Mechanism of Time-Dependent Deformation and Support Collaborative Failure in Water-Rich Red-Bed Soft Rock Tunnels

Abstract

Substantial time-dependent deformation and support failure in deep tunnels through water-rich red-bed soft rock present critical engineering challenges, yet the underlying mechanisms under hydro-mechanical coupling remain inadequately quantified. This study integrates wireless remote monitoring, laboratory testing, and theoretical analysis to investigate the stress-deformation behavior of surrounding rock and support structures. Results reveal that deformation evolves through four distinct stages as follows: sharp, slow, stable, and creep, with the creep stage—governed by pore-water pressure—accounting for over 40% of total displacement. Groundwater-induced clay mineral hydration and stress redistribution significantly weaken rock self-support capacity. Support elements exhibit degraded performance; rock bolts suffer interfacial bond failure, steel arches yield asymmetrically, and the secondary lining resists transmitted deformation pressure. A novel deformation rate-based failure criterion is proposed, revealing a progressive “local breakthrough-chain transmission–global instability” failure pathway. These findings provide a theoretical basis for stability control in deep buried tunnels under hydro-mechanical coupling.

1. Introduction

Neogene red-bed strata represent a distinctive geological formation extensively distributed across terrestrial sedimentary basins globally, notably within Europe’s North Sea Basin, the United States’ Colorado Plateau, and China’s traffic corridors in southern Gansu, southern Shaanxi, northern Sichuan, and eastern Yunnan [1,2,3]. Characterized by weak mud cementation, low diagenesis, and developed porosity, these strata exhibit significant susceptibility to water-induced softening, disintegration, and mechanical strength degradation [4]. Fundamentally, rock degradation in such settings arises from a combination of synergistic physicochemical processes: the hydration expansion of hydrophilic clay minerals (e.g., montmorillonite) weakens intergranular cementation; capillary-driven water infiltration propagates and interconnects microcrack networks; and prolonged seepage induces nonlinear strength deterioration through pore pressure elevation and effective stress reduction. Under water-rich conditions, groundwater infiltration further weakens the rock structure, triggering substantial tunnel surrounding rock deformation, tunnel face instability, and support system failure. Consequently, such strata pose considerable construction risks and incur elevated long-term maintenance costs, classifying them as a highly challenging tunnel surrounding rock type [5].

In China, major infrastructure projects, including the ‘Eight Vertical and Eight Horizontal’ high-speed railway network and the ‘Western Land–Sea New Corridor’, increasingly traverse topographically complex red-bed basins. The challenges to surrounding rock stability are exacerbated by increasing tunnel depths, adverse lithological conditions, and pervasive water-rich environments under deep burial scenarios [6,7,8]. Engineering experience confirms that Neogene red-bed strata possess higher water sensitivity and lower long-term strength compared to older formations, making them a primary cause of construction delays and economic losses in associated tunnel projects due to large deformations [9,10]. It is thus imperative to elucidate the progressive deterioration mechanism of Neogene red-bed surrounding rock under water-rich conditions, clarify the synergistic interaction within the anchor-steel arch-secondary lining support system, and establish a support design and control theory centered on deformation-stress-time evolution. This is crucial for ensuring the safety and cost-effectiveness of deep-buried red-bed tunnel engineering in China.

