Multifactor Coupling Effects on Permeability Evolution During Reinjection in Sandstone Geothermal Reservoirs: Insights from Dynamic Core Flow Experiments

Abstract

Efficient reinjection is critical for maintaining reservoir pressure and ensuring the sustainable development of sandstone geothermal systems. However, complex thermal–hydraulic–chemical (THC) interactions often lead to progressive permeability reduction, significantly impairing injection performance. This study systematically investigates the coupled effects of injection flow rate, temperature, and suspended particle size on permeability evolution during geothermal reinjection. Laboratory-scale core flow-through experiments were conducted using sandstone samples from the Guantao Formation in the Huanghua Depression, Bohai Bay Basin. The experimental schemes included graded flow rate tests, temperature-stepped injections, particle size control, long-term seepage, and reverse-flow backflushing operations. The results reveal that permeability is highly sensitive to injection parameters. Flow rates exceeding 6 mL/min induce irreversible clogging and pore structure damage, while lower rates yield more stable injection behavior. Injection at approximately 35 °C resulted in a permeability increase of 15.7%, attributed to reduced fluid viscosity and moderate clay swelling and secondary precipitation. Particles larger than 3 μm were prone to bridging and persistent clogging, whereas smaller particles exhibited more reversible behavior. During long-term seepage, reverse injection implemented upon permeability decline restored up to 98% of the initial permeability, confirming its effectiveness in alleviating pore throat blockage. Based on these findings, a combined reinjection strategy is recommended, featuring low flow rate (≤5 mL/min), moderate injection temperature (~35 °C), and fine filtration (≤3 μm). In addition, periodic backflushing should be considered when permeability loss exceeds 30% or a sustained injection pressure rise is observed. This study provides robust experimental evidence and practical guidance for optimizing geothermal reinjection operations.

1. Introduction

Sandstone geothermal reservoirs, as one of the most widespread forms of medium–low temperature geothermal resources, are commonly found in Mesozoic and Cenozoic rift basins [1,2]. In China, they are predominantly developed in the Bohai Bay Basin, the southern North China Basin, and the Songliao Basin, with the Neogene Guantao Formation sandstones serving as typical representatives. These reservoirs are characterized by shallow burial depth, wide spatial distribution, abundant thermal reserves, and favorable hydrothermal properties, making them the principal source of geothermal space heating in northern China [3,4]. However, during geothermal exploitation, insufficient reinjection of the produced geothermal water often results in reservoir pressure depletion, land subsidence, reduced resource sustainability, and environmental risks [5,6]. Therefore, enhancing reinjection efficiency has become a critical technical challenge for the sustainable and environmentally responsible utilization of sandstone geothermal reservoirs [7].

The recharge effect of sandstone reservoirs is affected by a variety of factors, and the recharge process not only involves complex physicochemical mechanisms, including fluid–rock interactions, mineral dissolution and precipitation, clay mineral particle migration, fine particle reorganization, vesicle obstruction, microbial film formation, etc. [4,7,8], but it is also affected by recharge parameters such as flow rate, temperature, and suspended particle size [9,10]. The synergistic effect of these factors determines the dynamic evolution of recharge efficiency and reservoir permeability, which in turn affects the long-term sustainability of geothermal resources.

In recent years, extensive experimental research has been conducted both domestically and internationally to investigate the clogging mechanisms induced by water–rock interactions during the reinjection process in sandstone geothermal reservoirs [11,12]. These studies have provided a more systematic understanding of macroscopic clogging behavior. It has been demonstrated that the compatibility between the particle size of suspended solids and the pore throat diameter is a critical factor influencing the extent of permeability reduction [13]. In the Guantao Formation of the Huanghua Depression, high clay content (e.g., up to 15% in the Wuqing Sub-depression) has been shown to significantly impair pore connectivity, resulting in substantial resistance to cold-front advancement during the early stages of reinjection and a marked decline in reinjection capacity within the first 48 h [14]. High-temperature and high-pressure dissolution experiments have revealed that at approximately 65 °C, precipitation of calcite and dolomite—formed through reactions involving Ca²⁺, Mg²⁺, and HCO₃⁻—is a major cause of permeability deterioration. Further studies indicate that chemical clogging can be effectively mitigated by lowering the reinjection tailwater temperature or by adjusting the pH and concentrations of Ca²⁺, Mg²⁺, and HCO₃⁻ prior to reinjection [15]. Displacement experiments and hydrogeochemical simulations have also shown that chemical clogging in the sandstone reservoirs of the Weihe Basin is primarily associated with the precipitation of carbonates, silicates, and iron-bearing minerals and that the severity of clogging increases with higher tailwater temperatures [16]. Laboratory column experiments simulating water–rock interactions during reinjection have revealed that mineral dissolution and precipitation predominantly occur in the frontal zone of the column. Cation exchange processes—particularly involving Na⁺, Ca²⁺, and Mg²⁺—were identified as significant contributors to chemical clogging [17]. In related studies conducted on sandstone reservoirs of the Guantao Formation in Tianjin, significant increases in concentrations of trace elements such as Sr, Li, and Ba were observed during reinjection, indicating active mineral dissolution. PHREEQC simulations further confirmed that these changes are closely related to complex mineral dissolution–precipitation equilibria [18]. According to the International Energy Agency (IEA), enhancing reinjection efficiency in sandstone geothermal reservoirs is critical for achieving large-scale and sustainable utilization of geothermal resources [19]. A key technical challenge lies in the quantitative characterization and predictive modeling of multi-factor coupled mechanisms during dynamic seepage processes [15].

While current research efforts have primarily focused on the optimization of reinjection parameters, including injection rate and temperature, as well as on identifying dominant clogging mechanisms, most laboratory investigations are limited to single-variable conditions. For instance, Zhang et al. conducted a series of seepage experiments on unconsolidated sandstones to investigate clogging behavior caused by grain migration and stress-induced compaction. Although the study highlighted the role of grain size distribution and mechanical stress in physical clogging, it did not address the potential interactions with thermal or geochemical processes [10]. Z. Zhao et al. carried out seepage experiments across a range of flow rates to determine the critical rate for particle clogging yet did not consider temperature–chemical coupling [20]. T. Zhong et al. focused solely on thermal effects on sandstone permeability evolution without accounting for flow conditions [21]. Ying Xu et al. used triaxial thermal–hydraulic–mechanical (THM) tests to evaluate permeability changes in argillaceous sandstone under elevated temperatures and found that illite expansion and particle release contributed to pore blockage [22]. In addition, a few studies have begun to explore reinjection processes under the coupling of two or more factors [23,24,25,26,27]; however, such investigations remain limited, particularly with respect to the synergistic effects of thermal, hydraulic, and particle-related mechanisms. Given these research gaps, the present study, which addresses three-factor coupling, offers valuable insights into the permeability evolution during geothermal reinjection.

