Introduction

According to the China Mine Geological Environment Survey Report by the China Geological Survey, China has a large number of mines, but relatively few large and medium-sized operations, with a predominance of small-scale mines. Non-metallic mines such as those producing building materials are abundant, whereas energy and metallic mines are less common. Among them, small-scale mines for building materials, metals, and chemical minerals account for the vast majority, and the goaf areas of such mines are mostly composed of hard rock. Therefore, a considerable number of hard rock mined caverns exist in China1,2,3. Compared with coal mine goaf areas, these hard rock mined caverns generally have irregular shapes and smaller scales. Due to significant differences in overburden conditions, burial depth, and stress fields, the currently mature evaluation methods and control measures for coal mine goafs are not directly applicable to hard rock goaf chambers. In addition, due to the lack of comprehensive standards for the evaluation, treatment, and monitoring of such chambers, a large number of hard rock goaf chambers remain long-term undeveloped or abandoned, posing potential geological hazard risks4,5,6,7,8. The surrounding rocks of these chambers are usually composed of brittle and high-strength rocks such as limestone and granite. As a result, hard rock goaf chambers typically exhibit limited settlement and deformation, and lower risks of large-scale collapse, and are often exposed. However, under external disturbances such as weathering, rainfall, and earthquakes, these chambers may undergo progressive instability, which makes their hazard evolution pathways highly complex9,10,11,12. In the context of increasingly scarce land resources in China, these chambers also possess substantial resource reuse and spatial development potential Therefore, research on the geological hazard evolution and stability analysis of hard rock goaf chambers is not only of theoretical significance but also of urgent practical value for hazard prevention and sustainable land use13,14,15.

Regarding the stability of hard rock goaf chambers, previous studies have made certain progress, primarily focusing on the stability evaluation of chambers formed by the extraction of metal minerals and stone materials. For instance, Li et al.16 pro-posed a novel “separation + interface” simulation approach and conducted physical model tests to simulate the fracturing process of underground hard rock. The experiments showed that bedding planes in hard rock layers hinder the downward transmission of overlying rock loads, revealing the influence patterns of different fracture types on the support structures of goaf chambers. Gong et al. (2021)17 analyzed mining operations and geological data to identify key factors influencing the stability of hard rock roofs and, through numerical simulations and field tests, determined reasonable parameters for deep and shallow hole combined pre-splitting blasting. Li et al.18 used UAV-acquired orthophotos to analyze the deformation characteristics of landslides in karst cave areas and employed geophysical methods to assess the damage characteristics of the overlying strata above hard rock chambers. They also applied UDEC numerical modeling to analyze the deformation responses of hillsides under underground mining and found that tensile motion along hard sandstone joints and shear slip along weak mudstone layers were the main causes of landslides. Ma et al. (2023)19 adopted an optimized Discontinuous Deformation Analysis (DDA) method to evaluate the stability of actual underground cavern groups under different excavation schemes, finding that stability issues were mainly related to crack development in the roof of the surrounding rock. Zhang et al. (2023)20 used FLAC3D software along with single-factor testing (SFT), Plackett–Burman design (PBD), steepest ascent design, and response surface methodology to assess the influence of multiple variables on the deformation of hard rock goaf roofs. They developed a predictive model for roof deformation in lead–zinc hard rock goaf chambers and verified its accuracy. Feng et al.21 employed a newly developed large-scale three-dimensional physical modeling experimental system (PMES) to study geomechanical problems related to deep underground engineering. Their results revealed failure mechanisms of hard rock masses induced by high in-situ stress, special engineering geological structures, and excavation disturbances, and confirmed the system’s reliability. Additionally, several scholars have focused on the monitoring of hard rock goaf chambers. For example, Xie et al.22 developed a ground pressure monitoring and localization system based on Acoustic Emission (AE), incorporating fractal theory and Multifractal Detrended Fluctuation Analysis (MF-DFA) to process real-time AE data. Their study clarified the multi-scale and time-varying response characteristics of chamber sur-rounding rocks. Ding et al.23 monitored and analyzed microseismic events during mining under complex geological conditions, and discussed the occurrence patterns and precursors of rockbursts. They found that before a rockburst, the energy and frequency of microseismic events decreased, a phenomenon termed “microseismic event disappearance”, which may serve as a precursor to rockburst occurrence.