Significant research efforts, both domestic and international, have addressed the deterioration of soft rock tunnel surrounding rock and associated support behavior in response to these engineering challenges. A consensus identifies the complex coupling between the unique mineralogical composition of red-bed rock and its hydro-mechanical behavior as the root cause of engineering hazards [11,12]. Neogene argillaceous sandstones are notably rich in strongly hydrophilic clay minerals, such as illite and montmorillonite. During groundwater infiltration, montmorillonite adsorbs substantial water molecules within its interlayers, inducing significant lattice expansion and electrical double-layer thickening. This process rapidly attenuates, or even eliminates, intergranular cementation forces [13,14,15,16]. Concurrently, the rock’s internal microfracture network expands, interconnects, and coalesces under the combined influence of capillary suction and seepage pressure. Consequently, the rock structure evolves from a “microfracture-pore” system to a macroscopic “fracture-conduit” network, with permeability increasing by 2–3 orders of magnitude within a short timeframe. This permeability enhancement significantly accelerates groundwater circulation, intensifying the scope and magnitude of water-rock interactions [17,18,19,20,21]. Experimental studies by Tomor et al. [22], Guo et al. [23], Wu et al. [24], and Zhang et al. [25] through comprehensive rock mechanics testing have confirmed that saturated argillaceous sandstone exhibits >45% reduction in uniaxial compressive strength (UCS) and >60% reduction in elastic modulus (E) compared to dry conditions. Their laboratory investigations further established that both UCS and E undergo pronounced nonlinear deterioration with increasing moisture content. Crucially, red-bed rock masses under water-rich conditions display substantial time-dependent deformation characteristics. Systematic creep testing reveals that when the deviatoric stress acting on the rock exceeds its long-term strength threshold (under prevailing water-rock interaction conditions), it undergoes a characteristic three-stage creep process: primary (decelerating), secondary (steady-state), and tertiary (accelerating). The cumulative deformation during this phase imposes a sustained and substantial load on the support structure [26,27].

Complementing research on surrounding rock deterioration, studies on soft rock tunnel support structures have yielded significant findings. For rock bolts, traditional full-length bonded types are known to improve surrounding rock stress states via suspension, combined beam, and arching mechanisms [28,29,30,31]. However, Zhang et al. [32] and Yao et al. [33] have experimentally demonstrated that water-rich conditions severely degrade interfacial bond strength, leading to anomalous axial force fluctuations and localized fracture. Steel arches, typically configured as arched ribs in initial support systems, serve as critical reinforcing elements due to their high yield strength and considerable deformation capacity. These components constitute load-bearing skeletons that effectively constrain early-stage convergence through compressive resistance and ductile yielding. Steel arches, serving as the initial support skeleton, effectively constrain early deformation through their yield capacity. Nevertheless, under high creep pressures in water-rich weak rock masses, these arches typically enter plastic accumulation within seven to fifteen days of installation. Crucially, the Arch Springing and Arch Waist regions yield first, precipitating structural eccentric instability [34,35]. The secondary lining serves as a vital safety reserve, primarily resisting deformation pressure. Field monitoring reveals that secondary lining-primary lining contact pressure peaks during formwork removal. Furthermore, sustained surrounding rock creep in water-rich environments induces gradual contact pressure recovery, often causing tensile cracking in the lining [36,37,38].

Advances in tunnel monitoring research, particularly the integration of wireless remote monitoring systems and fiber optic sensing, have led to significant improvements in data acquisition capabilities. These technologies facilitate the acquisition of high-resolution and high-spatial-resolution data, crucial for studying rock-support interaction mechanisms at critical locations like the tunnel Crown, Arch Shoulder, Arch Waist, Arch Springing, and Invert [39,40,41]. Concurrently, advancements in numerical simulation methodologies, such as FLAC3D and ABAQUS, enable more detailed analysis of support response under complex water-rock coupling conditions [42,43].

However, significant limitations persist in current research. Primarily, monitoring processes often focus predominantly on single parameters, exhibiting inadequate capacity to concurrently capture the dynamic response of the complex water-rock-support multi-field coupling system. Furthermore, the analysis of internal stress redistribution evolution within support structures during the critical creep stage remains insufficiently addressed [44,45]. A pressing scientific challenge also lies in effectively transforming revealed time-dependent deformation laws of the surrounding rock into quantifiable, operational support design indices. Current research exhibits fragmentation: the coupling mechanism between surrounding rock deterioration and support mechanical response lacks sufficient quantitative observation and characterization concerning the specific dynamics of ‘water-rock’ interaction and spatiotemporal support behavior [46,47]. The collaborative mechanism among support components (e.g., rock bolt, steel arches, secondary lining), particularly under the long-term (aging) effects prevalent in water-rich environments, remains unclear. Crucially, the progressive failure path (‘local breakthrough-chain transmission- global instability’) under such conditions requires further mechanistic explanation [48,49]. Moreover, research specifically addressing the stress-deformation behavior of support systems in distinctive red-bed soft rock tunnels, such as those encountering Neogene cement-rich sandstone, is notably scarce. This deficiency renders existing empirical formulas and normative recommendations ineffective for guiding engineering practice in such challenging geology [50], severely limiting the scientific basis and reliability of support design for water-rich red-bed soft rock tunnels.

