Quantifying the Microscopic Distribution of Different Fluid Types in the Multiscale Pore-throat Structure of the Chang 7 Reservoir in the Longdong Area, Ordos Basin

Abstract

Fluid type and content directly control fluid mobility in tight reservoirs. At present, there are two ways to classify fluid types. One is to classify fluids into movable fluids (MF), capillary-bound fluids (CAF), and clay-bound fluids (CLF). The other is to classify fluids into free fluids and adsorbed fluids. However, the intrinsic relationship between the two fluid classification schemes is still unclear. In order to investigate the pore structure and fluid type characteristics, a series of experiments were performed on the Chang 7 tight sandstone in the Longdong area. The full-scale pore size distribution (PSD) can be obtained by combining low-temperature nitrogen adsorption (LTNA) with nuclear magnetic resonance (NMR). The PSD of Chang 7 tight sandstones primarily ranges from 1 nm to 20 μm. Based on the fractal characteristics, pore system is divided into macropores (mainly >150.8 nm), and micropores (mainly <150.8 nm). MF, CLF, and CAF constitute 11.1–49.7% (avg. 34.6%), 17.3–40.1% (avg. 26.7%), and 32.9–48.8% (avg. 38.7%) of total fluid, respectively. Additionally, macropores are positively correlated with MF and negatively correlated with CAF and CLF, whereas micropores show the opposite trend. By integrating NMR data with theoretical modeling, a clear correspondence between the two fluid classification approaches was established: MF and CAF closely correspond to free fluids, while CLF is strongly associated with adsorbed fluids. Notably, in samples with low porosity and extremely fine pore throats, the combination of centrifugation and theoretical modeling may underestimate the actual free fluid content. MF shows a weak negative correlation with quartz and clay minerals; CAF and CLF are weakly positively correlated with quartz and clay but negatively correlated with feldspar. The occurrence patterns of different fluid types within various pore-throat structures were established, revealing the relationships among mineral composition, pore size, pore-throat structure, and fluid distribution. These findings provide valuable insights into the pore structure and fluid distribution of tight sandstone reservoirs, enhancing the understanding of fluid behavior in unconventional systems.

1. Introduction

With the growing global demand for energy, unconventional tight oil resources have gained increasing importance. , Due to their widespread distribution and substantial resource potential, these resources are expected to supplement conventional oil resources in the future. China possesses vast continental shale oil (or tight oil) reserves and has made significant advancements in the exploration and development of shale oil in recent years. − In 2023, China’s shale oil production reached 4 million tons, with 221 × 10⁴ tons annually sourced from the Chang 7 member of the Ordos Basin. , Despite these achievements, both production volume and recovery rates from the Chang 7 shale oil remain relatively low, with fluid mobility being a critical factor limiting production and recovery. The content and distribution of various fluid types play a crucial role in determining fluid mobility within reservoirs. However, the Chang 7 shale oil reservoirs are tight and characterized by complex pore-throat structures, making it difficult to accurately characterize fluid content and distribution. Therefore, precisely defining the types and distribution of fluids is essential for understanding the mobility mechanisms in Chang 7 shale oil reservoirs.

Previous studies have classified reservoir fluid types into movable fluid (MF) and bound fluid based on logging analysis. Some scholars have further subdivided bound fluids in tight reservoirs into capillary-bound fluids (CAF) and clay-bound fluids (CLF). Among these, MF can overcome capillary pressure and can move freely, whereas CAF is confined by capillary pressure within the pores but still retains the potential for flow. CLF, which is electrochemically bound to the surface of clay particles, remains immobile and is considered to be an unextractable portion. − In the shale oil system, fluids are typically categorized as either free fluids or adsorbed fluids, depending on their state of occurrence. However, the relationship between free fluids, adsorbed fluids, MF, CAF, and CLF remains unclear. Moreover, the connection between these two fluid classification schemes has received limited attention. This paper aims to investigate and clarify these relationships.

Reservoir fluid mobility is a crucial factor in evaluating oil resources and has become a prominent research topic. It is considered a more effective characterization parameter than traditional metrics, such as porosity and permeability, as it provides a better foundation for predicting reservoir productivity and guiding development strategies. − Experimental methods, including Computed Tomography (CT), Nuclear Magnetic Resonance (NMR), high-speed centrifugation, microflow displacement, and spontaneous imbibition, in conjunction with theoretical molecular simulations, have been employed to study fluid mobility. ,−

CT scanning can be used to construct a three-dimensional pore-throat network model, enabling quantitative structural characterization and microscopic fluid detection in complex pore-throat systems. However, in the Chang 7 member of the study area, the pore throats in the dense sandstone are relatively fine, and the proportion of nanopores is high. One limitation of CT scanning is its inability to simultaneously achieve high resolution while covering large sample sizes.

Moreover, CT scanning is typically used alongside other techniques, such as Nuclear Magnetic Resonance (NMR), displacement, gas injection (CO₂ flooding/huff-n-puff, N₂ flooding), water injection, chemical flooding, or spontaneous imbibition experiments. However, its application is limited to characterizing only movable and bound fluids. Dong et al. and Xu et al. combined displacement and centrifugation–NMR methods to evaluate fluid mobility characteristics in different pore spaces of tight sandstone reservoirs. However, the choice of centrifugal and displacement forces may lead to varying results. Additionally, these methods are not capable of effectively characterizing CAF and CLF. In terms of molecular simulation, Li et al. combined NMR centrifugation experiments with molecular simulation data to establish a theoretical model for quantifying free water and determining the distribution of adsorbed and free water in shales. However, no similar research has been conducted on tight sandstone reservoirs. Zhao et al. used molecular simulations to interpret the adsorption effects of clay minerals on fluids at the nanoscale. They classified the nanoconstrained fluids into three regions: completely adsorbed, strongly influenced by the pore wall, and weakly influenced, thereby distinguishing between free fluids and adsorbed fluids. However, similar research has not yet been applied to the Chang 7 tight sandstone reservoir.

