Pore structure and fluid mobility of tight carbonate reservoirs in the Western Qaidam Basin, China

Tight carbonate reservoirs in the Western Qaidam Basin have complex lithologies and pore structures. The oil–water mobility law in reservoirs has not yet been completely determined, restricting the formulation of rational reservoir development methods. To bridge this gap, in this study, we used several test methods, such as casting thin sections, mercury intrusion, and nuclear magnetic resonance, to obtain the pore structure and oil–water displacement characteristics of tight carbonate reservoirs in the Western Qaidam Basin. The pore structures of the reservoirs could be categorized into three types: microfractures + dissolved pores + micropores (MFD), microfractures + micropores (MF), and matrix (M). The characteristics of single‐phase oil seepage and water flooding in reservoirs with various pore structures differed evidently. For the MF‐ and M‐types, the water‐locking effect caused by the strong capillary force affected oil charging in the micropores. The effect of the pressure drop on the MFD‐type algal limestone was less than that on the MF‐type limestone (dolomite) because of the occurrence of a non‐Darcy flow. The MFD‐type, which contained microfractures, had preferential seepage channels, resulting in obvious fluid channeling and low water displacement efficiency. Oil−water displacement mainly occurred in the dissolved pores and microfractures, suggesting that starting oil accumulation in the micropores was crucial. This study will assist in efficient development of tight carbonate reservoirs in the Western Qaidam Basin.


| INTRODUCTION
2][3][4][5] The Dafengshan region is located in the western-northern part of the Qaidam Basin, adjacent to the Mangya Depression in the south.Specifically, it includes many geological structures, such as the Jiandingshan, Heiliangzi, Changweiliang, and Dafengshan, with areas of approximately 4200 km 2 (Figure 1). 6,7The Nanyishan region includes a tertiary structure in the Nanyishan anticline belt in the Mangya Depression subarea of the western depression in the Qaidam Basin, Qinghai Province.The Fengxi and Nanyishan blocks of the Qinghai oilfield contain thinbedded algal limestone and limestone reservoirs with low porosity, low permeability, and strong heterogeneity.Oil productivity in the Nanyishan region is relatively high, while the Fengxi region is a potential area for reserve growth and production.However, to date, the seepage F I G U R E 1 Location and structural division of the Western Qaidam Basin. 6haracteristics of these reservoirs have not been sufficiently studied.Therefore, developing a method to enhance oil recovery through seepage experiments is necessary.
][10][11][12] The coexistence of small and large pores results in a heterogeneous pore distribution and high fractal dimension. 13Reservoir quality increases with the degree of complexity of microscopic pore structures, and multiple approaches have been used to characterize the continuous distribution of pore networks in heterogeneous dolostones. 14The characteristics of both the pore structure and movable fluid are significant properties for controlling the flow regularity in pores in tight reservoirs. 15Low-velocity non-Darcy flows exist in shale and tight reservoirs because of rock-liquid interactions. 168][19][20][21] Different pore types in carbonate reservoirs exhibit different two-phase seepage characteristics. 227][28] In these carbonate reservoirs, the water flowing through these channels displaces only some amount of the oil in the connected vugs. 29However, the oil-water relationship in this type of algal limestone + limestone carbonate reservoir, which has a complex lithology and pore structure, has not been completely determined; additionally, determining effective measures to enhance oil recovery remains difficult.These issues need to be urgently addressed by studying the flow mechanisms of carbonate reservoirs in the Western Qaidam Basin.This can improve our understanding of such reservoirs and their developmental effects.Therefore, based on the pore structure characteristics of the carbonate reservoirs in the Western Qaidam Basin, we conducted an experimental study of oil-water flow and investigated the effect of pore structure on fluid flow to improve our understanding of the characteristics of oil production in different-sized pores with different pore structure types.

| Samples and materials
Core samples were obtained from carbonate reservoirs in the Fengxi and Nanyishan regions of Western Qaidam Basin.The core samples from the Fengxi region were acquired from Shangganchaigou and Xiayoushashan formations at depths of 3276-4015 m.The samples from the Nanyishan region were from Shangganchaigou and Shangyoushashan formations at depths of 1031-1519 m.The diameter and length of the cores were 2.5 and 5.0 cm, respectively.Thermoplastic pipes were used to wrap the cores to ensure the integrity of the experimental samples.The fluids used in the experiments were simulated formation water and simulated oil. 30,31The ionic composition of the simulated formation water is listed in Table 1.The simulated oil comprised crude oil and kerosene.Nitrogen was used as the gas source for the gas permeability tests.

