Quantitative Identification of Main Controlling Factors for Tight Sandstone Reservoir Sensitivity Based on PLS: A Case Study of the Yanchang Formation in the Xunyi–Yijun Area, Southern Ordos Basin

Abstract

This study aims to evaluate the controlling factors of tight sandstone reservoir sensitivity in the third member of the Yanchang Formation, Xunyi–Yijun area, southern Ordos Basin. Based on core samples from 12 wells, we established a partial least squares regression (PLS) model through thin section observation, SEM, XRD, high-pressure mercury injection, and sensitivity flow experiments, to quantitatively analyze the relationship between reservoir sensitivity and its controlling factors. The results show that the study area reservoirs are dominated by feldspathic sandstone with moderate compaction, characterized by low porosity (4.4–17.8%, avg. 10.93%), low permeability (0.104–2.33 mD, avg. 0.82 mD), and heterogeneous distribution of clay minerals (mainly chlorite, illite, kaolinite, and illite/smectite mixed layer). The reservoirs generally show weak to moderately weak sensitivity. The PLS model reveals that reservoir sensitivity is controlled by the coupled effects of multiple factors, with no single absolute dominant factor for any sensitivity type. Porosity is the most influential variable for overall reservoir sensitivity, followed by feldspar, illite, and illite/smectite mixed layer, and porosity exerts the strongest control on most sensitivity types via VIP score analysis. This study provides a theoretical basis for reservoir damage prevention in the study area and a technical reference for quantitative sensitivity evaluation of similar tight sandstone reservoirs.

1. Introduction

With the continuous deepening of the exploration and development of conventional oil and gas resources worldwide, low-porosity and low-permeability tight oil and gas have become a key strategic area for global oil and gas reserve growth and production enhancement. Among them, continental tight sandstone oil and gas is regarded as an important target for global oil and gas exploration and development in the next 10 to 20 years [1,2,3]. However, such reservoirs are generally characterized by complex mineral compositions, fine pore-throat structures, and strong heterogeneity. During the whole processes of drilling, completion, fracturing, and development, they are prone to various types of reservoir damage, including pore-throat plugging, hydration swelling, particle migration, and stress sensitivity, resulting in irreversible deterioration of seepage capacity and decline in well productivity [4,5,6].

Reservoir sensitivity evaluation is a core approach for revealing reservoir damage mechanisms, quantitatively characterizing damage degrees, and establishing a full-process damage prevention and control technical system [7,8,9]. Through systematic analysis of the evolution laws of reservoir petrology, pore structure, and fluid–rock interactions under various development conditions, the dominant controlling factors of sensitivity damage can be identified, providing theoretical support for the optimization of oil and gas field development plans and reservoir protection. The sensitivity of tight sandstone reservoirs generally includes five types: velocity sensitivity, water sensitivity, acid sensitivity, alkali sensitivity, and stress sensitivity, whose intensities are strictly controlled by the whole process of sedimentary and diagenetic evolution [10]. Studies have shown that reservoir petrological characteristics (clastic composition, interstitial material content), clay mineral types and occurrence states, and pore structure characteristics are the three core internal factors determining the types and damage intensities of reservoir sensitivity, while cementation and dissolution are the key causes leading to strong reservoir heterogeneity and differences in sensitivity [11].

Currently, scholars worldwide have conducted extensive research on the sensitivity of typical tight sandstone reservoirs, revealing the damage mechanisms of individual sensitivities, the dominant role of clay minerals in water sensitivity, and the control of pore-throat structures over stress sensitivity. Furthermore, univariate correlation analyses have been employed to explore the relationships between reservoir sensitivity and key geological parameters. However, existing studies still exhibit three scientific shortcomings: First, research on the composite damage mechanisms of multiple types of sensitivity remains insufficient, lacking systematic quantitative characterization of reservoir parameters and sensitivity under multi-genetic coupling. Second, traditional univariate analysis struggles to address multicollinearity issues, making it difficult to quantify the independent effects and coupled interactions of multiple factors. Third, there is a lack of refined and quantitative sensitivity studies in local potential zones, which hinders the provision of evaluation bases for the development of these areas [12,13].

The Ordos Basin is a representative continental tight oil and gas basin in China. Its tight sandstone reservoirs in the Yanchang Formation are characterized by low porosity, low permeability, and strong heterogeneity, making them prone to multiple types of composite damage during development. Reservoir protection has thus become a core bottleneck restricting development efficiency [14,15,16,17]. The Chang 3 Member reservoir in the Xunyi area exhibits favorable conditions for tight oil accumulation. However, systematic experimental evaluations of reservoir sensitivity and quantitative analyses of multi-factor coupling in this area remain insufficient. This study, based on core observation, thin-section identification, SEM analysis, XRD whole-rock and clay mineral analysis, mercury injection experiments, and laboratory sensitivity tests, introduces a partial least squares regression (PLS) model to quantitatively analyze the relationships between key reservoir parameters and different types of sensitivity. The findings aim to provide theoretical and technical support for the prevention and control of reservoir damage during the efficient development of tight oil.

2. Regional Geological Setting

The Xunyi–Yijun area is located on the southern margin of the Ordos Basin, tectonically situated at the junction of the southern Yishan Slope and the Weibei Uplift (Figure 1). It is adjacent to the Qinling Orogenic Belt to the south and is indirectly influenced by the Weihe Graben and its associated fault systems. The Ordos Basin, as a whole, is a large, multi-cycle, superposed intracratonic basin with relatively stable internal structures. However, its southern margin experienced significant differential uplift and tectonic modification during the Mesozoic and Cenozoic, endowing the area with dual characteristics of stable basin sedimentation and marginal tectonic adjustment. The southern Ordos Basin presently comprises secondary tectonic units such as the southern Yishan Slope and the Weibei Uplift. This structural framework governs the spatial variability in the preservation of strata since the Triassic, the direction of sediment supply, and the intensity of subsequent diagenesis and tectonic modification [18].