To address these research gaps, the Xijialiang Tunnel, constructed in Neogene cement-rich sandstone, is selected as a representative case to systematically investigate the interactions among water, rock, and support structures. The research objectives are threefold: first, to elucidate the four-stage time-dependent deformation and strength deterioration mechanisms of water-rich red-bed surrounding rock under hydro-mechanical coupling; second, to quantify the stress evolution and collaborative responses of support components, including rock bolts, steel arches, and the secondary lining, under saturated conditions, and to establish a deformation-rate-threshold-based criterion for support failure; finally, to reveal the synergistic failure pathway that links progressive rock deterioration, structural yielding, and global instability. By integrating in situ wireless remote monitoring, laboratory experiments, theoretical analysis, and numerical simulations, this study provides a scientific foundation for enhancing stability control in deep tunnels traversing water-rich red-bed formations.

2. Materials and Methods

2.1. Study Subjects

The Xijialiang Tunnel case study is located in Pingliang City, Gansu Province, Northwest China. The inlet portal occupies the right bank of Shanzhai River, while the outlet portal is located on the southern slopes of Xijialiang Mountain. Geomorphologically, the area comprises tectonically denuded low-to-medium mountains. Inlet slopes exhibit moderate inclination (45–50°) with terraced topography, overlain by Pleistocene silty clay forming the lower-middle Upper Pleistocene loess horizon underlain by Neogene sandy mudstone. The region features pronounced relief with maximum elevation difference of 96 m. Summit areas display colluvial silty clay forming loess cappings. The outlet section presents steeper slopes (65–70°), similarly covered with colluvial deposits. Tunnel elevation ranges 1744.3–1841.5 m ASL, featuring dense vegetation and multiple alluvial gullies with perennial flows demonstrating significant rainy-season discharge amplification. Exposed strata consist of Lower Cretaceous (K₁) red-bed conglomeratic sandstone exhibiting brick-red coloration, weak argillaceous cementation, and poorly sorted coarse-to-fine grained texture. Quartz and feldspar dominate the mineral assemblage, with subangular particles predominantly 2–10 mm diameter. Sandy mudstone interbeds occur locally. The rock mass displays low lithification, poor weathering resistance, high water-softening susceptibility, and sensitivity to mechanical disturbance, which frequently causes structural damage. Core samples are predominantly fragmented; however, localized grayish green units exhibit notably high impact resistance and hammer fracture resistance, as shown in Figure 1. Mineralogical analysis reveals that the primary mineral constituents are quartz (trigonal system) and feldspar (monoclinic or triclinic systems), while clay minerals, including montmorillonite and illite (generally monoclinic), form the weak cementing matrix.

Bedrock fissure water primarily occurs within bedrock fractures, recharged by atmospheric precipitation. Groundwater runoff exhibits gradual flow characteristics, discharging through gully networks into Quaternary colluvial deposits in lower topographic zones. The tunnel site demonstrates favorable hydrogeological conditions attributed to three factors: limited hydraulic connectivity in sandy mudstone units, terminal positioning within gully systems, and efficient bedrock fracture drainage. Drilling data reveal significant water accumulation at rock-soil interfaces, though no stable phreatic surface was detected in tunnel-aligned boreholes, indicating generally low saturation conditions. Localized densely fractured zones exhibit enhanced stability with reduced environmental sensitivity. Groundwater development along the tunnel alignment is negligible, with no persistent water table observed. Maximum water influx was predicted using the precipitation infiltration method, with detailed results presented in

Table 1. Calculation summary of predicted normal groundwater inflow in tunnel using the atmospheric precipitation infiltration method.

