The oil-bearing reservoirs in the Chang 3 Member are gray and gray–brown fine sandstones, with particle sizes mainly ranging from 0.06 to 0.25 mm. In addition, the clastic particles are mostly subangular to sub-round with moderate sorting and roundness. The lithologies of the reservoir are mainly feldspar lithic sandstone and lithic feldspar sandstone (Figure 3). Among the mineral components, the content of quartz is the largest (55.54%), followed by feldspar (24.09%) and debris (19.31%) (Figures 4A,B). According to the observation of microscopic thin sections, the composition of the Chang 3 Member comprises magmatic, metamorphic, and sedimentary rock debris. In the detrital composition, Q/F is 2.30 and Q/(F+R) is 1.29, indicating that the Chang 3 reservoir has high compositional maturity. In addition, the interstitials are dominated by carbonate cement and muddy matrix, with an average content of 8.28 and 3.80%, respectively; the content of clay minerals is low (average content of 3.8%) and dominated by kaolinite and illite (Figures 4C,D).

According to the observation results of the cast thin sections, the average face ratio of the Chang 3 Member is 7.61%, and its surface area is relatively low (the average value is 0.32 μm⁻¹). Statistics show that the shape factor of the target layer is relatively low (average 0.46), indicating that the pores are regular in shape. Moreover, the average coordination number of pore throats is 0.48 and the sorting coefficient is 9.78, indicating that the number of throats connected to pores is large and the pore heterogeneity is strong (Table 1). In addition, the proportions of three types of pores, namely, intergranular, intragranular dissolution pores, and residual intergranular pores, are 76.15, 18.15 and 5.7%, respectively. This shows that the pore types of the tight sandstone in the target layer are mainly intergranular and intragranular dissolution pores, and the proportion of residual intergranular pores is the lowest (Figures 4C,D).

The petrophysical properties and pore structure parameters of the samples obtained based on the high-pressure mercury intrusion experiment are shown in Table 2. The porosity of the samples ranges from 2.20 to 14.97%, with an average value of 10.17%; the permeability ranges from 0.09 × 10⁻³ to 1.58 × 10⁻³ μm², with an average value of 0.44 × 10⁻³μm². Referring to the classification criteria for ultra-low permeability reservoirs proposed by Yang and Fu (2012 ), we divided the samples into three categories with permeability (K) of 0.3 × 10⁻³ and 0.5 × 10⁻³ μm² as the boundaries: (Type I) K ≥ 0.5 × 10⁻³ μm², (Type II) 0.5 × 10⁻³μm²>K ≥ 0.3 × 10⁻³ μm², and (Type III) K < 0.3 × 10⁻³ μm².

According to statistics, the proportions of Types I, II, and III samples in the Chang 3 Members are 31.4, 20.0, and 48.6%, respectively (Figure 5). From Type I to Type III reservoirs, with the decrease of the petrophysical properties of the samples, the average value of the maximum pore throat radius, median pore throat radius, and mercury withdrawal efficiency gradually decreased; and the average value of displacement pressure, median pressure, and residual mercury saturation gradually increased. Among them, the mercury withdrawal efficiency corresponds to the recovery factor in oilfield development, and the residual mercury saturation is the mercury saturation that remains in the rock sample when the injection pressure drops to a minimum value (Fan et al., 2020). It shows that the smaller the pore throat radius of the sample the greater the capillary resistance and the worse the petrophysical properties. Such reservoirs will show low recovery and high residual oil content in the actual development process.

From the correlation between permeability and porosity (Figure 6), it can be seen that there is a certain positive correlation between porosity and permeability, but the correlation is poor and the correlation coefficient R ² is only 0.33. Some samples with less porosity can also obtain larger permeability, indicating that the porosity–permeability relationship of tight sandstone is significantly different from that of conventional sandstone. The research on the differences in petrophysical properties of this type of tight reservoirs should start from the perspective of microscopic pore throat structures (Nelson, 2009; Ryazanov, et al., 2014; Li et al., 2020).

The capillary pressure curve is the relationship between capillary pressure and mercury saturation, and a certain capillary pressure corresponds to a certain pore throat radius. From the perspective of pore structure, the capillary pressure curve can reflect the distribution of the pore and throats of the rock (Wang, 2008; Qiao et al., 2020). The shape of the capillary pressure curve is mainly controlled by the sorting of the pore throats and size of the throats. Sortability refers to the degree of dispersion of throat size (Wang L. et al., 2018; Zhou et al., 2021). The more concentrated the distribution of throat size, the better the sorting, the longer the flat section in the middle of the capillary pressure curve, and the closer it is to parallel with the abscissa (Yang and Fu., 2012).

Moreover, the distribution interval and distribution shape of pore throat radius represent the proportion of the corresponding pore throat radius and strength of seepage capacity, respectively, and the pore structures represented by different peak shapes are also different (Yang et al., 2017). A single peak indicates that the pore type is relatively single, such as intergranular dissolution pores, intragranular dissolution pores, residual intergranular pores, or microcracks. The double peaks indicate that there are two main pore types combined, which together control the storage and permeability of the rock (Wang et al., 2017; Zhong, 2017; Wang R. et al., 2018). Due to the differences in the dominant pore type, the reservoir and percolation capacities of single peak–type or double peak–type reservoirs are also different.

