Mechanical Properties and Gas Permeability Characteristics of Airtight Concrete for Sealing Layer in Compressed Air Energy Storage Underground Caverns

Abstract

Compressed air energy storage (CAES) is a promising large-scale energy storage technology, but its efficiency and safety depend critically on the sealing performance of underground caverns. From a building materials science perspective, a key challenge is to develop cement-based sealing materials that maintain low gas permeability and mechanical stability under coupled thermo-pressure cycles. However, the gas permeability evolution and microstructural mechanisms of sulphoaluminate cement-based airtight concrete under such conditions remain insufficiently understood. To address this problem, this study developed an airtight concrete for CAES cavern sealing layers, based on sulphoaluminate cement with fly ash, slag, basalt fiber, and chemical admixtures. Mechanical properties, gas permeability, and microscopic mechanisms were evaluated through uniaxial compression and tensile tests, gas permeability tests under different thermal-pressure cycles and pore pressure conditions, and SEM/DEM analysis. The material exhibits a compressive strength of 62.59 MPa and good airtightness. Under 60 °C cycling, Permeability remained on the order of 10⁻¹⁸ m², following the Klinkenberg gas slippage model. Microscopic analysis suggested a dense structure with limited visible pore connectivity. COMSOL Multiphysics preliminarily indicated that, under assumed design conditions (e.g., the Three Gorges Ulanqab project), the sealing layer remained in triaxial compression with an estimated gas leakage rate below 1%, demonstrating satisfactory sealing performance and structural stability within the investigated operational period. This study provides a theoretical basis and experimental support for sealing material design in CAES caverns.

1. Introduction

As traditional mineral energy reserves deplete, renewable energy is destined to become the dominant pillar of future energy structures. This trend is irreversible. However, the intermittency and volatility of renewable energy sources affect their integration into power grids [1]. Consequently, power systems face an increasingly urgent demand for large-scale, long-duration energy storage technologies [2]. Compressed Air Energy Storage (CAES), offering advantages such as large storage capacity [3,4,5,6], relatively low construction costs [7], and environmental friendliness [8], has emerged as one of the most promising large-scale energy storage technologies. Within a CAES system, the underground gas storage cavern is a core component, and its operational sealing performance [9,10] directly impacts the system’s operational efficiency, safety, and stability.

Currently, the main types of underground CAES storage caverns [11] include salt caverns, hard rock caverns, porous media aquifers, and abandoned mines and tunnels. Compared to other storage types, the greatest advantage of hard rock caverns [12] lies in the wide variety and distribution of hard rock formations meeting construction requirements, making site selection relatively easy. To reduce construction costs, engineered linings and flexible sealing layers can be added based on natural hard rock caverns or abandoned hard rock mines. Under high-pressure underground gas storage conditions, the cavern lining structure must not only withstand periodically varying thermo-pressure loads [13] but also possess extremely low gas permeability [14,15] to prevent compressed air leakage, thereby ensuring the economic viability and environmental compatibility of the energy storage system. Therefore, developing high-performance sealing materials for CAES underground caverns and systematically investigating their mechanical and permeability characteristics under complex coupled thermal–mechanical–fluid environments [16,17] holds significant engineering application value and scientific importance.

The sealing structures of gas storage caverns mainly consist of steel liner sealing layers and polymer sealing layers [18]. Currently, the composite structure of steel liner and reinforced concrete has become the most common sealing solution. However, steel liners are prone to bending deformation, leading to structural failure. If surrounding rock conditions are poor, the thickness of the steel liner must be increased, dramatically raising sealing costs [19]. In contrast, polymer materials exhibit good deformability, better buffering deformations caused by drastic changes in internal air pressure and temperature, thereby significantly reducing cyclic thermal-pressure stresses transmitted to the concrete lining and surrounding rock. In this context, polymer materials such as rubber and polyethylene have attracted widespread attention. Natural rubber possesses excellent flexibility, but contains a large number of unsaturated double bonds in its structure, causing it to gradually age under oxygen and high-temperature conditions [20], losing its performance and shortening the cavern’s service life. High-density polyethylene (HDPE) features low cost and easy processability but tends to soften at high temperatures, accompanied by a sharp increase in gas permeability [21]. Therefore, developing a sealing layer material with low permeability, stable properties, and relatively low cost has become a key focus of current research. Microstructural analysis has confirmed that sulfoaluminate cement [22,23], through the formation of AFT crystals, fills micro-cracks and refines the pore structure, reducing harmful porosity (pore radius > 1 μm) by 14.8% [24]. Although sulfoaluminate cement slightly affects workability, overall results indicate it helps mitigate degradation caused by temperature variations. Furthermore, under sub-zero curing conditions, the bond strength of sulfoaluminate cement concrete specimens is higher than that of ordinary Portland cement concrete [25]. The primary reason is that sulfoaluminate cement hydrates rapidly, forming a dense structural network that enables the concrete to quickly develop a compact microstructure even at below-freezing temperatures.

To address these gaps, this study adopts airtight concrete as the sealing layer material, employing a composite cementitious system composed of sulfoaluminate cement, fly ash [26,27,28,29], slag [30,31,32,33], and basalt fibers [34,35,36,37]. Despite the potential of sulfoaluminate cement-based materials, several research gaps remain. First, the gas permeability evolution of airtight concrete under coupled thermo-pressure cycling and varying pore pressures has not been systematically quantified. Second, the microscopic mechanisms by which hydration products and pore structure control airtightness are not fully understood. Third, the sealing performance of such materials at the cavern scale under CAES operating conditions has rarely been evaluated. These gaps define the scientific problem of this study: to understand the coupled thermo-hydro-mechanical [38] effects on the gas permeability and microstructural evolution of a sulfoaluminate cement-based composite system. The applied problem is to provide a low-permeability, thermally stable, and cost-effective sealing material for CAES underground caverns. The overall objective of this study is to develop and evaluate an airtight concrete sealing layer material for CAES underground gas storage caverns. The specific objectives are: (1) to design and characterize the mechanical properties of the material; (2) to quantify its gas permeability under different thermal-pressure cycles and pore pressures; (3) to reveal the microstructural mechanisms governing airtightness using SEM/EDS; and (4) to preliminarily assess the sealing performance at the cavern scale through coupled numerical simulation [39]. The results are expected to provide a theoretical basis and experimental support for the design and optimization of sealing layers in CAES underground gas storage caverns.

2. Materials and Methods

2.1. Materials Used

The sealing layer material used in this experiment is gas-tight concrete, prepared by mixing sulfoaluminate cement, fly ash, slag, basalt fibers, quartz sand, water reducer, retarder, and tap water. The sulfoaluminate cement is a rapid-setting and high-early-strength cement, sourced from Zibo Yunhe Cement Co., Ltd., Zibo, China. The fly ash was obtained from Shijiazhuang Shang’an Power Plant, Shijiazhuang, China [40,41,42]. The slag is S105 slag powder, supplied by Gongyi Longze Water Purification Materials Co., Ltd., Gongyi, China. The gradation curves of these three materials are shown in Figure 1, and their chemical compositions are presented in Table 1. The basalt fibers, sourced from Shanghai Chenqi Chemical Technology Co., Ltd., Shanghai, China, have a density of 2.69 g/cm³, a tensile strength of ≥2000 MPa, and a tensile elastic modulus of ≥85 GPa. The quartz sand is produced from Nanzhao Dingcheng Calcium Industry Co., Ltd., Nanyang, China with silica (SiO₂) as its main component (99.6% content) and a specific gravity of 2.66 g/cm³. In this study, a polycarboxylate superplasticizer supplied by Sichuan Dongrun Baisheng New Materials Co., Ltd., Chengdu, China, and a special retarder for specialty cements were selected, with the retarder supplied by Shandong Fuzhu Building Materials Co., Ltd., Zhucheng, China. Table 1 presents the main chemical compositions of the sulfoaluminate cement, fly ash, and slag. The mix proportion of the airtight concrete was optimized using the Response Surface Methodology (RSM). The water-to-binder ratio, the content of supplementary cementitious materials (fly ash and slag), and the basalt fiber content were selected as the independent variables, while compressive and tensile strength, and gas permeability were set as the response targets. Based on the RSM optimization results, the optimal combination of these parameters was determined, and the corresponding mix proportion is presented in Table 2 [43,44]. This approach ensures that the designed mix achieves a balanced performance in terms of mechanical strength, airtightness, and constructability.