Starting and Ending Pile Numbers	Tunnel Length, m	Mean Annual Precipitation, mm	Infiltration Coefficient	Catchment Area, km2	Predicted Tunnel Water Inflow, m3/d
YK5+476-YK5+920	340	565.9	0.20	0.084	26
YK5+920-YK6+120	200	565.9	0.2	0.57	176.76
YK6+120-YK6+352	180	565.9	0.25	0.36	139.55
Total	720	1697.7	0.65	1.014	342.31

.

Hydrogeological analysis predicts a normal water inflow of 342.31 m³/d per tunnel section. Given precipitation-dependent groundwater dynamics exhibiting significant discharge variability, rainy season construction may substantially increase inflow. Comprehensive data review confirms peak rainy season inflow reaches triple normal rates, yielding maximum predicted inflow of 1026.93 m³/d per tunnel section.

Geologically, the site occupies the southern Helan fold belt within the Dianwa-Taitongshan-Daitaizi compound anticlinal zone, characterized by NW-SE trending fold axial surfaces and thrust faults. Field mapping and borehole data reveal no significant discontinuities proximal to the alignment, consistent with regional tectonic stability. Sonic logging and cross-hole seismic testing in exploration boreholes enabled quantitative rock mass quality assessment through P-wave velocity measurements, providing a basis for strength classification and lithological stratification as shown in

Table 2. Xijialiang Tunnel (left line) rock mass classification table.

Start and End Mileage	Section Length, m	Surrounding Rock Designation	Saturated Uniaxial Compressive Strength, MPa	Rock Mass Integrity Index	BQ	Rock Mass Classification	[BQ]
ZK5+474~ZK5+920	446	Argillaceous sandstone	0.19	0.55	192	V	152
ZK5+920~ZK6+120	200	Argillaceous sandstone	4.95	0.63	254	V	234
ZK6+120~ZK6+376	256	Argillaceous sandstone	0.19	0.55	192	V	152

.

2.2. Infinite Remote Monitoring System for Tunnels

The tunnel structural health monitoring system utilized in this investigation is a wireless remote monitoring system, designated as the “Infinite” system. This system enables continuous, long-term, and real-time data acquisition, eliminating the need for wired connections and thus avoiding associated logistical constraints. The Xijialiang Tunnel is a typical horseshoe-shaped highway tunnel with an excavation span of approximately 13.5 m and a height of 10.2 m. The support system comprises a primary lining and a secondary lining. The primary lining, erected promptly following excavation, includes shotcrete with a thickness of 25–30 cm, steel arches (I20b section, spaced at 0.6 m intervals), and rock bolts (3–4 m in length, arranged in a 1.0 m × 1.0 m pattern). Its principal role is to offer timely support and stabilize the surrounding rock during construction. The secondary lining, generally cast after stabilization of the primary support, is constructed of reinforced concrete with a thickness of 45 cm. It functions as the permanent load-bearing component, ensuring long-term structural safety and waterproofing. Critical monitoring cross-sections are defined at the Crown (top centerline), Arch Shoulder (transition between vault and sidewall, approximately 45° from the vertical axis), Arch Waist (upper segment of the sidewall), Arch Springing (lower sidewall near the haunch), and the Invert (tunnel bottom). The nomenclature “Infinite” denotes the system’s capability to deliver robust and persistent monitoring performance under harsh environmental conditions, including adverse weather, while ensuring stable and uninterrupted data transmission. This system enables real-time stability assessment and early warning capabilities, ensuring operational safety through continuous structural evaluation. Remote online monitoring functions remain operational under adverse weather conditions, providing stable data acquisition in harsh environments. The persistent monitoring capability delivers uninterrupted 24 h observation, effectively capturing subtle structural trend evolution. The strain sensors utilized in this study, comprising both surface-mounted and embedded types (detailed in

Table 3. Monitoring items and measurement points for tunnel surrounding rock.