From the perspective of the regional stratigraphic framework, the study area develops Precambrian, Lower Paleozoic, Upper Paleozoic, Mesozoic, and Cenozoic strata, among which the Silurian, Devonian, Lower Carboniferous, and Upper Cretaceous are generally absent in the southern basin. The Mesozoic Triassic Yanchang Formation is a significant lacustrine basin sedimentary filling unit in the Ordos Basin, and is divided into ten oil-bearing members, from Chang 10 to Chang 1, in ascending order. Among these, the Chang 7 period represents the peak of lacustrine transgression, marking the maximum expansion stage of the basin’s lake system, while the Chang 6 Member and above entered the Late Triassic evolutionary phase. Existing chronostratigraphic studies indicate that the middle to upper part of the Yanchang Formation spans the Middle to Late Triassic, with an important paleoenvironmental transition interface between the Chang 7 and Chang 6 Members. This context is critically significant for the adjustment of the basin’s sedimentary system and the distribution of reservoirs [19].

During the Late Triassic depositional period of the Yanchang Formation, the Ordos Basin was generally in the evolutionary stage of a large intracontinental depression lake basin, with a widely developed fluvial-delta-lacustrine depositional system within the basin. Numerous studies have shown that the sedimentary filling of the Yanchang Formation was jointly controlled by the gently sloping paleotopography of the basin basement, fluctuations in lake level, continuous sediment supply, and peripheral tectonic activities. This resulted in a lake basin-delta filling pattern characterized by alternating phases of multi-stage progradation and retreat. On a basin scale, the maximum flooding surface was most developed during the Chang 7 period. In contrast, the Chang 3 period was generally marked by enhanced delta progradation against a background of lake basin contraction, with sand bodies continuously advancing into the lake basin interior. This provided an important sedimentary foundation for the formation of tight sandstone reservoirs [20].

The study area, located on the southern margin of the basin near the sediment source input zone, primarily features delta front subfacies deposits in the Chang 3 Member. Common sedimentary bodies include subaqueous distributary channels, mouth bars, and associated sheet sands, with localized characteristics of multi-stage sandbody superimposition and lateral migration. Previous studies on the sedimentary architecture of the Yanchang Formation delta front in the Ordos Basin have indicated that distributary channels in the delta front serve as critical conduits for the transport and redistribution of terrigenous clastics into the lake basin. Their geometry, sandbody connectivity, and stacking patterns are jointly controlled by paleotopography and hydrodynamic conditions. This depositional configuration of “gentle slope lake margin—multi-stage progradation” determines that the sandbodies of the Chang 3 Member exhibit distinct planar zonation and display characteristics of multi-cycle superimposition and enhanced heterogeneity in the vertical dimension [21].

The Chang 3 Member reservoir in the Xunyi–Yijun area is primarily composed of fine- to medium-fine clastic rocks, with the sedimentary environment generally belonging to the transitional zone between the lake margin and the delta front. Influenced by sediment supply, micro-topographic variations, and lake-level fluctuations, the sandbody thickness, grain size, and connectivity exhibit strong lateral variations, resulting in pronounced reservoir heterogeneity. Concurrently, the area has undergone multi-stage diagenetic alterations during burial, including compaction, cementation, and dissolution. These processes led to a substantial loss of primary porosity and formed a tight reservoir system characterized by low porosity, low permeability, and micro-fine pore throats. Therefore, the Chang 3 Member in this area not only exhibits a typical sedimentary response of a southern margin delta front but also displays a geological basis for reservoir sensitivity under the dual control of sedimentation and diagenesis. This provides the regional geological premise for the subsequent quantitative analysis of the main controlling factors of reservoir sensitivity.

3. Materials and Methods

Building on previous research, the authors integrated data from drilling, thin-section petrography, whole-rock analysis, physical property measurements, and sensitivity tests to investigate the correlation between reservoir sensitivity and mineral assemblages in the third member of the Triassic Yanchang Formation in the Xunyi area of the Ordos Basin.

3.1. Sample Selection

For this study, core samples were collected from 12 wells in the Xunyi–Yijun area of the Ordos Basin, all from the Chang 3 Member of the Yanchang Formation. A total of 180 samples were initially collected. After quality control and data integrity screening, 120 samples were ultimately selected for analysis. The sample selection criteria included: 1. Samples were collected from similar sedimentary facies zones to ensure reservoir representativeness; 2. Samples were intact and free from breakage or weathering, preserving complete pore and mineral structures.

3.2. Microstructure Characterization

Thin-section analysis was conducted following the standard SY/T 5368—2016 “Methods for Identifying Rocks Under Microscope,” [22] using a Leica polarizing microscope (Model 009, Leica, Wetzlar, Germany) for observation. A total of 120 thin sections were prepared for grain composition statistics, grain diameter measurement, and point counting. Grain size and contact relationship analyses were employed to interpret the depositional environment and compaction degree. The results indicate that the Chang 3 Member of the Yanchang Formation in the Xunyi area, Ordos Basin, is predominantly composed of sandstone. The cementation types are primarily film-porous cementation and porous cementation. Detrital grains exhibit sub-angular roundness and moderate sorting. This represents a transitional feature typically found in sedimentary environments such as delta fronts or the middle parts of fluvial and alluvial fans, signifying transport channels or depositional areas with frequent energy fluctuations and unstable hydrodynamic conditions. Grain contacts are mainly point-to-line contacts, supplemented by line contacts, which generally reflect that the formation has experienced moderate compaction intensity [22].

3.3. Sample Composition Identification

X-ray diffraction (XRD) analyses of whole-rock and clay mineral compositions were performed following the Chinese petroleum industry standard SY/T 5163-2010 [23]. The XRD instrument used was a Bruker AXS D8-Focus X-ray diffractometer (Bruker, Karlsruhe, Germany). The samples were analyzed as solid powders under ambient conditions of 20 °C and 32% air humidity, with the analysis type being semi-quantitative phase determination. The data were processed using XROCK XRD^® software (Version 1.0). By identifying characteristic peaks in the diffractograms, the mineral contents and clay mineral compositions in the samples were estimated.

3.4. Sensitivity Measurement

In accordance with the Chinese petroleum industry standard SY/T 5358-2010, “Formation Damage Evaluation by Flow Test,” the rock mineral characteristics and potential sensitivity damage of the Chang 3 reservoir in the Yanchang Formation in the Xunyi area, Ordos Basin, were evaluated (Figure 2) [24,25,26,27,28,29]. Reservoir core sensitivity experiments were conducted as follows: Prior to the sensitivity experiments, natural core plug samples from the study area, measuring 6 cm in length and 2.5 cm in diameter, were selected to ensure the representativeness and authenticity of the experimental results. The samples were thoroughly cleaned and dried before the experiment to remove impurities from the core surfaces and guarantee the accuracy of the experimental data. The following key parameters were obtained from the experiments:

$$K_s={\displaystyle \frac{K_n}{K_{is}}}\times 100\%;D_s=1-K_s$$

Theorem 1.