To this end, an integrated experimental and theoretical approach is employed to investigate the coupled fluid–rock interaction mechanisms during reinjection in sandstone geothermal reservoirs. This study aims to elucidate how these interactions influence the evolution of reservoir physical properties under prolonged dynamic flow conditions. This study systematically investigates the effects of temperature (20–60 °C), flow rate (1–10 mL/min), and suspended particle size (1–5 μm) on the permeability and reinjection performance of Guantao Formation sandstone. A multi-factor, stepwise core flow experiment was designed to simulate key processes during geothermal reinjection. The results reveal the characteristic trends of permeability evolution and highlight the typical features of clogging behavior under different injection conditions. The aim is to provide an experimental basis for optimizing reinjection strategies, enhancing reservoir responsiveness, and reducing clogging risks. The findings offer valuable technical insights for the efficient and sustainable development of geothermal energy in sandstone-dominated systems.

2. Samples and Experimental Methods

2.1. Geological Background and Sample Selection

The Huanghua Depression is located in the Bohai Bay basin of North China and is geographically situated in the eastern part of Hebei Province and the southern part of Tianjin City. The region boasts a complex geological structure primarily formed by sedimentation since the Neogene period. It is rich in geothermal resources predominantly consisting of sandstones from the Minghuazhen Formation and Guantao Formation of the Neogene system. These resources are characterized by shallow burial depth, widespread distribution, large reserves, high porosity, and low permeability. The thermal reservoirs in the Guantao Formation are buried at a depth of approximately 1600–2000 m, with water temperatures ranging from 48 to 82 °C. The geothermal gradient within the Huanghua Depression varies significantly, generally ranging from 3.27 to 3.57 °C/100 m, indicating strong geothermal activity [28,29,30].

The experimental investigation utilized sandstone reservoir cores retrieved from the KC-01C geothermal well within the Neogene Guantao Formation of the Huanghua Depression, located in the Binhai New Area, Tianjin (its structural location and cross-sectional profile are shown in Figure 1 and Figure 2). This tectonically active region is characterized by well-developed fault systems, with geothermal geological conditions primarily controlled by the Cangdong Fault (western boundary) and Zengfutai Fault (southern boundary).

The core specimen, retrieved from a depth of approximately 1324.5 m in the Guantao Formation, is characterized by grayish-white coloration and composed primarily of fine- to medium-grained sandstone with minor coarse-grained gravel. The grains exhibit moderate sorting and subangular to subrounded morphology, while petrographic observations indicate significant intergranular porosity and weak siliceous cementation, reflecting good reservoir quality. To minimize the influence of sample heterogeneity and ensure representativeness, core samples used in injection experiments were selected from this narrow depth interval based on consistent visual appearance, sedimentary texture, and pore structure. This selection strategy ensured that the samples were typical of the target reservoir facies, providing a basis for evaluating fluid–rock interaction behavior. Cylindrical samples (Φ2.5 cm × L5–8 cm) were prepared for permeability testing, and geothermal fluids were obtained from the same formation used for injection experiments (Figure 3). All cylindrical sandstone core samples (Φ2.5 cm × L5–8 cm) were oven-dried at 60 °C to a constant weight and subsequently saturated with geothermal formation water through low-rate infiltration to simulate reservoir brine saturation and establish stable single-phase flow conditions throughout the injection experiments. Mineralogical composition was quantified via X-ray diffraction (XRD) through Rietveld refinement of diffraction orientations (2θ range: 3–45°) and intensity profiles. X-ray diffraction patterns and quantitative phase analysis are shown in Figure 4 and

Table 1. Mineral content of sandstone cores.

Minerals	Quartz	Potassium Feldspar	Albite	Clay Mineral			Amphibole
				Kaolinite	Montmorillonite	Illite
Mass fraction/%	31.0	28.1	21.5	3.4	10	2.6	3.4

.

The geothermal fluid from Well KC-01C is of the Cl-Na type, with a total mineralization degree of 5159 mg/L. The ion concentrations were determined by inductively coupled plasma atomic emission spectroscopy and ion chromatography (

Table 2. Concentration of major ions in geothermal fluid from Well KC-01C.

Test Items	Concentration (mg/L)	Test Items	Concentration (mg/L)
Na+	1850	Cl−	2662
K+	11.9	HCO3−	153
Ca2+	140	SO42−	362
Mg2+	24.3	NO3−	0.13
Fe2+	0.003	F−	1.4
Ba2+	0.165	Br−	2.69
H2SiO3	36.6	Ammonia nitrogen	1.47

).

2.2. Experimental Instruments

The instrument used in the dynamic core seepage experiment was a high-temperature and high-pressure core flooding apparatus (Origin: Huaxing Petroleum Instrument Co., Ltd., Nantong, China). The device and structure diagram are shown in Figure 5. This instrument can test the permeability of fractured core under a constraint pressure load and obtain the core permeability and other parameters. It is mainly composed of a constant pressure and constant-flow pump, intermediate container, core support, balance device, etc. The remaining experimental equipment includes an oil-free vacuum pump, filter membranes, and a digital thermostatic water bath used, respectively, for vacuum filtration of the injected fluid and precise temperature control to ensure consistent experimental conditions. In order to ensure the integrity of the core structure and avoid the risk of brittle fracture under the coupling effect of high pressure and high temperature for a long time, the short-period, high-precision observation mode is used to replace the traditional continuous loading scheme, breaking through the traditional continuous loading paradigm. Through finite-time, high-density monitoring (sampling interval of 15 s) combined with numerical extrapolation, the dual goals of risk aversion and full cycle behavior prediction are achieved.

