Pore-scale analysis of gas huff-n-puff enhanced oil recovery and waterflooding process

Fuel 215 (2018) 561–571

Contents lists available at ScienceDirect

Fuel journal homepage: www.elsevier.com/locate/fuel

Full Length Article

Pore-scale analysis of gas huff-n-puff enhanced oil recovery and waterflooding process

T

⁎

Tao Wana, , Weimin Wangb, Jiaqi Jiangb, Yuandong Zhangb a b

Faculty of Petroleum, China University of Petroleum – Beijing at Karamay, China Department of Information Science and Technology, Peking University, China

G RA P H I C A L AB S T R A C T

A R T I C L E I N F O

A B S T R A C T

Keywords: Tight oil formations Gas huff-n-puff Waterflooding NMR investigation

Understanding the complex pore network and fractures are crucial to efficiently producing tight gas and oil reservoirs. A better understanding of gas flooding recovery mechanisms will lead to improved successes of EOR (enhanced-oil-recovery) practices in tight oil formations. In this paper, the gas huff-n-puff efficiency performance in tight oil formations is under surveillance by NMR technology. It is of our interest to estimate the recoverable or movable oil saturation by waterflooding and gas flooding at different types of pore sizes. Fortunately, NMR measurements provide an avenue for calculating the recoverable reserves in different types of pore system (micropores, mesopores and macropores). The NMR T2 relaxation time closely correlates with the pore sizes. The NMR technique was used to analyze the mechanisms of gas flooding and waterflooding in shale formations from a microscopic scale view. In this paper, a series of nitrogen huff-n-puff experiments were conducted on tight cores and NMR was used in the whole huff-n-puff process to observe the gas flooding efficiency at different cycles. The NMR relaxation spectrum reveals that most of the oil production happened in the first few cycles, less oil is recovered in the subsequent cycles. The recoverable oil of this field falls into a range of 1–100 ms T2 relaxation pore size system. Oil production only occurs in certain type of pores. Due to the nanometer or micrometer scales of the pores and pore throats, cycle depletion time has considerable effect on oil recovery from tight oil reservoirs in the first few cycles. The literature lacks a study of the NMR investigation of

⁎

Corresponding author. E-mail address: [email protected] (T. Wan).

https://doi.org/10.1016/j.fuel.2017.11.033 Received 7 September 2017; Received in revised form 19 October 2017; Accepted 10 November 2017 0016-2361/ © 2017 Elsevier Ltd. All rights reserved.

Fuel 215 (2018) 561–571

T. Wan et al.

gas huff-n-puff effect in tight formations. The purpose of this current study is to illustrate the application of NMR technique in interpreting the effect of cyclic nitrogen injection in tight oil reservoirs.

distribution, NMR techniques was used to assess the movable fluid saturation for 264 cores from the Yanchang Ordos basin [11]. The lithology of these cores is mostly comprised by lithic feldspar sandstone, whose permeability ranges from 0.11 mD to 2.8 mD. The T2 cutoff value lies at about 5.34–20 ms. The average movable fluid percentage for a total of 264 core samples is 48.35%. The measured T2 distribution is dominant by the bimodal distribution in ultra-low permeability reservoirs. The typical T2 value of the cores falls in a range of 1–10 ms for the left peak and 10–100 ms of the right peak. In nanopores, the T2 distribution measured in shales may not reflect an indication of the pore size distribution when the length scale of NMR diffusion is comparable to the relaxation time of pores [12–14]. NMR T2 test and MRI were used to examine the oil mobilization in different pore sizes of tight matrix during CO2 injection process [15]. Experimental results showed that oil mobilization by CO2 injection in the tight matrix is affected by the exposure time. Oil mobilization in pores with radius smaller than 1 μm happens in a slow manner in which the oil recovery improves with an increase of exposure time. Gannaway presented a series of NMR experiments on Barnett shales [16]. NMR measurements were performed to characterize the pore network distribution in gas shales by conducting the Mncl2 solution imbibition process after fully dodecane saturated. All the shale samples exhibited a unimodal T2 distribution with a peak at 0.2 ms (clay bound water porosity) in their native state. By comparing the native state, dodecane saturated and Mncl2 imbibed NMR distributions, the classification of effective porosity, inorganic porosity and organic porosity was discussed in the paper. It was found that displacement process took place at dodecane-filled pores (> 5 ms) in the Mncl2 imbibition process and the clay-bound water filled pores at 0.2 ms was also displaced by water invasion. Recent studies indicate that significant T2 relaxation time in Haynesville shales is smaller than a few milliseconds due to the fact that the high surface-to-volume ratios and the adsorption of methane on the pore surface result in large reduction of the relaxation times from their bulk values [23]. Studies on the Barnett gas shales by Songergold et al. also demonstrated an indication of a dominate range in pore body radius from 5 nm to 150 nm on the NMR spectra and most of the signal lies below 3 ms [25]. Tinni et al. presented a study of NMR responses in shale samples from Haynesville, Barnett, and Woodford [17]. Different brine-saturation pressures at 1000 psi, 2000 psi, 3000 psi, 4000 psi, 5000 psi, to 7000 psi were applied at the received state core plugs. The brine saturated cores show essentially one peak with T2 values that range from 0.1 to 10 ms. For dodecane saturation after brine-saturated, an increase in amplitude at T2 values from 1 to 100 ms is observed but there is no significant change of the water wet porosity peak. A broader distribution of T2 relaxation time shifted to longer times is shown after dodecane saturation at exerted pressures which suggests that dodecane did not enter the water-wet pores but the hydrocarbon wet pores. NMR response in shale rocks is complicated by the presence of organic matter, maturity, and clay content [18,19]. The coexistence of different scale of pores (micro-pores and nano-pores) renders the interpretation process more difficult. The challenge of NMR interpretation in shales attributes to the heterogeneous distribution of pore sizes [20]. The NMR experimental design in cyclic gas injection avoids the perturbation of water signals without introducing the water component. In the cyclic nitrogen injection process, the only medium that

