrum curves shown in Figure 11, we were able to obtain the relationship between the remaining oil in different pore sizes and CO 2 huff-n-puff cycle numbers, as shown in Figures 12 and 13. The remaining oil saturation was the highest, and the producing degree was the lowest at end of each gas injection cycle in the micropores. The remaining oil saturation was still as high as 87.63% and 84.59% after the entire experiment in the two cores. The remaining oil saturation values of the medium and smallpores were between those of the micropore and macropore, and they obviously dropped in the first three cycles before basically stabilizing at 26.77~45.03%. The remaining oil saturation was only 2.74%, and the producing degree was very high in the macropores. Additionally, the total remaining oil saturation continuously dropped with the cycle numbers, and the average percentages of the remaining oil saturation in the small, medium, and large pores were 8.35%, 8.90%, and 0.46%, respectively, but the percentage in the micropores gradually increased to 82.29%. This means that, after the CO 2 huff-n-puff experiment for continental shale oil, the remaining oil was mainly in the micropores, a little was in the medium and smallpores, and an even smaller amount was in the macropores. This showed that there were great differences in the characteristics of mobilization and remaining oil micro-distribution among different sizes of pores. Consequently, the critical problem of EOR for Chang 7 continental shale oil is how to efficiently develop and reduce the remaining oil in the micropores.

Process and Mechanism Analysis of CO 2 Huff-n-Puff
Alfarge et al. [33,34] summarized several mechanisms for injecting CO 2 to improve oil recovery in unconventional reservoirs through laboratory experiments and numerical simulations, as shown in Table 5. For one CO 2 huff-n-puff cycle, there are three stages: huff, soak, and puff. During huff, CO 2 flows into the reservoir and passes through the fractures, and the rock matrix is encompassed by CO 2 . Then, the CO 2 permeates into the rock by differential pressure and concentration. Some oil is carried into the matrix pores by the flow of CO 2 , but the oil also swells and some is simultaneously extruded out of the pores. During the soaking period, oil further swells and viscosity is further reduced, and the oil migrates toward the fractures. During puff, the bulk fluid (miscible or immiscible oil and CO 2 ) in the matrix flow towards the fracture, and the oil moves from the pores into the fracture via CO 2 diffusion caused by the differential pressure and concentration of the gradually reducing CO 2 . Lab and simulation 9: Oil Viscosity reduction Lab and simulation 10: Combination of more than one mechanism from above -However, we analyzed the results of multiple-cycle CO 2 huff-n-puff experiments for the Chang 7 continental shale oil reservoir with an ultra-low matrix permeability and a complex pore structure in this study. Then, the processes and mechanisms could be indicated as follows.
(1) In the first huff-n-puff cycle, CO 2 was preferentially injected into the macropores with smaller capillary resistance under the pressure differential effect. Additionally, CO 2 was dissolved in the crude oil to generate oil swelling, pressurization, and viscosity reduction. Then, the oil was rapidly produced. Secondly, a small amount of CO 2 entered the mesopores and smallpores to produce a little oil. Finally, a tiny amount of oil was obtained by CO 2 with diffusion in micropores.
(2) With the cycle numbers increased, most of the crude oil in the macropores drained, and then the flow resistance and CO 2 consumption in the macropores reduced. Therefore, the concentration of the CO 2 that contacted and reacted with the remaining oil in the mesopores and smallpores was relatively higher, and the crude oil could be further extracted. However, as shown in Figure 14 and Table 4, the ORF and the percentage of oil production in meso and smallpores both sharply reduced after three injection cycles in this experiment. Analyses have shown that the proportion of adsorbed oil in such pores is higher than that in macropores [36,37], and after several cycles of CO 2 huff-n-puff, the free oil basically discharged but the adsorbed oil was hard to extract. (3) For the micropores that accounted for more than half of the total pore volume, the shale oil recovery factor was generally at a low level. As shown in Figure 14, though the ORF was found to be the highest in the second cycle, it was also only 3.27% and 4.14% in the first and second cycles, respectively. An analysis showed that the capillary resistance was very large due to the tiny pore size of the micropores. Then, the CO 2 mainly entered into the micropores through diffusion to interact with shale oil. Thus, the total amount and rate of CO 2 entering the micropores were both lower. Meanwhile, with the production of shale oil in other pores, more CO 2 flowed towards the micropores, and the amount of CO 2 diffused into the micropores increased, leading to the free oil with a low proportion to quickly discharge. However, the adsorption effect of micropores was stronger, and the remaining adsorbed oil was more difficult to extract. This resulted in the shale oil recovery factor significantly reducing after two cycles of CO 2 huff-n-puff. Figure 14. ORF of different pore scales versus cycle number.