The experiment utilized Darcy’s law to calculate the core permeability, which states that the seepage flow rate Q is directly proportional to the difference in upstream and downstream water heads (H₁ − H₂) and the cross-sectional area A perpendicular to the direction of flow and is inversely proportional to the seepage length L. Due to the relatively loose lithology of the sandstone, the core was placed in a rubber sleeve for fixation during the experiment. During the core flow experiment, the entire cross-section of the core was regarded as a flow cross-section, and the pressure and liquid flow rate at the inlet and outlet ends of the core were measured using a high-temperature high-pressure core flooding apparatus to calculate the equivalent permeability k₁ of the rock mass.

Darcy’s law is as follows:

$$Q=KA{\displaystyle \frac{H_1-H_2}{L}}$$

(1)

$$\rho \mathrm{g}H=P$$

(2)

The relationship between permeability k and permeability coefficient K is expressed as:

$$\mathrm{k}=K{\displaystyle \frac{\mu }{\rho \mathrm{g}}}$$

(3)

$${\mathrm{k}}_1={\displaystyle \frac{Q\mu L}{A_1(P_1-P_2)}}$$

(4)

In the formula, Q represents the volumetric flow rate of liquid through the core (m³/s), K denotes the permeability coefficient (m/d); k₁ signifies the equivalent permeability (m²); A₁ stands for the cross-sectional area of the core (m²); H indicates the water head (m); L is the length of the core (m); ρ is the liquid density (kg/m³); g is the acceleration of gravity (m/s²); P₁ represents the inlet high pressure (Pa); P₂ represents the outlet low pressure (Pa); and μ represents the dynamic viscosity coefficient of the liquid (Pa·s) (

Table 3. Temperature-dependent viscosity of water.

Temperature (°C)	20	25	30	35	40	45	50	55	60
Viscosity (Pa·s)	0.001027	0.000917	0.000818	0.000738	0.000669	0.000619	0.000559	0.000519	0.000479

) [31,32].

While local pore-scale heterogeneity or potential edge effects may influence flow distribution, previous studies [33] have shown that such uncertainties typically introduce permeability measurement errors within a range of ±10–15%, which are acceptable for interpreting core-scale injectivity behavior in geothermal systems. In this study, care was taken to ensure full sample saturation and tight core–sleeve sealing, and pressure–flow data were interpreted based on the assumption of effective axial flow through the bulk cross-section.

2.3. Experimental Methods

This experiment focuses on the thermal reservoir core of sandstone from the Guantao Formation in the Huanghua Depression and designs a multi-parameter coupled indoor experimental recharge simulation to systematically analyze the dynamic response mechanism of permeability. The experimental conditions are set as shown in

Table 4. Design of experimental conditions.

Case	Core Number	Diameter/cm	Length/cm	Flow Rate (mL/min)	Reinjection Fluid Temperature (°C)	Suspended Particle Size (μm)	Confining Pressure (MPa)	Reservoir Temperature (°C)
1	C-1	2.54	5.63	1.67–2.50–3.33–4.17–5	25	3	5	59
2	C-2	2.54	6.77		35
3	C-3	2.55	7.06		59
4	C-4	2.48	6.78	3.33–4.17–5	25
5	C-5	2.53	7.13		35
6	C-6	2.45	7.12		59
7	C-7	2.49	7.29	1.67–2.50–…–8.30–9.99–8.30–…–1.67	25
8	C-8	2.50	6.72	9.99–8.30–…–1.67
9	C-9	2.50	7.34	5	20–25–…–50–60	3
10	C-10	2.53	8.28		25	1–3–5–raw water (>5)
11	C-11	2.47	7.16	5 (long duration)	25	3
12	C-12	2.43	7.15	5 (flow reversal)	59	3

:

2.3.1. Advantageous Flow Rate Screening and Initiation Strategy Optimization

Clogging and permeability decline during geothermal reinjection are highly sensitive to injection flow rate [34]. In this study, a preliminary screening of the advantageous flow rate range was first conducted using positive and negative gradient flow rate variation experiments. Based on the screening results, comparative experiments on different flow rate initiation strategies were subsequently carried out to further optimize the injection protocol. The specific experimental design is as follows:

(1)

Flow rate gradient experiments (preliminary screening of advantageous flow rates):

Under isothermal conditions (25 °C), two flow variation modes were employed to observe permeability responses under dynamic flow regimes:

①

Positive gradient flow mode: the injection flow rate was gradually increased from 1.67 mL/min to a peak of 9.99 mL/min, followed by a symmetrical decrease to the initial value. Each step involved a change of 0.83 mL/min, with the system maintained at each level until the inlet pressure stabilized.

②

Negative gradient flow mode: the injection began at 9.99 mL/min and was progressively reduced to 1.67 mL/min using the same stepwise pattern.

By comparing permeability responses and blockage onset points under both modes, the safe and advantageous flow rate range for reinjection at low temperature was preliminarily identified.

(2)

Initiation strategy comparison (refinement of advantageous flow rate):

Within the advantageous flow rate range identified above, further experiments were conducted at three constant temperatures (25 °C, 35 °C, 59 °C) to evaluate two different flow rate initiation strategies:

①

Low-flow initiation: the injection started at 1.67 mL/min, with stepwise increments of 0.83 mL/min after pressure stabilization.

②

High-flow initiation: the injection began at 3.33 mL/min, with subsequent stepwise increases using the same protocol.

2.3.2. Advantageous Temperature Screening

Based on the previously identified advantageous flow rate, a constant injection rate was maintained while gradually increasing the injection temperature from 20 °C to 59 °C in 5 °C increments. Permeability was monitored at each temperature stage to identify temperature-sensitive inflection points and assess the thermal sensitivity of the formation.

The selected temperature gradient was designed with reference to the thermal conditions of Guantao Formation sandstone, which serves as the primary reservoir in the study area. Previous hydrothermal field data and numerical models have shown that the Guantao Formation typically exhibits reservoir temperatures in the range of 50–65 °C, depending on burial depth and geothermal gradient [35]. The upper limit of 59 °C thus approximates the in situ temperature of mid-depth geothermal reservoirs in the Bohai Bay Basin, where the Guantao Formation is widely distributed and extensively developed for space heating. The lower limit of 20 °C represents surface or shallow groundwater temperature, simulating the initial state of reinjected fluids. The incremental increase of 5 °C enables detection of nonlinear permeability responses while ensuring experimental control and thermal equilibrium.