Nomenclature A Swc Sor T2

NMR signal amplitude Connate water saturation Irreducible oil saturation for gas-liquid table T2 relaxation time

1. Introduction Due to the prevailing of nano-scale pores in shale formations, gas flooding has the advantage to access the trapped hydrocarbons in nanopores compared to waterflooding. Experimental studies of CO2 huff-npuff EOR in shale oil core plugs include investigation of injection pressure, injection period, soaking time and the cycle efficiency [1–3]. Wan et al. [4] implemented a laboratory-scale evaluation of the nitrogen huffn-puff in shale formations. Laboratory and numerical simulation results showed the role of diffusion mechanism in the mobilization and production of oil in very low permeability shale oil or gas reservoirs. The dispersive-convective flux through nano-pores during gas injection process in shale oil reservoirs is modeled. This paper extends the previous research work by using NMR spectrometer to investigate the recovery performance of cyclic gas injection in nanometer sized pores [4]. Better understanding of recovery mechanisms of water flooding and gas flooding will lead to improved successes of EOR projects in tight oil formations. In this paper, we will discuss the methods that can characterize the reservoir pore structures, reservoir heterogeneity and which type of pores mostly contribute to the ultimate oil recovery. Advances in NMR logging use diffusion-based NMR methods to identify and characterize the fluids in hydrocarbon bearing reservoirs invaded by the oil-based mud filtrate [5]. Fluid analysis is made based on the contrasts in 2D distribution of relaxation times and molecular diffusion rates. NMR measurement of differences in the relaxation time at different water-saturation states provides an estimate of water saturation in chalk cores [6]. Freedman and Heaton compared the T2 distribution for the partially oil-saturated rocks with the bulk oil. It was found that the NMR surface relaxation rate depends on the wetting phase saturation in mixed-wet rocks [7]. During NMR monitoring the process of water imbibition in shales, water firstly enters into the micropores (0.01–5 ms). After hydrocarbons in the micropores has been swept, water then infiltrates into less preferential mesopores [8]. Cracks were observed to appear on the surface of the shales when the core samples were in imbibition process [8]. Comparative measurements of the helium porosity with the NMR method, it shows that the helium porosity is higher than the NMR measured porosity [9]. The echo time exerts a strong influence on NMR measurements, and a longer echo time produces a smaller NMR porosity. NMR core experimental examination of the Haynesville shale also showed that the NMR measured effective porosity is lower than density porosity [10]. The NMR porosity measurement is affected by the pyrite concentration and kerogen content, which decreases with an increasing of pyrite. The measured average air permeability of the Haynesville shale cores is 0.268 mD, with an average porosity of 4.35%. The T2 relaxation time distribution of the micropores ranges from 0.8 to 2 ms, in contrast, large pores corresponding to 2–200 ms. In addition to measure the pore size

562

Fuel 215 (2018) 561–571

T. Wan et al.

acquisition of reliable NMR signals is essential to determine the movable oil saturation in water flood projects. 4. In whole waterflooding process, the only fluid that can produce signals is water. 5. Make NMR T2 measurements of the cores at irreducible water saturation Swir. At this state, only residual water can produce the NMR signals, whose amplitude is the lowest. 6. The fluorocarbon in the plugs was then displaced by water until the NMR signals were stable, which means the no more oil will be swept. The computation of oil displacement is obtained by subtracting the NMR porosity of water-flooded state from initial watersaturated porosity (as shown in Eq. (1)).

contains hydrogen protons is hydrocarbons, which can be measured by NMR spectrometer. The signal change of the unimodal peaks in cyclic gas injection process clearly reflects the recovery efficiency by nitrogen injection (as shown in Fig. 9). 2. Experimental materials and methods SPEC-RC2 NMR spectrometer was used to conduct measurements with an echo time TE = 150 μs. Experiments were performed at reservoir temperatures (178 °F). Synthetic oil that simulated reservoir crude oil properties (such as oil viscosity and density) was used in the waterflooding process with the density of 0.882 g/cm3, but reservoir crude oil was employed in the cyclic gas injection process. Oil viscosity is measured as 11.2 cp at 122 °F. Core plugs are from Xinjiang field tight oil cores in China (Fig. 1). The reservoir oil properties was measured in the lab, presented in Table 1. The oil density is high because the reservoir oil contains asphaltene. The collection of 35 samples in this field showed a range of porosity from 6.09 to 20.7% with a medium 10.9%, and permeability ranges from 0.001 to 100 mD with a medium at 0.01mD. Fig. 2 shows the SEM micrograph at 46.56 K magnification of core samples of pay zones. Dark features are considered to be pores. Fig. 3 shows the distribution of the pore sizes by CT scan. It shows the average pore radius is about 6 μm. We have a good knowledge of the rock’s constituent minerals. The size dimensions of the pores are in good agreement with those inferred from CT scan measurements. Some samples we collected show clear fractures. The presence of fracture system in the core as shown in Fig. 4 is a cause of high permeability measurement. A series of laboratory NMR core-analysis experiments were performed to achieve a basic understanding of waterflooding and gas flooding in tight oil formations. We will not attempt to explain the principle of nuclear magnetic resonance. We provide a brief theoretical introduction of NMR application in determination of movable oil saturation. The hydrogen protons have an inclination to align themselves parallel to an imposed magnetic field. Measurement of the time taken for aligned protons to decay is called T2 relaxation time. In a waterflooding case, unfortunately, both water and hydrocarbon within the pore space can produce signals. The NMR interpretation during waterflooding process is interfered by the presence of hydrocarbons. In order to distinguish the nuclear magnetic signals in the water from oil, there are two approaches that can be used to solve this problem: (1) fluorocarbon that exhibits no NMR signals is used as the oil phase. In this case, any increase of NMR signals above bound-water state is caused by water entering the pores; (2) the heavy water (D2O) can be used to replace the ordinary water (H2O) during displacement experiments. However, in cyclic gas flooding process, it is not necessary to replace crude oil with fluorocarbon. The use of actual crude oil can be measured by NMR technology because injected gases (N2 or CO2) contain no hydrogen nucleus. The reservoir crude oil was used in the cyclic nitrogen injection process. All the saturated core plugs were scanned by NMR to establish the base T2 spectrum. The experiment set up in Fig. 5 allows NMR scans at any point during core flooding. The temperature of the flooding system is remained constant as reservoir temperature. The following procedures of NMR application in waterflooding process were followed:

T

MOS =

∫0 2max (Awf −Awir ) dt T

∫0 2max (Aw −Awir ) dt

(1)

In Fig. 6, Aw represents the NMR signal amplitude of water saturated cores; Awf represents the NMR signal amplitude of water flooded cores; Awir represents the NMR signal amplitude of oil saturated (at irreducible water state, fluorocarbon produces no NMR signals) cores; The total porosity is represented by the integration of the water-filled T2 distribution curve. Since fluorocarbon yields no NMR signals, any amplitude reduction observed in the NMR spectrum is stemmed from water saturation changes in the core plugs. The increase of water peak intensity results from an increase of water saturation because the fluorocarbon was displaced by water. The movable oil saturation in waterflooding process can be calculated from Eq. (1) [21,22]. 3. Results and discussion 3.1. Comparison of waterflooding in high permeability and low permeability reservoirs In the waterflooding process, different water injection rates (Q = 1, 0.1, 0.01 ml/min) were performed to examine the effect of rates on flooding efficiency. It is seen from both Figs. 7 and 8 the higher injection rate achieves a higher oil recovery. In water-saturated state, the samples have the highest peaks ranging from 1 to 1000 ms. In contrast, the transverse T2 values of the tight oil formations in Fig. 8 range from 1 to 100 ms which has a narrower distribution than Fig. 7. The reason is

1. Measure the permeability and porosity of the rock using gas injection method. 2. The core plugs initially 100% saturated with water. Obtain the T2 distribution of fully water-saturated cores. The T2 distribution under water-saturated cores is qualitatively representing the pore size distribution. 3. Use fluorocarbon to displace water and reduce the water to residual water saturation. It should be noticed that fluorocarbon contains no NMR signals but it has similar properties as reservoir oil. The

Fig. 1. Cores drilled from the tight formations.

563

Fuel 215 (2018) 561–571

T. Wan et al.

Table 1 Lab measured reservoir oil properties at different wells. Layer

Cored wells

Depth, m

Density g/cm3

Oil API

Asphaltene, %

Viscosity measured at 50 °C (mpa.s)

Upper layer 1

Well #23 Well#30 Well#31 Well#28 Well#171 Well#172_H

2309 4018 2707 3198 3074 3133

0.871 0.882 0.888 0.877 0.890 0.882

30.95 28.03 27.84 29.84 27.48 28.93

32.2 0.3 11.05 7.566 10.718 9.3

156.6 11 51.8 30.61 49.13 38

Down layer 2

Well#31 Well#36_H Well#37 Well#36 Well#33

2875 4391 2830 4209 3664

0.908 0.909 0.882 0.897 0.898

24.33 24.33 30.5 26.24 26.07

3.85 3.69 0.5 5.01 3.445

132.9 141.6 11.2 80.46 76.63

Fig. 2. SEM scanning of pay zones.

EHT=20 KV Mag = 46.56 KX

m

m

EHT=20 KV Mag = 36.12 KX

0.16

amplitude than the light blue curves with square markers (Q = 0.01) which implies that larger injection rates produce a better flooding efficiency because more volume of water is injected at higher injection rates. Nevertheless, the waterflooding efficiency at Q = 1 is very close to Q = 0.1. Once the injection rates reached a certain level, an increase of injection rate would not cause pronounced effect on ultimate oil recovery. We divide the pores and throats into three categories: micropores (T2 < 10 ms), mesopores (10 ms < T2 < 100 ms) and macro-pores (T2 > 100 ms). The T2 values in the relaxation time distribution directly reflect the pore-size diameters, where longer T2 correspond to signals from large pores [5]. The T2 relaxation time should not be larger than 100 ms in tight oil reservoirs. Tables 2 and 3 summarize the results from Figs. 7 and 8. The water flooding efficiency at different type of pores was measured correspondingly at injection rates of Q = 1, 0.1 and 0.01 ml/min. The pores with shorter relaxation times (T2 < 10 ms) is classified as micropores. Longer relaxation times (T2 > 100 ms) correspond to large pores. Determination of these cut-off values depends on the formation permeabilities, but this simplified classification only provides us a concept about which type of pores contribute mostly to oil recovery in EOR

0.14

Frequency

0.12 0.1 0.08 0.06 0.04 0.02 0 0

5 10 15 20 25 30 35 40 45 50 55 60

Radius, micro-meter Fig. 3. Distribution of the pores by CT scan.

that the average pore sizes in tight reservoirs are smaller than the high permeability reservoirs. This translates into the shorter transverse T2 times in Fig. 8. The signal amplitude of waterflooding at Q = 1 is much higher than the initial oil-saturated condition (at irreducible water) because the water saturation in the core is building up as the progress of waterflooding. The black dotted lines (Q = 1) give a higher signal

Fig. 4. The core image with fractures.

564

Fuel 215 (2018) 561–571

T. Wan et al.

Fig. 5. Lab NMR apparatus set up for cyclic gas injection EOR process.

180

Aw

Waterflooded

140

Amplitude

formations. The lower permeability formation has shorter T2 values. For k = 1.25 md, only 2.7% of pores is consisted by large pores. Recovered oil mostly comes from the micropores and mesopores because large pores contain a small portion of oil. The area difference under the T2 distribution between the water saturated state and water-flooded state represents the oil being displaced.