K_s is the permeability ratio; K_is (i stands for v (velocity), w (water), ac (acid), and al (alkali) respectively); D_s is the permeability damage ratio; K_n is the permeability value under the current experimental conditions; K_i is the original permeability of the formation.

Water sensitivity test procedure: The purpose of the water sensitivity test is to evaluate the effect of water with different salinities on core permeability. First, the core is saturated with water having the same salinity as the formation water. It is then subsequently treated with brine whose salinity is half that of the formation water and with distilled water. These fluids with different salinities are injected into the core sequentially at a flow rate of 0.1 m³/min. After each injection, the test fluid is allowed to react with the core sample for more than 12 h, and the change in core permeability is then measured. Finally, the water sensitivity damage rate is calculated based on the initial permeability and the permeability measured after water injection.

Figure 2. Schematic diagram of experimental apparatus and workflow 1. Advection Pump 2. Intermediate Container 3. Filter 4. Pressure Gauge 5. Multi-Port Valve Base 6. Confining Pressure Pump 7. Core Holder 8. Back Pressure Valve 9. Outlet Flowmeter.

Figure 2. Schematic diagram of experimental apparatus and workflow 1. Advection Pump 2. Intermediate Container 3. Filter 4. Pressure Gauge 5. Multi-Port Valve Base 6. Confining Pressure Pump 7. Core Holder 8. Back Pressure Valve 9. Outlet Flowmeter.

The velocity sensitivity experiment is used to evaluate the permeability changes of a core under different flow rates, reflecting the impact of fluid flow velocity on core permeability. The experimental procedure is as follows: Water with the same salinity as the formation water, or another appropriate fluid, is used for the test. The initial flow rate was set to 0.05 mL/min and then gradually increased until the fluid flowing through the core generates a pressure gradient of 2 MPa/cm. Pressure and flow rate data are recorded at each flow rate. At each flow rate, after stabilization, the change in core permeability is measured. Using a calculation method similar to that in the water sensitivity experiment, the permeability changes at different flow rates are determined to assess the flow rate sensitivity of the core.

Acid Sensitivity Experiment Procedure: The purpose of the acid sensitivity experiment is to evaluate the impact of the reaction between acid and core minerals on permeability. The experimental procedure is as follows: First, the permeability of the core sample before acid treatment is measured using potassium chloride (Shanghai Aladdin Biochemical Technology Co., Ltd., Shanghai, China) solution with the same salinity as the formation water. Then, 0.5 to 1 pore volume of acid is injected in the reverse direction through the core. Injection is stopped, the valves at both the inlet and outlet are closed, and the acid is allowed to react for 1 h. Finally, potassium chloride solution with the same salinity as the formation water is injected in the forward direction to measure the permeability of the core sample after acid treatment.

Alkali Sensitivity Experiment Procedure: Prepare a potassium chloride solution and adjust its pH value using NaOH (Shanghai Aladdin Biochemical Technology Co., Ltd., Shanghai, China), increasing the pH value stepwise by 1.5 units each time. Potassium chloride solutions with different pH values are sequentially injected into the core, with an injection volume of 10–15 times the pore volume of the core. After injection, the core is left static for 12 h to ensure full contact between the core and the alkaline solution. Following the 12 h static period, the change in core permeability is measured to evaluate the impact of the alkaline solution on core permeability.

Stress Sensitivity Experiment Procedure: The experiment sets an initial displacement pressure of 0.5 MPa, a confining pressure of 4 MPa, and an initial net stress of 3.5 MPa. If no gas is produced, the initial displacement pressure is increased. Starting from the initial net stress, the confining pressure is increased stepwise until no gas is produced. At this point, the displacement pressure is increased until the confining pressure.

3.5. High-Pressure Mercury Intrusion Porosimetry

The samples were analyzed in accordance with the Chinese petroleum industry standard GB/T 29171-2023 [17]. The instrument used was a Micromeritics 9505 high-pressure mercury intrusion porosimeter (Norcross, GA, USA). Four core samples from the study block were selected for high-pressure mercury intrusion experiments to characterize the pore structure of the study area. Mercury is a non-wetting phase for most rocks. When the applied mercury pressure is greater than or equal to the capillary pressure of the pore throats, mercury overcomes the capillary resistance and enters the pores. Based on the percentage of pore volume invaded by mercury and the corresponding capillary pressure, the relationship curve between capillary pressure and mercury saturation of the samples can be obtained.

3.6. Establishment of the PLS Model

Partial Least Squares (PLS) regression was employed because the predictor set integrated mineralogical and petrophysical variables that are not statistically independent, and the objective of this study was to explain multiple response variables simultaneously. Compared to individual bivariate fitting or ordinary multiple linear regression, PLS is more suitable for datasets with small sample sizes and multicollinearity, as it extracts latent variables from the predictor matrix while maximizing their covariance with the response matrix.

PLS analysis of the data was performed using SIMCA (version 14.1, Sartorius Stedim Data Analytics AB, Umeå, Sweden). The X matrix comprised seven predictor variables: the relative contents of mixed-layer illite/smectite (I/S), kaolinite, chlorite, and illite, as well as the absolute contents of quartz and feldspar, and measured porosity. The Y matrix included five response variables, namely the damage rates of water sensitivity, velocity sensitivity, acid sensitivity, alkali sensitivity, and stress sensitivity. All variables were standardized using z-scores prior to modeling to eliminate dimensional effects and make the regression coefficients directly comparable.

Before model fitting, data quality screening was performed on the predictor matrix. Variance Inflation Factor (VIF) analysis indicated moderate multicollinearity among the clay mineral variables, which supported the use of PLS over ordinary least squares regression. Grubbs’ test did not identify any statistically significant outliers in the analytical dataset, so all samples were retained for subsequent modeling.

The optimal number of latent variables was determined using leave-one-out cross-validation based on the Predicted Residual Sum of Squares (PRESS) and the Q² statistic. Two latent variables were ultimately retained. Under this configuration, the cumulative explained variance for the Y matrix was 0.3433. This result indicates that the model’s predictive capacity is limited; therefore, it should primarily be interpreted as a tool for identifying relative factor importance and multivariate control patterns, rather than as a high-precision prediction model for individual sensitivity indices.