| Experimental setup and conditions
The design of conventional core flow equipment is shown in Figure 2. Details of the equipment composition can be found in a previous study. 32According to calculations based on the test data, the overburden pressure reached 30 MPa.Thus, the maximum confining pressure in the experiment was set to 30 MPa to simulate actual reservoir conditions.

| Experimental procedures
Before the flow experiments, the samples were sent to laboratory for analysis.We analyzed the microscopic pore throat structures using scanning electron microscopy (SEM), casting thin section observations, highpressure mercury intrusion (HPMI) tests, and nuclear magnetic resonance (NMR) measurements.The NMR T 2 relaxation time and pore radius r were found to be | 3399 related. 33,34According to a previous study, 14 we determined the oil-water displacement characteristics of different types of pore spaces in this study.
The flow experiments were conducted as follows: (1) The core samples were saturated with simulated formation water using a vacuum pump after washing and drying the samples.(2) Simulated oil was used to displace the simulated formation water in the cores.(3) The NMR T 2 spectra of the cores maintaining a bound water state were obtained.(4) The fluid mobility in the core was measured under different confining pressures, and the flow curve of the core was plotted to analyze the threshold pressure gradient and study the flow characteristics of the oil phase.
(5) A water flooding experiment was conducted to investigate two-phase oil-water flow.During displacement, the NMR T 2 spectra of the cores were obtained for different injection quantities.

| Lithology
The lithological classification of carbonate rocks is complex.In this study, we adopted the classification method of a predecessor, which can be classified according to composition and structure of reservoirs. 35,36s shown in Figure 3, the reservoir lithology mainly comprised algal limestone, limestone, dolomite, mudstone, and sandstone.The petrophysical properties of limestone and dolomite varied considerably.The limestone surface was rough and gray, whereas the dolomite surface was smooth and mostly white.Moreover, the grain structure of dolomite was finer than that of limestone.In addition, most dolomites were generated from the dolomitization of calcite, which could generate abundant nano-scale intercrystalline pores in oil reserves.The Cenozoic algal limestone in the Western Qaidam Basin was believed to be formed mainly in a coastal and shallow lake environment. 6,37In this study, we determined that the lithology in the Fengxi region was mainly calcite dolomite, algal limestone, and calcilutite, whereas that in the Nanyishan region was mainly limestone, algal limestone, and dolomite.

| Physical characteristics
As shown in Figure 4, the permeabilities and porosities of the samples from the Fengxi region were in ranges of 0.00294-18.6× 10 −3 μm 2 and 1.3%-12.7%,respectively, while those from the Nanyishan region were 0.0196-120 × 10 −3 μm 2 and 8.8%-22.7%,respectively.Overall, the reservoir permeability in the study area had a wide distribution range.The permeabilities of samples with different lithologies differed significantly (Figure 5).The permeabilities of limestone and algal limestone were higher than those of calcitedolomite and dolomite.Fractures developed in some of the sandy mudstones; therefore, the permeability of the mudstone was high.