Figure 1. Particle size distribution curves of sulphate aluminate cement, fly ash and slag.

Table 1. The main chemical components of sulphate aluminate cement, fly ash and slag [45].

Material	Mass Fraction/%
	CaO	SiO2	Al2O3	Fe2O3	MgO	TiO2	SO3	P2O5
Sulfoaluminate cement	45.3	7.23	18.6	4.3	1.35	0.87	12.5	-
Fly ash	5.7	43	24	2.5	0.93	-	0.8	-
Slag	37.5	36.6	17.7	1.6	5.3	-	1.3	-

Table 2. Sealing layer proportions [45].

Specimen	Cement (kg/m3)	Fly Ash (kg/m3)	Slag (kg/m3)	Basalt Fiber (kg/m3)	Quartz Sand (kg/m3)	Water Reducer (kg/m3)	Retarder (kg/m3)	Water (kg/m3)
Sealing layer	473.4	80	40	5.1	900	3	3.6	180

The preparation process for the gas-tight concrete specimens includes specimen preparation and curing, as described below: weigh dry powders, fibers, water reducer and water according to the mix proportion. Mix dry powders until uniform; manually separate basalt fiber clusters, sprinkle evenly into dry powders, and mix until no fiber clumps remain; mix water reducer with water, add to dry powders, mechanically stir for 2 min, scrape walls, and stir again until uniform; load the mixture into molds in two layers, vibrate each layer for 60 s, then vibrate until slurry bleeds and no bubbles escape; place molds in a standard curing box (20 ± 2 °C, ≥95% RH) for 24 ± 2 h; carefully demold, continue curing for 14 days, then cut specimens to a height of 100 ± 0.2 mm with a flat surface. The process is illustrated in Figure 2.

Figure 2. Experimental Flowchart.

2.2. Test Instruments and Test Methods

2.2.1. Uniaxial Compression Test

To obtain stable and reliable full stress–strain curves and peak strength, this study employed an electro-hydraulic servo loading method for the uniaxial compression tests. Cylindrical specimens with a diameter of 50 mm and a height of 100 mm were used. The uniaxial compressive strength testing [46,47,48] was conducted using the RMT-150B rock mechanics testing system (Institute of Rock and Soil Mechanics, Chinese Academy of Sciences, Wuhan, China), as shown in Figure 3. This system is an electro-hydraulic servo loading platform capable of performing conventional mechanical tests such as compression on rock or concrete materials. The equipment has an axial loading capacity of up to 1000 kN and can be equipped with a confining pressure module for triaxial tests. Under uniaxial compression conditions, the RMT-150B typically uses displacement control to achieve stable loading. Axial deformation measurements were taken using displacement sensors (Shanghai Guangyang Ouke Electronic Technology Development Co., Ltd., Shanghai, China), and displacement loading was performed at a rate of 0.002 mm/s. Stress–strain data were collected in real time by a computer.

Figure 3. RMT-150B rock mechanics testing system.

2.2.2. Uniaxial Tensile Test

The tensile strength test was conducted using a WAW-series electro-hydraulic servo universal material testing machine (Zhejiang Lixian Testing Instrument Co., Ltd., Shaoxing, China). The test instrument and specimen molds are shown in Figure 4. This equipment enables stable displacement-controlled loading and real-time acquisition of load and deformation data. Dumbbell-shaped specimens were used for testing, with uniform dimensions of 450 mm × 100 mm × 25 mm. The central uniform cross-section had a length of 150 mm and a width of 60 mm. During the test, the tensile load was obtained using a high-precision force sensor (Shenzhen Xinjingcheng Sensing Technology Co., Ltd., Shenzhen, China), and specimen deformation was measured using a displacement sensor (Shanghai Guangyang Ouke Electronic Technology Development Co., Ltd., Shanghai, China) to obtain a complete stress–strain curve. The loading method adopted displacement control, with a loading rate controlled at 0.5 mm/min. Throughout the test, parameters such as load, displacement, and time were recorded in real time by a computer system, thereby capturing the full-response process of the material under uniaxial tension.

Figure 4. Electro-hydraulic servo universal material testing machine and specimen mold.

2.2.3. Permeability Test

The gas permeability test of the material in this study was conducted using a fully automatic geotechnical triaxial servo apparatus manufactured by TOP INDUSTRIE, Vaux-le-Pénil, France. This apparatus consists of different components, including an automatic control system, data acquisition system, pressurization system, pressure stabilization system, as well as servo devices for confining pressure, axial pressure, and water pressure. The test equipment is shown in Figure 5. It is capable of performing conventional compression tests, water or gas permeability tests, various rheological tests, and ultrasonic tests. The pressure chamber has a maximum confining pressure of 150 MPa, a maximum pore pressure of 60 MPa, and an axial pressure capacity of up to 1000 MPa. It enables stable loading and operational pressure stabilization control under multi-field coupling conditions. In this study, cylindrical specimens with a diameter of 50 mm and a height of 100 mm were used. The quasi-static method was used for testing. The basic procedure is as follows: the upper and lower ends of the specimen are connected to the atmosphere and the pressure chamber, respectively. The gas pressure is supplied and regulated by the pressure chamber. After a stable seepage state is formed inside the specimen, the readings of a high-precision pressure gauge are recorded using CCS30 software (version 1.1, KELLER Druckmesstechnik AG, Winterthur, Switzerland), and the gas permeability is further calculated.

Figure 5. Fully automatic geotechnical triaxial servo apparatus.

Figure 6 shows the gas permeability testing system. The control panel used for the gas permeability test is shown in Figure 6a. One side of the panel is connected to a high-pressure argon gas cylinder. The gas is reduced in pressure by a valve before entering a small gas storage tank, allowing for safer adjustment and control of pressure. The other side is connected to the air inlet pipe of the triaxial testing apparatus pressure chamber. A high-precision pressure gauge with an accuracy of up to 10⁻³ bar (see Figure 6b) is installed in between, allowing the permeability pressure to be adjusted according to the test requirements. The triaxial testing apparatus is equipped with two valves: VM06 is used to control the gas inlet channel, and VM09 is used to control the gas outlet channel. During the permeability test, VM06 needs to be closed and VM09 opened.

Figure 6. Gas permeability testing system.

(1)

Experimental Scheme for Thermo-Pressure Cyclic Permeability Testing

To investigate the effect of temperature cycling on the permeability characteristics [49,50] of gas-tight concrete, the specimens were subjected to temperature cycling treatment at 60 °C. Gas permeability was measured using a fully automatic geotechnical triaxial servo apparatus under conditions of a confining pressure of 10 MPa and a pore pressure of 7.5 MPa [51]. First, the gas permeability of the specimens under constant temperature conditions at room temperature was tested, serving as the gas permeability result for 0 cycles. Subsequently, the specimens were subjected to thermo-pressure cyclic loading treatment, and a gas permeability test was conducted after each completed temperature cycle. By comparing the changes in permeability under different numbers of cycles, the influence pattern of thermo-pressure cycling on the concrete’s permeability performance can be analyzed. A complete stress-temperature cycle process is shown in Figure 7. After each completed stress-temperature cycle, a gas permeability test was conducted to obtain the gas permeability of the material under different numbers of cycles.