To clarify this limitation, model fit statistics, including R² and RMSE, are reported for each dependent variable in

Table 1. PLS model-fitting statistics for each sensitivity response variable.

Reservoir Sensitivity	R2	RMSE
Water sensitivity	0.0445	8.12%
Velocity sensitivity	0.3654	5.93%
Acid sensitivity	0.5118	4.87%
Alkali sensitivity	0.5656	4.21%
Stress sensitivity	0.2292	6.75%

. The low to moderate R² values suggest that the measured sensitivity damage rates are jointly controlled by several weakly expressed factors and may also contain experimental and geological noise that cannot be fully captured by the current set of predictors.

Variable Importance in the Projection (VIP) was used to rank the contribution of each predictor variable to the multivariate model. In this study, VIP values greater than 1.0 were considered highly influential, values between 0.8 and 1.0 were considered moderately influential, and values below 0.8 were considered weakly influential. This criterion is used only as an interpretive guide. Notably, if no variable achieves VIP > 1, the appropriate geological interpretation is not that a single factor dominates the sensitivity response, but rather that reservoir sensitivity reflects the combined effect of multiple interacting factors. Consequently, porosity is interpreted as the variable with the highest relative importance among the tested predictors, rather than being the sole dominant control.

4. Result

4.1. Reservoir Petrological Characteristics

Statistical analysis of microscopic thin sections indicates that the Chang 3 Member of the Yanchang Formation in the Xunyi–Yijun area of the Ordos Basin is mainly composed of sandstones with varying grain sizes. A ternary diagram (Figure 3) was constructed using the four-component sandstone classification system to categorize the sandstone types. The results show that the target interval is predominantly composed of feldspar sandstones and feldspathic litharenites. Statistical analysis of sandstone types reveals that feldspar sandstones account for approximately 70.6%, while feldspathic litharenites account for about 29.4%

Microscopic observations (Figure 4) indicate that the Chang 3 Member of the Yanchang Formation in the Xunyi–Yijun area of the Ordos Basin is dominated by sandstone. The cementation types are mainly film-pore cementation and pore cementation. The clastic grains exhibit subangular roundness and moderate sorting—an indicator of transitional sedimentary environments, typically occurring in delta fronts, fluvial settings, or mid-alluvial fan areas. Such characteristics represent transport channels or depositional zones with frequent energy fluctuations and unstable hydrodynamic conditions. The grain contact mode is predominantly point-line contact, supplemented by line contact, which generally reflects moderate compaction intensity experienced by the formation.

4.2. Clay Mineral Composition

The clay mineral composition of the study area mainly includes kaolinite, illite, chlorite, montmorillonite, and illite-smectite (I/S) mixed layers, with their contents presented in Figure 4. Among these minerals, kaolinite is derived from feldspar alteration, and its occurrence is predominantly pore-filling. It has a loose structure and mostly appears as irregular flakes (Figure 4), with a relative content ranging from 14% to 60% (Figure 5). Chlorite accounts for 18% to 80% of the total clay minerals (Figure 5), mostly filling pores in a film-like form (Figure 4), and exhibits strong acid sensitivity but weak water sensitivity. Illite has a relative content of approximately 2% to 24% in the samples (Figure 5); this mineral shows strong water sensitivity and tends to absorb water, swell, or migrate, thereby affecting the migration of oil and gas. The illite-smectite (I/S) mixed layer also responds strongly to water, accounting for about 8% to 48% of the samples (Figure 5).

4.3. Pore-Permeability Structure Characteristics

Petrophysical properties of the reservoir control the reserves and productivity of oil and gas reservoirs, and the pore-permeability structure of the reservoir characterizes the quality of its physical storage capacity. Results of gas-measured permeability show that the porosity of the Chang 3 Member of the Yanchang Formation in the Xunyi–Yijun area, Ordos Basin, ranges from 4.4% to 17.8% with an average of 10.93%, while the permeability varies from 0.104 mD to 2.33 mD with an average of 0.82 mD (Figure 6).

4.4. Reservoir Sensitivity Characteristics

Water sensitivity refers to the phenomenon where clay minerals in the reservoir absorb water and swell upon contact with low-salinity fluids, thereby blocking pore throat channels and damaging reservoir permeability. The experimental results (

Table 2. Maximum and Minimum Values of Reservoir Sensitivity for Samples in the Chang 3 Member of Yanchang Formation, Xunyi–Yijun Area.

Formation	Types of Sensitivity	Maximum Value	Minimum Value
Chang 3	Water sensitivity	51.1%	6%
	Velocity sensitivity	68.4%	5.98%
	Acid sensitivity	40%	3.86%
	Alkali sensitivity	43.4%	0.26%
	Stress sensitivity	69.2%	9.5%

) indicate that the water sensitivity damage rate for the Chang 3 Member of the Yanchang Formation in the Xunyi–Yijun area, Ordos Basin, ranges from 6% to 51.1%, generally exhibiting weak to moderately weak water sensitivity.

Velocity sensitivity refers to the phenomenon where, when the fluid flow rate entering the reservoir is too high, authigenic mineral particles within the reservoir detach and migrate, thereby blocking pores and ultimately deteriorating reservoir physical properties. The experimental results show that the velocity sensitivity damage rate for the Chang 3 Member of the Yanchang Formation in the Xunyi–Yijun area, Ordos Basin, ranges from 5.98% to 68.4% (Table 2), indicating generally weak velocity sensitivity.

Acid sensitivity is one of the fundamental properties evaluated for reservoir sensitivity. It manifests as reactions between acidic fluids and authigenic minerals within the reservoir after acid injection, generating precipitates or causing mineral detachment, which leads to changes in reservoir physical properties. As shown in the table, the acid sensitivity permeability damage rate for the Chang 3 Member of the Yanchang Formation in the Xunyi–Yijun area, Ordos Basin, ranges from 3.86% to 40% (Table 2), mainly characterized by weak to moderately weak acid sensitivity.

Alkali sensitivity evaluation aims to quantitatively determine the extent of damage to reservoir physical properties caused by injecting high-pH fluids into the formation. The experimental results (Table 2) indicate that the alkali sensitivity permeability damage rate for the Chang 3 Member of the Yanchang Formation in the Xunyi–Yijun area, Ordos Basin, ranges from 0.26% to 43.4%, exhibiting overall weak to moderately weak alkali sensitivity. This suggests that high-pH fluids cause slight damage to the reservoir.