Monitoring Items	Device Name	Device Model	Monitoring Location	Quantity	Equipment Photograph
Surrounding rock pressure	Earth pressure cell	FS-TY08/20 (JMZX-5020Am)	Crown Arch shoulder Arch waist Arch springing Mid-span of invert	1 2 2 2 1
Contact pressure between primary support and secondary lining			Crown Arch shoulder Arch waist	1 2 2
Internal and external forces acting on the steel frame	Surface strain gauge	FS-BM30 (JMZX-212AT)	Crown Arch shoulder Arch waist Arch springing Mid-span of invert	1 2 2 2 1
Initial shotcrete stress of primary support	Embedded strain gauge	FS-NM30 (JMZX-215AT)	Crown Arch shoulder Arch waist	1 2 2
Stress in secondary lining reinforcement	Rebar stress meter	FS-GJ22/25 (JMZX-422A)	Crown Arch shoulder Arch waist Arch springing Mid-span of invert	1 2 2 2 1
Displacement	Single-Point displacement gauge	FS-B-LG10 (JMDL-3110A)	Crown Arch shoulder Arch waist Arch springing	1 2 2 2

), were aligned to measure strains along the circumferential (hoop) direction of the tunnel. This alignment was selected to enable direct quantification of bending moments and hoop stresses within the support structures—key mechanical parameters for evaluating structural response to surrounding rock pressure.

This implementation synthesizes prior geotechnical investigations with current construction progress to enable continuous remote monitoring of the following: surrounding rock pressure, structural internal forces, and deformations at Crown, Arch Shoulder, Arch Waist, Arch Springing, and Invert locations, plus rock mass temperature-humidity parameters. The specific monitoring items and the number of measurement points are shown in Table 3, and the sensor arrangement scheme is shown in Figure 2.

This approach eliminates operator-dependent errors inherent in manual inspections while generating comprehensive quantitative datasets. Continuous remote operation facilitates automated monitoring with enhanced data traceability and repeatable measurement protocols. The surface-mounted strain gauges (Model: FS-BM30/JMZX-212AT, manufactured by Nanjing Jitai Civil Engineering Instrument Co., Ltd., Nanjing, China) used for monitoring steel arches have a gauge length of 30 mm and a measurement range of ±1500 µε, with a resolution finer than 1 µε. The relatively small strain values observed in Figure 3 and Figure 4 are indicative of the structural response during the initial loading phase and remain well within the sensors’ high-resolution detection capability. The remote wireless monitoring system deployment at Xijialiang Tunnel (Jinghua Expressway) sections ZK5+890 (existing support) and ZK5+905 (new excavation) integrates comprehensive engineering-geological and hydrogeological evaluations of Neogene red-bed soft rock. Data acquisition was conducted automatically by the integrated loggers of the “Infinite” system at preset intervals: 1–10 Hz for capturing dynamic responses and 0.1–1 Hz for monitoring long-term trends. Raw voltage signals were converted into physical values (stress, strain, pressure) using factory-provided calibration coefficients with integrated temperature compensation. All time-series data were assigned precise timestamps, transmitted wirelessly in real-time to a remote server, and archived for subsequent processing and analysis.

To systematically investigate the hydro-mechanical behavior of the surrounding rock and the synergistic response of the support system in the water-rich red-bed soft rock tunnel, this study adopts an integrated methodology combining field monitoring, laboratory experimentation, and theoretical analysis. The research framework is structured as follows: First, a comprehensive wireless remote monitoring system (“Infinite” system) was deployed in selected tunnel sections (ZK5+890 and ZK5+905) to continuously record spatiotemporal data on surrounding rock pressure, structural strains, displacements, and contact pressures at critical locations (Crown, Arch Shoulder, Arch Waist, Arch Springing, and Invert). Second, laboratory tests were conducted on core samples to characterize the mineralogical composition, saturation-induced strength deterioration, and creep properties of the Neogene cement-rich sandstone. Third, the field and laboratory data were synthesized to analyze the deformation stages, support performance degradation, and stress redistribution mechanisms under water-rock coupling. Finally, a theoretical model incorporating time-dependent deformation and a deformation rate-based failure criterion was proposed to explain the progressive failure pathway from local breakthrough to global instability. This multi-scale, multi-method approach ensures a holistic understanding of the complex interactions among water, rock, and support structures, providing a robust basis for stability control in deep-buried tunnels under hydro-mechanical coupling.