In summary, although current studies have made progress in the simulation of mechanical behavior and stability analysis of hard rock mined cavern, there remain several limitations that hinder practical risk assessment and safe utilization in engineering ap-plications. Firstly, most studies focus on the stability of single, idealized chambers with regular geometries. There is a lack of in-depth investigation into the overall stability of large-scale cavern groups with complex spatial structures and coupled multi-chamber interactions. The mutual influence mechanisms among chambers and their synergistic effects on geological hazard development have not been effectively revealed24,25. Secondly, advanced engineering geological mapping techniques such as UAV photogrammetry and 3D laser scanning have not been widely applied to the study of cavern stability. There is still a lack of three-dimensional geological models capable of characterizing the spatial distribution and geometry of complex cavern groups and their coupling relation-ships with geological structures and hydrogeological conditions, which limits the precision of subsequent stability modeling and hazard simulations. In addition, the develop-mental mechanisms of geological hazards in hard rock cavern groups remain poorly understood. The formation mechanisms, spatiotemporal evolution paths, and controlling factors of typical disaster types, such as crack propagation, pillar failure, and roof collapse, have not yet been unified under a theoretical framework. Classification systems and corresponding evolutionary models for different failure modes are urgently needed. Therefore, systematic research integrating engineering geological mapping, hazard evolution mechanism identification, failure mode classification, and stability evaluation framework construction is required to provide theoretical support and technical guidance for the scientific development and disaster prevention of hard rock mined cavern groups26,27,28,29.

Based on the above analysis, this paper takes a hard rock mined cavern group formed by artificial quarrying in China as a case study. Through a combination of field engineering geological surveys and numerical simulations, the geological hazard development characteristics and failure patterns of the hard rock mined cavern group were systematically analyzed. A 3D geological model of the cavern group was constructed using ANSYS numerical simulation software, and the stress distribution and deformation characteristics of different profile directions were obtained. The failure mechanisms of the hard rock cavern group were further explored. The findings provide a scientific reference for disaster prevention and mitigation in hard rock mined caverns and offer insights into the mechanical behavior analysis and prediction of cavern group surrounding rocks.

Geological and environmental setting of the study area

The cavern group in the study area was formed by ancient quarrying activities and has since been developed into a tourist attraction. The terrain of the region is characterized by significant undulation, with elevations ranging from approximately 100–400 m, and is primarily composed of hilly landforms. The regional tectonic setting is relatively simple, with undeveloped fault structures. The bedrock mainly consists of basalt and andesite, interbedded with tuff, all of which are characterized by high strength but abundant joints and fissures. The joint sets are predominantly steeply dipping, with NE–SW and NW–SE orientations, exerting strong control on potential failure modes. Volcanic rocks are widely distributed throughout the area. The exposed strata in and around the cavern group mainly belong to the Middle and Lower Jurassic (Table 1), while portions of the foothill zones are covered by Quaternary slope deposits. In some cavern walls, weathered zones with loosened rock and partially filled fissures can be observed, further weakening local stability. A geological profile of the area is shown in Fig. 1. The region experiences a rainy season from May to June, and is significantly affected by typhoons between July and September, during which landslides and rockfalls are particularly prone to occur. The local groundwater primarily consists of pore water and fractured bedrock water, with groundwater levels showing marked seasonal fluctuations. Recharge is mainly derived from atmospheric precipitation. The basic seismic intensity in the area is classified as Grade VI, and no strong seismic events have been recorded in nearby areas in recent decades, indicating that the region lies within a geologically stable crustal zone.

Table 1 Summary of strata in the cavern group area.
Full size table
Fig. 1
figure 1

Geological cross-section of the study area.