Water-saturated

160

Oil-Saturated

120 100

Awf

80 60 40 20

3.2. The NMR response of cyclic nitrogen injection in low permeability reservoirs

Awir

0 0.1

1

10

100

1000

The core plugs were initially 100% saturated with reservoir crude oil and later displaced by nitrogen for 6 cycles. Since nitrogen gas will not introduce extra hydrogen protons, any reduction of NMR signals during cyclic nitrogen injection is going to be a result of oil displaced out of pores by injected nitrogen. A confining pressure of 2500-psi was applied when the gas injection pressure was 1000-psi. The confining pressure was maintained much higher than the injection pressure to prevent leakage past the rubber sleeve. One must be aware of the amplitude difference between nitrogen injection and water injection process. Reduction of amplitude during N2 injection process was caused by

Relaxation time T2, ms Fig. 6. Schematic diagram of waterflooding NMR T2 distribution.

process. The results in Table 2 illustrate that most of oil is recovered from mesopores and macropores for the case K = 53.4 md because the fraction of micropores is very small in high permeability formations. Thus, limited oil resided in micropores can be recovered. In contrast, less oil is recovered from the large pores in the low permeability 250

Water-saturated Q=1 Q=0.1 Q=0.01 Irreducible water saturated

K=53.4 md, Por=0.142

Amplitude

200

Micropores Mesopores

150

Macropores

100

50

0 10-1

100

101

101

102

Relaxation time T2,ms 565

103

Fig. 7. NMR T2 distribution of waterflooding at different rates (Q = 1, 0.1, 0.01 ml/min).

Fuel 215 (2018) 561–571

T. Wan et al.

Fig. 8. NMR T2 distribution of tight cores in water flooding process.

350

K=1.25 md, Por = 0.125

Water-saturated Q=1 Q=0.1 Q=0.01 Irreducible water saturated

300

Amplitude

250

200

150

100

50

0 -1 10

0

10

10

1

10

2

10

3

10

4

Relaxation time T2,ms

Fig. 9, the solid blue curve represents the porosity distribution at fully oil saturated condition; Black dotted line shows the measurements after nitrogen injection with 0.5 h depletion; Dashed black triangle curve represents the measurements after 1 h depletion; Solid black dots represent the NMR response at the end of first cycle depletion. At any point in the oil production stage, NMR relaxation measurements can be implemented to observe the dynamics of oil recovery process. We firstly consider the NMR measurements for oil production within one cycle but at different production times. The signal peaks continuously decrease from the 0.5 h, 1 h cycle depletion time to the end of the first cycle depletion. It implies that the cycle depletion time has considerable effect on oil recovery from tight oil reservoirs. Due to the nanometer or micrometer scale of pores sizes, liquid production in tight oil reservoirs requires significant amount of time to complete a cycle depletion. The oil extraction process in tight oil reservoirs is dramatically different from oil mobilization in conventional reservoirs in which oil production is dominant by convective flux. However, the contribution of convection to cumulative oil production in tight oil formations is restricted by the high capillary resistance to flowing in micropores. Diffusive process occurs in a relatively slower manner than advection flux [24]. It provides a good explanation for the response of oil recovery process in the first cycle with cycle production time. The NMR measurements were conducted at the end of each cycle (6 cycles in total). A decrease in peak intensity of the second cycle compared to the first cycle corresponds to a decrease in oil saturation. The amplitude change of the peaks in cyclic gas injection process essentially reflects the oil displacement process by nitrogen injection. Reduction of amplitude in nitrogen injection process was caused by oil mobilization in the pores. Fig. 9 shows a unimodal T2 distribution when the samples were initially oil-saturated. The distributions are peaked around 10 ms for all samples. The NMR amplitude in Fig. 9 is dimensionless. The NMR amplitude can be normalized. The magnitude by itself is not important. The relative area difference under the T2 distribution between two processes determines the oil recovery as illustrated by the following equation:

Table 2 Movable oil saturation distribution in different pores by waterflooding (K = 53.4 md). Pore types

micropores mesopores Large Total

Percent

0.117 0.358 0.525 1

Displacement efficiency Q = 1 ml/min

Q = 0.1

Q = 0.01

0.032 0.312 0.207 0.551

0.061 0.263 0.177 0.501

0.074 0.232 0.109 0.415

Table 3 Movable oil saturation distribution in different pores by waterflooding (K = 1.25 md). Pore types

micropores mesopores Large Total

Percent

0.641 0.332 0.027 1

Displacement efficiency Q = 1 ml/min

Q = 0.1

Q = 0.01

0.286 0.145 0.003 0.434

0.275 0.14 0.002 0.417

0.199 0.078 0.0 0.277

a decrease of oil saturation in the pore space because injected nitrogen contains no hydrogen protons, while in water flooding process extra hydrogen protons are introduced in the system which results in an increased amplitude than oil-saturated state. Nitrogen was injected at a rate of 1 cc/min until the system pressure arrived at 1000 psi. The system then was subject 1 h of soaking period to allow the gas to diffuse into tight core matrix. The effect of soak duration during cyclic gas injection process on nano-permeable shales production response is reported by Wan et al. (2015). [4] We will not repeat the discussions in this paper. After the soaking, oil production phase begins by depressurizing the system to atmospheric pressure. NMR T2 measurements of the cores were made every 15 min in the production period to observe the dynamics. We did not show every T2 curves in the graphs for clarity. NMR core analysis were used to monitor the cyclic nitrogen injection process of the core samples from the Xinjiang tight oil formations. The role of injection pressure on cyclic injection performance was evaluated as shown in Fig. 9 (1000-psi, 1500-psi, 2000-psi, 2500psi and 3000-psi). The transverse T2 relaxation time distributions for each core sample during the cyclic nitrogen injection process are illustrated in Fig. 9. In