Stress sensitivity primarily studies the characteristics of reservoir permeability changes under varying confining pressure conditions. The experimental results show that the stress sensitivity damage rate for the Chang 3 Member in the Xunyi–Yijun area, Ordos Basin, ranges from 9.5% to 69.2% (Table 2). The reservoir mainly exhibits weak to moderately weak stress sensitivity.

4.5. Pore-Throat Structure Characteristics

According to high-pressure mercury intrusion experiments, the median pore-throat radius of the Chang 3 Member reservoir in the Xunyi–Yijun Area is greater than 0.08 μm, the average mercury ejection efficiency is higher than 25%, and the average maximum mercury intrusion saturation exceeds 85% (Figure 7).

4.6. Controlling Factors of Reservoir Sensitivity

Variable Importance in the Projection (VIP) is used to evaluate the comprehensive contribution of each predictor variable within the multi-response PLS framework. Generally, variables with VIP > 1 are considered highly influential, those between 0.8 and 1 are considered moderately influential, and those with VIP < 0.8 are considered weakly to moderately influential. In this study, all VIP values are considerably below 1, indicating that no single variable can be regarded as an absolutely dominant control on reservoir sensitivity. Instead, the sensitivity response reflects the combined influence of reservoir physical properties, mineral composition, and diagenesis.

$$VI{P}_j=\sqrt{{\displaystyle \frac{p}{hS{S}_Y}}{\sum }_{k=1}^{h}S{S}_Y \left(t_K\right){\omega }_{kj}^2}$$

Theorem 2.

p = 7 (number of independent variables), h = 2 (number of latent variables), SS_Y is the total sum of squares of variation of Y, SS_Y (tk) is the sum of squares of variation of Y explained by the k-th latent variable, and w_kj is the j-th element of the weight vector of the k-th latent variable.

As shown in

Table 3. Quantitative Results of Controlling Factor Analysis.

Independent Variables	I/S	K	C	I	Quartz	Feldspar	Porosity
VIP	0.1745	0.1563	0.1741	0.1845	0.0849	0.2120	0.3115

, porosity exhibits the highest VIP value (0.3115), followed by feldspar (0.2120) and illite (0.1845), while quartz shows the lowest VIP value (0.0849) (Table 3). These values should be interpreted comparatively rather than absolutely: they establish a relative ranking of variables within the tested dataset, but they do not imply that porosity alone determines all types of sensitivity. This finding aligns with the explanatory power of the PLS model (Section 3.6) and indicates that the reservoir sensitivity of the Chang 3 Member is governed by multiple interacting factors rather than by a single dominant variable.

Therefore, the role of the PLS results in this study is to provide an integrative framework for ranking factor importance and constraining geological interpretations of individual sensitivity types.

5. Discussion

Based on the multivariate quantitative analysis results of the PLS model, none of the five types of sensitivity damage—velocity sensitivity, water sensitivity, acid sensitivity, alkali sensitivity, and stress sensitivity—in the tight sandstone reservoir of the Chang 3 member in the Xunyi–Yijun area of the Ordos Basin is governed by a single dominant controlling factor. Instead, they result from the coupled effects of multiple factors, including reservoir physical properties, detrital mineral composition, and clay mineral types and contents. By extracting latent variables, the PLS model effectively addresses the issue of multicollinearity among variables, which cannot be resolved by traditional single-factor analysis, and achieves a quantitative characterization of the independent relative contributions of seven independent variables within a multi-response variable system.

The ranking of Variable Importance in Projection (VIP) values from the model (as shown in Figure 8) indicates that the variable with the highest contribution to the comprehensive reservoir sensitivity is porosity (VIP = 0.3115), followed by feldspar (VIP = 0.2120), illite (VIP = 0.1845), mixed-layer illite/smectite (I/S, VIP = 0.1745), chlorite (VIP = 0.1741), and kaolinite (VIP = 0.1563), while quartz exhibits the weakest overall contribution (VIP = 0.0849). The fact that all VIP values are less than 1 further corroborates the multi-factor coupling nature of reservoir sensitivity in the study area, indicating that no single variable can fully explain the overall pattern of sensitivity damage. The subsequent content of this chapter will be driven by the quantitative conclusions of the PLS analysis outlined above, integrating single-factor correlation analysis, scanning electron microscopy (SEM) micro-observation, and diagenetic evolution analysis to systematically elucidate the controlling mechanisms of tight sandstone reservoir sensitivity under a multi-factor coupling framework. In this context, single-factor linear correlation analysis serves only as an auxiliary validation tool for the PLS findings, rather than as a basis for determining the dominant controlling factors.

5.1. Differentiated Regulation Mechanisms of Clay Minerals on Various Sensitivities Under a Multivariate Coupling System

Clay minerals serve as the core carriers of fluid–rock interactions in reservoirs. Quantitative results from the PLS model reveal that different types of clay minerals exhibit distinctive contribution characteristics within the multivariate control systems governing various sensitivities, and their effects on sensitivity are consistently regulated by coupling with macroscopic reservoir parameters such as porosity.

5.1.1. Coupled Regulation of Water Sensitivity by Clay Minerals

The coefficient of determination R² for water sensitivity in the PLS model is only 0.0445, indicating that no single independent variable can effectively explain the overall characteristics of water sensitivity damage, which is most significantly governed by multi-factor coupling. The model results (Figure 9) show that porosity and I/S exhibit relatively higher importance within the multi-factor system for water sensitivity damage, suggesting that water sensitivity is more likely the result of the coupling between reservoir space constraints and the hydration swelling potential of I/S. In contrast, illite, chlorite, and kaolinite show weak independent contributions to water sensitivity. This finding directly indicates that water sensitivity in the study area cannot be explained by the content of a single clay mineral but is instead the result of the coupling between clay mineral hydration swelling potential and reservoir space constraints.