| Microscopic analysis
According to the thin-section and SEM analysis data (Figure 6), the pore system of the tight carbonate reservoirs comprised nano-intercrystal, microintergranular, oomoldic, and nanometer-to millimeterscale dissolution pores and microfractures.These results have been reported in previous studies as well. 38From a lithological perspective, pore structures can be divided into two types: pore structures developed by algal texture traces of algal limestone, and pore structures dominated by horizontal bedding of limestone, dolomite, and mudstone or sandstone.The pore types of the algal limestone were mainly of the micropores + microfractures + dissolved pores (MFD) type.The pore types in limestone, dolomite, and mudstone or sandstone were mainly micropores + microfractures (MF).The physical properties of the algal limestone depended on the development degree of microfractures and dissolved pores.The physical properties of limestone, dolomite, and mudstone or sandstone depended on the microfracture development degree.By comparing several pore  throat sizes, we found that microfractures and dissolved pores constituted the main seepage space.Micropores were generally smaller than 1 μm; thus, they were on the nanometer to micrometer scale.Compared to microfractures and dissolved pores, micropores accounted for the main reservoir space.Therefore, microfractures and dissolved pores were the main seepage channels in the study area.The degree of development of microfractures and dissolved pores determined the physical properties of the reservoir.
Because large dissolved pores developed mainly in the algal limestone, where algal texture traces were retained, the number and distribution of the dissolved pores were limited compared with those of the fractures.Therefore, in this study, we focused on the microfractures.Micron-level fractures had the highest effect on the seepage ability of the reservoir rocks; thus, we selected representative casting thin sections of samples from the Fengxi and Nanyishan reservoir rocks and used an Olympus stereomicroscope to observe the degree and characteristics of the fracture development over the entire thin section under ×40 magnification.At this magnification, the fractures above 2 μm could be observed clearly (Figures 7 and 8).The observations revealed that the fractures in the algal limestone were often multidirectional, and that the connectivity between the fractures was relatively good.In other lithologies, such as micrite, calcite dolomite, and mudstone, fractures most often developed along the stratum plane, but few vertical or oblique fractures existed.The connectivity between the fractures was related to their density; that is, the fewer the number of fractures, the poorer the connectivity between them.

| HPMI analysis
HPMI can be used to quantitatively study the pore structures of porous media.Figure 9 shows the pore throat size distributions obtained through HPMI measurements conducted on various lithologies.The results  indicated that the main pore type of the algal limestone in the study area was MFD, whereas that of the limestone (dolomite) was MF.The differences between these two types of pore structures are clearly illustrated in the mercury intrusion curves.According to the mercury intrusion saturation, the volume of the microfractures accounted for a maximum of approximately one-eighth of the total pore volume, and the micropore size was mainly 0.003-0.1 μm, while the size of the dissolved pores was mainly 0.1-2 μm, and that of the microfractures was mainly 3-100 μm, which was consistent with the casting thin-section observations.

| NMR analysis and pore structure classification
NMR can be used to record the relaxation times of hydrogen protons in static and pulsed magnetic fields and to evaluate the petrophysical properties (porosity, pore size distribution, permeability, and fluid status) of rocks. 21,39,40If oil and water are used directly in the experiment, distinguishing the contributions of oil and water would be difficult to the NMR signal.In this study, manganese chloride was added to the simulated formation water for shielding.A fluid comprising simulated formation water and manganese chloride was used to obtain the NMR T 2 spectra.The amount of manganese chloride added was 8 wt%. Figure 10 shows the test results obtained using the simulated formation water and manganese chloride fluid.The nuclear magnetic signal weakened after manganese chloride was added to the simulated formation water.This suggested that manganese chloride exhibited a good shielding effect on the nuclear magnetic signal of hydrogen protons in the simulated formation water.
As shown in Figure 11, the positions and numbers of peaks in the NMR T 2 spectra of the algal limestone, limestone (dolomite), and dolomite from the Nanyishan and Fengxi regions showed certain differences.The T 2 spectra of the Nanyishan and Fengxi core samples were characterized by bimodal distributions and were mostly left-skewed, with the main peak primarily located at 1-10 ms.We used NMR to evaluate the pore structures of the rocks.According to the related theories of NMR and HPMI, 21 the relationship between NMR T 2 and the pore throat radius can be expressed as where C is the conversion coefficient for nonlinear conversion.The parameters C and n can be obtained from a power law regression between the transverse relaxation time T 2 and the pore throat radius r.Sample NQ157 was used as an example to demonstrate the correlation of T 2 with r (Figures 12 and 13).Table 2 shows the regression results of the core plugs obtained using the power law regression models.The low T 2 components (<10 ms) corresponded to the <0.2 µm micropores.The other peak was located at 10-100 ms, which mainly corresponded to larger dissolved pores and microfractures with sizes of 0.2-6.9μm.According to our research results, the relaxation time of microfracture signals in cores was longer than that of pore signals.However, microfracture porosity was generally low; therefore, the amplitude of the relaxation signal of the microfractures was low.As indicated by the NMR results of our study, the signals of the long T 2 components (>50 ms) for the analyzed carbonates originated from the microfractures.
As previously mentioned, the pore types of the algal limestone from the Fengxi and Nanyishan regions were mainly MFD.The pore type of limestone (dolomite) was mainly MF.In addition to an extremely small percentage of limestone (dolomite) and mudstone, the shale content was high, pores were not developed (porosity < 3%), and the permeability was poor.This type of pore structure is referred to as matrix type (M).Therefore, the pore structures of the carbonate reservoirs in the Fengxi and Nanyishan regions could be categorized into three types: MFD, MF, and M. Furthermore, microfractures were developed in all three types of pore structures.However, the degree of microfracture development may have differed, which had an important effect on the seepage characteristics.The pore structure division described above may not adequately explain this seepage law.Therefore, in this study, the degree of microfracture development was characterized using the response characteristics of the NMR T 2 spectra and the results of mercury intrusion experiments.The permeability (K) and physical property (K•φ) of the core samples were used as the criteria for determining the microfracture development in the study area.The criteria used to determine the degree of microfracture development are listed in Table 3.