Full size image

Geological hazard development characteristics of the cavern group

Types and distribution characteristics of geological hazards

Engineering geological investigations conducted in the study area indicate that fault structures are poorly developed, and the dominant structural features are joints and fractures. The joints are mainly compressive-shear in nature, with some forming conjugate sets. Based on joint measurements taken from bedrock slopes, caverns, and other outcrops during the survey, as well as regional geological data, it was found that individual joints generally extend 5–50 m, are closed to slightly open in form, with widths of 2–10 mm, and exhibit uneven spatial densities. A rose diagram of joint orientations was constructed based on the field survey results (Fig. 2), showing that joints trending nearly northward (N) and south-southwest (SSW) are the most developed, while those trending southeast (SE) are also relatively prominent. The bedrock in the study area is mainly composed of Jurassic andesitic-rhyolitic and trachy rhyolitic pyroclastic welded tuff with breccia and tuffaceous breccia. The rock masses are dense and of high strength, exhibiting distinct brittle behavior. Due to the high content of vitric fragments and the heterogeneous structure, the rock mass is prone to joint and fracture opening and penetration under weathering and disturbance, forming weak structural zones and reducing overall stability.

Fig. 2
figure 2

Rose diagram of joint orientations in the study area.

Full size image

In addition, field investigations identified two types of geological hazards in the area: large-scale collapses and local rockfalls. The mountain mass containing the cavern group has numerous excavated voids due to extensive historical quarrying. The supporting pillars or arch roofs bear concentrated stress and were partially damaged during mining, which can lead to roof rock collapse and a chain reaction resulting in extensive large-scale structural failure (Fig. 3). The estimated collapsed volume in this location is approximately 20,000 m3. This type of collapse indicates that in hard rock masses, once local structures are weakened due to stress concentration or excavation-induced disturbance, their overall load-bearing capacity may drop sharply. Particularly in critical locations such as rock pillars and arch roofs, when stress exceeds the strength limit, local failure can quickly propagate to adjacent structures, inducing cascade structural instability and exhibiting a typical progressive failure mechanism. Moreover, due to the varying sizes and large number of interconnected caverns, the spatial structure is highly complex. In the early stages of instability, effective stress isolation is difficult to achieve, which facilitates the development of more extensive secondary failures and significantly increases both the extent and suddenness of hazards30.

Fig. 3
figure 3

Collapsed caverns in the study area.

Full size image

In addition to large-scale collapse, local rockfalls (Fig. 4a) are also a significant geological hazard in the cavern group. Under the influence of rainfall infiltration, weathering, and the steep surfaces formed by quarrying, favorable structural conditions for rockfall development have been created—especially on the rock surfaces subjected to later blasting (Fig. 4b)31,32. Local rockfalls typically present as shallow sliding or block detachment, with development locations controlled by the density and orientation of joints and fractures. These failures exhibit a certain periodicity and spatial clustering. At present, the cavern group’s surrounding rock remains relatively intact, with few natural structural planes. Most of the existing fractures are secondary joints caused by quarry-induced cracking, and are mainly distributed within stress-concentrated areas inside the caverns. Based on the integrity and enclosure characteristics of the caverns, the study area has been functionally zoned (Fig. 5). Potentially hazardous rock masses are point-distributed with some localized concentration, mainly in the northwestern part of Zone F and the northern part of Zone Y33. The spatial distribution of these unstable rock masses is closely related to the heterogeneity of the rock structure, the intensity of underground disturbance, and hydrogeological conditions, making these areas key zones for future hazard risk prevention and control.

Fig. 4
figure 4

Rock mass failure in caverns: (a) local rockfall, (b) unstable rock mass on cavern roof slope.

Full size image
Fig. 5
figure 5

Functional zoning of the cavern group and distribution of unstable rock masses.

Full size image

Destruction mechanisms and patterns of unstable rock masses in the cavern group

Numerous potentially unstable rock masses are widely distributed within the study area and may trigger collapse-type geological hazards under natural or anthropogenic disturbances. According to field geological survey results, a total of 26 unstable rock masses are primarily distributed throughout the cavern group. These rock masses vary in scale and exhibit diverse failure patterns. The characteristics of the rock masses are summarized in Table 2, and their distribution is illustrated in Fig. 5.

Table 2 Characteristics of unstable rock masses.
Full size table

In summary, the failure modes of unstable rock masses exhibit diverse characteristics, and the formation mechanisms are governed by the combined influence of structural types of the rock mass, spatial distribution, joint connectivity, degree of weathering, hydro-meteorological conditions, and human disturbance. A comprehensive comparison of these influencing factors suggests that the failure modes can be categorized into four fundamental types: sliding, toppling, tensile cracking, and tensile cracking–toppling34.