Single-factor correlation analysis shows that the water sensitivity damage rate exhibits a moderate positive correlation with increasing I/S content (Figure 10), with a coefficient of determination R² = 0.5523, which aligns with the strong hydration swelling capacity of the smectite layers within I/S, verifying the mineralogical trigger role of I/S for water sensitivity identified by the PLS model. However, this univariate correlation still cannot fully explain the overall characteristics of water sensitivity damage without the macroscopic constraints of porosity, which is consistent with the low R² (0.0445) of the PLS model for water sensitivity, further demonstrating that I/S content alone, without the macroscopic constraints of porosity, cannot fully capture the true pattern of water sensitivity damage. In the multivariate framework of the PLS model, porosity exhibits the highest contribution to water sensitivity primarily because porosity determines the space for fluid intrusion, the contact area for water-rock reactions, and the migration and plugging efficiency of swelling products: in low-porosity reservoirs, pore throats are inherently narrower, and even limited clay hydration swelling can rapidly block effective flow pathways, leading to more severe permeability damage, whereas high-porosity reservoirs possess greater buffering capacity, and the same content of I/S hydration swelling results in significantly reduced damage.

Furthermore, the PLS model shows that illite and chlorite have extremely low independent contributions to water sensitivity (Figure 9), and single-factor correlation analysis reveals no significant linear relationship between these two minerals and water sensitivity, indicating that they are not the primary triggers for water sensitivity damage in the study area. However, SEM observations reveal that flaky and fibrous illite, along with pore-filling chlorite, modify pore throat connectivity and alter the mechanical stability of fine particles, thereby indirectly modulating the ultimate degree of water sensitivity damage during the hydration swelling of I/S [30]. This also represents an important reason why the correlation between I/S content alone and water sensitivity is relatively weak.

5.1.2. Regulatory Role of Clay Minerals on Velocity Sensitivity

The coefficient of determination R² for velocity sensitivity in the PLS model is 0.3654, indicating a moderate explanatory capacity of the model. The results show that among the multi-factor system for velocity sensitivity damage, illite and chlorite exhibit relatively higher contributions among clay minerals, while porosity still plays an important comprehensive modulating role; kaolinite shows a weak independent contribution and exhibits a weakly negative correlation with velocity sensitivity (Figure 9). These findings suggest that velocity sensitivity damage is more likely associated with the detachment, migration, and bridging of fine particles under fluid scouring, with illite and chlorite serving as important sources of migratable fine particles, while porosity determines the plugging efficiency of particle migration [31].

Single-factor correlation analysis verifies a significant positive correlation between the velocity sensitivity damage rate and the combined content of illite and chlorite (Figure 10), which is highly consistent with the quantitative conclusions of the PLS model. From a geological mechanism perspective, SEM observations reveal that illite in the study area predominantly occurs as fibrous or flaky forms attached to particle surfaces or bridging between pore throats, while chlorite develops abundant pore-filling needle-like aggregates. Both minerals exhibit weak bonding strength with the particle substrate, making them prone to detachment under increased fluid flow rates, forming migratable fine particles that create bridging plugs at narrow pore throats, resulting in irreversible permeability damage. In contrast, kaolinite mostly occurs as book-like aggregates filling larger intergranular pores and is less likely to undergo bulk detachment and migration under fluid scouring, which accounts for its weak contribution and weakly negative correlation in the PLS model.

Meanwhile, the core regulatory role of porosity in velocity sensitivity is reflected in the fact that low-porosity reservoirs have smaller throat radii, where even a small amount of fine particle migration can easily form effective plugs, whereas high-porosity reservoirs have wider throats and better connectivity of flow pathways, allowing fine particles to be more readily carried out by the fluid without causing sustained permeability damage. This also represents the fundamental reason why the correlation between clay mineral content and velocity sensitivity is limited in traditional single-factor analyses.

5.1.3. Differential Effects of Clay Minerals on Acid Sensitivity and Alkali Sensitivity

The coefficient of determination R² for acid sensitivity in the PLS model is 0.5118, representing the second-highest explanatory power among the five types of sensitivities, indicating that the selected independent variables can effectively characterize the controlling factors for acid sensitivity damage. The model results show that within the multi-factor system for acid sensitivity damage, chlorite exhibits relatively greater mineralogical importance, while porosity also exerts a strong modulating effect; the independent contributions of other clay minerals and detrital minerals are relatively weak (Figure 9). These findings support the interpretation that chlorite is more likely an important mineralogical trigger for acid sensitivity damage in the study area, while porosity significantly influences the degree of acid sensitivity expression, with their coupled interaction jointly controlling the overall characteristics of reservoir acid sensitivity.

Single-factor correlation analysis reveals a strong positive correlation between the acid sensitivity damage rate and chlorite content (Figure 11), with a coefficient of determination R² of 0.953, which is consistent with the relative importance of chlorite indicated by the PLS results. From a geological mechanism perspective, chlorite in the study area is predominantly iron-bearing chlorite. Its reaction with acid releases large quantities of iron ions, which, with variations in fluid pH, readily form secondary precipitates such as iron hydroxide, clogging pore throats and causing permeability damage. The high contribution of porosity to acid sensitivity in the PLS model explains the fundamental reason for the significant variation in acid sensitivity damage rates among samples with identical chlorite content: high-porosity reservoirs have wider pore throats and better flow connectivity, allowing precipitates generated by acid reaction to be more readily carried out by the fluid without forming effective plugs, whereas low-porosity reservoirs have microfractures and narrow throats that are easily completely blocked by secondary precipitates, potentially leading to severe permeability damage even with relatively low chlorite content. Additionally, the PLS model shows that feldspar also has a certain relative contribution to acid sensitivity, which is closely related to the acid dissolution characteristics of feldspar: acid dissolution of feldspar can, on one hand, generate secondary porosity and improve reservoir flow capacity; on the other hand, its dissolution products may also undergo secondary precipitation, synergistically exacerbating pore throat blockage with the acid dissolution precipitates from chlorite.

For alkali sensitivity, the coefficient of determination R² in the PLS model is 0.5656, representing the highest explanatory power among all sensitivity types. However, the model results indicate that all clay mineral variables exhibit weak independent contributions to alkali sensitivity, with no single dominant clay mineral. Single-factor correlation analysis also verifies that kaolinite, illite, chlorite, and I/S show no significant linear correlation with the alkali sensitivity damage rate, which is consistent with the experimental results indicating overall weak to moderately weak alkali sensitivity in the study area. This suggests that the reaction degree between clay minerals and alkaline fluids is relatively low, and alkali sensitivity damage is not determined by the dissolution of a single clay mineral but is instead regulated by the coupled interaction of multiple factors including pore structure and mineral surface properties.