F I G U R E 12
Cumulative saturation with pore radius r measured using high-pressure mercury intrusion (HPMI) and cumulative saturation with transverse relaxation time T 2 measured using nuclear magnetic resonance (NMR) for sample NQ157.
F I G U R E 13 Regressed correlations between T 2 and r for Sample NQ157.
T A B L E 2 Correlations for T 2 and r obtained using power regression models.

| Single-phase oil seepage
We tested the characteristics of single-phase oil seepage in four core samples by simulating the pressure of the overlying strata in an actual reservoir.The relationship curve between the permeability and pressure gradient at each pressure point was obtained using the Darcy equation (Figure 14).The results showed that the oil-phase permeability of the reservoirs gradually decreased as the pressure gradient decreased, indicating that single-phase oil seepage in the Fengxi and Nanyishan reservoirs can be characterized by non-Darcy seepage.For example, for the limestone (dolomite) sample FX412 collected from the Fengxi region, when the pressure gradient decreased from 0.8 to 0.04 MPa/cm, the oil-phase permeability decreased from 0.0035 × 10 −3 μm 2 to 0.002 × 10 −3 μm 2 , thus, suggesting that well spacing should be appropriate during exploitation.If the well spacing is too large during development, it may cause poor seepage of some amount of the oil or the oil may not flow.Additionally, the variation in permeability with the pressure gradient was complicated, which has been discussed further in the next section.

| Analysis of NMR results
We conducted NMR analysis of the four core samples collected from the Fengxi and Nanyishan regions during displacement.The T 2 spectra and NMR images of the initial oil, 0.5 PV water injection, and residual oil states are shown in Figures 15 and 16, respectively.In the NMR T 2 spectra, the black line represents the state of the initial oil under the condition of bound water, the red line represents the state of water injection at 0.5 PV, and the blue line represents the residual oil state.In the NMR images, different colors represent the amount of saturated oil in the core, and the amount of oil decreases from red to blue.The experimental results showed that the T 2 spectrum of Sample NQ157 had two and three peaks before and after water flooding, respectively.The spectrum of Sample FX1185 exhibited two peaks, and the area of the left peak accounted for a small proportion.After water flooding, the two peaks changed significantly, with the left and right peaks increasing and decreasing, respectively.The T 2 spectra of Samples NQ725 and FX1613 exhibited approximately unimodal distributions before and after water flooding.As shown in Figure 16, the distribution of the oil phase in the sample changed significantly with increasing injection time, and the oil content generally decreased.The oil distribution in all sample types differed to some extent.Particularly, the oil distribution in Sample NQ157 was relatively uneven under the initial oil, 0.5 PV injection, and residual oil states.This suggested that the seepage channel exhibited strong anisotropism, whereas for Samples FX1185 and NQ725, the oil-phase distribution in each stage was relatively uniform.

| Oil charging and displacement efficiency
The oil saturations under bound-water conditions are listed in Table 4. Oil charging in the samples was insufficient.Contrastingly, the degree of oil-charging in the Fengxi Reservoir was higher than that in the Nanyishan Reservoir, and the degree of oil-charging in limestone (dolomite) was higher than that in algal limestone.The water displacement efficiencies of the samples were also analyzed in this experiment, and the corresponding results are presented in Table 4.The water displacement efficiencies of the Fengxi and Nanyishan samples differed significantly.The water displacement efficiency of the Nanyishan samples ranged from 24.3% to 33.2%, with an average of 28.8%, whereas that of the Fengxi samples ranged from 23.8% to 39.0%, with an average of 31.4%.In terms of lithology, the displacement efficiency of the algal limestone NQ157 sample was relatively low, whereas those of the limestone (dolomite) NQ725 and FX1185 samples were relatively high.Moreover, the displacement efficiency of the limestone (dolomite) was significantly higher than that of the sandstone.