To distinguish bound fluids, the combination of NMR with heating or spontaneous imbibition is commonly employed. Yuan et al. discussed the impact of temperature changes on pore fluids and determined the T ₂ cutoff value for CLF in the Permian Carynginia shale. However, heating experiments may induce irreversible changes in the pore structure of samples, and the applicability of this method to tight sandstones remains unverified. Tian et al. designed a series of NMR experiments (saturation, centrifugation, and spontaneous imbibition) to determine the distribution characteristics of different fluid types in the conglomerate of the Triassic Baikouquan Formation in the Junggar Basin. Nevertheless, the conditions of spontaneous imbibition experiments may significantly differ from actual production conditions, potentially impacting water-sensitive minerals, altering the pore structure, and affecting the experimental results. This method has also not been thoroughly studied in the Chang 7 tight sandstone reservoir. Liu et al. summarized the application of T ₂ cutoff values, T _2C1 and T _2C2, to distinguish MF, CAF, and CLF. Other scholars have also considered additional factors influencing T ₂ spectra and proposed methods for determining T _2C values. − However, T _2C2 values vary across different study areas. Moreover, in tight reservoirs, there is considerable overlap between CAF and CLF on the T ₂ spectra, and MF is almost evenly distributed across pores of varying sizes. ,, Therefore, the T _2C method is not suitable for evaluating the distribution of different fluid types in tight sandstones.

Currently, research on the characteristics of fluids in the Chang 7 tight oil reservoirs is limited, and the primary factors influencing the distribution of movable fluids remain unclear. Further investigation into the characteristics of movable fluids in these tight reservoirs is therefore essential. This study employs various rock physical properties, including casting thin section (CTS), X-ray diffraction (XRD), FE-SEM, low-temperature nitrogen adsorption (LTNA), high-pressure mercury injection (HPMI), and NMR of saturated samples, to investigate the pore structure and conduct comprehensive pore-size characterization. Based on a series of NMR experiments (saturation, centrifugation, and heating), MF, CAF, and CLF are differentiated. By analyzing NMR data under varying centrifugal forces and integrating theoretical models, the distributions of free and adsorbed fluids are obtained. Finally, the interconnections between these two classification schemes are explored, and the factors influencing the distribution of different fluid types are discussed. The results aim to provide a theoretical foundation for the exploration and development of tight sandstone reservoirs.

2. Samples and Experimental Methods

2.1. Samples

Six fine-grained sandstone samples were collected from the Chang 7 Member of the Upper Triassic Yanchang Formation in the Longdong area of the Ordos Basin, provided by PetroChina Changqing Oilfield Company, with depths ranging from 1630.01 to 2317.68 m. The Chang 7 Member in this region is characterized by the development of sandy debris flow deposits and turbidity current deposits, formed within semideep to deep lacustrine subfacies environments. Stratigraphically, the Chang 7 Member is subdivided into three submembers: Chang 7₁, Chang 7₂, and Chang 7₃. The Chang 7₁ and 7₂ submembers are dominated by tight sandstones, reflecting relatively high-energy depositional conditions associated with gravity-driven sediment transport, including sandy debris flows and turbidity currents. In contrast, the Chang 7₃ submember consists primarily of organic-rich black shale and functions as a major hydrocarbon source rock. The tight sandstones of the Chang 7₁ and 7₂ submembers constitute the primary focus of this study. The collected core samples were cut into cylindrical specimens approximately 2.5 cm in diameter, with each sample drilled parallel to the bedding. These samples were subsequently subdivided for various analytical techniques, including X-ray diffraction (XRD), low-temperature nitrogen adsorption (LTNA), high-pressure mercury intrusion (HPMI), field emission scanning electron microscopy (FE-SEM), and nuclear magnetic resonance (NMR). Prior to analysis, the core pieces were rinsed with distilled water to remove drilling mud and subjected to a three-day Soxhlet extraction using a dichloromethane–methanol mixture (9:1 v/v) to eliminate residual hydrocarbons.

2.2. Experimental Methods

2.2.1. XRD

XRD analysis was performed using the Bruker AXS D8 Discover X-ray diffractometer at 40 kV and 30 mA, with 2θ range of 3°–85° under Cu Kα radiation conditions, following the Chinese Oil and gas industry standard SY/T 5163-2018. The mineral content of the whole rock and the relative contents of clay minerals were determined.

2.2.2. FE-SEM

The samples were prepared into a block with a width of 1 cm and a thickness of 0.5 cm along the direction perpendicular to the bedding. An argon ion beam was used to polish the observation surface with an Ilion + II 697C argon ion mill. Then, FE-SEM tests were performed on an FEI Quantum model 450, following the GB/T 16594-2008.

2.2.3. LTNA

Powdered samples were subjected to N₂ adsorption on an ASAP 2460 physisorption instrument. The relative pressure (P/P ₀) in this LTNA experiment ranged from 0.009 to 0.995. Referring to Tian et al. for details, the Barrett–Joyner–Halenda (BJH) and Brunauer–Emmett–Teller (BET) methods were employed to acquire the pore size distribution (PSD), pore volume (PV), specific pore volume (SPV), and specific surface area (SSA)

2.2.4. HPMI

The samples used in the HPMI experiment were dry, and approximately 2.5 cm. An Autopore IV 9500 automatic mercury poros-imeter manufactured by American Instruments was used for testing, with a maximum pressure of 413 MPa and a pore-throat test range of 0.003–800.000 μm. We placed each sample into a closed dilatometer to be vacuumed for testing. The mercury injection process used a staged pressure boost, and the cumulative mercury injection saturation under a certain pressure was recorded after the pressure reached stability. To ensure that the pore space of the sample was not damaged by the mercury pressure, the maximum mercury pressure in this experiment was set at 200.0 MPa, following the GB/T 29171-2012.

2.2.5. NMR

NMR experiments were conducted using a Niumag MesoMR23-060H-I instrument (Suzhou, China) by Suzhou Niumag Analytical Instrument Corporation, according to the CNPC enterprise standard SY/T 6490-2014. T ₂ analyses were performed on the tight sandstones under the following conditions: (1) dried state, (2) saturated state, (3) centrifugal state, and (4) heating state. The T ₂ spectra of the dried state were used as background signals, while the T ₂ spectra of the saturated state were used to analyze the full-scale PSD. The T ₂ spectra of the centrifugal state were used to analyze fluid mobility. The experimental parameters were as follows: sequence name: CPMG; SF (MHz): 21; the waiting time (T _w): 6000 ms; echo time (T _E): 0.06 ms; number of echoes (NECH): 80000; number of scans (NS): 32.