T

MOS =

2max (Ainitial −Aj ) dt ∫T2min

T

2max Ainitial dt ∫T2min

T

=

566

2max (Ainitial −Aj ) × h × dt ∫T2min

T2max Ainitial × h × dt ∫T2min

,h is a constant. (2)

Fuel 215 (2018) 561–571

T. Wan et al.

25

25 Initial signal 1st cycle after 0.5 hour 1st cycle after 1 hour end of 1st cycle 2nd cycle after 1 hour end of 2nd cycle end of 3rd cycle end of 4th cycle end of 5th cycle end of 6th cycle

15

10

20

Amplitude

Amplitude

20

Initial signal 1st cycle after 0.5 hour 1st cycle after 1 hour end of 1st cycle 2nd cycle after 1 hour end of 2nd cycle end of 3rd cycle end of 4th cycle end of 5th cycle end of 6th cycle

#a, P= 1000 psi

15

10

5

5

0 -3 10

10

-2

10

-1

10

0

10

1

10

2

10

3

10

0 -3 10

4

10

-2

10

-1

25

1

10

10

2

10

3

10

4

25 Initial signal 1st cycle after 0.5 hour 1st cycle after 1 hour end of 1st cycle 2nd cycle after 1 hour end of 2nd cycle end of 3rd cycle end of 4th cycle end of 5th cycle end of 6th cycle

15

Initial signal 1st cycle after 0.5 hour 1st cycle after 1 hour end of 1st cycle 2nd cycle after 1 hour end of 2nd cycle end of 3rd cycle end of 4th cycle end of 5th cycle end of 6th cycle

#c, P=2000 psi 20

Amplitude

20

Amplitude

0

10

Relaxation time T2,ms

Relaxation time T2,ms

10

15

10

5

#d, P=2500 psi

5

0 10

-3

10

-2

10

-1

10

0

10

1

10

2

10

3

10

0 -3 10

4

10

-2

10

Relaxation time T2,ms

40 35 30 25 20

10

0

10

1

10

2

10

3

10

4

25 Initial signal 1st cycle after 0.5 hour 1st cycle after 1 hour end of 1st cycle 2nd cycle after 1 hour end of 2nd cycle end of 3rd cycle end of 4th cycle end of 5th cycle end of 6th cycle

Initial signal 1st cycle after 0.5 hour 1st cycle after 1 hour end of 1st cycle 2nd cycle after 1 hour end of 2nd cycle end of 3rd cycle end of 4th cycle end of 5th cycle end of 6th cycle

#e, P=2500 psi 20

Amplitude

45

-1

Relaxation time T2,ms

50

Amplitude

#b, P= 1500 psi

15

10

# e, P=3000 psi

15

5

10 5 0 -4 10

10

-2

10

0

10

2

10

4

0 10

-3

10

-2

10

-1

Relaxation time T2,ms

10

0

10

1

2

10

10

3

10

4

Relaxation time T2,ms

Fig. 9. NMR measurements of tight cores in cyclic nitrogen process.

was set much higher than injection pressure. In previous work [4] we have examined the injection pressure effect on the performance of cyclic gas injection process in detail. In this paper, we implement the nitrogen huff-n-puff for the same core plug at different operating pressures. Earlier studies on cyclic CO2 injection suggest that higher injection pressures yield a higher oil recovery. Estimation of the oil recovery was determined by measuring the weight difference of the core plugs at various stages. From Tables 4 and 5, it is observed that the calculated oil recovery by NMR method is fairly close to lab measured values. There are no noticeable errors found in the experimental process. For core plug #a operating at 1000-psi reservoir

Due to the ultra-low permeability of the cores, the saturated water content in the cores is low which results in producing weak NMR signals. The image quality of the core flooding by NMR is blurring. Light areas in the Fig. 10 represents the oil rich portion of the cores. It is observed that the light regions were gradually turned into dark with the progress of nitrogen injection. As shown in Fig. 10, there is a distinct dark area at the top inlet of the cores as oil was displaced by the nitrogen. It reflects that oil was continuously displaced by gas injection with an increasing number of cycles. At the inlet of the core, the displacement efficiency is considerably higher than other region. Gas channeling is unnoticeable in Fig. 10 because the confining pressure 567

Fuel 215 (2018) 561–571

T. Wan et al.

Fig. 10. NMR image of tight cores (#a) in cyclic nitrogen process.

pressure, most of the oil is recovered at the first two cycles. As core plug #e operating at 3000-psi, although the first cycle produced 19.2% of oil, additional oil recovery in subsequent cycles was observed but with a lower rate. With an increase of gas injection cycles, the incremental oil recovery at each cycle is declined. The decrease in peak intensity of the relaxation-time mode corresponds to the decrease of oil saturation in the core plugs and an increase in injected nitrogen saturation. Comparing the core plug #b and #e, which have similar rock properties such as permeability and porosity, a significant decrease of NMR intensity happens in the first cycle. While for #e operating at a higher pressure, the recovery profile is not flatten out after the first two cycles but it still keeps growing in subsequent cycles. The recovery ratio of different pore intervals at different displacement stages was included in Table 6. It is seen that the primary contribution to tight oil recovery comes from the pores with the relaxation time ranging from 10 to 100 ms, while oil mobilization in the micropores ranging from 1 to 10 ms has also been detected by gas injection at different cycles. The effect of injection pressure on cyclic flooding performance is presented in Table 7 and Fig. 9#e. By implementing different injection pressures on the same core, the experimental results shows that a higher injection pressure leads to an increasing in oil recovery. This is consistent with the findings from previous studies [4,26]. It also noticed that the rate of recovering oil is declined with more cycles progressing. As shown in Fig. 9#e, the amplitude profile all exhibits uni-modal distribution with a dominant peak centered at 10-ms. The NMR T2 response for the two cases is very similar because the measurements were conducted on the same core. It is observed in Fig. 9 that the cycle depletion time has remarkable effect on the cycle oil recovery. From the NMR T2 measurement result, it implies that a longer cycle depletion time is needed in order to attain fully equilibrium in the depletion stage. The analysis