Figure 7. A complete stress-temperature cycle process.

Based on the operation conditions of compressed air energy storage caverns and the available experimental equipment, 10 cycles of stress-temperature cyclic tests were conducted. By analyzing the test results, the evolution law of internal structural damage of the sealing material under thermo-pressure cyclic conditions can be further investigated, and the influence mechanism of thermo-pressure coupling on the material’s permeability performance can be explored.

(2)

Permeability Tests under Different Pore Pressures

During the tests, the confining pressure was maintained at 10 MPa. The gas permeability of the specimens under different pressure conditions was measured by gradually varying the pore pressure. The pore pressure was set to 7.5 MPa, 5.5 MPa, 3.5 MPa, and 1.5 MPa, respectively.

2.2.4. Microscopic Characterization

Field Emission Scanning Electron Microscopy (SEM) (TESCAN GROUP, a.s., Brno, Czech Republic) [52,53,54] combined with Energy-Dispersive X-ray Spectroscopy (EDS) (Oxford Instruments, Oxford, UK) [55,56,57] can be used to observe the microscopic morphology of gas-tight concrete, reveal the reaction mechanisms among basalt fibers, mineral admixtures, and sulfoaluminate cement, analyze element concentrations, and infer reaction products. The test equipment used was the TESCAN MIRA4 Field Emission Scanning Electron Microscope system (TESCAN, Brno, Czech Republic), equipped with an EDS energy-dispersive spectroscopy system. This instrument features high-resolution imaging capability and enables observation of the surface microstructure of samples under high-vacuum conditions, as well as qualitative and semi-quantitative analysis of elemental compositions in micro-areas.

3. Test Results

3.1. Uniaxial Compressive Mechanical Properties

Uniaxial compression tests were conducted on the gas-tight concrete specimens using the RMT-150B rock mechanics testing system. Three sets of parallel tests were performed, and the average values were taken. The test results of the uniaxial compressive stress–strain curve of the gas-tight concrete sealing layer are shown in Figure 8. At the initial stage of loading, the stress increased approximately linearly (segment o–a). In the next stage, the stress–strain curve entered a plastic rising segment, where the stress gradually increased and reached its peak (segment a–b). The third stage is the steep drop segment of the stress–strain curve, i.e., the failure stage (segment b–c). The specimen corresponding to the strain at point c had completely failed. The turning point (i.e., the peak) between the stress–strain curves of segment a-b and segment b-c was taken as the uniaxial compressive strength, which was 62.59 MPa. The elastic modulus was determined as the secant modulus corresponding to 40% of the peak stress (i.e., 25.04 MPa) on the stress–strain curve, yielding a value of 21.77 GPa.

Figure 8. Uniaxial compressive stress–strain curve of gas-tight concrete.

3.2. Uniaxial Tensile Mechanical Properties

The uniaxial tensile stress–strain curve results of the gas-tight concrete are shown in Figure 9. At the initial stage of loading, the stress increased approximately linearly, with a tensile strain of 0.05% and a tensile stress reaching 1.11 MPa (segment o–a). As the load further increased, the specimen entered the stress hardening stage. Cracks began to appear on the specimen surface, and the stress increased nonlinearly to its peak value. At the peak, the ultimate tensile strain reached 0.2%, and the ultimate tensile stress reached 3.58 MPa (segment a–b). As the load continued to increase, the specimen entered the strain softening stage. Surface cracks propagated into macroscopic cracks, and the load-bearing capacity of the specimen continuously decreased until fracture failure occurred.

Figure 9. Uniaxial tensile stress–strain curve of gas-tight concrete.

3.3. Effect of Thermo-Pressure Cycle Number on Permeability

To investigate the influence of thermal cycling on the permeability characteristics of concrete, temperature pretreatment of the specimens was first carried out. The temperature was set to 60 °C, held constant for 1 h, and then cooled to room temperature before specimen installation. Subsequently, a triaxial loading pre-test was conducted, and the gas permeability was measured using a fully automatic geotechnical triaxial servo apparatus under a confining pressure of 10 MPa and a pore pressure of 7.5 MPa. After the test, the specimen was unloaded, and the above process was repeated. To intuitively reflect the influence of thermo-pressure cycling on permeability, the variation curve of permeability with cycle number was plotted based on the test results, as shown in Figure 10. As can be seen from the Figure, at the initial stage of thermo-pressure cycling, the gas permeability of the specimens showed a slight decrease. For example, during the 1st to 2nd cycles, the permeability decreased from 1.78 × 10⁻¹⁸ m² to 1.72 × 10⁻¹⁸ m². This phenomenon indicates that at the initial stage of temperature cycling during the triaxial seepage test, the specimens were subjected to a confining pressure of 10 MPa. The confining pressure compresses the internal pore structure of the concrete, causing some micro-cracks to close, thereby reducing pore connectivity.

Figure 10. Gas permeability test results under thermo-pressure cycling.

As the number of temperature cycles increased, the test results showed a gradual recovery trend in gas permeability. For example, after the 5th cycle, the permeability gradually increased from 1.82 × 10⁻¹⁸ m² to 1.94 × 10⁻¹⁸ m². This phenomenon suggests that under repeated temperature variations, possibly due to the different thermal expansion coefficients between the aggregate and the cement paste, thermal deformation incompatibility occurs between them during temperature cycling, generating thermal stresses in the interfacial transition zone. When temperature cycles are applied repeatedly, these thermal stresses may induce new micro-cracks in the interfacial region or cause the propagation of existing micro-cracks, leading to an increase in gas permeability. As the number of temperature cycles further increased, the magnitude of permeability change gradually diminished. This indicates that the development of micro-cracks within the concrete gradually stabilizes, and the internal structure of the material gradually reaches a new equilibrium state. At this stage, the rate of change in pore structure decreases significantly, and thus the gas permeability gradually stabilizes. Overall, the effect of 60 °C temperature cycling on the gas permeability of concrete is relatively limited, with permeability changes remaining within the order of magnitude of 1–2 × 10⁻¹⁸ m². This indicates that under lower-temperature cyclic conditions, no significant thermal damage occurs to the internal structure of the concrete, only minor adjustments in the microstructure.

3.4. Effect of Pore Pressure on Permeability

Gas permeability tests under different pore pressure conditions were conducted under a certain confining pressure. During the tests, the confining pressure was maintained at 10 MPa, and the gas permeability of the specimens under different pressure conditions was measured by gradually varying the pore pressure. The test results obtained are shown in Table 3.

Table 3. Gas permeability test results of specimens under different pore pressures.

Confining Pressure (MPa)	Pore Pressure (MPa)	Permeability k/(10−18m2)
		Specimen 1	Specimen 2	Specimen 3	Average
10	7.5	1.74	1.79	1.81	1.78
	5.5	2.63	2.68	2.70	2.67
	3.5	5.54	5.61	5.65	5.6
	1.5	10.52	10.61	10.67	10.6

From the test results, it can be observed that under a confining pressure of 10 MPa, when the pore pressure increased from 1.5 MPa to 7.5 MPa, the gas permeability of the specimens gradually decreased from 10.6 × 10⁻¹⁸ m² to 1.78 × 10⁻¹⁸ m², showing a significant downward trend. This variation pattern indicates that under a certain confining pressure, pore pressure has a notable effect on the gas permeability of concrete. To more intuitively reflect the relationship between pore pressure and permeability, a curve showing the relationship between gas permeability and pore pressure was plotted based on the test data, as shown in Figure 11.

Figure 11. Relationship curve between gas permeability and pore pressure of specimens.