The NMR tests primarily measured transverse relaxation time (T ₂). First, T ₂ spectra under dried conditions were recorded and used as background signals. Subsequently, T ₂ spectra were measured under three different conditions: water saturation, centrifugation, and heating. Background signals were removed from the T ₂ spectra of the saturated and centrifugal states for analysis. Water was used as the saturating fluid. The samples were placed in a pressure saturation instrument, vacuumed for 12 h to remove air, and then pressurized for saturation. The saturation pressure was 17 MPa, and the saturation time was 24 h. Centrifugal experiments were conducted on the saturated samples using a GL-21 M centrifuge at rotational speeds of 5000 rpm, 8000 rpm, and 12,000 rpm for 16 h. The temperature during the heating experiments ranged from 50 to 100 °C, with intervals of 10 °C.

2.3. Analytical Methods

To systematically characterize pore structure, fluid distribution, and fractal behavior of the samples, a series of analytical techniques were applied, including NMR, centrifugal displacement experiments, and fractal analysis. Detailed descriptions of the theoretical principles, experimental procedures, and calculation methods are provided below to ensure reproducibility and clarity.

2.3.1. NMR Principle and Pore Size Calculation

According to the principle of NMR, the expression between T ₂ and pore diameter can be expressed as

$$D={\rho }_2{F}_{\mathrm{s}}{T}_2=C{T}_2$$ 1

where D is the pore diameter, μm, ρ₂ denotes the transverse surface relaxivity, μm/s; F _s represents the geometry morphologic factor; and C is the conversion coefficient, μm/s. This relationship forms the basis for calculating the pore-size distribution from the T ₂ spectra. The specific instrument information and experimental parameters used in the NMR tests are described in detail in Section .

2.3.2. Fractal Theory

Fractal theory has been widely applied in predicting reservoir fractures, analyzing pore structures, and studying heterogeneity. The closer the fractal dimension (D) is to 2, the more homogeneous the reservoir, while a value closer to 3 indicates greater complexity in the pore structure. Based on fractal theory, Shao et al. developed a fractal dimension model using NMR data, and Li et al. applied this algorithm to classify the pore structure of the Ordos Chang 7 tight sandstone, revealing that it exhibits three-stage fractal characteristics.

$${\log }_{10}(S_d)=(3-D){\log }_{10}(d)+(D-3){\log }_{10}(d_{\max })$$ 2

where S _d is the cumulative PV with pore diameter less than d, and d _max is the maximum pore diameter, μm. In our analysis, T ₂ distributions were logarithmically converted into pseudopore size distributions and used to derive fractal dimensions for different pore scales.

2.3.3. Centrifugal Experiment and MF Estimation

Comparing the NMR Spectra at saturated conditions with those after centrifugal experiments is helpful to analyze the pore fluid migration during the centrifugation process.

The centrifugal experiments were performed at rotation speeds of 5000, 8000, and 10,000 rpm for 16 h. In this study, the rotation speed (n) and the corresponding centrifugal pressure (ΔP) were calculated by the following formula

$$\mathrm{\Delta }P=1.097\times {10}^{-9}\mathrm{\Delta }\rho L_{\mathrm{c}}{n}^2(R_{\mathrm{c}}-L_{\mathrm{c}}/2)$$ 3

In which ΔP is the centrifugal pressure, MPa; Δρ denotes the density difference between air and water, g/cm³; L _c stands for the length of cores, cm; and R _c represents the radius of the revolving core, cm. Therefore, the corresponding centrifugal pressure considered in this study was 0.9 MPa, 2.4 MPa, 5.4 MPa.

2.3.4. Estimation of Maximum MF by Modeling

Theoretically, continuously increasing the centrifugal force can gradually convert CAF into MF, thereby maximizing the amount of MF. However, under the current experimental conditions, it is impractical to fully convert all CAF into MF by infinitely increasing the centrifugal force. Since it is not feasible to directly measure the amount of MF through experiments, exploring the functional relationship between centrifugal force and free water content becomes particularly important. By employing this approach, we can conduct a series of feasible centrifugation experiments in the laboratory, enabling us to theoretically estimate the maximum amount of MF.

The relationship between movable water content and centrifugal force is described as follows

$$Q_{\mathrm{m}}=\frac{Q_{\mathrm{f}}\mathrm{\Delta }\mathrm{P}}{\mathrm{\Delta }\mathrm{P}+{\mathrm{\Delta }\mathrm{P}}_{\mathrm{L}}}$$ 4

$$\frac{1}{Q_{\mathrm{m}}}=\frac{{\mathrm{\Delta }P}_{\mathrm{L}}}{Q_{\mathrm{f}}}\frac{1}{\mathrm{\Delta }P}+\frac{1}{Q_{\mathrm{f}}}$$ 5

where Q _m is the measured movable fluid content, Q _f is the maximum movable water amount when the centrifugal force reaches infinity, that is, the free water amount, mg/g; ΔP is the centrifugal pressure, MPa; ΔP _L is the median displacement pressure, Mpa. A linear regression of 1/Q _m vs 1/ΔP allows determination of Q _f.

2.3.5. Calculation of Adsorbed and Free Water Distribution

The weight ratio of adsorbed water to total water (r _a) is

$$r_{\mathrm{a}}=\frac{Q_{\mathrm{a}}}{Q_{\mathrm{f}}+Q_{\mathrm{a}}}$$ 6

where Q _a is the adsorbed water amount per gram of rock, mg/g.