Table 5 NMR measured cumulative oil recovery (%) at different cycles. Cumulative Recovery

cycle 1

cycle 2

cycle 3

cycle 4

cycle 5

cycle 6

a (1000 psi) b (1500 psi) c (2000 psi) d (2500 psi) e (3000 psi)

15.6 19.6 21.6 5.1 19.2

28.0 33.7 28.2 11.9 32.4

29.7 35.7 28.9 19.3 37.1

29.8 35.3 29.0 29.1 39.29

29.8 35.7 29.2 33.4 41.6

29.8 35.7 29.6 35.8 42.9

of Fig. 9 suggests that the oil transport in low permeability cores occurs in a slow manner. 3.3. The effect of injected pore volume on cyclic injection efficacy The effect of nitrogen injection pore volume on cumulative oil recovery is shown in Fig. 11. The amplitude peaks for 2 PVI and 4 PVI almost overlapped. As shown in Fig. 12, the cumulative oil recovery increases with the injected nitrogen pore volumes, but the oil recovery flattens out at 1.5 PVI indicating that an increase of nitrogen PVI more than this value would not have significant effect on improved oil recovery. This experimental results are very helpful in providing some insights about the recovery mechanisms at different pore sizes by cyclic gas injection in tight oil formations. Due to the experimental designs and condition variations (such as reservoir permeability, injection pressure, clay minerals and anisotropy), it is difficult to find exactly similar experiments and compare the cyclic N2 injection performance. However, it is interesting to compare a similar case study and identify the factors that have an impact on nitrogen injection efficacy in tight or shale oil formations. Yu

Table 4 The basic properties of core plugs and corresponding measured oil recovery. Sample

Perm (mD)

Por

Dry weight (g)

Saturated Weight (g)

Final weight

Displaced Oil (g)

Measured Recovery

#a #b #c #d #e

0.3042 0.1243 0.0682 0.0742 0.1599

0.147 0.135 0.113 0.105 0.121

78.06 77.25 84.9 101.73 93.63

83.25 81.95 87.96 105.89 97.99

81.69 80.35 86.99 104.41 96.19

1.55 1.6 0.97 1.48 1.8

30.0% 34.0% 31.0% 35.0% 41.0%

568

Fuel 215 (2018) 561–571

T. Wan et al.

Table 6 NMR cumulative oil recovery for different pore intervals at different stages. Pressure

a (1000 psi)

b (1500 psi)

c (2000 psi)

d (2500 psi)

Cycle stage

T2 < 10

10 < T2 < 100

T2 < 10

10 < T2 < 100

T2 < 10

10 < T2 < 100

After 0.5 h After 1 h End 1st cycle After 1 h End 2nd cycle 3 cycle 4 cycle 5 cycle 6 cycle

0 0.7 1.5 3.7 3.7 3.7 3.7 3.7 3.7

1.7 4.3 10.4 15.3 17.3 18.6 18.8 18.8 18.8

0.8 2.2 4.3 4.8 8.6 9.1 9.1 9.1 9.1

0.6 6.2 13.6 15.3 22.9 24.4 24.4 24.4 24.4

0.5 1.9 4.9 5.7 6.7 7 7.2 7.3 7.3

0.4 6.2 14.4 16.2 17.9 18.2 18.4 18.4 18.5

T2 < 10

e (3000 psi) 10 < T2 < 100

1.7

3.4

4.6 8 12 13.6 14.6

7.3 11.3 17 19.8 21.1

T2 < 10

10 < T2 < 100

1.9 3.4 5.3 7.6 10.1 12.1 13.2 15.2 15.8

2.1 5.7 13.6 18.2 22.1 24.7 25.8 26.2 26.8

30

Table 7 Effect of injection pressure on cyclic nitrogen flooding performance (in Fig. 9#e).

100 Cumulative oil recovery Oil saturation

Cumulative Recovery

cycle 1

cycle 2

cycle 3

cycle 4

cycle 5

cycle 6

#e (2500 psi) #e (3000 psi)

17.86 19.2

27.86 32.4

31.10 37.1

34.01 39.29

36.41 41.6

36.73 42.9

K=0.56 md, Por=0.11

0 PV 0.5 PV 2 PV 4 PV

14

80

Oil Saturation

16

Oil Recovery

20

10

12

Amplitude

10

8

0

0

1

2

3

4

5

60

PV

6

Fig. 12. Cumulative oil recovery vs nitrogen PVI.

4

cyclic gas injection method. A combination of NMR technique application with enhanced oil recovery method can lead to a better understanding of the recovery mechanisms in ultra-low permeability reservoirs.

2

0 -3 10

-2

10

-1

10

0

10

1

10

2

10

3

10

3.4. Comparison of the cyclic CO2 with nitrogen injection performance

Relaxation time T2,ms A simulation study was performed to compare the nitrogen and CO2 injection performance in the Xinjiang tight oil formations. The gas flooding process efficacy strongly depends on the phase behavior of the injected gas and crude oil mixtures. CMG-GEM was used to simulate compositional effects of injected fluid during enhanced oil recovery processes. Firstly, the simulation model was calibrated by history matching the experimental core flooding data as shown in Fig. 13. The relative permeability curves has substantial effect on the gas flooding recovery predictions. The simulation result shows that the scenario 1 (endpoint saturation of oil phase 1-Swc-Sor = 0.5) predicts a higher oil recovery than using lower end point values. There are slightly deviations between the simulation computed results with the observed data by NMR measurement. Our simulation case used a homogeneous core model which has not taken into account the impact of complex pore structure and pore size distribution on simulation results. It is essential to adopt a history matched model to predict the recovery performance by different injection fluids such as nitrogen and carbon dioxide. The predicted recovery by using an end point value of 0.45 matched closely with observed laboratory data (in Fig. 13). The calibrated model was then used to predict the recovery