It can be seen from the figure that as the pore pressure increases, the gas permeability of the concrete gradually decreases, exhibiting a relatively obvious nonlinear trend. Theoretically, under a given confining pressure, when the pore pressure increases, the internal pore structure of the specimen may undergo a certain degree of lateral expansion, enlarging the pore space and thereby making the seepage channels more open, which could lead to an increasing trend in permeability. However, the test results indicate that under the conditions of this study, the permeability of the specimens gradually decreases with increasing pore pressure, showing a clear regularity. This phenomenon suggests that in low-permeability porous media, gas seepage behavior is influenced not only by changes in pore structure but also significantly by the gas slippage effect [58,59]. Under low pore pressure conditions, the mean free path of gas molecules is relatively large, causing gas molecules to slip along the pore wall surfaces, thereby increasing the gas flow velocity and manifesting as an increase in the apparent gas permeability. As the pore pressure gradually increases, the mean free path of gas molecules gradually decreases, and the gas slippage effect gradually weakens, so the measured apparent gas permeability also decreases accordingly.

To describe this phenomenon, Klinkenberg [60,61] proposed a modified permeability model that accounts for the gas slippage effect:

$${\mathrm{k}}_{\mathrm{g}}={\mathrm{k}}_{\mathrm{m}}(1+{\displaystyle \frac{\mathrm{b}}{\mathrm{p}}})$$

(1)

where kg is the apparent gas permeability, km is the intrinsic gas permeability, b is the Klinkenberg coefficient, and p is the average pore pressure of the sealing layer.

For concrete materials, the internal pore structure [62,63] typically consists of capillary pores, gel pores, and micro-cracks. Since these pore sizes are usually on the micrometer or even nanometer scale, gas flow within the pores is susceptible to the slippage effect. When the pore pressure is low, the apparent gas permeability is significantly higher than the intrinsic permeability. Given that the pore pressure in underground caverns fluctuates within the range of 3–10 MPa, the influence of the gas slippage effect on permeability measurements must be considered when investigating the gas permeability characteristics of materials used in compressed air energy storage underground caverns. The reciprocal of the average pore pressure was calculated based on the experimental data, and the relationship curve between permeability and the reciprocal of the average pore pressure was plotted. The fitting results are shown in Figure 12.

Figure 12. Relationship between the reciprocal of average gas permeability pressure and permeability of specimens.

It can be seen from the figure that there is a good linear relationship between permeability and the reciprocal of the average pore pressure, indicating that the experimental results conform to the basic principles of the Klinkenberg modified model. According to the Klinkenberg model equation, the fitting equation obtained by incorporating the experimental data is:

$${\mathrm{k}}_{\mathrm{g}}=0.0168\left(1+{\displaystyle \frac{966.55}{\mathrm{p}}}\right)$$

(2)

The fitting results indicate that under low-pressure conditions, the gas slippage effect is relatively pronounced, and the apparent gas permeability is significantly higher than the intrinsic permeability of the material. As the pore pressure gradually increases, the slippage effect gradually weakens, and the apparent gas permeability progressively approaches the absolute permeability of the material. This trend is generally consistent with previous research findings, further demonstrating that for concrete as a low-permeability porous material, its gas seepage behavior is significantly influenced by the slippage effect under low-pressure conditions.

3.5. Microscopic Testing and Mechanism Analysis

3.5.1. Analysis of SEM Test Results

By selecting three typical specimens for scanning electron microscopy observation, the intrinsic mechanism of the influence of each component on the performance of the sealing material can be revealed at the microscopic level, as shown in Figure 13. Figure 13a,b show the microscopic conditions of the specimens, where a remarkably dense microstructure can be observed, with extremely low porosity and most pores being isolated closed pores. Hydration products (especially C-S-H gel) are generated in large quantities and form a continuous, dense spatial network, tightly cementing the unreacted particles together. Figure 13c clearly shows ettringite (AFt), the core hydration product of the sulfoaluminate cement system, which appears as slender needle-like or rod-shaped crystals. These crystals interlock and overlap with each other, forming a spatial skeleton structure that supports the early-age strength. Figure 13d shows the C-S-H gel, which primarily provides later-age strength. It exhibits a “honeycomb-like,” network structure without a regular geometric shape. It can be observed wrapping around all other materials, filling nanoscale pores, and forming a densified matrix.

Figure 13. SEM micrographs of specimens.

3.5.2. Analysis of EDS Test Results

To reveal the types and spatial distribution characteristics of hydration products in sulfoaluminate cement concrete incorporating mineral admixtures and basalt fibers, two different micro-areas of the sample were selected for EDS surface scanning and quantitative analysis, with results expressed as mass fractions. The results are shown in Figure 14.

Figure 14. EDS test results. In the figure, “电子图集” denotes “Electronic Atlas”, “分布图总数谱图” denotes “Total number spectrum of the distribution map”, and “技术支持来自 Tru-Q” denotes “Technical support was provided by Tru-Q”.

In addition to the major elements, a certain amount of C was detected in both measurement areas. Since the samples were gold-sputtered, the detected C did not originate from a carbon coating layer. Nevertheless, it may still be influenced by sample carbonation, surface adsorption, conductive adhesive, resin, or admixture residues. In the discussion of this paper, C is treated as a possible “contamination” signal and is not used to determine the dominant phase composition. The mass fraction ratios obtained after excluding C are shown in Table 4.

Table 4. EDS element mass fraction ratios of each measurement point in the specimen after excluding carbon.

Element	O	Ca	Si	Al	S	Fe	Mg	K	Na	Cl	Ti
Analyzed zone 1	38.6	25.5	5.5	9.7	5.7	2.6	0.7	0.5	0.3	0.3	0.3
Analyzed zone 2	32.1	23.4	18.4	5.6	4.5	2.8	0.5	0.5	0.3	0.5	0.4

Measurement Area 1 exhibits distinct calcium-rich characteristics, with a Ca content of 25.5%, while the Al content reaches 9.7%, the S content is 5.7%, and the Si content is only 5.5%. The relative enrichment of Ca-Al-S indicates that this micro-area is closer to the typical hydration product composition of sulfoaluminate cement. During the early hydration process of sulfoaluminate cement, the aluminate phase and sulfate readily form Ca-Al-S-O system products such as ettringite (AFt) under Ca²⁺-sufficient conditions, possibly accompanied by the formation of monosulfoaluminate or monocarboaluminate hydrate phases (AFm). Considering the coexistence of high Al and S contents along with Ca in this measurement area, it can be concluded that Area 1 is a region relatively enriched in sulfoaluminate hydration products. Its microstructure typically exhibits an interwoven morphology of needle-like, columnar crystals and gel phases, which significantly contributes to the early-age strength and the formation of the skeleton structure of the material. At the same time, localized agglomeration of AFt/AFm crystals or their concentrated precipitation in pores may also lead to micro-scale compositional heterogeneity, a common phenomenon in the microstructure of cement-based materials. Measurement Area 2 shows prominent differences compared to Area 1: the Si content increases to 18.4%, the Ca content is 23.4% (slightly lower than that in Area 1), while Al decreases to 5.6% and S decreases to 4.5%. The substantial increase in Si typically indicates an increased proportion of silicate gel phases or proximity to the reaction layer of mineral admixtures, as well as the presence of unreacted particles. Considering that fly ash and slag are incorporated into the system, both can provide reactive SiO₂ and Al₂O₃ and undergo pozzolanic and latent hydration reactions under alkaline conditions, generating C-S-H or C-A-S-H type gel products. Such gels typically exhibit a continuous or flocculent distribution, capable of effectively filling pores, wrapping around particles, and improving the interfacial transition zone structure, thereby enhancing the later-age strength and durability of the material [64,65]. Therefore, Area 2 can be regarded as a micro-area relatively enriched in gel phases, reflecting the role of mineral admixtures in participating in reactions within the sulfoaluminate cement system and forming a densified structure. Compared with Area 1, the decreased Al and S contents in Area 2 also indirectly indicate that the proportion of sulfoaluminate crystalline products in this region is relatively lower, with silicate gels being the main contributor. The Fe contents in the two measurement areas are 2.6% and 2.8%, respectively, and the Mg contents are 0.7% and 0.5%, along with small amounts of trace elements such as Na, K, Cl, and Ti. These can be attributed to the intrinsic trace components of fly ash or slag, aggregate impurities, or components introduced from basalt fibers. Their low contents generally do not alter the dominant hydration phase types of the system, but may act as nucleation sites or participate in the formation of composite gel structures in localized regions.