| Non-Darcy flow
We found that micropores developed in the Feng Xi and Nanyishan samples.Although microfractures were also well developed, they were sensitive to effective stress; as the effective stress increased, the microfractures would close.Thus, under low-pressure conditions, micropores play an important role in the fluid flow; particularly, micropores with a smaller pore size result in stronger rock-fluid interactions. 41,424][45][46] Non-Darcy flows have also been observed in the carbonate reservoirs in the Yingxi region. 32Under a low-pressure gradient, the fluid boundary layer in the micropore channel was extremely thick, and marginal amount of free internal fluid participated in the flow.The thickness of this layer decreased with an increasing pressure gradient, and more free fluid in the micropore channel participated in the flow.Thus, the fluid seepage capacity increased as the pressure gradient increased.Moreover, the variations in the permeabilities of the different types of pore structures differed.For the MF-type samples, the permeability decreased rapidly as the pressure gradient decreased during the low-pressure stage.However, for the MFD-type samples, the permeability slowly decreased as the pressure gradient decreased.In the limestone (dolomite) samples, the variation in seepage capacity was mainly affected by the interaction forces between the fluid and micropores under low-pressure gradients.In addition to micropores, dissolved pores developed in the algal limestone.The dissolution pores were relatively large.Thus, under a low-pressure gradient, the dissolved pores significantly affected the seepage capacity, weakening the effect of the boundary layer fluid on seepage.This finding also suggested that the effect of the pressure drop on the MFD-type algal limestone was less than that on the MF-type limestone (dolomite) because of the non-Darcy flow.

| Oil and water displacement characteristics
According to the NMR images, the oil content distribution in all core samples differed to some extent.This difference was mainly due to the lithology and pore structure.Sample NQ157 was an MFD-type algal limestone with well-developed microfractures.The dissolved pores and microfractures were well-developed.During water flooding, the fluid entered the microfractures or dissolved pores that formed the channels with the least resistance, and channeling occurred easily, resulting in an oil distribution that exhibited relatively strong heterogeneity.Samples FX1185 and NQ725 were MFtype limestones (dolomite) with poor microfracture development.The fluid distribution was relatively uniform.Sample FX1613 was an M-type with poor microfracture development.Due to the insufficient pore development of the sample (Figure 16B), the oil saturation of this sample was limited; therefore, the NMR images for each stage did not change significantly.Variations in the NMR T 2 spectra during water flooding reflected the characteristics of production in different pore spaces.As shown in Figure 15, the peaks of the T 2 spectra were lower after water flooding than before water flooding, except for the T 2 spectra between 0.1 and 1 ms.For example, the T 2 spectra peaks of samples FX1185 and NQ157 in the range of 0.1-1 ms were higher after water flooding than before water flooding.The reason for this anomaly was that the high fluid pressure during water flooding caused some of the micropores to be punctured.The oil entered the punctured micropores, increasing the corresponding peaks in the T 2 spectra and inevitably reducing the oil displacement efficiency.To gain a better understanding of the characteristics of oil production in different pore sizes, we divided the relaxation times into three different zones: 0.1-10, 10-100, and 100-1000 ms.As shown in Figure 17, the overall water displacement efficiency of the 0.1-10 ms relaxation time interval in the T 2 spectrum was the lowest.This result indicated that the water displacement efficiency in the micropores was the lowest during water flooding.Thus, during water flooding in the study area, initiation of oil accumulation in the micropores should be prioritized.