2.3.3. Advantageous Particle Size Screening

Under constant temperature (25 °C) and advantageous flow rate conditions, the injection fluid was pretreated using filter membranes of varying pore sizes (1 μm, 3 μm, and 5 μm), with untreated raw water (>5 μm) used as a control.

The pretreated fluids were injected into the rock core in continuous flow experiments. By analyzing the size distribution and concentration of suspended solids in the effluent, and correlating these results with the evolution of permeability, the most appropriate filter membrane accuracy was determined.

2.3.4. Long-Term Seepage and Reinjection Simulation Under Advantageous Flow Rate

A long-duration dynamic seepage experiment was conducted at the previously identified critical advantageous flow rate and a constant temperature of 25 °C to simulate the long-term operation of geothermal reinjection.

Following this, the core was reversed in the holder, and the injection fluid temperature was raised to the reservoir temperature (59 °C) while maintaining the same flow rate. This stage was designed to simulate the uplift operation and evaluate its effectiveness in restoring permeability by altering the flow direction.

By comparing permeability before and after the experiments, along with changes in fluid chemistry and core structure, the dominant clogging mechanisms and their influence on reinjection efficiency were assessed.

3. Experimental Results and Discussion

3.1. Effect of Injection Flow Rate on Permeability Evolution and Comparison of Initiation Strategies

3.1.1. Positive Gradient Flow Mode:

The results of the gradient flow injection showed that under the starting temperature of 20 °C, the initial injection was performed at a small flow rate (1.66 mL/min), and then the flow rate was gradually increased, the permeability increased significantly at the high-flow-rate stage (≥6.67 mL/min), and the corresponding inlet pressure increased significantly; when the flow rate was increased to the highest (9.99 mL/min), the permeability reached the maximum, and at the same time, the inlet pressure decreased significantly. Then, the inlet pressure decreased rapidly, and the permeability decreased during the subsequent stepwise reduction in the flow rate (Figure 6).

The graded flow injection experiment revealed a nonlinear and asymmetric response of core permeability to changes in injection flow rate. Specifically, during the low-flow-rate stage (1.66–6.67 mL/min), permeability exhibited a slight decrease followed by a gradual increase. In contrast, during the high-flow-rate stage (6.67–9.99 mL/min), permeability increased sharply and then declined upon flow rate reduction. A critical flow velocity threshold (~6.67 mL/min) was identified, beyond which permeability enhancement transitioned into clogging-induced decline, demonstrating a clear hysteresis effect in the permeability–pressure relationship.

Experiments showed that in the low-flow-rate injection stage (1.66–6.67 mL/min), the core penetration process was mainly controlled by adsorption–desorption interactions. At this stage, the fluid scouring effect is limited, and the fluid shear stress is not enough to overcome the van der Waals force and electrostatic force binding of the particles in the pore space, resulting in the restricted migration of the particles [36]; the scouring efficacy of the fluid on the pore channels is low, it is difficult to effectively remove the blockage in the pore space, and the prolongation of the fluid residence time under the low flow rate contributes to the formation of an adsorption equilibrium of solute molecules on the surface of the minerals. The surface adsorption behavior of clay minerals is dominant, resulting in a further reduction in the pore space and permeability. The surface adsorption behavior of clay minerals dominates, resulting in further reduction in pore space and a decrease in permeability, which is consistent with the microporous filling behavior predicted by adsorption potential theory [37,38]. With the increase in flow rate step by step, the fluid scouring ability is enhanced, and the clogged material is gradually taken away from the pore channels, the contact time between the fluid and the surface of the rock is shortened, the adsorption is weakened, the desorption is enhanced, pore space availability is improved, and permeability is therefore shown to be lower than that of the pore. The availability of pore space is improved, the permeability therefore shows a gradual increase, and the inlet pressure is synchronized with a significant increase.

In the high-flow-rate stage (6.67–9.99 mL/min), the permeation process was dominated by the fluid inertia effect and particle rearrangement mechanism, consistent with insights from granular thermodynamic modeling that describe particle restructuring under coupled thermal–hydro–mechanical conditions [39]. When the flow rate reaches a high level, the inertia effect of the fluid is significantly enhanced, and the kinetic energy of the fluid is enough to drive the blocked particles to cross the pore throat barrier, which restores the connectivity of the originally blocked pores. In addition, the permeability shows an exponential growth; however, at this time, due to the increase in flow resistance, a higher pressure is needed to drive the flow of the fluid, and the inlet pressure is relatively large, which triggers the stress-induced deformation of the rock particles at high pressures and changes the contact relationship between particles; furthermore, a certain degree of displacement and rearrangement of particles occurs. Under high pressure, the stress-induced deformation of rock particles is triggered, the contact between particles is changed, the particles are displaced and rearranged to a certain extent. In addition, the pore structure is optimized to enhance the permeability network connectivity, which further improves the permeability.

When the flow rate reached the maximum (9.99 mL/min), the system exhibited turbulent energy dissipation. The development of turbulent vortices reduced the effective hydraulic resistance, causing a notable drop in inlet pressure (ΔP = 0.13 MPa), despite continued permeability increase.

However, during the subsequent flow reduction phase, remobilized particles lost suspension energy and were redeposited within the pores, forming secondary blockages. Meanwhile, stress-induced particle rearrangement partially reversed, degrading pore connectivity and causing a decline in permeability. This irreversible permeability loss reflects the memory effect of the system and confirms the presence of dynamic hysteresis in fluid–rock interactions.

After approximately 6 h of graded flow injection, the mineral composition of the core exhibited significant alterations (

Table 5. Comparison of major mineral compositions between the original sandstone (protolith) and the post-injection core sample (C-7) after long-term reinjection under variable flow conditions.