Additionally, r _a can also be evaluated by a theoretical adsorption ratio equation developed by Li et al. which describes the coexistence characteristics of adsorbed and free fluids in porous media

$$r_{\mathrm{a}}=\frac{1}{1+\frac{{\rho }_2}{{\rho }_1}(\frac{V_{\mathrm{w}}}{S_{\mathrm{w}}H}-1)}$$ 7

Combining with eq , we can obtain the following form

$$\frac{V_{\mathrm{w}}}{S_{\mathrm{w}}}=\tau H+\tau H\frac{{\rho }_1}{{\rho }_2}\frac{Q_{\mathrm{f}}}{Q_{\mathrm{a}}}$$ 8

V _w is the volume of water-bearing pores, 10^–3 cm³/g. S _w is the specific surface area of water-bearing pores, m²/g, τ is a dimensionless correction coefficient, which is greater than or equal to one. ρ₁ and ρ₂ are the mean densities of adsorbed and free water, g/cm³. According to Li et al. and Zhao et al., H = 1.55 nm is the adsorbed water thickness, and ρ₁, ρ₂ are densities of adsorbed and free water (1.66 and 1.005 g/cm³, respectively).

Finally, the volume fraction of adsorbed water is

$$r_{\mathrm{a}\mathrm{v}}=\frac{{r_{\mathrm{a}}\rho }_2}{{r_{\mathrm{a}}\rho }_2+(1-r_{\mathrm{a}}){\rho }_1}$$ 9

where r _av is the volume ratio of adsorbed water, fraction.

3. Results

3.1. Porosity, Permeability, Mineralogy, and Pore-Throat Characteristics

The reservoir quality of Chang 7 sandstone is relatively poor. The porosity and permeability measurement results for the six samples are shown in Table , which reflect the characteristics of tight reservoirs. The helium porosity ranges from 3.43 to 11.12%, with an average value of 7.71%, while permeability ranges from 0.0115 to 0.1040 mD, with an average of 0.0545 mD. A positive relationship exists between permeability and porosity (R ² = 0.80) (Figure a; Table ).

1. Porosity, Permeability, Mineral Composition, and Contents of the Chang 7 Tight Sandstones.

			XRD analysis (wt %)				clay minerals content
sample ID	Φ (%)	K (mD)	quartz	feldspar	carbonate minerals	clay minerals	I/S	I	K	C
LD-1	8.286	0.0606	38.3	46.2	6.0	9.5	63	23	5	9
LD-2	10.317	0.1040	62.5	14.2	9.3	13.0	56	35	3	6
LD-3	3.431	0.0115	36.4	35.6	9.4	18.6	35	18	9	38
LD-4	5.721	0.0209	63.1	16.9	6.4	13.6	32	53	4	11
LD-5	11.115	0.0798	53.9	26.5	11.6	7.6	64	30	2	4
LD-6	7.409	0.0504	65.0	21.5	4.3	9.2	58	32	3	7

1.

(a) Relationships between permeability and porosity; (b) whole-rock results; (c) the relative contents of clay minerals of the Chang 7 tight sandstones.

According to the XRD results, the Chang 7 tight sandstones are predominantly composed of quartz and feldspar (mainly plagioclase), with contents ranging from 36.4 to 65.0% (average of 53.2%) and from 14.2 to 46.2% (average of 26.8%), respectively. The carbonate mineral content varies from 4.3 to 12.0%, with an average of 8.1%, including calcite, dolomite, Fe-dolomite, and siderite. The clay mineral content ranges from 7.6 to 18.6%, with a mean value of 11.9% (Figure b; Table ). The clay minerals are primarily composed of Illite/smectite (I/S) mixed layers, with relative contents ranging from 32.0 to 64.0% (mean of 51.3%), followed by Illite (18.0 to 53.0%, mean of 31.8%), chlorite (4.0 to 38.0%, mean of 12.5%), and kaolinite (2.0 to 9.0%, mean of 4.3%) (Figure c; Table ).

3.2. Pore Types

According to the CTS and SEM images, the types of storage spaces are primarily residual intergranular pores (RIPs), dissolution pores (DPs), intercrystalline pores (IPs), and microfractures (Figure ). RIPs develop in tight sandstones with better grain sorting and are abundant in rigid grains. The shape of RIPs becomes highly irregular, mostly triangular, due to reservoir compaction and cementation. The pore size of RIPs mainly ranges from 1 to 30 μm (Figure a,c,d,f–i). DPs are formed by feldspar dissolution, rock fragment dissolution, and carbonate dissolution, with feldspar DPs being the most developed, while carbonate DPs are relatively rare. The pore size of DPs ranges from nanometers to tens of microns (Figure b,c,e). IPs are primarily micropores formed between or within grains by carbonate cements, siliceous cements, and clay cements. Clay IPs are the most common in the study area, with diameters generally less than 0.6 μm (Figure h). Two types of microfractures can be observed: structural fractures and diagenetic fractures. The shapes of microfractures are sheet-like and bent-sheet-like (Figure d,e). Structural fractures have widths up to the micrometer scale and generally extend over long distances, while diagenetic fractures are mostly nanometers in width and extend over short distances. To provide quantitative support for pore-type identification, 2D pore area fractions (image-based porosity) were calculated from representative CTS images using ImageJ. The results are as follows: Figure a: 2.56%, Figure b: 1.93%, Figure c: 3.84%, Figure d: 2.92%, Figure e: 3.21%, and Figure f: 1.70%.

2.

Pore features of the Chang 7 tight sandstone in the Longdong area. (RIPs: residual intergranular pores, DPs: dissolution pores, IPs: intercrystalline pores). (a) RIPs, CTS; (b) DPs, CTS; (c) RIPs, DPs and microfracture, CTS; (d) RIPs and DPs, CTS; (e) microfracture, CTS; (f) DPs and IPs, CTS; (g) RIPs, DPs and microfracture, SEM; (h) RIPs and DPs, SEM; (i) RIPS, SEM.

3.3. Pore Structure Characteristics

3.3.1. HPMI Results

The capillary pressure curves are presented in Figure and Table . Based on the morphological features of the mercury intrusion and extrusion curves, including maximum mercury saturation (S _max), the highest mercury withdrawal efficiency (W _e), and average pore-throat radius (R _a), the six samples can be classified into three types of pore-throat structures.

3.

Capillary pressure curves (a), and pore-throat size distribution curves (b) of the Chang 7 tight sandstones in the Longdong area.