| Effect of pore structure on oil charging and water flooding
As previously observed, the oil charging in the reservoir was incomplete.The degree of oil charging was affected by fractures, pore throat size, and capillary force.Micropores with various pore structures developed in tight carbonate reservoirs.Charging oil in a large number of intergranular micropores was difficult.The resistance provided by the capillary force increased as the oil entered the micropores, resulting in a significant waterlocking effect.Additionally, the presence of fractures increased the heterogeneity of oil charging in the micropores.Both these factors led to inadequate charging in carbonate reservoirs.For MF-and M-type lithologies, the first factor played a dominant role, whereas for MFDtype lithologies, the second factor was dominant.Water displacement efficiency is also related to the pore structure. 10,32We found that the water displacement efficiency of the MFD-type with developed microfractures was significantly lower than that of the MF-type with poor microfracture development.As the fluids flowed in the MFD-type lithologies with developed microfractures, the microfractures and dissolved pores constituted the main preferential seepage channel, and the fluid-channeling phenomenon became obvious (Figure 16C).Additionally, under high-pressure injection, some micropores were punctured; consequently, oil entered these punctured micropores.Thus, the water displacement effect was not ideal.Previous studies have also reported this phenomenon in fractured carbonate reservoirs and concluded that micropores have great potential. 23,24,29For MFD-type algal limestones with developed microfractures, initiating the residual oil in the micropores is essential.Recovery from a fractured core is significantly better under low-speed or lowpressure injection conditions. 47,48This is mainly because, under low-speed or low-pressure injection conditions, the oil in the micropores can be effectively displaced into the fractures through imbibition.During displacement and imbibition, the synergistic effects of the capillary and driving forces can be maximized to achieve highefficiency water displacement.This suggested that water flooding issues can enhance the displacement efficiency of algal limestone reservoirs in the Western Qaidam Basin.

| CONCLUSIONS
Insights into the oil-water mobility law support the ability to enhance oil recovery in tight carbonate reservoirs in the Western Qaidam Basin.In this study, we focused on the effects of lithology, pore structure, and fracture development on oil-water mobility.We observed that a non-Darcy flow occurred in tight carbonate reservoirs.Particularly, for the MF-type limestone (dolomite), appropriate well spacing was the key to avoiding poor seepage of some amount of the oil.Water displacement efficiency in the micropores was the lowest during water flooding.Thus, during water flooding in the study area, initiating oil accumulation in the micropores should be prioritized.To achieve high-efficiency water displacement, water flooding methods should be used in the field.Furthermore, adding surfactants to water can enhance oil recovery.Additionally, high-pressure injection resulted in fluid channeling and puncturing of the micropores.Therefore, moderate injection pressure is important.

F
I G U R E 4 Porosity versus permeability for the carbonate reservoirs.F I G U R E 5 Comparison of permeabilities of the core samples with different lithologies.

F
I G U R E 8 Typical pore structure and fracture distribution characteristics of the Nanyishan Reservoir.(A) Algal limestone, 1392.70 m, 13.7% porosity, 5.09 mD permeability; and (B) sandy mudstone interbedded with calcareous sandstone, 1397.19 m, 11.91% permeability, 4.92 mD permeability.F I G U R E 9 Pore throat size distributions of the various lithologies obtained via high-pressure mercury intrusion measurements.

F
I G U R E 10 Nuclear magnetic resonance T 2 spectra response characteristics of the fluid.Simulated formation water + manganese chloride for the (A) Fengxi and (B) Nanyishan reservoirs.F I G U R E 11 Nuclear magnetic resonance T 2 spectra of the core samples from the Fengxi and Nanyishan reservoirs under the initial oil saturation.

F
I G U R E 14 (See caption on next page).

F
I G U R E 14 Plot of the oil phase permeability versus pressure gradient.The permeability increases as the pressure gradient increases.(A) Sample FX412 is limestone (dolomitic) and MF-type with developed microfractures; (B) sample FX965 is limestone (dolomitic) and MF-type with poor microfracture development; (C) sample NQ159 and (D) Sample NQ386 are algal limestone and MFD-type with developed microfractures.F I G U R E 15 (See caption on next page).

F
I G U R E 15 Nuclear magnetic resonance T 2 spectra for different stages during water flooding.(A) Sample FX1185 is MFtype with poor microfracture development; (B) sample FX1613 is M-type with poor microfracture development; (C) sample NQ157 is MFD-type with developed microfractures; and (D) sample NQ725 is MF-type with poor microfracture development.F I G U R E 16 Nuclear magnetic resonance images of the different stages during water flooding.Different colors represent the amount of saturated oil in the core, and the amount of oil decreases from red to blue.
T A B L E 4 Experimental results of oil-water displacement.