Overall, the main elements in the sample are O, Ca, Si, Al, and S, reflecting the typical characteristics of a sulfoaluminate cement–admixture composite system. The O element content dominates in both measurement areas (38.6% in Area 1 and 32.1% in Area 2), indicating that the tested regions are primarily composed of oxygen-bearing hydration products, which is consistent with the fact that cement-based materials mainly consist of various oxygen-containing compounds after hydration. Ca, Si, Al, and S are all major structure-related elements, reflecting that this admixture-fiber reinforced sulfoaluminate cement system exhibits a multiphase characteristic of “sulfoaluminate hydration products (AFt/AFm) + silicate gels (C-(A)-S-H/C-A-S-H)” coexisting, with pronounced spatial heterogeneity. Area 1 is biased towards AFt/AFm enrichment, reflecting the advantage of early-age products brought by the rapid hydration of sulfoaluminate cement; Area 2 is biased towards a silica-rich gel structure, reflecting the pore-filling and densification effects of fly ash or slag after their participation in reactions. This synergistic mechanism of “early-age crystalline skeleton + later-age gel densification” provides a microscopic basis for the material to simultaneously achieve relatively rapid strength development and good structural stability.

4. Numerical Simulation

To ensure reproducibility without excessively increasing the length of the main text, comprehensive documentation of the numerical model is provided in Supplementary Material S1. This includes the governing equations, coupling relationships, constitutive assumptions, mesh information, convergence study, time-stepping scheme, and solver settings. A summary paragraph is provided below.

4.1. Engineering Background

The Three Gorges Ulanqab Huade County Source-Network-Load-Storage Integrated Green Power Supply Industrial Park Demonstration Project is located in Huade County, Ulanqab City, Inner Mongolia Autonomous Region [66]. The project has a roof burial depth of approximately 155 m, a gas storage pressure ranging from 3.0 to 10 MPa, and a gas storage volume of 40,000 m³. The surrounding rock is composed of granodiorite, with the engineering geological category of the surrounding rock being mainly Class II to III. The annular gas storage cavern has a length of 478.25 m and a tunnel diameter of 10 m. A 0.5 m thick C40 reinforced concrete lining is cast on the inner wall of the cavern, and a 0.1 m thick sealing layer is laid on the inner surface of the lining. The sealing layer adopts the experimentally formulated material, as shown in Table 2.

4.2. Physical Property Parameters and Boundary Conditions

Within the framework of COMSOL Multiphysics (version 6.4, COMSOL, Inc., Burlington, MA, USA), a numerical model coupling non-isothermal flow, fluid flow, and geomechanical deformation of the compressed air energy storage rock cavern was established using the Partial Differential Equation (PDE) solver [67,68,69]. Figure 15 shows the boundary conditions for the seepage field, temperature field, and stress field. The initial temperature of this CAES underground cavern is set to 286.15 K, and the initial pressure is set to 1.5 MPa based on the surrounding rock properties. Table 5 presents the numerical simulation parameters for sensitivity analysis. The model calculation domain extends to six times the cavern diameter. The cavern thermodynamics are applied as boundary conditions on the inner surface of the cavern. For the cavern structural deformation field, the normal displacement is constrained at the outer boundary of the calculation model, while the inner boundary is a free boundary. For the seepage field, the outer boundary of the calculation model is set as a constant pressure, and the inner boundary is set as the gas pressure inside the cavern. For the temperature field, the outer boundary of the calculation model is set as a constant temperature, and the inner boundary is set as the gas temperature inside the cavern. During each daily cycle of CAES operation, air is injected into the underground rock cavern at a constant mass flow rate for 8 h, followed by a 4 h storage period. Then, air is released from the underground rock cavern at a constant mass flow rate for 4 h, followed by another 8 h storage period.

Figure 15. Boundary conditions of the numerical calculation model.

Table 5. Value of the parameters in the sensitivity analysis.

Parameter (Variable, Unit)	Value	Source of Calculation Parameters
Cavern radius, $r_0$ (m)	5	Operating condition
Cavern length, H (m)	5.093 × 102	Operating condition
Cavern surface area, $A_c$ (m2)	8 × 104	Operating condition
Cavern volume, V (m3)	4 × 104	Operating condition
Cavern burial depth, D (m)	155	Operating condition
Lining thickness, $d_c$ (m)	0.5	Zhou et al. and Rutqvist et al. [70,71]
Sealing layer thickness, $d_m$ (m)	0.1	Zhou et al. and Rutqvist et al. [70,71]
Temperature of injected air, $T_i$ (K)	313.15	Operating condition
Ratio of discharge rate to charge rate, $C_D$	2	Zhou et al. [72]
Mass flow rate of injected air, $m_i$ (kg/s)	70	Zhou et al. and Rutqvist et al. [70,71]
Mass flow rate of released air, $m_e$ (kg/s)	35	Zhou et al. and Rutqvist et al. [70,71]
Heat transfer coefficient, $h_c$ (W/(m2·K))	50	Zhou et al. and Rutqvist et al. [70,71]
Initial cavern air pressure, $p_{a0}$ (Pa)	4.5 × 106	Zhou et al. and Rutqvist et al. [70,71]
Initial temperature, $T_{a0}$, $T_0$ (K)	293.15	Operating condition
Elastic modulus of rock, $E_r$ (GPa)	30	Operating condition
Poisson’s ratio of rock, ${\nu }_r$	0.3	Operating condition
Rock density, ${ \rho }_r$ (kg/m3)	2600	Operating condition
Coefficient of thermal expansion of rock, ${\alpha }_{Tr}$ (1/K)	8 × 10−6	Operating condition
Thermal conductivity of rock, ${ \mathsf{\lambda }}_{\mathrm{r}}$ (W/(m·K))	2.5	Operating condition
Specific heat capacity of rock, $C_r$ (J/(kg·K))	0.8 × 103	Operating condition
Elastic modulus of lining, $E_c$ (GPa)	34	Operating condition
Poisson’s ratio of lining, ${\nu }_c$	0.3	Operating condition
Density of lining, ${ \rho }_c$ (kg/m3)	2500	Operating condition
Coefficient of thermal expansion of lining, ${\alpha }_{Tc}$ (1/K)	1 × 10−5	Operating condition
Thermal conductivity of lining, ${ \mathsf{\lambda }}_{\mathrm{c}}$ (W/(m·K))	2	Operating condition
Specific heat capacity of lining, $C_c$ (J/(kg·K))	0.9 × 103	Operating condition
Porosity of lining, ${\varphi }_c$	0.12	Operating condition
Permeability of lining, $k_c$ (m2)	1 × 10−16	Operating condition
Elastic modulus of sealing layer, $E_m$ (GPa)	20	Xu et al. [73]
Poisson’s ratio of sealing layer, ${\nu }_m$	0.3	Zhou et al. and Rutqvist et al. [70,71]
Density of sealing layer, ${ \rho }_m$ (kg/m3)	2500	Xu et al. [73]
Coefficient of thermal expansion of sealing layer, ${\alpha }_{Tm}$ (1/K)	9 × 10−6	Zhou et al. and Rutqvist et al. [70,71]
Thermal conductivity of sealing layer, ${ \mathsf{\lambda }}_{\mathrm{m}}$ (W/(m·K))	1.6	Zhou et al. and Rutqvist et al. [70,71]
Specific heat capacity of sealing layer, $C_m$ (J/(kg·K))	0.9 × 103	Zhou et al. and Rutqvist et al. [70,71]
Porosity of sealing layer, ${\varphi }_m$	0.08	Zhou et al. and Rutqvist et al. [70,71]
Apparent permeability of sealing layer, $k_m$ (m2)	1.68 × 10−20	Test
Klinkenberg coefficient, b (N/m2)	966.55	Test
Initial reservoir pressure, $p_0$ (Pa)	1.0133 × 105	Zhou et al. [72]