Fig. 11. NMR measurements of gas huff-n-puff in tight cores.

and Sheng [26] used experimental methods to examine the role of soaking time and injection pressure on cyclic nitrogen injection performance in shale oil reservoirs. The significance of soaking time and injection pressure on shale oil recovery process is highlighted. The effect of injection pressure on shale oil recovery is consistent with our experimental results from NMR measurements. Li and Sheng conducted cyclic CH4 injection in the Eagle Ford Shale and their results indicated that the cyclic CH4 injection recovered more oil than cyclic nitrogen injection under the identical operation conditions [1]. There are many different factors that affect the cyclic injection efficacy in shale oil reservoirs. Hoffman and Evans observed that conformance control has been a major issue for cyclic CO2 injection in the Bakken formation. It suggests that some fractures connect between injectors and producers, which results in an early breakthrough and poor sweep efficiencies. There are extensive research in the literature on NMR characterization of pore structure of gas shales. The objective of this paper is to provide a new perspective to study the recoverable reserves at different pores by 569

Fuel 215 (2018) 561–571

T. Wan et al.

1

35 30 25 20 15 NMR measured data Simulation result scenario 1 Simulation result scenario 2 Simulation result scenario 3

10 5 0

1

2

3

4

5

Fig. 13. Simulation of nitrogen injection results compared with NMR measured data (core sample #e at 2500-psi injection pressure) Scenario 1: 1-Swc-Sor = 0.5, Scenario 2: 1-SwcSor = 0.45 and Scenario 3: 1-Swc-Sor = 0.4.

Krg Kro scenario 1 Krg Kro scenario 2 Krg Kro scenario 3

40

Relative permeability

Cumulative oil recovery factor, %

45

0.8

0.6

0.4

0.2

6

0 0.2

0.25

Number of cycles

0.3

0.35

0.4

0.45

0.5

Gas saturation

Cumulative oil recovery factor, %

performance by cyclic nitrogen and CO2 injection. The viscosity of crude oil is in cyclic change during cyclic nitrogen injection process as shown in Fig. 14. When the reservoir pressure is at 2500-psi at the injection period, the oil viscosity is reduced to 8.5 cp. In contrast, if the core was at depletion stage and nitrogen gas was produced, the viscosity of crude oil would increase to the original value at 11.2 cp. In comparison with nitrogen injection, the viscosity of crude oil under CO2 injection at 2500-psi is reduced dramatically to 0.07 cp calculated by the Pedersen viscosity model. The dissolved CO2 acts more than pressure maintenance, it also causes oil to swell. A comparison of cyclic nitrogen and CO2 injection performance at different injection cycle is shown in Fig. 15. The cyclic CO2 injection scenario is operated at the same conditions as the cyclic nitrogen injection. The rate of oil production for CO2 injection is substantially higher than cyclic nitrogen injection. The oil recovery profile by cyclic nitrogen injection flattens out from the 3rd cycle, but it is still increasing for cyclic CO2 injection at 5% for each cycle. It is more inclined for CO2 to form miscibility displacement with crude oil than nitrogen. The recovery mechanism by CO2 injection is well-established in the literature which includes oil swelling, viscosity reducing and hydrocarbon component extraction. More detailed discussion of comparing carbon dioxide recovery mechanisms with nitrogen can be referred to Masoud (2015) [27]. However, a higher injection pressure is required for the reservoir oil to attain miscibility with nitrogen [28]. The nitrogen could alternatively be used in immiscible form to push miscible front through the reservoir.

60 50 40 30 20 Cyclic nitrogen injection Cyclic CO2 injection

10 0

1

2

3

4

5

6

Gas injection cycles Fig. 15. Comparison of cyclic nitrogen and CO2 injection performance at different injection cycles.

4. Conclusions Laboratory examinations were conducted to improve our understanding on complicated dynamics of hydrocarbon production from tight formation rocks. The NMR response is capable of estimating the fraction of pore size distribution (micropores, mesopores and macropores) being occupied by the hydrocarbon and water.

Fig. 14. The pressure and oil viscosity changes with different nitrogen injection cycles.

1st cycle 2

3

4

5

6 cycle

570

Fuel 215 (2018) 561–571

T. Wan et al.

1. The NMR relaxation spectra reveals that most of the oil production happened in the first few cycles, less oil is recovered in the subsequent cycles. The recoverable oil of this field falls into a range of 1–100 ms with a peak centered on 10 ms. All gas injection NMR measurements show a dominant peak because signaled water was not introduced in the experiments. 2. For core plugs operating at a lower reservoir pressure, most of the oil is recovered at the first two cycles. While at a higher pressure operation, the recovery profile is not flatten out after the first two cycles but it keeps increasing in subsequent cycles. The contribution of convection flux to oil recovery is limited in tight oil reservoirs because liquid production is restricted by the high capillary resistance to flowing in micro pores. Due to the nanometer or micrometer scales of the pores, cycle depletion time has considerable effect on oil recovery from tight oil reservoirs. 3. The cumulative oil recovery increases with an increase of injected nitrogen pore volumes, but the oil recovery flattens out at a certain value indicating that an increase of nitrogen PVI more than a critical value would not have significant effect on improved oil recovery. A comparison of cyclic nitrogen and CO2 injection performance by simulation indicates that the cyclic CO2 injection is more favorable than nitrogen in tight formations.