3. Results

3.1. Mechanical Behavior Characteristics and Deformation Properties of Rock Bolt

The stress and strain data of rock bolts presented in Figure 3a–f align with the sensor arrangement detailed in Figure 2. Specifically, Figure 3b,c,f display monitoring results from the right arch shoulder (R-AS), right arch waist (R-AW), and left arch springing (L-ASP), respectively. Figure 3d,e provide comparative data between the left arch shoulder (L-AS) and left arch springing (L-ASP). The observed anomalous stress fluctuations and localized tensile failures at these critical locations-particularly at the arch waist and springing-suggest the action of significant non-uniform radial pressures and bending moments on the tunnel periphery over time. This asymmetric loading pattern, which is predominantly directed inward from the sidewalls and haunches, drives characteristic asymmetric deformation of the tunnel cross-section, manifesting as pronounced convergence at the arch waist and inward displacement at the springing.

The presence of groundwater markedly exacerbates these adverse mechanical responses. Moisture infiltration severely degrades the interfacial bond strength between the rock bolt grout and the surrounding rock. This degradation disrupts effective stress transfer, resulting in disordered fluctuations in axial force and a shift from a cohesive, distributed load-bearing mechanism to localized point loading at the bolt–rock interface. As a consequence, the reinforcement efficacy of the rock bolt system is significantly compromised, allowing intensified and asymmetric deformation of the surrounding rock mass.

Continuous wireless monitoring of rock bolt stress at critical tunnel locations—the right arch shoulder, right arch waist, left arch shoulder, and left arch springing—reveals significant mechanical anomalies in Xijialiang Tunnel’s initial support under water-rich conditions. The rock bolts used in this study were of the full-length bonded type, designed to reinforce the surrounding rock through suspension, combined beam, and arching mechanisms. These bolts, with a length of 3.0 m and a diameter of 22 mm, were fabricated from HRB400 steel, which has a nominal yield strength of 400 MPa and a tensile strength of 540 MPa. The steel primarily consists of iron (Fe), with a carbon (C) content below 0.25%, along with manganese (Mn), silicon (Si), and trace amounts of sulfur (S) and phosphorus (P) to improve mechanical properties and workability. To mitigate corrosion in water-rich environments, the rock bolt surface was treated with an epoxy zinc-rich primer. Nevertheless, interfacial bond degradation remains a critical concern under sustained hydro-mechanical coupling. Figure 3 illustrates highly irregular stress–strain time-series curves for full-length bonded rock bolts, characterized by disordered spatiotemporal fluctuations and localized stress concentrations. Such alternating tensile and compressive stresses, particularly pronounced at the left arch springing, suggest a complex stress state attributed to bending moments and non-uniform redistribution of surrounding rock pressure. The characteristic three-phase deformation evolution (rapid-slow-stable) is absent. Instead, rock bolt stress variations exhibit close synchronization with surrounding rock pressure redistribution. At structurally sensitive arch shoulders and arch waists, Figure 3b–d demonstrate chaotic stress oscillations with stochastic peak migration, showing no correlation with surrounding rock deformation values.

This behavior, consistent with findings by Ma et al. [51] and Li et al. [52], significantly diminishes support effectiveness relative to non-water-rich tunnels and prevents the formation of competent load-bearing rings. Figure 3e further reveals distinct tensile-compressive stress alternation phenomena at the left arch springing.