4.3. Model Results and Analysis

4.3.1. Variation Law of Cavern Pressure

Relevant studies have shown that the temperature field of the compressed air energy storage cavern reaches a steady state after 50 cycles [74]. Considering the computational time, calculations were performed for 60 cycles. The variations in temperature and pressure of the CAES underground cavern during 60 operating cycles (60 days) were investigated, and the pressure variations during the 1st week and the 60th cycle were selected for comparative analysis. Figure 16 shows the variation of air pressure inside the cavern. From the perspective of a single operating cycle, the air pressure inside the cavern gradually increases during the charging phase as the injected air volume increases, exhibiting an approximately linear increasing trend. After entering the storage phase, the cavern pressure shows a slight decrease due to the slow leakage of gas through the sealing layer into the surrounding rock. Subsequently, during the discharge phase, as the gas is released for power generation, the air pressure inside the cavern continues to decrease in an approximately linear trend. Before the next storage phase begins, the pressure rises to some extent again due to changes in system operating conditions. Therefore, within a complete operating cycle, the cavern pressure generally exhibits a periodic variation characteristic of “increase—slow decrease—rapid decrease—recovery”. From the perspective of operation results, as the number of operating cycles increases, the overall level of air pressure inside the cavern shows a slowly decreasing trend. This is mainly due to continuous gas leakage during the storage process, resulting in a gradual reduction of air mass inside the cavern. Over 60 operating cycles, the cumulative leaked air mass is approximately 2.54 × 10⁵ kg, corresponding to an air leakage ratio of about 0.21% of the cavern’s gas storage capacity. Affected by this, during the 60th operating cycle, the maximum air pressure inside the cavern is approximately 8.7 MPa, and the minimum pressure is 3.31 MPa, both within the designed operating pressure range of 3–10 MPa.

Figure 16. Variation of air pressure: (a) Week 1, (b) Day 60, (c) 60 cycles.

Overall, during cyclic operation, the average air pressure inside the cavern exhibits a gradually decreasing trend. This phenomenon is mainly related to two factors. On the one hand, as the gas mass inside the cavern decreases, the air pressure correspondingly decreases. On the other hand, as time progresses, some air gradually accumulates in the pores of the sealing layer, forming gas pressure, which to some extent slows down the subsequent air leakage rate, exerting a retarding effect on subsequent gas leakage.

4.3.2. Variation Law of Cavern Temperature

(1)

Variation Law of Air Temperature

To further analyze the thermodynamic characteristics of the CAES underground cavern during cyclic operation, the variation law of cavern air temperature over 60 operating cycles was investigated. The temperature variations during the 1st week and the 60th cycle were selected for comparative analysis, as shown in Figure 17. From the perspective of a single operating cycle, the air temperature inside the cavern increases rapidly during the charging phase as compressed air is continuously injected. This is due to the adiabatic compression effect that occurs after the compressed air enters the cavern, which increases the internal energy of the gas, resulting in a significant rise in air temperature. After entering the storage phase, heat exchange occurs between the air inside the cavern and the cavern structure, with part of the heat being transferred to the cavern, causing the air temperature to decrease to some extent. During the discharge phase, as the air expands and is output for power generation, the internal energy of the gas decreases, and the cavern air temperature further decreases. Subsequently, during the next storage phase, the temperature rises slightly due to heat exchange. Therefore, within a complete operating cycle, the cavern air temperature generally exhibits a periodic variation characteristic of “increase—slow decrease—significant decrease—recovery”.

Figure 17. Variation of air temperature: (a) Week 1, (b) Day 60, (c) 60 cycles.

From the perspective of operation, as the number of cycles increases, the overall variation of cavern air temperature gradually stabilizes. In the early stage of system operation, due to the large temperature difference between the cavern air and the cavern structure, heat exchange is more significant, and the temperature variation amplitude is relatively large. As the operation time extends, the cavern structure gradually absorbs some heat and forms a certain amount of heat accumulation, causing the temperature difference between the cavern air and the cavern structure to gradually decrease, and the temperature variation gradually enters a stable state.

During the 60th operating cycle, the variation amplitude of cavern air temperature remains basically consistent across cycles, and the temperature variation curve exhibits stable periodic fluctuations, indicating that the system has gradually reached a thermodynamic dynamic equilibrium state. Overall, under cyclic operating conditions, the cavern air temperature maintains regular periodic variations within each operating cycle, with its average temperature variation tending to stabilize. This is mainly due to the heat exchange between the cavern air and the cavern structure gradually reaching equilibrium. During the charging period, the air temperature rises from 32.3 °C to 65.6 °C, then slightly decreases to 55.15 °C during the first storage period. Subsequently, the air temperature rapidly decreases to approximately 5.12 °C during the discharge period, and then slightly increases to 32.25 °C during the second storage period.

(2)

Variation Law of Temperature in the Sealing Layer

To further analyze the temperature variation in the sealing layer of the CAES cavern during operation, the temperature variation at point P1, located at the middle position of the sealing layer on the symmetric boundary, was analyzed over 60 operating cycles. The variation processes during the 1st week and the 60th cycle were selected for comparison, as shown in Figure 18. It can be seen from Figure 18 that the temperature at the center of the sealing layer also exhibits distinct periodic variations with system operation, and its variation trend is generally consistent with that of the cavern air temperature, also exhibiting the periodic variation characteristic of “increase—slow decrease—significant decrease—recovery”. During the charging phase, as the air temperature inside the cavern increases, the sealing layer temperature gradually rises. During the discharge phase, due to the expansion and cooling of the air, the sealing layer temperature also decreases. As the number of operating cycles increases, the overall temperature inside the sealing layer shows a gradually increasing trend. This is because, during operation within the investigated 60-day period, the cavern air continuously transfers heat to the surrounding rock, resulting in gradual heat accumulation within the surrounding rock. At the same time, the amplitude of temperature variation inside the sealing layer is significantly smaller than that of the cavern air temperature. This is mainly due to the relatively large heat capacity and slower heat conduction process of the surrounding rock and sealing layer materials, which cause a certain lag in temperature variation. Therefore, the temperature variation inside the sealing layer is relatively gradual and tends to stabilize over cyclic operation.

Figure 18. Temperature variation within the sealing layer: (a) Week 1, (b) Day 60, (c) 60 cycles.

4.3.3. Analysis of Gas Leakage Law

To analyze the sealing performance of the CAES underground cavern during operation, the variation law of the gas leakage rate over 60 operating cycles (60 days) was investigated. The variations during the 1st week and the 60th operating cycle were selected for comparative analysis, as shown in Figure 19. The gas leakage rate is defined as the mass flow rate of air seeping outward through the sealing layer and the lining layer per unit time. A positive leakage rate indicates that air flows from the cavern into the sealing layer, while a negative leakage rate indicates that air flows back from the sealing layer into the cavern. The gas leakage ratio is defined as the proportion of leaked air mass to the injected air mass within a single cycle.