Conclusions
In this paper, the Chang 7 continental shale pore characteristics and radius distribution were analyzed by SEM and NMR T 2 spectrum curves, and a series of CO 2 huff-n-puff experiments and NMR tests were carried out to investigate the shale oil recovery factor with different cycles, characteristics of production, and amounts of remaining oil distribution in different size pores. The main conclusions based on the above experimental results are as follows.
(1) In order to study the pore structure of Chang 7 continental shale, a series of SEM and NMR tests were conducted. The SEM images showed that there were massive micro-smallpores and fractures, as well as a few large-scale pores. These results qualitatively characterized the complexity of the pore structure. Furthermore, the T 2 spectrum curves of saturated oil and remaining oil had two and three wave crests, respectively. We could quantitatively classify the pores into four types, namely micropore, smallpore, mesopore, and macropore. The proportions of the different pore volumes were 56.38% and 60.26%, 13.57% and 11.93%, 19.19% and 18.29%, 10.85%, and 9.53%, respectively, in the two samples. The above results indicated a strong heterogeneity of the Chang 7 continental shale.
(2) With the huff-n-puff cycle number, the cumulative ORF gradually increased in the natural logarithmic form, and the single-cycle ORF rapidly decreased in the exponential form. Additionally, the fifth cycle was only 1.37% and 1.29%, respectively, for the aforementioned forms. This shows that an economic production comprises about three-to-five cycles for the shale oil reservoir. Additionally, the higher the permeability was, the higher the ORF was found to be. However, each pore was mainly composed of micropores, and the total capillary force and oil flow resistance were higher, resulting in a difference of ORF that was far less than that of permeability between the two experimental samples. 3) A battery of NMR T 2 spectrum curves of the CO 2 huff-n-puff reflected the recovery factors and remaining oil content of different pores. After five cycles, for the experimental samples, the average cumulative ORF of macropores was greater than 97%, and the mesopores, smallpores, and micropores had average cumulative ORFs of 70.94%, 64.71%, and 13.89%, respectively. Meanwhile, the final remaining oil of micropores accounted for 82.29% of the total amount, and the smallpores, mesopores, and large pores accounted for 8.35%, 8.90%, and less than 0.5%, respectively. These results suggest that the remaining oil was mainly in micropores and rarely in macropores. This reflects that there were great differences in the characteristics of mobilization and the remaining oil micro-distribution among different sizes of pores. 4) The shale oil was usually stored freely and adsorbed in the pores. However, the proportion of adsorbed oil in the micropores was higher than that in macropores, and the adsorbed oil was hard to extract. Additionally, the CO 2 could more easily flow into macropores, mesopores, and smallpores under the pressure differential effect. However, only small amount of CO 2 could slowly diffuse into the micropores, resulting the differences of oil production and remaining oil micro-distribution. 5) For the Chang 7 continental shale, large-scale fracturing and acidizing may comprise foundation to promote a better contact and reaction between CO 2 and shale oil in micropores. Then, one could effectively develop shale oil, reduce remaining oil, and enhance shale oil recovery. Funding: This research was funded by National Natural Science Foundation of China (52004217; 51874239). We would like to express our appreciation to the other members of the laboratory for help provided in experiments and language editing.