5.2. Macro-Regulatory Role of Reservoir Petrological Characteristics on Sensitivity

The VIP ranking of all variables in the PLS model shows that porosity is the variable with the highest contribution to comprehensive reservoir sensitivity in the study area, and it exhibits significant modulating effects on various types of sensitivities such as velocity sensitivity, acid sensitivity, and stress sensitivity. Feldspar ranks second in overall contribution, while quartz shows the weakest overall contribution. These results clarify that reservoir petrological and physical characteristics constitute important macroscopic constraints regulating multiple types of sensitivity damage, and their influence should not be understood in isolation from individual clay minerals.

5.2.1. Macro-Regulatory Mechanism of Porosity on Various Sensitivities

In the PLS model, porosity demonstrates the highest relative importance among the overall variable system and exhibits extensive modulating effects across multiple sensitivity types. This is not because porosity directly triggers sensitivity damage, but because it controls the geometry, connectivity, and spatial buffering capacity of the pore-throat system, thereby determining whether processes such as clay hydration swelling, fine particle migration, secondary precipitation, and stress deformation can effectively translate into irreversible permeability damage. This also represents the fundamental reason why the coefficients of determination for correlations between various minerals and corresponding sensitivities are generally low in traditional single-factor analyses—without the macroscopic constraints of porosity, the content of a single mineral cannot fully capture the true pattern of sensitivity damage.

For different types of sensitivity, the regulatory mechanism of porosity exhibits consistent characteristics: for water sensitivity, porosity determines the plugging efficiency of hydration swelling products; for velocity sensitivity, porosity controls the bridging probability of fine particle migration; for acid sensitivity, porosity governs the retention and expulsion efficiency of secondary precipitates; for stress sensitivity, porosity directly controls the magnitude of permeability loss during rock compression [32]. The porosity of reservoirs in the study area generally ranges from 4.4% to 17.8%, with an average of only 10.93%, representing typical low-porosity tight reservoirs with a high proportion of micro-fine pore throats and extremely weak buffering capacity within the pore system. Minor changes in pore structure induced by various mineral-fluid reactions are amplified into significant permeability damage, which constitutes the core macroscopic background for the generally weak to moderately weak sensitivity exhibited by reservoirs in the study area [33].

5.2.2. Modulating Role of Detrital Minerals on Sensitivity

The comprehensive VIP value of feldspar in the PLS model reaches 0.2120, ranking second only to porosity among all variables. This result reflects the dual geological role of feldspar within the multi-factor system of reservoir sensitivity. On one hand, as detrital framework grains, feldspar content directly influences rock structural stability and pore space evolution; on the other hand, feldspar dissolution during diagenesis not only generates secondary porosity that improves reservoir physical properties but also forms authigenic clay minerals such as kaolinite and illite through dissolution-reprecipitation processes, altering the mineral composition and surface properties of the reservoir, thereby exerting indirect but widespread effects on various sensitivities including water sensitivity, acid sensitivity, and velocity sensitivity. This explains why feldspar shows high contributions in the multi-factor system while exhibiting weak linear relationships with various sensitivities in single-factor correlation analyses.

For quartz, the PLS model shows that its comprehensive VIP value is only 0.0849, the lowest among all variables, indicating that quartz is not a dominant controlling factor for comprehensive reservoir sensitivity in the study area. Single-factor correlation analysis verifies a weak negative correlation between quartz content and stress sensitivity damage rate (Figure 12), which is closely related to the stronger compression resistance of rock frameworks composed of rigid quartz particles. However, its modulating effect is limited to stress sensitivity alone. Meanwhile, quartz overgrowth cementation directly fills primary and secondary pores, reducing reservoir porosity and throat connectivity, indirectly promoting various types of sensitivity damage. The superposition of these positive and negative effects ultimately results in quartz exhibiting an extremely low comprehensive contribution in the PLS model.

5.3. Microscopic Modulation Mechanisms of Diagenetic Evolution and Mineral Occurrence on Sensitivity

The variable contribution ranking provided by the PLS model essentially represents the quantitative manifestation of the diagenetic evolution process in the study area. The Chang 3 member reservoir in the study area experienced moderate-intensity compaction, cementation, and dissolution, with these three diagenetic processes simultaneously modifying reservoir porosity, detrital mineral composition, and the type, content, and occurrence of clay minerals—precisely the core independent variables selected in this PLS model. In other words, the quantitative characterization of relative variable contributions in the PLS model is essentially a quantitative evaluation of the capacity of diagenetic end-products to regulate sensitivity [34].

5.3.1. Fundamental Control of Compaction on Sensitivity

Thin section observations reveal that detrital grains in the study area reservoir are predominantly in point-line contact, with local line contact development, indicating that the reservoir experienced moderate-intensity compaction (Figure 4). Compaction is the core diagenetic process responsible for primary porosity loss in the study area reservoir, directly determining the development degree of porosity—the core independent variable in the PLS model. Additionally, by compressing pore space, compaction exacerbates the filling effect of cementation on pore throats, representing the fundamental cause of the reservoir’s low-porosity, low-permeability characteristics [35].

In the PLS model, porosity exhibits the highest contribution to all sensitivity types, essentially reflecting the comprehensive manifestation of compaction’s modifying effect on the reservoir pore system. Moderate-intensity compaction significantly reduced primary pore space, forming a pore system dominated by micro-fine throats, greatly diminishing the reservoir’s flow buffering capacity. This makes processes such as clay hydration, particle migration, and secondary precipitation highly likely to cause irreversible permeability damage, providing the core geological background for the development of various sensitivities.

5.3.2. Differential Modulation of Sensitivity by Cementation

Cementation is the core diagenetic process regulating the contributions of clay mineral and quartz variables in the PLS model. SEM observations (Figure 13) show that two main types of cementation develop in the study area: clay mineral cementation and siliceous cementation, with the occurrence of different cements directly determining their relative contributions in the PLS model.

For clay mineral cementation, mixed-layer I/S fills pores in honeycomb-like aggregates (Figure 13), possessing a large specific surface area and hydration swelling capacity. This microscopic occurrence characteristic reasonably explains the geological basis for I/S showing relative importance for water sensitivity within the PLS framework. Chlorite develops two types of occurrence: coating-type and pore-filling needle-like forms (Figure 13). Coating-type chlorite exhibits tight bonding with the grain substrate, effectively inhibiting fluid erosion of grains and providing some mitigation of sensitivity damage, which helps explain why chlorite does not show higher relative importance than I/S in the comprehensive PLS ranking. Meanwhile, needle-like chlorite, with its large specific surface area, is prone to fragmentation and migration under fluid scouring and readily generates secondary precipitates when reacting with acid, representing an important mechanism exacerbating velocity sensitivity and acid sensitivity damage, providing microscopic explanation for chlorite’s relative influence within the PLS framework. Illite occurs as fibrous and flaky forms bridging between pore throats, highly susceptible to detachment under fluid scouring to form migratable particles. This occurrence characteristic strongly aligns with illite’s high contribution to velocity sensitivity in the PLS model.