Mineral Content/%	Quartz	Potassium Feldspar	Allbite	Kaolinite	Montmorillonite	Illite	Amphibole
Protolith	31.0	28.1	21.5	3.4	10	2.6	3.4
C-7	23.1	24.6	31.7	4.0	12.5	4.0	-
Variation	−25%	−12%	+47%	+17%	+25%	+54%	-

). The contents of quartz and potassium feldspar decreased notably, while the contents of sodium feldspar and various clay minerals—such as kaolinite, montmorillonite, and illite—increased markedly. These observations indicate that the mineralogical evolution was governed by a combination of potassium feldspar dissolution–sodium feldsparization, quartz dissolution, and clay mineral transformation, all of which were driven by the coupled effects of fluid chemistry, temperature, and seepage-induced transport.

Under the experimental conditions, the injected fluid was strongly alkaline (pH = 9.2), which substantially enhanced the solubility of both quartz and potassium feldspar. The dissolved SiO₂(aq) and Al³⁺ subsequently recombined to form secondary clay minerals through reprecipitation reactions. Simultaneously, in a high-salinity environment, potassium feldspar underwent cation exchange with Na⁺ in the solution, resulting in the formation of albite (sodium feldspar). The concurrent decrease in potassium feldspar and increase in sodium feldspar content thus suggest a coupled geochemical transformation pathway [40,41,42]. These synchronous mineralogical evolutions are supported by the corresponding changes in effluent ion concentrations (

Table 6. Ion concentrations (mg/L) in the effluent collected during stepwise graded injection under 25 °C conditions.

Test Items	Concentration (mg/L)	Test Items	Concentration (mg/L)
Na+	2030	Cl−	2863
K+	39.1	HCO3−	167
Ca2+	155	SO42−	429
Mg2+	20	NO3−	0.34
Fe2+	0.003	F−	1.51
Ba2+	0.06	Br−	3.68
H2SiO3	20.7	Ammonia nitrogen	1.94

), which reflect active fluid–rock interactions.

The main representative reactions involved in this process are as follows:

SiO₂ + 2OH⁻ → SiO₃²⁻ + H₂O;

2KAlSi₃O₈ + 2H⁺ + H₂O → Al₂Si₂O₅(OH)₄ + 4SiO₂ + 2K⁺;

KAlSi₃O₈ + Na⁺ → NaAlSi₃O₈ + K⁺;

Al³⁺ + 3SiO₂ +2H₂O → Al₂Si₂O₅(OH)₄ (kaolinite);

Al³⁺ + SiO₂ + H₂O → illite/montmorillonite;

These reactions collectively reflect the mineralogical reorganization occurring during alkaline fluid–rock interaction under dynamic seepage conditions. However, they were likely constrained by pore-scale heterogeneity, moderate temperature conditions, and the short experimental timescale and thus did not result in extensive bulk alteration.

3.1.2. Negative Gradient Flow Mode:

The experiment was initiated with the maximum flow rate, and then the flow rate was gradually reduced. The results showed that permeability showed a tendency to stabilize after being gradually reduced. When the flow rate was reduced to 5 mL/min, the permeability gradually stabilized, and when the flow rate was further reduced to 2.5 mL/min, the permeability slightly decreased; the injection pressure peaked at the end of 8.3 mL/min, and with the reduction in the flow rate, the pressure was also reduced step by step (Figure 7).

The experiment reveals a pronounced hysteresis effect between permeability and injection pressure during the process of flow rate reduction. At the initial stage, a high flow rate (>8.3 mL/min) was applied, during which strong fluid turbulence enhanced the scouring capacity of the injected water. This turbulence effectively dislodged fine particles adhered to the core surface, which then migrated deeper into the pore structure, leading to blockage of pore throats and the formation of a compact filter cake. This accumulation increased flow resistance and significantly elevated the injection pressure.

As the flow rate decreased, the flow regime transitioned from turbulent to laminar or transitional flow. The boundary layer effect became more pronounced, and fluid motion near the pore walls exhibited slip-flow characteristics, effectively reducing the available flow area. Meanwhile, the reduced kinetic energy of the fluid was insufficient to maintain particle suspension, causing sedimentation to dominate. Since the previously formed filter cake was not fully removed, permeability continued to decline.

The results also suggest that under high-flow conditions, fluid shear stresses may partially compress the pore structure, reducing pore radius. Upon decreasing the flow rate, limited elastic recovery of the compressed pores contributes to an irreversible loss of permeability.

3.1.3. Flow Rate Initialization Strategy and Temperature-Coupled Permeability Response

• Low-flow-rate start:

Comparison of the low-flow-rate start-up injection experiments conducted at different injection temperatures (

Table 7. Permeability evolution in Guantao Formation sandstone cores under low-flow initiation at different injection temperatures (25 °C, 35 °C, and 59 °C).

Temperature (°C)	25 (C-1)	35 (C-2)	59 (C-3)
Initial permeability (mD)	5.38	6.19	8.95
Final permeability (mD)	9.45	6.07	8.47
Average permeability (mD)	8.31	6.40	8.87

, Figure 8) reveals that permeability evolution exhibits a relatively smooth trend across all three temperature conditions. Notably, at 25 °C, permeability increases significantly—by approximately 75% after injection. In contrast, at 35 °C and 59 °C, only slight decreases in permeability are observed—approximately 2% and 5%, respectively—with no notable fluctuations during the injection process.

This behavior suggests that under low initial flow conditions, the driving pressure gradient is insufficient to activate finer pore networks, allowing only larger pores and coarser channels to contribute to flow. As a result, permeability in this stage is primarily governed by the macropore structure. In the low temperature range (25–35 °C), the thermal expansion of matrix particles may induce slight pore compression, offsetting any enhancement in permeability from increased molecular thermal motion. Although 59 °C approaches reservoir conditions, the thermal slip effect remains limited at low flow rates and does not significantly improve fluid mobility within micropores. Consequently, permeability shows minimal variation across temperatures under low-flow conditions.

• High-flow-rate start

The results of high-flow rate start-up injection experiments under varying temperature conditions reveal significant differences in permeability and injection pressure responses. At 25 °C, as the flow rate is increased stepwise, real-time permeability exhibits a gradual decline, accompanied by a progressive increase in injection pressure, until both stabilize. A similar trend is observed at 35 °C. However, at 59 °C, permeability initially decreases at the first flow stage but then increases at the second stage, where injection pressure also peaks. Upon further increasing the flow rate to the third stage, permeability rises significantly while injection pressure decreases sharply. This inverse relationship between the fitted permeability and pressure curves (Figure 9) suggests that fluid–rock interactions are strongly temperature-dependent, particularly at elevated temperatures.