2. Pore Structure Parameters Obtained From Various Experimental Results.

	HPMI experiments					NA experiments
sample	R max	R a	S p	S max	W e	SSA (m2/g)	PV (mL/100 g)
LD-1	0.361	0.117	2.489	74.020	20.809	5.7141	0.9477
LD-2	0.547	0.146	1.663	77.272	21.467	2.0109	0.7056
LD-3	0.270	0.047	2.476	42.292	33.679	4.7433	1.0950
LD-4	0.134	0.029	1.511	62.079	27.612	5.3351	1.4415
LD-5	0.179	0.052	1.346	70.969	29.961	3.6054	1.1165
LD-6	0.361	0.105	2.381	73.284	31.737	4.8561	1.4152

Type I samples (LD-1, LD-2, LD-5, and LD-6) exhibit capillary pressure curves with a long and gentle section, located below the mercury injection pressure of 5 MPa (Figure a). This indicates that the pore-throat structure of Type I has a larger degree of pore-throat development. The average pore-throat radius (R _a) ranges from 0.05 to 0.15 μm, with a mean value of 0.11 μm. As a result, mercury can more easily penetrate the sample at the initial stage. PSD of the Type I pore-throat structure is unimodal (Figure b), with peak values primarily between 0.01 and 0.1 μm. The pore-throat combinations of Type I are mainly composed of RIPs, DPs, and sheet-like and bent-sheet-like throats (Figure ). This type exhibits the highest average maximum mercury saturation (S _max) of 73.89% and the lowest mercury withdrawal efficiency (W _e) of 26.00%, indicating that the Type I pore-throat structure has better pore-throat connectivity.

Type II sample (LD-4) exhibits a steeper capillary pressure curve, located above the mercury injection pressure of 5 MPa, and shows a unimodal PSD curve. The left peak ranges from 0.005 to 0.03 μm, indicating a higher proportion of small and large pores and throats. The average pore-throat radius (R _a) is 0.029 μm. The maximum mercury saturation (S _max) for Type II is 62.08%, while the mercury withdrawal efficiency (W _e) is 27.61%.

The Type III sample (LD-3) has the steepest capillary pressure curve, positioned above the mercury injection pressure of 2 MPa. Its PSD curve is also unimodal, with the peak shifted to the left compared to Type I, primarily ranging from 0.001 to 0.07 μm. The average pore-throat radius (R _a) is 0.047 μm, with a maximum mercury saturation (S _max) of 42.29%, and the highest mercury withdrawal efficiency (W _e) of 33.68%.

3.3.2. LTNA Results

The LTNA experiment characterizes pores in the 1–200 nm range. According to the International Union of Pure and Applied Chemistry (IUPAC) classification scheme, , the LTNA adsorption curves exhibit a type IV isotherm, with an overall inverted S-shape (Figure ). When the pressure (P/P ₀) exceeds 0.45, the predominant hysteresis loop types are H3 (Figure b,d–f), indicating that slit-shaped pores are dominant.

4.

N₂ adsorption/desorption isotherms of the Chang 7 tight sandstones in the Longdong area. (a) Is from Sample LD-1, (b) is from Sample LD-2, (c) is from Sample LD-3, (d) is from Sample LD-4, (e) is from Sample LD-5, (f) is from Sample LD-6.

Based on the BJH model, PV ranges from 0.7056 to 1.4415 mL/100 g (average: 1.1203 mL/100 g), and SSA, calculated using the BET model, ranges from 2.01 m²/g to 5.71 m²/g (average: 4.38 m²/g). − Samples exhibiting H2 and H3 hysteresis loops show similar PSD characteristics. The PSD obtained from the nonlocal density functional theory (NLDFT) model is unimodal (Figure a), with the main peaks distributed between 30 and 50 nm.

5.

Pore size distribution (PSD) derived from (a) LTNA experiments and (b) NMR T ₂ spectra.

3.3.3. NMR Results

NMR is a nondestructive technique. The NMR T ₂ spectra in a water-saturated state under a low-frequency magnetic field can effectively reflect the characteristics of the entire pore space (Figure b). The resulting curve may be either unimodal or bimodal. The T ₂ spectra of most samples are bimodal, with peaks at 0.01–0.3 ms (P1), 0.3–5 ms (P2), and 5–300 ms (P3). Based on the NMR results, the Chang 7 tight sandstone samples were classified into three categories.

The T ₂ spectra of Type I samples show higher P2 peaks and lower P1 peaks, while Type II samples exhibit a unimodal distribution. Type III samples, on the other hand, display higher P1 peaks and lower P3 peaks. A higher T ₂ value corresponds to larger pore sizes, making the T ₂ spectra useful for pore size characterization. This suggests that Type I samples are dominated by larger pores, while Type III samples are dominated by smaller pores. Type II samples, in contrast, exhibit a more balanced pore structure. Referring to the pore characteristics shown in Figure , we can infer that the P3 peak primarily reflects RIPs and DPs, while the P1 peak mainly corresponds to IPs.

3.4. Full PSD Characteristics

Compared to the NMR T ₂ spectra, the PSD obtained from the LTNA experiment primarily characterizes smaller pores (Figure a) and struggles to capture larger pores (>200 nm). Previous studies have successfully applied various transformations of NMR T ₂ spectra into reservoir pore-throat sizes, based on LTNA data, achieving good results in pore-throat structure characterization. To overcome this limitation, the PSD derived from the LTNA experiment was used to transform the NMR movable fluid T ₂ spectra, allowing for the full PSD of the tight sandstone reservoirs to be characterized (Figure a,b).

The main peak of the LTNA PSD corresponds to peak P1 in the T ₂ spectra (Figure a). Based on this relationship, the T ₂ spectra were calibrated and converted into a PSD following the procedure outlined by Li et al. The calculated C values range from 0.055 μm/ms to 0.096 μm/ms, with an average of 0.071 μm/ms. The NMR-derived PSD for the tight sandstone in this area primarily ranges from 1 nm to 20 μm (Figure b), which is consistent with the PSD observed in the FE-SEM images (Figure g–i). In this study, SEM images were qualitatively compared with NMR-derived pore size distributions to illustrate the general agreement in pore size characteristics. This qualitative comparison supports the interpretation of NMR data trends but does not serve as a direct quantitative validation of the pore size distributions.