For siliceous cementation, it primarily develops in the form of quartz overgrowths (Figure 13), directly reducing reservoir porosity and throat connectivity by filling primary and secondary pores. From a diagenetic mechanism perspective, quartz overgrowth continuously consumes pore space, explaining its indirect adverse impact on the reservoir pore system and sensitivity response. This is also the fundamental reason why quartz, despite its rigid compression resistance, exhibits the lowest comprehensive VIP value in the model—its destructive effect on the pore system far outweighs its weak mitigating effect on stress sensitivity [36,37,38,39].

5.3.3. Dual Effects of Dissolution on Sensitivity

Dissolution in the study area is predominantly acidic dissolution of feldspar (Figure 13), exerting significant dual effects on the independent variables in the PLS model, which helps explain why feldspar exhibits relatively high comprehensive importance in the model. On one hand, secondary pores generated by feldspar dissolution can improve reservoir porosity and enhance the buffering capacity of the pore system, mitigating various types of sensitivity damage. On the other hand, dissolution products of feldspar can undergo reprecipitation, forming authigenic clay minerals such as kaolinite and illite, altering the reservoir’s clay mineral composition. Simultaneously, rough particle surfaces created by dissolution can exacerbate fine particle attachment and migration, thereby promoting sensitivities such as velocity sensitivity and water sensitivity.

This superposition of positive and negative effects makes it impossible for single-factor analysis to clarify a unidirectional influence of feldspar on sensitivity, yet feldspar exhibits relatively high comprehensive importance in the PLS multivariate system, fully demonstrating the core advantages of the PLS model in handling multi-factor coupling and bidirectional effects.

5.4. Multi-Factor Coupling Model of Reservoir Sensitivity Damage and Methodological Implications

Integrating PLS multivariate quantitative analysis with geological genetic analysis enables the construction of a comprehensive model linking diagenetic evolution, reservoir physical properties and mineral composition, multi-factor coupling, and sensitivity damage for the Chang 3 member tight sandstone reservoir in the Xunyi–Yijun area of the Ordos Basin: moderate-intensity compaction and cementation jointly shaped the macroscopic background of low porosity and low permeability, determining the core regulatory role of porosity across all sensitivity types; the dual effects of feldspar dissolution simultaneously modified reservoir pore structure and clay mineral composition, explaining why feldspar exhibits relatively high importance second only to porosity in the overall VIP ranking; clay minerals of different types and occurrences serve as important mineralogical triggers for various sensitivity damages, with their influence on sensitivity consistently constrained macroscopically by porosity.

Traditional single-factor linear correlation analysis can only reflect co-variation trends between two variables, cannot eliminate multicollinearity interference from other variables, and struggles to achieve simultaneous analysis of multiple response variables, thus failing to fully reveal the essence of multi-factor coupling control in tight sandstone reservoir sensitivity. The PLS model introduced in this study effectively addresses the core limitations of traditional methods, achieving quantitative characterization of relative contributions from multiple factors and providing a reliable methodological framework for quantitative identification of dominant controlling factors in tight sandstone reservoir sensitivity.

Meanwhile, the relatively low cumulative explained variance of the PLS model also indicates that reservoir sensitivity in the study area is further influenced by additional geological factors not included in the model, such as pore throat heterogeneity, spatial variations in diagenetic intensity, and fluid properties. Future research can improve the independent variable system to enhance model explanatory power, providing more precise theoretical support for reservoir damage prevention and efficient development in the study area [40].

6. Conclusions

This study quantitatively analyzed the sensitivity of tight sandstone reservoirs in the Chang 3 Member of the Yanchang Formation in the Xunyi–Yijun area, Ordos Basin, by employing a Partial Least Squares Regression (PLS) model. Integrating experimental data from core observations, thin-section analysis, SEM observations, and XRD mineral composition analysis, the following core conclusions are drawn:

• The PLS model results indicate that reservoir sensitivity is controlled by the combined effects of multiple factors. Specifically, porosity, feldspar, and mixed-layer illite/smectite (I/S) minerals contribute significantly to different types of sensitivity. Through VIP analysis for various sensitivity types, this study confirms the relatively dominant role of porosity in reservoir sensitivity, particularly concerning velocity sensitivity and acid sensitivity. Furthermore, the influence of I/S minerals on water sensitivity is prominent, reflected by their relatively high VIP value in the PLS model, suggesting that their swellability has a direct impact on reservoir water sensitivity.
• This study reveals the complex relationship between the sensitivity of tight sandstone reservoirs and the occurrence state and evolutionary characteristics of minerals during diagenesis. By combining SEM observations with the PLS model, this research further validates the role of the honeycomb-like texture of I/S minerals in promoting hydration swelling and correlates this with their contribution to water sensitivity in the PLS model. This finding provides a new methodological framework for evaluating the sensitivity of tight sandstone reservoirs and offers theoretical support for developing relevant reservoir protection strategies. The study also indicates that the impacts of quartz overgrowth and feldspar dissolution on the reservoir are dual-faceted. While secondary pores generated by feldspar dissolution can improve porosity, the reprecipitation of dissolution byproducts can form pore-blocking cements, further influencing reservoir permeability.
• Although this study effectively analyzed reservoir sensitivity using the PLS model, the relatively low cumulative explained variance of the model suggests that interactions among multiple factors and unconsidered geological variables may affect a complete interpretation of reservoir sensitivity. Limitations in sample size and the local adaptability of experimental conditions could also influence the generalizability of the model results. Future research can build upon this study by expanding the sample size, considering tight sandstone reservoir sensitivity in different regions and sedimentary settings, and delving deeper into the mechanisms of mineral dissolution-reprecipitation during diagenesis and their control on reservoir sensitivity. Such efforts will contribute to a more comprehensive understanding of the multi-factor regulatory mechanisms in tight sandstone reservoirs and provide more accurate predictions and protection strategies for practical development.