The pore throat distribution of the Chang 3 Member reservoir is in the form of single peak and double peak, and double peak is dominant, with double peak samples accounting for 74% (Figure 7). A single peak–type pore throat distribution represents a single pore–type reservoir. The pore types of the single peak samples are mainly intergranular or intragranular dissolution, and the pore throat sorting is better and the average sorting coefficient is low (0.86). When the pore throat distribution is bimodal, it means that the Chang 3 reservoir mainly develops pore throat spaces with two pore diameter ranges. This type of reservoir mainly develops intergranular and intragranular dissolution pores, which are poorly sorted and have a high average sorting coefficient (1.44). In addition, the average petrophysical properties of the double peak samples are worse than those of the single peak samples, indicating that tight reservoirs with a single pore type have better petrophysical properties.

Comparing the typical capillary pressure curves of the abovementioned three types of samples (Figure 8), it can be found that the capillary curves of the samples of Types I, II, and III are getting more and more smoother, indicating that the sorting of pore throats is gradually getting better. The sorting coefficients corresponding to these three types of reservoirs also decreased gradually from 0.3427 to 0.1018. That is, the smaller the pore throat sorting coefficient (the better the sorting), the worse the petrophysical properties of the samples. In addition, with the gradual improvement of the sorting properties of the samples, the pore throat radius distribution of the samples gradually changed to a single double peak shape. The double peak samples contain two pore types: intergranular and intragranular dissolution pores.

In addition, the lower limit of the pore throat radius of these three types of samples is about 0.0013 μm. However, only when the pore throat radius is greater than 0.004 μm, a certain amount of mercury exists in the pore throat. Mercury in Type I samples mainly exists in the pore throats controlled by the radius of 0.004–3.25 μm, and the permeability is the highest. Mercury in Type II samples mainly exists in the pore throats controlled by the radius of 0.004–1.3 μm, and the permeability is moderate. Mercury in Type III samples mainly exists in the pore throats controlled by the radius of 0.004–0.21 μm, and the permeability is the lowest. Samples with larger large pore throats have the best percolation performance.

Comparing the relationship among mercury injection amount, permeability contribution, and pore throat radius, it is found that as the pore throat radius decreases, the pore throat radius corresponding to the peak of permeability contribution is always larger than the pore throat radius corresponding to the peak mercury injection amount. Moreover, the pore throat radius covered by the peak mercury injection amount is larger. This shows that the reservoir permeability is contributed by a small part of pore throats with large radius, while the reservoir space is mainly contributed by a large number of pores connected with small throats.

In this study, the high-pressure mercury intrusion experiments were used to obtain not only the petrophysical properties and pore throat structure parameters of the reservoir but also parameters such as structure coefficient and geometric factor (Table 3). Among them, the microscopic homogeneity coefficient represents the concentration degree of pore throats, and the pore structure coefficient represents the detour degree of fluid seepage in the pores. The larger the pore structure coefficient, the stronger the degree of bending and tortuousness of the pores (Zhong, 2017; Ren et al., 2019; Liu et al., 2020). Mercury injection and withdrawal tortuosity reflect the tortuosity of the pore throat morphology experienced by the nonwetting phase (mercury) when it enters and exits the cores, respectively (Song and Kovscek., 2016; Zhu et al., 2018; Shi et al., 2019). From the range of these characteristic parameters and the relationship between the mercury injection amount/permeability contribution value and pore throat distribution (Figures 7, 8), it can be found that due to the severe densification of the Chang 3 reservoir, the pore structure parameters are not much different, so the reservoir has good sorting. Because the pore throat radius tends to be small as a whole, the number of effective throats in the reservoir is relatively small. The permeability of tight sandstone reservoirs is mainly contributed by a small fraction of larger pore throats. Ultimately, it is difficult for mercury to enter into the core and even more difficult to exit the core (the tortuosity of mercury withdrawal is greater than that of mercury injection).

From the correlation between the characteristic parameters of the pore structure of the Chang 3 Member and the petrophysical properties, it is found that the characteristic parameters of the pore structure that have a good correlation with the petrophysical properties include median pressure, mercury injection tortuosity, pore throat radius corresponding to the maximum permeability contribution rate, maximum pore throat radius, and sorting coefficient. Among them, the porosity is negatively and positively correlated with the median pressure and mercury injection tortuosity, respectively (Figure 9). This is because when the pore throat space is small, the capillary resistance is large, and the median pressure is also large. Mercury injection tortuosity reflects the degree of tortuosity of the pore throat morphology experienced by the nonwetting phase (mercury) when entering the core. When the tortuosity of mercury injection is large to a certain extent, the connectivity of pore throats will be impaired, but the increase of pore space will not be affected. The pore throat radius, maximum pore throat radius, and sorting coefficient corresponding to the maximum permeability contribution rate are positively correlated with the permeability (Figure 9). This is because the larger pore throat radius corresponding to the maximum permeability contribution rate and larger maximum pore throat radius are beneficial to the fluid flow in the pore throat space. The larger the sorting coefficient, the higher the probability of pore throats with larger radius and the higher the rock permeability.