A comprehensive analysis indicates that at lower temperatures (25 °C and 35 °C), thermal expansion of matrix particles is the dominant influence. Although higher flow velocity can activate finer pore throats, the thermal expansion induced by rising temperature compresses the pore structure, reducing effective flow channels. Moreover, the yield stress of the fluid remains high at low temperatures, and the boundary layer thickness increases with decreasing temperature. These factors, combined with enhanced viscous resistance at the fluid–solid interface, inhibit permeability improvement. Additionally, although high initial flow may dislodge some surface-adsorbed particles, the shear force is insufficient to fully mobilize them, leading to localized disturbance, migration, and accumulation—ultimately forming dynamic clogging and significantly reducing permeability.

At 59 °C, the initial injection stage is marked by strong fluid-induced shear forces that dislodge loosely bound rock particles, causing their migration and accumulation in pore throats. This leads to rapid localized clogging, reflected in a sharp permeability decline and a corresponding rise in injection pressure. The peak pressure likely indicates the maximum extent of clogging and reflects early-stage fluid–particle interaction. In subsequent stages, elevated temperature reduces fluid viscosity and surface tension, alters flow resistance, and induces thermal expansion in the rock framework, promoting adjustments in the pore structure. These combined effects enhance the displacement of previously retained particles, allowing partial recovery and stabilization of permeability, along with a gradual decline in injection pressure.

3.1.4. Comparison of Flow Rate and Initiation Strategies on Permeability

Flow sensitivity experiments revealed that when the injection rate was maintained below 6 mL/min, the system exhibited relatively stable pressure responses and permeability behavior, indicating that pore structure integrity and fluid transmission capacity were effectively preserved. In contrast, during the initial high-flow stage (>8.3 mL/min), strong shear forces and turbulence facilitated particle detachment and deep migration, leading to localized clogging and sustained permeability decline. To mitigate flow path restriction and irreversible permeability degradation caused by filter cake formation, it is recommended that the injection flow rate not exceed 6 mL/min in practical applications. This threshold is supported by the results of the gradient flow rate tests.

However, the effectiveness of shear-driven mechanisms strongly depends on the looseness of the particle structure and whether the applied shear stress is sufficient to overcome interparticle cohesion. In dense formations or under low-flow conditions, such mechanisms may be limited or entirely ineffective. Moreover, the positive-gradient injection experiment demonstrated that when the flow rate exceeded 6 mL/min, a sharp increase in injection pressure was observed, accompanied by irreversible permeability loss—suggesting permanent damage to the pore structure. Therefore, shear-enhanced injection strategies should be carefully evaluated based on site-specific reservoir conditions to avoid structural degradation.

Regarding injection startup strategies, while high-flow initiation may aid in permeability recovery under elevated temperatures, it is associated with higher operational risks, including early-stage structural damage and system instability. In contrast, low-flow startup yields more controllable and stable permeability responses across various temperature regimes, thereby preserving pore structure integrity and enabling gradual activation of flow pathways. In summary, it is recommended that geothermal reinjection systems adopt a low initial injection rate, followed by a gradual increase to an optimal range (5–6 mL/min). The injection temperature should be adjusted in accordance with subsequent operational or experimental conditions to ensure sustained injectivity and long-term reservoir performance.

3.2. Impact of Injection Temperature on Permeability Variation

3.2.1. Permeability Response to Temperature Changes

The results of the constant-rate variable-temperature injection experiments show that the apparent permeability of the core generally exhibits a trend of gradual decline followed by stabilization as the injection temperature increases. Specifically, permeability decreases significantly between 20 °C and 30 °C and then remains relatively stable from 30 °C to 35 °C. As the temperature continues to rise, permeability decreases slightly, followed by a modest recovery at the end of the high-temperature stage. Meanwhile, the injection pressure shows a steady increase, with a slight reduction at the final stage (Figure 10).

3.2.2. Mechanisms of Temperature Effects on Permeability

At the low-temperature stage (20–25 °C), the high viscosity of the fluid limits its flow capacity within the porous medium, resulting in increased resistance and a significant rise in injection pressure. This leads to a notable decline in apparent permeability (−44.42%). Moreover, at lower temperatures, the adsorption between water molecules and rock surfaces becomes stronger, which may reduce the effective pore volume and promote the release and transport of fine particles. These processes contribute to partial pore clogging. In addition, the lower reaction kinetics at this stage hinder the redissolution of previously precipitated minerals, exacerbating pore blockage and flow resistance.

As the injection temperature increased to the intermediate range (30–40 °C), the viscosity of the water gradually decreased to a relatively stable level, enhancing the fluid’s mobility within the pore medium. However, experimental results show that the effective permeability exhibited a slight increase (+15.7%) during this stage and then stabilized. This suggests that the pore structure remained in a dynamic state of evolution, where effects such as particle migration and retention, as well as colloidal clogging, persisted and were only partially alleviated by the temperature increase. Additionally, the elevated temperature may have intensified the hydration swelling of clay minerals (e.g., montmorillonite), further compressing the pore space.

With continued heating, thermal perturbation triggered the rearrangement of rock framework particles and subtle modifications of the pore structure, leading to a slight decline in permeability. The temperature was within the range of 40 °C to 50 °C, and the permeability decreased by 18%. Similarly, water–rock reactions are accelerated at elevated temperatures, especially in silicate and clay-rich systems, potentially resulting in the precipitation of secondary minerals or the swelling of clays such as montmorillonite, both of which reduce permeability [43]. In the final stage, a modest increase in permeability was observed, accompanied by a reduction in injection pressure, possibly associated with the redissolution of colloidal materials or precipitates and partial reopening of pore throats. However, the contribution of this process to overall permeability enhancement was limited, indicating that under constant-flow-rate conditions, permeability evolution is primarily governed by changes in pore structure and increased pressure drop rather than by improvements in fluid physical properties alone.