org/10.2118/26471-PA. [7] Freedman R, Heaton N. Fluid characterization using nuclear magnetic resonance logging. Petrophysics 2004;45:241–50. [8] Meng M, Ge H, Ji W, Shen Y, Su S. Monitor the process of shale spontaneous imbibition in co-current and counter-current displacing gas by using low field nuclear magnetic resonance method. J Nat Gas Sci Eng 2015;27:336–45. [9] Xu H, Tang D, Zhao J, Li S. A precise measurement method for shale porosity with low-field nuclear magnetic resonance: a case study of the carboniferous–Permian strata in the Linxing area, eastern ordos basin, China. Fuel 2015;143:47–54. [10] Tan M, Mao K, Song X, Yang X, Xu J. NMR petrophysical interpretation method of gas shale based on core NMR experiment. J Petrol Sci Eng 2015;136:100–11. [11] Gao H, Li H. Determination of movable fluid percentage and movable fluid porosity in ultra-low permeability sandstone using nuclear magnetic resonance (NMR) technique. J Petrol Sci Eng 2015;133:258–67. [12] Fleury M, Romero-Sarmiento M. Characterization of shales using T1–T2 NMR maps. J Petrol Sci Eng 2016;137:55–62. [13] Cai Y, Liu D, Pan Z, Yao Y, Li J, Qiu Y. Petrophysical characterization of Chinese coal cores with heat treatment by nuclear magnetic resonance. Fuel 2013;108:292–302. [14] Fleury M, Soualem J. Quantitative analysis of diffusional pore coupling from T2store-T2 NMR experiments. J Colloid Interface Sci 2006;336:250–9. [15] Wang H, Lun Z, Lv C, Lang D, Ji B, Luo M, et al. Measurement and visualization of tight rock exposed to CO2 using NMR relaxometry and MRI. Sci Rep 2017;7:44354. http://dx.doi.org/10.1038/srep44354. [16] Gannaway G. NMR investigation of pore structure in gas shales. Paper SPE 173474 presented at SPE Annual Technical Conference and Exhibition, Amsterdam, The Netherlands, October 27–29; 2014 http://dx.doi.org/10.2118/173474-STU. [17] Tinni A, Odusina E, Sulucarnain I, Sondergeld C, Rai CS. Nuclear-magnetic-resonance response of brine, oil, and methane in organic-rich shales. SPE Reservoir Eval Eng 2015;18:400–6. [18] Fleury M. Characterization of shales with low field NMR. In: Presented at the international symposium of core analysts, Avignon, France, September 8–11; 2014; SCA2014-014. [19] Washburn EK. Relaxation mechanisms and shales. Concept Magn Reson 2014;Part A 43:57–89. [20] Mehana M, Elmonier I. Shale characteristics impact on Nuclear Magnetic Resonance (NMR) fluid typing methods and correlations. Petroleum 2016;2:138–47. http://dx. doi.org/10.1016/j.petlm.2016.02.002. [21] Li H, Guo H, Liu Q, Meng Z, Wang X, Sun Y. NMR experimental study of water displacing oil of tight oil reservoir. J Cent South Univ 2014;45:4370–6. In Chinese. [22] Zheng Y, Wan D, Ayaz M. Utilizing NMR mud logging technology to measure reservoir fundamental parameters in well site. Energy Power Eng 2013;05:1508–11. [23] Kausik R, Minh CC, Zielinski L, Vissapragada B, Akkurt, R, Song Y, et al. Characterization of gas dynamics in kerogen nanopores by NMR. Presented at the SPE annual technical conference and exhibition; 2011; http://dx.doi.org/10.2118/ 147198-MS. [24] Wan T, Sheng JJ. Compositional modelling of the diffusion effect on EOR process in fractured shale-oil reservoirs by gas flooding. J Can Pet Technol 2015;54:107–15. http://dx.doi.org/10.2118/2014-1891403-PA. [25] Sondergeld CH, Ambrose RJ, Rai CS, Moncrieff J. Micro structural studies of gas shales. Presented SPE-131771 at SPE unconventional gas conference, Pittsburgh, Pennsylvania, USA, February 23–25; 2010. [26] Yu Y, Sheng JJ. A comparative experimental study of IOR potential in fractured shale reservoirs by cyclic water and nitrogen gas injection. J Petrol Sci Eng 2016;149:844–50. [27] Masoud M. Comparing carbon dioxide injection in enhanced oil recovery with other methods. Austin Chem Eng 2015;2(2):1019. [28] Todd HB, Evans JG. Improved oil recovery IOR pilot projects in the Bakken Formation. In, SPE 180270 presented at Low Perm Symposium, 5–6 May, Denver, Colorado, USA; 2016.

Acknowledgements The experimental work supported by Peking University is highly appreciated. This work was completed when the author worked with Petrochina. The support from Science Foundation of China University of Petroleum-Beijing at Karamay (No. KL01JB201700003) is highly appreciated, and supported by 973 Program (2015CB250902). We highly appreciate the detailed and valuable comments from two anonymous referees. References [1] Li L, Sheng JJ. Experimental study of core size effect on CH4 huff-n-puff enhanced oil recovery in liquid-rich shale reservoirs. J Nat Gas Sci Eng 2016;34:1392–402. http://dx.doi.org/10.1016/j.jngse.2016.08.02. [2] Meng X, Sheng JJ, Yu Y. Experimental and numerical study on enhanced condensate recovery by gas injection in Shale gas condensate reservoirs. SPE Reservoir Eval Eng 2016. http://dx.doi.org/10.2118/183645-P. Preprint. [3] Song C, Yang D. Experimental and numerical evaluation of CO2 huff-n-puff processes in Bakken Formation. Fuel 2017;190:145–62. [4] Wan T, Yu Y, Sheng JJ. Experimental and numerical study of the EOR potential in liquid-rich shales by cyclic gas injection. J Unconventional Oil Gas Resour 2015;12:56–67. [5] Freedman R. Advances in NMR logging. J Petrol Technol 2006;58:60–6. http://dx. doi.org/10.2118/89177-JPT. [6] Howard JJ, Spinler EA. Nuclear magnetic resonance measurements of wettability and fluid saturations in Chalk. SPE Adv Technol Ser 1995;3:60–5. http://dx.doi.

571