Figure 19. Variation of gas leakage rate: (a) Week 1, (b) Day 60, (c) 60 cycles.

From the perspective of a single operating cycle, the variation of the leakage rate is closely related to the variation of air pressure inside the cavern. During the charging phase, as the air pressure inside the cavern gradually increases, the pressure difference across the sealing layer increases, causing gas to seep outward, and the leakage rate gradually increases. During the discharge phase, due to the rapid decrease in air pressure inside the cavern, the leakage rate gradually decreases, exhibiting a trend of “increase—decrease—decrease—increase”. The leakage rate becomes negative during certain time periods because when the pressure inside the cavern is lower than the pore pressure on the outer side of the sealing layer, the direction of the seepage driving force changes, and some gas flows back into the cavern through the pores of the sealing layer.

During the 60th operating cycle, the variation of the leakage rate has essentially stabilized, and the variation pattern within each cycle is generally consistent. Within this cycle, the maximum leakage rate is approximately 0.132 kg/s, and the minimum value is approximately −0.062 kg/s. By integrating the leakage rate over time, the total leaked mass during the first cycle is approximately 8964 kg, corresponding to an air leakage ratio of approximately 0.44%, which is less than the specified minimum leakage ratio of 1% [71,73], indicating that the underground gas storage cavern has good sealing performance under the current sealing layer parameters.

From the overall trend, the gas leakage rate exhibits a pronounced periodic variation characteristic during cyclic operation, with a slight decreasing trend as the number of operating cycles increases, and the overall variation amplitude is relatively small. In the early stage of operation, due to the rapid increase in air pressure inside the cavern and the relatively low initial pore pressure in the sealing layer and lining layer, the pressure difference across the sealing layer is large, resulting in a relatively high gas seepage rate at the initial stage. As cyclic operation proceeds, some air gradually enters and accumulates in the pores of the sealing layer, causing the pressure difference across the sealing layer to gradually decrease, thereby exerting a certain retarding effect on further gas seepage. Therefore, during operation, the leakage rate overall shows a trend of slowly decreasing and gradually stabilizing, indicating that the sealing layer has a good control effect on gas leakage.

In summary, the numerical simulations within the investigated 60-day operational period indicate that the gas leakage rate exhibits a clear periodic pattern that closely follows the cavern pressure variation. The peak leakage rate during the 60th cycle is approximately 0.132 kg/s, and the total leakage ratio in the first cycle is about 0.44%, which is below the 1% allowable value adopted in this study. The leakage rate gradually decreases and stabilizes as the number of cycles increases, suggesting that the sealing layer effectively limits gas leakage under the assumed operating conditions. However, these results are based on idealized assumptions and have not been validated against field measurements. Therefore, they should be regarded as trend estimations rather than precise predictions. The following section further discusses the parameter influences, engineering implications, and limitations of the numerical model.

5. Discussion

5.1. Comparison with Previous Studies

The gas permeability of the developed airtight concrete is on the order of 10⁻¹⁸ m², which is comparable to or lower than that of conventional concrete linings used in CAES caverns. Zhou et al. [72] and Wu et al. [75] reported that a concrete lining with permeability below 1.0 × 10⁻¹⁸ m² can limit air leakage to less than 1% under operating pressures of 5–8 MPa. Our results are consistent with this criterion.

Compared with UHPC/UHPFRC, the proposed material exhibits a gas permeability approximately one to two orders of magnitude higher. Tan et al. [76] reported intrinsic gas permeability coefficients of UHPC matrix in the range of 0.11 × 10⁻¹⁹ to 1.32 × 10⁻¹⁹ m² at 90 d, and those of UHPFRC containing steel, polypropylene, or coarse aggregates in the range of 0.196 × 10⁻¹⁹ to 0.716 × 10⁻¹⁹ m². In contrast, the gas permeability of the proposed airtight concrete is approximately 1.78 × 10⁻¹⁸ m², which is about one to two orders of magnitude higher than that of UHPC/UHPFRC. This difference is mainly attributed to the lower binder content and the absence of special curing in the proposed material, which is intentionally designed to reduce cost and improve compatibility with conventional concrete construction. Nevertheless, the permeability of the proposed material remains within the threshold required for CAES caverns, and it offers a more cost-effective alternative to UHPC while maintaining satisfactory airtightness.

In terms of cracking resistance, both UHPC and UHPFRC generally exhibit denser microstructures and fewer connected cracks than the proposed material, owing to their very low water-to-binder ratio, high binder content, and fiber bridging effect. Fan et al. [77] showed that the incorporation of steel, PE, or basalt fibers effectively controls crack width and reduces water transport in UHPC. In the present study, basalt fibers play a similar role in bridging microcracks and limiting pore connectivity, although the crack-bridging efficiency is lower than that of UHPC/UHPFRC because of the lower fiber content and the multicomponent binder system. This is consistent with the observed permeability and microstructural results.

In terms of mechanical properties, the proposed airtight concrete has a uniaxial compressive strength of 62.59 MPa, a uniaxial tensile strength of 3.58 MPa, and an elastic modulus of 21.77 GPa, which are lower than those of UHPC and UHPFRC. Fan et al. [77] reported compressive strengths of 138.74–150.11 MPa, tensile strengths of 5.13–8.91 MPa, and elastic moduli of 46.33–49.22 GPa for UHPC/UHPFRC. Liao et al. [78] also showed that FRP confinement can significantly enhance the compressive strength and ductility of UHPC columns. However, the proposed material is not intended for high-strength structural applications; rather, it is designed as a sealing layer for CAES caverns, where airtightness, thermal stability, and cost-effectiveness are the primary requirements. The lower strength is therefore acceptable as long as the sealing layer remains in a triaxial compressive state with a sufficient safety margin, as confirmed by the numerical simulations in this study.

In terms of cost and constructability, UHPC/UHPFRC typically require a very low water-to-binder ratio, high cement content, large amounts of silica fume, and special curing, which significantly increase material and construction costs. In contrast, the proposed material uses a multicomponent system of sulfoaluminate cement, fly ash, slag, and basalt fibers, which reduces the binder content and eliminates the need for special curing. This composition improves compatibility with conventional concrete construction and reduces cost, while still achieving permeability on the order of 10⁻¹⁸ m². Compared with polymer sealing layers such as HDPE, the proposed cement-based material avoids the sharp increase in permeability at elevated temperatures [21] and does not suffer from aging or softening.

However, the proposed airtight concrete differs from existing low-permeability cementitious sealing materials [79,80] in several fundamental ways. First, unlike UHPC, which typically relies on a very low water-to-binder ratio, high cement content, and special curing, the proposed material uses a multicomponent system of sulfoaluminate cement, fly ash, slag, and basalt fibers. This composition reduces cost, improves compatibility with conventional concrete construction, and still achieves permeability on the order of 10⁻¹⁸ m². Second, unlike ordinary concrete linings, the proposed material maintains stable permeability under 60 °C thermal-pressure cycling and varying pore pressures, which are critical for CAES operation. Third, unlike polymer sealing layers such as HDPE, it avoids the sharp increase in permeability at elevated temperatures and does not suffer from aging or softening. Fourth, compared with other sulfoaluminate cement-based materials reported in the literature, the present work systematically quantifies the gas permeability evolution under coupled thermo-pressure cycling and reveals the underlying microstructural mechanism.

5.2. Advantages of the Developed Material

The main advantages of the proposed material are threefold. First, the combined use of sulfoaluminate cement, fly ash, slag, and basalt fibers provides a synergistic enhancement of early-age strength and microstructural densification. Second, the material exhibits a compressive strength of 62.59 MPa and low gas permeability on the order of 10⁻¹⁸ m², with satisfactory thermal stability under 60 °C cycling. Third, the material is cost-effective and compatible with conventional concrete construction, making it suitable for large-scale CAES cavern sealing.