6.

Comparison of PSDs obtained from LTNA and NMR (a); conversion of T ₂ spectra to PSD by calibrating with LTNA PSD (b).

3.5. Fractal Characteristics of Pore Throats

In this study, the pores are categorized into micropores, and macropores, using demarcation point, d _c (Figure ). The value of d _c ranges from 47.4 to 254.1 nm (average 150.8 nm). The fractal dimensions for micropores, and macropores are designated as D ₁ and D ₂, respectively.

7.

Fractal characteristics of full-scale pore structure from NMR data. (a) Is from Sample LD-1, (b) is from Sample LD-2, (c) is from Sample LD-3, (d) is from Sample LD-4, (e) is from Sample LD-5, (f) is from Sample LD-6.

Table shows the fractal dimensions of macropores and micropores, as well as the correlation coefficients of linear fitting models. D ₁ ranges from 1.0394 to 1.5583, with a mean value of 1.3315. D ₂ ranges from 2.6961 to 2.9724, with an average of 2.8457. The value of D ₂ is significantly larger than that of D ₁, suggesting that macropores are more heterogeneous than micropores. All linear fitting models exhibit correlation coefficients greater than 0.8, indicating that the studied tight sandstones generally possess fractal characteristics and can be effectively characterized using fractal theory. In addition, the correlation coefficients of the linear fitting models for macropores (0.8149–0.9910, average 0.9077) are generally higher than those for micropores (0.8260–0.8957, average 0.8701), suggesting that the macropore network in the tight sandstone exhibits more pronounced fractal behavior compared to the micropore network.

3. Fractal Dimensions of Chang 7 Tight Sandstones.

		r < r c			r > r c
sample	r c (nm)	K	D 1	R 2	K	D 2	R 2
LD-1	254.10	1.6038	1.3962	0.8260	0.183	2.8170	0.9128
LD-2	161.51	1.4417	1.5583	0.8596	0.2456	2.7544	0.9619
LD-3	157.10	1.7309	1.2691	0.8770	0.0276	2.9724	0.9013
LD-4	98.80	1.8147	1.1853	0.8913	0.0555	2.9445	0.8149
LD-5	47.38	1.9606	1.0394	0.8957	0.3039	2.6961	0.9910
LD-6	185.70	1.4594	1.5406	0.8708	0.1100	2.8900	0.8640

The porosity and distribution of different pore types were determined using NMR-PSD. The Chang 7 tight sandstone exhibits a comparable contribution from macropores and micropores. The porosity of macropores ranges from 0.79 to 6.21%, contributing approximately 17.45 to 73.69% of the total porosity. The porosity of micropores ranges from 1.86 to 5.16%, accounting for about 25.63 to 79.73% of the total porosity (Figure ).

8.

Total porosity (a), NMR porosity (b), and proportion of different types of pores (c).

3.6. NMR Results under Different Treatment Conditions

We designed a series of experiments to determine the MF, CAF, and CLF of experiment samples. The analysis is performed based on the comparisons of T ₂ responses measured for fully saturated, centrifuged, and heated samples. We selected 6 samples to measure NMR T ₂ spectra at different centrifugation speeds and selected three of them to measure NMR T ₂ spectra under different heating temperatures.

3.6.1. NMR-Derived PSD Spectra under Centrifugal Conditions

After centrifugation, the P3 peaks of samples LD-1, LD-2, and LD-5 showed a significant decrease, especially during the initial stage, where the T ₂ spectral signal values experienced the greatest change. As the centrifugal force gradually increased, the rate of decline in the P3 peak slowed, indicating that most of the free water in the pores had been expelled. Meanwhile, the corresponding P1 and P2 peaks did not exhibit significant changes with increasing centrifugal force, suggesting that water in small- to medium-sized pores was either rarely or not discharged by centrifugation. The same phenomenon was observed in samples LD-3, LD-4, and LD-6. Overall, the P2 values in the NMR T ₂ spectra from the centrifugation experiment decreased significantly. As the centrifugation speed increased, the P1 peak also showed a gradual decrease, though this trend was less pronounced than the decrease in the P3 peak (Figure ). This indicates that water filling larger pores has been expelled from the samples, while water in micropores has been only minimally excluded or discharged.

9.

NMR PSD Characteristics under water saturation and centrifugation. (a) Is from Sample LD-1, (b) is from Sample LD-2, (c) is from Sample LD-3, (d) is from Sample LD-4, (e) is from Sample LD-5, (f) is from Sample LD-6.

3.6.2. NMR-Derived PSD Spectra under Heating Conditions

For all heated samples, a significant decrease in the T ₂ spectra was observed as the temperature increased. Before reaching 60 °C, the P3 peak in the NMR spectra showed a notable reduction. The P1 and P2 peaks also gradually decreased, though the decline was less pronounced than that of the P3 peak. As the temperature continued to rise, the P3 peak nearly reached its minimum value and then stabilized.

On the other hand, the P1 peak continued to decline until reaching 90 °C, after which the decrease slowed, and the value remained almost constant. For the LD-2 and LD-4 samples, which had relatively prominent P2 peaks, the P2 peak also decreased significantly before 60 °C. However, as the temperature continued to rise, the change in the signal value became less pronounced, although the P1 peak still exhibited a downward trend. This behavior is attributed to the evaporation of water from the samples, which caused a decrease in the signal values detected by the NMR. The T ₂ spectra of the heated samples show a clear shift of the T ₂ peak toward lower T ₂ times, centered around 0.1 ms for all samples (Figure a–c). The plot of NMR cumulative porosity for oven-dried samples versus temperature reveals two distinct trends: a significant downward trend below 90 °C, followed by stabilization after 90 °C (Figure d). This observation is consistent with the findings of Testamanti and Rezaee.

10.

NMR PSD after centrifugation and drying (a–c), and NMR porosity vs temperature (d). (a) Is from Sample LD-1, (b) is from Sample LD-2, (c) is from Sample LD-4.