Sliding mode

This failure type (Fig. 6a) primarily occurs in rock masses cut by gently dipping joints that form weakly connected bodies with the bedrock. The shear strength of the unpenetrated portion of the joint maintains rock mass stability. The dip direction of the rock mass aligns with the slope direction, making the upper unstable blocks prone to sliding under gravity toward the free face. During heavy rainfall, earthquakes, or other disturbances, pore water pressure in joints increases rapidly, reducing shear strength and inducing slip failure along structural planes, ultimately leading to shear-sliding dominated instability.

Fig. 6
figure 6

Typical failure modes of unstable rock masses: (a) sliding failure of rock mass Y12, (b) toppling failure of rock mass F5, (c) tensile cracking failure of rock mass F2, (d) tensile cracking–toppling failure of rock mass F8.

Full size image

Toppling mode

Toppling rock masses (Fig. 6b) are cut by two or more joint sets. Slope-parallel joints play a controlling role in stability, and the rock mass connects only locally to the base bedrock. The rock mass has poor lateral stability and a vertically elongated shape. Weathering, rainfall, and gravity promote joint extension and eventual penetration. Mining activities intensify damage to rock mass integrity, shifting the center of gravity and causing the mass to rotate outward along the base, resulting in rotational toppling failure. This type of failure is often sudden and difficult to predict, making it one of the most hazardous modes.

Tensile cracking mode

These rock masses (Fig. 6c) are intersected by multiple joint sets that are nearly fully penetrating, with only partial attachment to the bedrock. The upper portion forms an overhanging block with well-developed structural or weathering joints. Over time, gravity causes joints to propagate until failure occurs. The lower part often consists of relatively weak and weathered layers, intensifying the suspension of the overlying mass and pushing it toward a critical unstable state. Influenced by seepage and seismic disturbance, these masses are highly prone to tensile-induced collapses.

Tensile cracking–toppling mode

This hybrid failure mode (Fig. 6d) typically occurs in rock masses cut by multiple tensile joints, often suspended from steep rock walls with low self-stability. As pre-existing fractures propagate, the rock mass’s shear strength becomes insufficient to maintain its attachment to the bedrock. Failure initiates as tensile detachment, followed by rotational toppling due to a shifted center of gravity. This mode is common in areas with complex joint structures under extensional stress and is characterized by abrupt onset and severe destructiveness.

Analysis of geological hazard influencing factors in the cavern group

Numerous factors influence geological hazards in the cavern group, which can be broadly divided into two categories: intrinsic controlling factors (e.g., slope geometry, rock structure) and external triggering factors (e.g., rainfall, earthquakes, anthropogenic activities). The influencing factors specific to the hard rock cavern group can be categorized into the following three types35,36:

Rock type and structure

Lithological characteristics vary significantly across formations of different geological periods. Different lithological combinations exert considerable impact on the development and failure modes of collapses. The study area is dominated by Jurassic tuff, which is mechanically strong but highly jointed and penetrated.

Climatic and hydrogeological conditions

Diurnal and seasonal temperature fluctuations accelerate weathering. Rainfall is a major triggering factor, infiltrating the rock mass through surface joints, increasing slope self-weight and both static and dynamic water pressure, thereby weakening the strength of weak structural planes. The soaking–softening process during rainfall leads to infill material loss, enhances joint development, and generates potential shear-slip surfaces. Additionally, capillary action and water infiltration significantly reduce the friction resistance of joint sidewalls37.

Nearby human engineering activities

In addition to natural factors, human activities are a non-negligible subjective contributor to geological hazard initiation. The study area experiences intensive anthropogenic activities. Although quarrying promotes economic growth, it also alters geological conditions and induces numerous hazards by disturbing slope stability. Stress unloading in surrounding rock leads to joint opening and propagation, eventually forming through-going fractures and serving as a trigger for rock mass instability.