5.3. Limitations and Risks

It should be acknowledged that only one optimized mixture was tested in this study. Although the mix proportion was determined through RSM, the individual and synergistic effects of sulfoaluminate cement, fly ash, slag, and basalt fiber could not be experimentally separated. Future work will involve testing a wider range of mix proportions to quantify the contribution of each component. Furthermore, only three parallel specimens were tested for the compressive and permeability measurements. Due to the limited number of specimens, the standard deviation and coefficient of variation were not reported. This may affect the statistical reliability of the results. Future work will increase the number of parallel specimens to improve the repeatability assessment.

The microstructural interpretations in this study are primarily based on SEM/EDS observations from limited areas. These observations are insufficient to confirm three-dimensional pore connectivity or uniquely identify all hydration phases. Advanced characterization techniques such as XRD, TG, MIP, or XCT would provide more robust evidence. Therefore, the statements regarding specific phases and pore connectivity should be regarded as exploratory interpretations rather than definitive conclusions.

In addition, the experimental programme consists of only ten thermo-pressure cycles, and the numerical simulation covers 60 operating days. These durations are insufficient to fully assess long-term degradation effects such as creep, shrinkage, thermal fatigue, cracking, interface deterioration, and aging. The conclusions regarding sealing performance should be strictly limited to the investigated 60-day operational period.

Furthermore, the numerical model lacks independent validation against field measurements. The model was applied to the Three Gorges Ulanqab project under idealized assumptions, but in situ monitoring data are currently unavailable. The material parameters were calibrated solely based on laboratory tests, and scale effects may introduce deviations. Therefore, the numerical results should be interpreted as trend estimations rather than precise predictions.

The main risks include potential long-term permeability degradation due to thermal fatigue, interface debonding between the sealing layer and lining, and uncertainties in scaling up from laboratory to field conditions.

5.4. Application Ranges and Engineering Recommendations

The developed airtight concrete is suitable for sealing layers in hard rock caverns and abandoned mine tunnels for CAES, particularly where cost-effective and thermally stable sealing is required. It may not be suitable for salt caverns or highly corrosive environments without additional protective measures. For engineering design, a permeability threshold of 1.0 × 10⁻¹⁸ m² is recommended to ensure a leakage rate below 1%. Future work should focus on long-term field monitoring, interface performance, and optimization of fiber content and curing conditions.

5.5. Performance Requirements for Airtight Concrete in CAES Caverns

Based on the literature review and the specific operating conditions of CAES caverns, the target performance requirements for airtight concrete sealing layers are established. These requirements include target permeability, mechanical strength, deformation capacity, thermal stability, durability, and acceptable leakage.

Target permeability. The gas permeability of the sealing layer is the most critical parameter. Rutqvist et al. [71] and Kim et al. [81] demonstrated that a concrete lining with permeability below 1.0 × 10⁻¹⁸ m² results in an acceptable air leakage rate of less than 1% under operating pressures of 5–8 MPa. Therefore, the target intrinsic permeability for airtight concrete should be ≤1.0 × 10⁻¹⁸ m².

Mechanical strength. The sealing layer must withstand cyclic thermo-pressure loads without structural failure. Based on existing CAES sealing materials, a minimum uniaxial compressive strength of ≥40 MPa and a tensile strength of ≥3 MPa are recommended [44]. The material should also maintain sufficient elastic modulus to limit deformation.

Deformation capacity. The sealing layer must accommodate deformation induced by thermo-pressure cycling without cracking. A tensile strain capacity of ≥0.5% is suggested, with an elastic modulus in the range of 10–30 GPa to balance stiffness and flexibility [44].

Thermal stability. The sealing layer is subjected to air temperatures ranging from approximately 5 °C to 80 °C during CAES operation [75]. The permeability should remain stable (variation < 20%) within this temperature range, and no significant strength degradation should occur after thermal cycling.

Durability. The sealing layer must maintain its properties over thousands of charging–storage–discharging cycles (design life of 30–50 years). Key durability concerns include creep, shrinkage, thermal fatigue, and interface degradation [82].

The developed material achieves a permeability of approximately 1.78 × 10⁻¹⁸ m² (below the 1 × 10⁻¹⁸ m² threshold), a compressive strength of 62.59 MPa (above 40 MPa), and a tensile strength of 3.58 MPa (above 3 MPa). The thermal stability within 60 °C cycling and the estimated leakage rate of 0.21–0.44% (below 1%) indicate that the material satisfies the primary engineering requirements, although long-term durability and interface performance require further verification.

6. Conclusions

In this study, an airtight concrete sealing layer material for underground caverns in compressed air energy storage (CAES) systems was developed based on sulphoaluminate cement, incorporated with fly ash, slag, basalt fiber, and chemical admixtures. The mechanical properties, gas permeability characteristics, microstructural features, and sealing performance of the material were systematically investigated through laboratory experiments and COMSOL Multiphysics-based thermo-hydro-mechanical coupled numerical simulations. The main conclusions are drawn as follows:

(1) The developed airtight concrete exhibits satisfactory mechanical properties. Its uniaxial compressive strength reaches 62.59 MPa, its uniaxial tensile strength is 3.58 MPa, and its elastic modulus is 21.77 GPa. Preliminary numerical simulations under assumed operating conditions based on design parameters indicate that the sealing layer is in a triaxial compressive state (hoop stress ≈ −5.5 MPa, radial stress ≈ −9.8 MPa). The maximum compressive stress accounts for only 15.3% of the material’s compressive strength, and a preliminary assessment suggests a considerable safety margin. However, it should be noted that a more comprehensive structural safety assessment should also consider tensile cracking, cyclic thermal stresses, and interface stresses.

(2) The material demonstrates satisfactory gas tightness. Under a confining pressure of 10 MPa and a pore pressure of 7.5 MPa, the gas permeability is approximately 1.78 × 10⁻¹⁸ m². Under thermal-pressure cycling at 60 °C, the permeability remains within the order of 10⁻¹⁸ m² with only minor fluctuations, indicating satisfactory thermal stability. The permeability decreases with increasing pore pressure, which follows the Klinkenberg gas slippage model.

(3) The material possesses a relatively dense microstructure. SEM observations suggest that the hydration products are dominated by needle/rod-like ettringite (AFt), which provides early-age strength, and honeycomb-like C-S-H gel, which provides later-age strength, with limited visible pore connectivity. EDS analysis suggests the possible multi-phase coexistence of “sulphoaluminate hydration products (AFt/AFm) + silicate gel (C-A-S-H)”, and it is speculated that this forms a synergistic enhancement mechanism of “early crystal skeleton + later gel densification”. However, definitive phase identification and three-dimensional pore structure characterization require further verification using complementary techniques such as XRD, MIP, or XCT.

(4) Numerical simulations within 60 operational cycles (60 days) estimate that the cumulative leaked air mass is approximately 2.54 × 10⁵ kg, corresponding to a leakage rate of 0.21–0.44%, which is below the design allowable value of 1%. However, it should be noted that the lack of field measurement data is a major limitation of this model. These results should be regarded as trend estimations under idealized assumptions rather than precise quantitative predictions, and future validation against field monitoring data is required.

(5) Future research should focus on optimizing the mix design, conducting longer-term thermo-pressure cycling tests, applying complementary microstructural characterization techniques (XRD, MIP, XCT), validating the numerical model with field monitoring data, and extending the material to other CAES cavern types. These efforts will support the engineering application and long-term reliability assessment of airtight concrete sealing layers.