4. Discussion

4.1. Fluid Type Identification Based on NMR Results under Saturation, Centrifugation, and Heating Conditions

The MF distribution can be determined by comparing the T ₂ spectra under saturated and centrifugal conditions (Figures and a,c). MF predominantly occurs in macropores, with the proportion of MF ranging from 11.1 to 49.7%, and an average value of 34.6% (Figure ).

11.

Determination of different types of fluids (taking sample LD-1 as an example), (a) saturation and centrifugation conditions; (b) centrifugation and heating conditions; (c) MF spectrum; (d) CLF spectrum; (e) BF spectrum; (f) CAF spectrum. (MF: movable fluid, CLF: clay-bound fluid, BF: bound fluid, CAF: capillary-bound fluid).

12.

Distribution characteristics of different types of fluids (a–c), and proportions of different types of fluids in pores of different sizes (d,e). (a,d; b,e; c,f; correspond to LD-1, LD-2, and LD-4, respectively, MF: movable fluid, CAF: capillary-bound fluid, CLF: clay-bound fluid).

The plot of NMR cumulative porosity for the oven-dried samples as a function of temperature revealed two distinct trends (Figure d). The intersection of these trendlines was calculated for all three samples, indicating the minimum temperature at which the sample must be heated to ensure that only CLF remains (Figure b). At 90 °C, the NMR porosity was considered irreducible, ranging from 1.82 to 3.36% (mean: 2.49%). The proportion of CLF ranged from 17.3 to 40.1%, with an average value of 26.7% (Figure ).

After determining the distribution of CLF, the CAF distribution was established by comparing the T ₂ spectra of CLF with those under centrifugal conditions (Figure d–f). The proportion of CAF ranged from 32.9 to 48.8%, with a mean value of 38.7% (Figure ). Moreover, a strong correlation was observed between the proportion of MF and pore throats above a certain size. MF content exhibits the highest correlation with pore throats larger than 200 nm in diameter (R ² > 0.9).

4.2. Fluid Type Identification Using NMR Centrifugation Experiments and Theoretical Model

By fitting the experimental data and establishing a linear relationship between 1/Q _m and 1/ΔP, the quantity of free water was determined (eqs –). The adsorbed water is then calculated as the difference between the saturated water and free water, from which the adsorption ratio can be derived based on the amounts of adsorbed and free water.

The results of the saturation-centrifugation experiment are shown in Figure a, illustrating the movable water content under different centrifugal force conditions. The volume of movable water increases gradually with rising centrifugal force, although the growth rate slows down significantly, which is consistent with the trend observed in Figure . This suggests that as centrifugal force increases indefinitely, the movable water amount approaches a constant value, representing the free fluid. According to the relationship between the reciprocal of the movable fluid amount and the reciprocal of the centrifugal force (Figure b), a strong linear correlation is observed (R ² > 0.90).

13.

Results of centrifugation experiments and microdistribution characteristics of movable and bound water. (a) Relationship between movable water content and centrifugal force; (b) relationship between the reciprocal of centrifugal force and the reciprocal of movable water content; (c) weight ratio; (d) volume ratio.

Using linear relationships, the maximum movable fluid content, also referred to as free fluid content, can be estimated. Under the experimental conditions, the free fluid content of the studied samples ranges from 4.45 to 29.15 mg/g, with an average of 19.02 mg/g. The proportion of free fluid varies between 22.3 and 89.5%, with an average of 63.1%.

Previous studies have also shown that intermolecular forces play a critical role in unconventional oil and gas reservoirs. In shale oil and gas, these forces manifest as capillary pressures and adsorption forces. These forces are greater than buoyancy, preventing fluid migration and facilitating the accumulation and preservation of tight oil and gas.

The proportions of free fluid r and adsorbed fluid in this area were calculated using eqs –, and the results are presented in Figure c,d. Small pores (less than 10 nm) are fully adsorbed and immovable due to the dominance of adsorption forces. Medium pores (10–200 nm) are influenced by both adsorption and capillary forces, containing a small amount of free fluid. The proportion of free fluid increases as the pore diameter increases. Large pores (greater than 200 nm) are primarily affected by capillary forces, exhibiting better mobility and a higher proportion of free fluid.

4.3. Comparison of Two Classification Schemes

To further evaluate the correspondence between the two fluid classification schemes, a comparative analysis was conducted between the total proportion of MF and CAF derived from NMR and the theoretically calculated maximum proportion of free fluid (Figure ). This comparison was based on three representative core samples LD-1, LD-2, and LD-4.

14.

Relationship between free water and different types of fluids.

LD-1 and LD-2 represent Type I reservoirs, characterized by relatively high porosity (>8%), moderate permeability (>0.06 mD), and larger average pore throat radii (R _a = 0.11 μm). They also exhibit high mercury saturation values (S _max > 74%), indicative of well-connected pore networks. For these samples, the measured MF + CAF contents were 25.12 mg/g and 30.15 mg/g, respectively, while the calculated free fluid values were 29.15 mg/g and 28.09 mg/g. The absolute errors were 4.03 mg/g and 2.06 mg/g, with a mean absolute error of 3.05 mg/g and a standard deviation of 1.39 mg/g. The relative errors were 13.82 and −7.33%, yielding a mean relative error of 3.25%, an absolute mean relative error of 10.58%, and a standard deviation of 10.58%. In terms of volume fractions, MF + CAF and free fluid were 77.15 and 82.60% for LD-1, and 89.52 and 77.04% for LD-2. These results indicate a relatively good agreement between the two fluid classification systems, though deviations are observed, likely due to measurement uncertainties or variations in fluid distribution within the pore system.

In contrast, LD-4 represents a Type II reservoir, with significantly lower porosity (5.7%), reduced permeability (0.02 mD), smaller pore throat radii (R _a = 0.029 μm), and lower S _max (62.08%), indicating poorer reservoir quality. In this sample, a marked discrepancy is observed between the theoretical free fluid proportion and the NMR-derived MF + CAF content. The MF + CAF value for LD-4 was 12.97 mg/g, whereas the calculated free fluid value was only 4.83 mg/g, leading to an absolute error of 8.14 mg/g and a relative error of −62.78%. The volume fractions for LD-4 were 59.86% for MF + CAF and 22.31% for free fluid, indicating a substantial mismatch. This deviation is primarily attributed to the dominance of micropores (accounting for over 70%) and finer pore throats (average radius < 30 nm), which intensify capillary forces and impede the transition of CAF to MF. As a result, a considerable portion of the capillary-bound fluid remains effectively immobilized, thereby reducing the actual volume of free fluid that can be displaced.