Stability analysis of cavern groups via numerical simulation

Establishment of the numerical model

To quantitatively assess the stress distribution and stability of the cavern group within hard surrounding rock under self-weight load, a three-dimensional finite element model was developed using ANSYS 12.0 software. The model geometry was constructed strictly based on measured data: a digital elevation model (DEM) of the ground surface was obtained through UAV photogrammetry, and the internal cavern structure was accurately reconstructed via terrestrial 3D laser scanning. The two types of point cloud data were registered and filtered to generate a complete 3D solid model. In order to eliminate the boundary effect, the model area should be 3 ~ 5 times larger than the research area (Fig. 7). The surrounding rock was identified as moderately to slightly weathered tuff. Based on laboratory test results on uniaxial compressive strength, tensile strength, shear strength, and elastic modulus, as well as the field investigation report of the study area, the rock mass was assumed to be a homogeneous and isotropic elastic material. Different mechanical parameters were assigned based on the degree of weathering. Model parameters are listed in Table 338.

Fig. 7
figure 7

3D Numerical model: (a) 3D view; (b) top view.

Full size image
Table 3 Physical and mechanical properties of the rock.
Full size table

Results and analysis of numerical simulation

Results and analysis of typical sections

Numerical simulations were conducted for 11 sections (P01–P11, as shown in Fig. 8). Mesh generation and calculations were carried out for each section, and the maximum stress values are summarized in Table 4. Four representative sections, P01, P02, P07, and P10, were selected for stress and displacement contour visualization, as shown in Fig. 9.

Fig. 8
figure 8

Distribution of simulated sections.

Full size image
Table 4 Maximum stress values from numerical simulation of typical sections.
Full size table
Fig. 9
figure 9figure 9figure 9

Stress and displacement contours of typical sections: (a)–(d) section P01; (e)–(h) section P02; (i)–(l) section P07; (m)–(p) section P10.

Full size image

Simulation results indicate that in section P01 (Fig. 9a–d), the overall stress distribution of the surrounding rock is relatively complex, with the vertical stress field exhibiting significant zoning characteristics. The maximum tensile vertical stress is primarily concentrated at the arch of Cavern B, where the local rock mass is under tension, suggesting an upward extension of tensile stress. Conversely, the maximum compressive vertical stress appears in the pillar area within Cavern A, indicating compression due to the combined effect of overburden and lateral confinement, exhibiting pronounced stress concentration. Horizontal stress is mainly concentrated in the connecting rock mass between Caverns A and B, suggesting strong shear stress under structural transition conditions. Displacement contours show noticeable vertical subsidence at the arch of Cavern B and substantial horizontal displacement in the pillar between Caverns A and B, implying a shift in the stress path and weakening of structural continuity, indicating potential instability.

Section P02 (Fig. 9e–h) reflects the typical stress pattern of multi-cavern coupling zones. Simulation results show maximum vertical stress concentrated on the lateral edge of the pillar between Caverns A and C, while maximum horizontal stress occurs at the arch of Cavern C. This stress distribution indicates that the arch of Cavern C is a key area of stress redistribution due to spatial discontinuity and cavern interaction. Displacement analysis reveals a vertical settlement of 39.83 mm at the top of Cavern C and its connecting pillar with Cavern A. A horizontal slip of up to 40.62 mm occurs above Cavern C, indicating a tension-shear coupled deformation mode. This suggests that the region is already in the advanced stage of deformation evolution, posing a high risk of failure39.

In section P07 (Fig. 9i–l), located in the Guanxinqiao area, the surrounding rock is affected by topographic undulation and superimposed cavity structures, resulting in strong variability in stress and displacement responses. Maximum vertical compressive stress and maximum horizontal stress are concentrated on the bridge body and the upper-right pillar. The displacement contours reveal significant movement in the upper-right rock mass, with a clear shear-slip trend. This indicates considerable safety risks due to the combined effects of gravitational loading and structural weakening, which should be validated by in-situ monitoring.

Section P10 (Fig. 9m–p), corresponding to the Shouxing Viewing Point area with prominent elevation differences and complex structures, exhibits strong tensile stresses in both vertical and horizontal directions. The simulation shows a maximum vertical tensile stress of 1.61 MPa and a horizontal tensile stress of 2.48 MPa, concentrated at the base of protruding pillars. This forms a typical tensile-shear coupled stress field. The upper-left region of the section shows the largest vertical displacement (92.54 mm) along with 57.86 mm of horizontal slip, further indicating significant gravity-driven deformation and high instability risk.