Therefore, based on the experimental observations, 35 °C emerges as a relatively optimal injection temperature for geothermal reinjection under the tested conditions. At this intermediate temperature, the fluid viscosity is sufficiently reduced to enhance flow capacity, while the hydration swelling and secondary mineral precipitation effects remain relatively moderate compared to higher-temperature stages. Although the permeability did not exhibit a substantial rebound, it still showed a moderate increase (+15.7%), indicating that the pore structure underwent a degree of self-adjustment and that particle migration reached a relatively stable state. In contrast, both lower temperatures (e.g., 25 °C) and higher temperatures (e.g., 59 °C) led to pronounced permeability losses, likely due to intensified clogging or thermally induced structural disturbances. Therefore, injecting geothermal fluids at approximately 35 °C is expected to achieve an optimal balance among thermal efficiency, injection stability, and reservoir integrity.

3.3. Effect of Suspended Particle Size on Clogging Behavior and Permeability Response

3.3.1. Impact of Particle Size on Clogging and Permeability

Under a constant flow rate (5 mL/min) and temperature (25 °C), permeability tests conducted with suspensions of varying particle sizes revealed a non-monotonic response. As the particle size increased from 1 μm to 5 μm, the core permeability exhibited a gradual decline. However, upon injecting unfiltered geothermal water containing particles larger than 5 μm, permeability increased markedly, accompanied by a significant rise in injection pressure. When the fluid was subsequently switched to a finer particle suspension (<3 μm), permeability partially recovered, though the pressure drop continued to rise. Re-injection of the 1 μm suspension caused another drop in permeability along with a slight decrease in pressure, ultimately resulting in a clearly nonlinear and fluctuating permeability profile (Figure 11).

3.3.2. Mechanisms of Particle Size Effects on Clogging and Permeability

Such permeability evolution reflects the complexity of particle size-dependent mechanisms governing flow dynamics during geothermal reinjection. Fine particles (1–3 μm), due to their dimensional match with pore throats, tend to migrate through the pore network and induce micro-scale clogging. However, these particles are also more susceptible to shear-induced resuspension, which can locally clear blockages and reopen flow paths, producing alternating cycles of clogging and unclogging that manifest as permeability fluctuations. When particle size increases to 5 μm, bridging, retention, and aggregation at pore constrictions become dominant, leading to more stable clogging structures and a significant reduction in permeability. The concurrent drop in effluent particle concentration further indicates increased particle retention within the system (

Table 8. Suspended solids’ concentration in the water produced after injection of fluids filtered by membranes with different particle size cutoffs (1 μm, 3 μm, and 5 μm), as well as by unfiltered raw water (>5 μm). The results indicate that smaller filter sizes do not necessarily correspond to lower suspended solid content in the effluent, suggesting complex particle migration and mobilization behaviors during porous media flow.

Suspended Particle Size (μm)	1	3	5	Raw Water (>5 μm)
Suspended solids’ concentration (mg/L)	16	24.3	7.75	20

).

Injection of raw geothermal water containing large, suspended particles results in a mechanical dislodging effect under elevated injection pressure, potentially breaking pre-existing clogging structures and expanding preferential flow paths. This explains the observed permeability increase despite a sharp rise in pressure drop. Notably, when fine particles were reintroduced, the interplay between residual clogging structures, cumulative particle deposition, and increased fluid viscosity caused only partial recovery of permeability, while injection pressure continued to climb, indicating increased overall flow resistance. In the final injection phase with 1 μm particles, further permeability reduction occurred, likely due to secondary deposition on existing blockages. The accompanying pressure decrease may be attributed to localized shear disturbance that restructured clogging configurations or altered the system’s flow regime.

These results suggest that in multi-sized particulate systems, suspended solids undergo cyclic accumulation and release within the pore structure, driven by the coupled effects of mechanical behavior, surface adsorption, colloidal aggregation, and hydrodynamic perturbation. The resulting permeability evolution is nonlinear and non-monotonic. The dimensional compatibility between suspended particles and pore throat size is a fundamental factor governing permeability decline during geothermal reinjection, especially due to the preferential clogging of dominant flow paths. Experimental results demonstrate that particles larger than 3 μm have a significantly higher tendency to bridge and accumulate at pore constrictions, leading to persistent blockage. In contrast, finer particles (<3 μm) are more likely to remain mobile or be resuspended under moderate shear, resulting in less severe and more reversible clogging behavior. To mitigate the risk of permeability degradation caused by particulate retention, it is recommended that reinjection fluids undergo pretreatment to limit the particle size of suspended solids to below 3 μm. Such filtration measures can effectively reduce the likelihood of clogging in critical flow pathways and help ensure the long-term sustainability of reservoir injectivity.

3.4. Permeability Evolution During Long-Term Reinjection and Reverse Injection Under Advantageous Flow Rate Conditions

The results of the prolonged advantageous flow rate seepage experiments demonstrate that the core permeability declined over time to varying degrees. The rate of decline gradually decreased and eventually stabilized. Following the initial observation of slight permeability reduction, a reverse injection operation was conducted. This effectively removed potential particulate blockages, improved pore connectivity, and significantly enhanced permeability performance (Figure 12).

The early stage of the experiment shows a decrease in permeability, which is mainly attributed to the slow dissolution reaction of potassium feldspar (KAlSi₃O₈), which releases K⁺ and SiO₄²⁻ into the fluid system.

Table 9. Mineralogical composition of sandstone cores (C-11 and C-12) after long-term injection and reversed reinjection compared with the original reservoir protolith. C-11 represents the post-injection core under forward flow, while C-12 corresponds to the reversed core subjected to high-temperature reinjection.

Mineral Content/%	Quartz	Potassium Feldspar	Allbite	Kaolinite	Montmorillonite	Illite	Amphibole
Protolith	31.0	28.1	21.5	3.4	10	2.6	3.4
C-11	36.4	13.9	25.0	2.6	13.1	6.7	2.2
Variation	+17%	−50%	+16%	−23%	+31%	+157%	−35%
C-12	25.3	29.6	24.5	3.4	11.7	3.3	2.1
Variation	−18%	+5%	+14%	-	+17%	+27%	−38%

(C-11) shows a 50% reduction in potassium feldspar content, indicating substantial dissolution that altered the original pore structure. Concurrently, montmorillonite and illite contents increased by 31% and 157%, respectively, suggesting that Al³⁺ released from dissolution was adsorbed by clay minerals, promoting their swelling and colloidal behavior. This led to the formation of aggregates that blocked pore throats and further reduced permeability. The 17% increase in quartz content may be related to deposition of colloidal material enriched in quartz or fragmentation and rearrangement of existing quartz particles under shear flow.