These findings suggest that the agreement between the two fluid classification schemes is highly dependent on pore structure and reservoir quality. In well-connected, macropore-dominated systems such as LD-1 and LD-2, theoretical models and NMR measurements are consistent. However, in micropore-rich systems like LD-4, significant discrepancies arise due to reduced fluid mobility and lower capillary release efficiency, which leads to an underprediction of free fluid content by the theoretical models.

4.4. Effect of Physical Properties and Minerals on the Movable Fluid Parameters

To better interpret the relationships among fluid types, pore structure, petrophysical properties, and mineral composition, Pearson correlation analysis was conducted using the Correlation Plot module in Origin 2024 Pro. The results are visualized using circle size to represent correlation strength and color to indicate direction (positive or negative), with a blue-to-red gradient corresponding to correlation values from −1 to +1. Here, r denotes the Pearson correlation coefficient, which ranges from −1 to +1 and reflects the strength and direction of linear relationships between variables.

MF exhibits a strong positive correlation with both porosity and permeability (r > 0.93), indicating that higher reservoir quality is associated with increased movable fluid content. In contrast, CAF and CLF show strong negative correlations with these properties (r > 0.90). Similarly, macropores exhibit a strong positive correlation with MF and negative correlations with both CAF and CLF (r > 0.98), whereas micropores show the opposite trend, with a significant negative correlation with MF (r > 0.94) and positive correlations with CAF and CLF (r > 0.91).

In terms of mineralogy, MF shows a weak negative correlation with quartz and clay minerals, and nearly no correlation with feldspar. By contrast, CAF and CLF are weakly positively correlated with quartz and clay, and weakly negatively correlated with feldspar. Free fluids display a trend similar to MF, while adsorbed fluids follows the trend of the combined CLF and CAF (Figure ).

15.

Correlation heatmap of pore size, fluid type, physical properties, and mineral composition. (Por: porosity, Perm: permeability, MF: movable fluid, CAF: capillary-bound fluid, CLF: clay-bound fluid).

4.5. Model of the Occurrence Characteristics of Different Types of Fluids in Different Pore Throats

Based on the above analysis, we established the occurrence patterns of different fluid types in various pore throats (Figure ). In pore throats smaller than 10 nm, fluid distribution is predominantly controlled by strong molecular forces and adsorption, resulting in a complete CLF state. These pores are primarily interlayer pores (IPs) within clay minerals. When the pore throat ranges from 10 to 200 nm, both molecular and capillary forces govern fluid behavior. Under these conditions, CLF and CAF are the dominant fluid types, with only a small amount of MF. The corresponding pores are mainly IPS, along with a small proportion of feldspar-derived DPs. When the pore throat exceeds 200 nm, capillary forces become the dominant factor. In this regime,, the main fluid types are CAF and MF, with the proportion of MF gradually increasing as the pore throat size increases. When the pore throat reaches the micrometer scale, it is almost entirely MF, with the corresponding pores mainly being larger RIPs.

16.

Model of the occurrence characteristics of different types of fluids in different pore throats.

5. Conclusions

Through the investigation of rock physical properties, including CTS, XRD, FE-SEM, LTNA, HPMI, and a series of NMR experiments, this study reveals the pore structure and provides a comprehensive PSD. It also differentiates MF, CAF, and CLF. By integrating theoretical models, the free fluid and adsorbed fluid are quantified, and the interrelation-ships between the two classification schemes are explored. Finally, the factors influencing the distribution of different fluid types are discussed. The main conclusions are as follows:

• (1)The full-scale PSD can be obtained by combining LTNA with NMR. The PSD of Chang 7 tight sandstones primarily ranges from 1 nm to 20 μm. Based on the fractal inflection point of the full-scale PSD, the pore system is categorized into macropores, and micropores. The critical values separating micropores from macropores fall within 47.4 to 254.1 nm (average 150.8 nm).
• (2)By integrating NMR data with theoretical modeling, a clear correspondence is established between the MF/CAF/CLF and free/adsorbed fluid classification systems: MF and CAF together correspond closely to free fluids, while CLF aligns well with adsorbed fluids. Notably, in low-porosity samples with extremely fine pore throats, this theoretical approach may underestimate the actual free fluid content.
• (3)Macropores show a strong positive correlation with MF and free fluids, and a negative correlation with CAF and CLF. Micropores display the opposite trend. Mineral composition also plays a role: MF is weakly negatively correlated with quartz and clay minerals, while CAF and CLF show weak positive correlations with quartz and clay, and weak negative correlation with feldspar. These trends further support the correspondence between the two fluid classification schemes and highlight the influence of pore structure and mineralogy on fluid occurrence.
• (4)Fluid distribution patterns in different pore throats were established. CLF dominates in pores <10 nm due to strong molecular forces, primarily within IPs in clay minerals. In pores ranging from 10 to 200 nm, CLF and CAF are predominant under the combined influence of molecular and capillary forces, with limited MF present; these pores are mainly IPs, accompanied by a small proportion of feldspar-derived DPs. For pores >200 nm, including larger DPs and RIPs, capillary forces become dominant, and the proportion of MF increases progressively. At the micrometer scale, the fluid is almost exclusively MF.

Glossary

Abbreviations

Name

Abbreviations

MF

movable fluids

CLF

clay-bound fluids

LTNA

low-temperature nitrogen adsorption

CT

computed tomography

XRD

X-ray diffraction

PV

pore volume

SSA

specific surface area

DPs

dissolution pores

CAF

capillary-bound fluids

PSD

pore size distribution

NMR

nuclear magnetic resonance

CTS

casting thin section

HPMI

high-pressure mercury injection

SPV

specific pore volume

RIPs

residual intergranular pores

IPs

intercrystalline pores

The authors declare no competing financial interest.