A comprehensive comparison of these typical sections reveals that compressive stress is mainly concentrated at load-bearing pillars, arch feet, and corner regions of cavern structures—critical stress transfer zones of mechanical weakness. Tensile stress generally appears at structurally weak connecting zones between caverns, where geometric discontinuities induce stress concentration and release, easily leading to tensile cracking and reduced structural stability. Displacement responses are primarily governed by topographic relief and cavern layout: in elevated regions, rock masses are more prone to vertical subsidence and horizontal slip under self-weight, showing a coupling instability trend of "gravity-driven and structure-released" deformation. This characteristic is especially evident in high-elevation caverns and transition zones. Therefore, stability assessment and mitigation design for hazardous rock masses in the study area should focus on stress concentration and large deformation zones, emphasizing structural reinforcement at pillars, arch feet, and corner connections. Particular attention is needed for the deformation evolution of high-elevation rock masses under the combined effects of self-weight and topography40,41,42.

Simulation results and analysis of typical caverns

For caverns with complex structures and high safety requirements, it is necessary to construct a three-dimensional numerical model for quantitative analysis to better elucidate the stress mechanism of the surrounding rock and identify potential stability issues. Based on structural identification and functional importance evaluation, the column-supported system composed of the G cavern and three adjacent rock pillars, as well as the arch-shaped bridge representing a typical cantilevered arch structure, are recognized as representative configurations of vertical bearing and lateral cantilever modes in the study area, and are of typical engineering relevance and research value43,44.

Incorporating the measured 3D topographic data and structural characteristics of the caverns, a 3D numerical model with structural consistency and similar boundary conditions was established by applying moderate simplification according to stress characteristics and boundary conditions. Simulation results indicate a significant stress concentration effect in the surrounding rock of the G cavern, with stress primarily concentrated within the three rock pillars, particularly at the transition zones between pillar bases and bedrock, where a high-intensity compressive stress zone is formed (Fig. 10). This area bears the self-weight of the overlying cavern and adjacent mountainous mass, compounded by structural stress transfer among the pillars, leading to a marked increase in localized compressive stress. Such stress accumulation has been widely recognized in hard rock cavern stability studies as a key factor that provides the mechanical conditions conducive to the initiation and propagation of compressive fractures45,46.

Fig. 10
figure 10

G cavern model and stress cloud diagram: (a) geometric model, (b) mesh generation, (c) effective stress cloud diagram.

Full size image

Under long-term weathering and variable loading, the pillar base zones are prone to shear-compression and exfoliation-type failures, exhibiting a potential evolution path characterized by “bottom compaction–fracture propagation–overburden disintegration”33. This feature corresponds well with field observations of localized bulging of rock pillars and fracture extension at their ends, thereby validating the simulation results.

As a typical high-position self-supporting structure, the arch-shaped bridge shows a stress field (Fig. 11) where a significant bending-tension and shear-coupled stress concentration zone appears in the mid-lower part of the bridge. Subjected to lateral compression from both ends and its own self-weight, this zone undergoes stress redistribution and uplift of local tensile stress zones, constituting a potential structural softening area.

Fig. 11
figure 11

Arch-shaped bridge model and stress cloud diagram: (a) geometric model, (b) mesh generation, (c) effective stress cloud diagram.

Full size image

Field investigation confirms the development of three deep-penetrating vertical dike structures within the rock mass at this location (Fig. 12b). Additionally, slope surfaces and surrounding cutting fractures of the bridge exhibit a progressive opening trend (Fig. 12c and d), and the orientations of these fractures strongly coincide with the high-stress regions in the simulation. Although the overall rock mass remains relatively intact, its stability is significantly controlled by structural mechanical boundaries and natural joint systems, depending more on the integrative behavior of the structure rather than the strength of individual rock bodies. The simulation-identified shear-sensitive zones and fracture development trends provide key guidance for subsequent reinforcement design and early warning system deployment41,42.