In the late stage of the experiment, the permeability recovered, and with the increase in ambient pH (9.2→9.52), the Al(OH)₃ colloid or clay mineral precipitation formed in the early stage partially redissolved, and the pore connectivity was partially restored. The system finally entered the stabilization stage in which the dissolution–precipitation rates were balanced with each other, and permeability tends to progress to a steady state with time.

To further investigate the effect of backflushing on the reversibility of formation blockage, reverse injection tests were conducted by reversing the core direction and restarting injection when a decline in permeability was observed during long-term injection to simulate the common blockage phenomenon encountered in geothermal reinjection processes. Reverse injection significantly promoted the recovery of permeability (Δk = +98%), and the concentration of suspended particulate matter in the effluent was measured to be 17.5 mg/L, indicating that the shear stress generated by reverse flow effectively loosened the accumulated Al(OH)₃ colloids and clay aggregates in the pore throats, thereby alleviating the extent of formation blockage.

Mineralogical analyses support these findings (Table 9: C-12). A 18% decrease in quartz content was observed after reverse injection, likely due to colloid detachment from particle surfaces or particle fragmentation. Potassium feldspar increased by 5%, while sodium feldspar increased by 14%, suggesting that albite formation through K-feldspar alteration may have continued locally, but the overall geochemical process was not uniformly progressive. The slight enrichment of potassium feldspar may indicate secondary reprecipitation or reflect spatial heterogeneity in feldspar dissolution–precipitation dynamics under reversed flow. Compared with continuous unidirectional seepage, the reverse flow likely redistributed reactive zones and reduced the effective residence time of the fluid, which inhibited complete feldspar transformation. Meanwhile, the increases in montmorillonite (17%) and illite (27%) were notably lower than those observed under single-direction flow, implying that shear-induced disturbance and reduced fluid–rock equilibrium limited clay mineral growth and structural reorganization.

Combined with dynamic seepage tests and pore structure characterization, the influence of prolonged fluid–rock interaction and flow reversal on the pore system was assessed. As shown in

Table 10. Changes in specific surface area of rock for lengthy dynamic reactions.

Core Number		BAT (m2/g) Specific Surface	BJH Cumulative Specific Surface (m2/g)	BJH Total Pore Volume (cm3/g)	Average Pore Diameter (nm)
Protolith	-	15.277	18.355	0.032	4.081
Long-term response of critical flow velocity	C-11	14.644	17.247	0.027	4.119
Magnitude of change		−4.1%	−6.0%	−15.6%	+0.9%
Flow reversal	C-12	11.392	13.264	0.02	4.092
Magnitude of change		−25%	−27.7%	−37.5%	+0.2%

, the BET specific surface area decreased from the original rock value (15.275 m²/g) to 14.644 m²/g (−4.1%) after extended reaction and further to 11.392 m²/g (−25.4%) following the uplift operation. This reduction is associated with secondary mineralization processes such as micropore closure from interlayer water in clay minerals (e.g., montmorillonite) [44], colloid coverage, particle migration [45] and retention, and mineral phase recombination.

A concurrent decrease in total pore volume (0.032 cm³/g → 0.020 cm³/g, −37.5%) reflects both compaction from diagenesis and dynamic effects of particle transport. The expansion of clay minerals and dissolution–reprecipitation of silicate/carbonate cements narrowed pore throats and reduced connectivity. Under low temperature and high salinity, hydration-induced expansion of montmorillonite compressed pore spaces, while particles migrated and lodged at pore throats, causing localized blockage. Although shear force during flow reversal partially cleared blockages, it also induced grain rearrangement and pore collapse, further reducing effective porosity. The average pore diameter remained relatively stable (4.081 nm → 4.092 nm), indicating that smaller pores were more susceptible to clogging, whereas larger pores remained open due to fluid inertia—consistent with the bimodal distribution of pore sizes (Figure 13).

These findings reveal a nonlinear response in permeability evolution, where mineral dissolution–precipitation processes are intertwined with physical plugging and jointly influenced by flow direction and shear perturbation.

4. Conclusions

4.1. Key Findings and Engineering Implications

Through a series of laboratory experiments and comprehensive analysis of permeability evolution under varying flow rates, temperatures, and particle sizes during geothermal reinjection, the following conclusions can be drawn:

(1)

Under varying flow conditions, the experiments revealed that when the injection rate is kept below 6 mL/min, the permeability remains relatively stable, and the pore structure is better preserved. In addition, an injection fluid temperature around 35 °C and a suspended particle size below 3 μm effectively mitigated clay swelling and secondary mineral precipitation while reducing the risk of pore throat bridging. These conditions significantly improved permeability and overall system stability. Therefore, for practical geothermal reinjection operations, it is recommended to adopt a combined strategy of low flow rate (≤5 mL/min) + moderate temperature (≈35 °C) + fine particle filtration (≤3 μm) to achieve stable injectivity responses, controllable system risks, and improved long-term reinjection performance.

(2)

During long-term injection, permeability exhibited a rapid initial decline followed by stabilization. Reverse injection notably enhanced permeability, with a maximum recovery of up to 98%, indicating its effectiveness in removing pore blockages and restoring connectivity. In practice, periodic reverse flushing is recommended when permeability loss exceeds 30%, injection pressure continuously rises, or severe particle clogging is observed. However, it should be avoided in formations with weak structural integrity or significant mineral precipitation, as frequent flow reversal may trigger particle rearrangement and structural instability.

4.2. Limitations and Outlook

Although this study provides insights into the mechanisms of permeability variation and optimization strategies during geothermal reinjection, several limitations remain. First, the experimental duration was relatively short, limiting its ability to fully capture long-term mineralogical evolution. Second, while core-scale experiments offer valuable representation, they cannot entirely replicate the geological heterogeneity of real reservoirs. Lastly, the geochemical processes were not dynamically monitored, and some experimental groups lacked repetition, which may affect the statistical robustness of the conclusions. Future work should integrate multi-scale numerical simulations, dynamic monitoring of fluid–rock reactions, and long-term field reinjection data to further enhance the understanding and precision of injectivity prediction and management strategies.