Fig. 12
figure 12

Dike and fracture development of the arch-shaped bridge: (a) overall appearance, (b) dike development, (c) vertical cutting fractures, (d) horizontal cutting fractures.

Full size image

Comprehensive stability evaluation of artificially excavated cavern groups in hard rock

The study area features a typical hard rock geological setting, with the main caverns developed in medium- to thick-bedded tuff, forming a large-scale cavern group system excavated by artificial stone mining. These caverns are widely distributed and structurally complex, and the associated geological hazard risks are characterized by both concealment and suddenness. The geological hazard risk and the structural stability of the cavern group are intrinsically two facets of the same engineering geological issue. The comprehensive stability evaluation of the cavern group in the study area is based on geological surveys, engineering mapping, laboratory tests, numerical simulation, and field verification, and systematically analyzes the evolution mechanism of geological hazards and the stability status of the cavern group47,48,49.

The comprehensive evaluation of cavern group stability is as follows: The main surrounding rocks of the cavern group consist of relatively intact hard limestone, and under current conditions, the overall structural system remains stable, with clearly defined load-bearing frameworks and no signs of large-scale failure. However, due to early unregulated quarrying, complex structural plane combinations, and uneven gravity field distribution, evident stress concentration zones are present at local cavern–cavern transition zones, arch roofs, and pillar bases. In some of these areas, micro-deformation signs such as fracture propagation and rock exfoliation have been observed. The high consistency between numerical simulation and field investigation results indicates that the local structural stability degradation has already emerged and warrants close attention50,51,52.

Additionally, dense joint and fracture systems are developed in some cavern group zones, often appearing as sheet-like discontinuities, reducing rock mass integrity and creating conditions conducive to small-scale collapses. Analysis of the spatial distribution of quarrying remnants and topographic characteristics reveals that collapse hazards exhibit strong spatial discreteness and randomness, mostly isolated and point-distributed. Potentially hazardous zones are mainly concentrated in steep slopes with well-developed fractures and areas of intense mining disturbance. Simulation and remote sensing identification results indicate that these steep slope areas and zones above cavern entrances are primary aggregation zones of collapse risk53,54.

Overall, the artificial cavern group system developed in hard rock formations currently maintains basic structural stability. However, its safety exhibits significant heterogeneity and spatial variability. Under the influence of external triggers such as earthquakes, rainfall, and weathering, or further stress release, local regions remain at risk of progressive failure or sudden collapse. It is therefore recommended to implement intensive monitoring at critical structural locations, especially enhancing dynamic stability assessments of arch roofs, pillar connection zones, and fractured areas around cavern entrances, while establishing early-warning protocols and maintenance guidelines, and designing targeted reinforcement or remediation strategies. These measures will provide scientific support for protective remediation and safe, efficient future spatial reuse of abandoned caverns.

Conclusions

Due to their unique geological settings and formation mechanisms, hard rock mined caverns differ markedly from soft rock and coal mine goafs in terms of stability and hazard evolution. This study, based on engineering geological investigations and 3D numerical simulations of a typical cavern group formed by artificial quarrying, yields the following main conclusions:

  1. (1)

    The surrounding rock mass generally maintains high integrity, lacking the mechanical conditions for large-scale ground subsidence or “three-zone” development. The dominant hazard is localized instability, concentrated in structurally complex and joint-dense zones, particularly in the northern sections of Caverns F and Y. Failure is controlled by discontinuity configurations, primarily manifesting as tensile and sliding modes, with occasional toppling and composite failures.

  2. (2)

    Stress concentrations occur at the bases of rock pillars, cavern footings, and arch-shaped inter-cavern connections. Tensile stresses dominate the crown areas, while compressive stresses prevail in pillars. These zones represent structural weaknesses and should be prioritized in stability assessments. Elevation differences further amplify deformations, making upper steep-slope rock masses critical targets for hazard prevention.

  3. (3)

    Considering rock mass structure, weathering, hydrogeological conditions, and human activity, progressive failure is likely under combined stress concentration and structural weakening. Targeted monitoring and reinforcement of key caverns are recommended to improve hazard prevention and ensure safe, sustainable site utilization.