This stress reversal likely results from bending moments induced by the asymmetric deformation of the surrounding rock, which promotes localized yielding and interfacial debonding. Furthermore, certain rock bolts exhibited stress recovery after yielding, potentially due to strain hardening or partial re-bonding at the bolt–grout interface caused by ongoing rock deformation. Under water-rich conditions, however, such recovery is generally transient and soon followed by further degradation, as persistent creep and water-induced softening of both the rock mass and grout continue to weaken the system. These observations are consistent with the interfacial degradation and stress fluctuations reported by Cui et al. [53] in water-sensitive rock masses. Moreover, the alternating tensile and compressive stresses induced by bending—resulting from non-uniform rock pressure and structural asymmetry-align with the numerical simulations of Shi et al. [54], who emphasized their role in triggering localized yielding and debonding. Notably, the stress recovery observed in steel arches following significant stress reduction may be attributed to a combination of work hardening and stress redistribution from the surrounding rock, rather than purely elastic recovery-a phenomenon similarly documented by Ma et al. [55] in tunnels under high horizontal stress. In the present study, however, the presence of groundwater and associated corrosion-microstructure interactions likely reduced the effectiveness of such hardening mechanisms, leading to accelerated long-term degradation.

Field measurements identify a critical dimensional constraint: rock bolt lengths (2.6–3.3 m) consistently fall below the 3.5–4.2 m thickness of the surrounding rock’s loosened zone. This confinement restricts reinforcement efficacy to shallow rock masses, preventing anchorage penetration beyond plastic zones into stable elastic regions and inducing premature yielding during large-deformation stages. This anomalous behavior correlates directly with grout-rock interfacial bonding loss, indicating stress transfer shifts to the rebar upon interfacial failure. The absence of load decay phases indicates persistent stress redistribution without stabilization, consistent with creep-driven rock mass movements.

Comparative analysis demonstrates that traditional rigid rock bolts (ultimate extension < 200 mm) fail predominantly through interfacial bond degradation under coupled rock strength decay and creep-induced stress redistribution. Conversely, constant-resistance large-deformation rock bolts (extension capacity ≥ 1000 mm) maintain consistent load-bearing capacity throughout stress redistribution processes.

3.2. Mechanical and Deformation Characteristics of the Initial Support

Field monitoring reveals significant spatiotemporal variability and progressive yielding in steel arch frames for initial support within water-rich red-bed tunnels. As Figure 4 demonstrates, stress–strain time-series curves exhibit non-uniform response to surrounding rock pressure: at left Arch Springing and Arch Waist locations, steel arches experience elevated pressure immediately after installation, causing rapid stress escalation approaching yield strength and progressive plastic deformation. Figure 4a illustrates strain comparisons between the left arch shoulder and left arch springing positions (refer to Figure 2 for sensor layout). Figure 4b presents comparative temporal strain curves measured at the invert bottom and the left arch waist of the primary lining. Meanwhile, Figure 4c depicts the strain evolution recorded at the crown section of the primary lining. This subsequent stress increase is attributed primarily to work hardening of the steel material and ongoing stress redistribution within the surrounding rock, as opposed to a purely elastic recovery mechanism. Conversely, the crown, left arch shoulder, and invert sections initially sustain lower stress levels that gradually increase through a characteristic three-stage evolution toward delayed yielding.

This differential response causes localized sections to reach ultimate bearing capacity first. The asymmetric deformation patterns, particularly at the arch waist and springing, are indicative of bending-dominated failure mechanisms, corroborating observations by Palka et al. [56] in similar hydrogeological settings. Post-yield strain curves show slow progression where deformation development synchronizes with rock mass displacement rates until local buckling or weld-seam tearing triggers abrupt bearing capacity reduction.

Microstructural aspects, such as grain refinement and phase transformations in the steel, may enhance yield strength and extend the elastic regime under ideal conditions; however, in water-rich tunnels, these benefits are often overshadowed by accelerated corrosion and stress corrosion cracking.

3.3. Mechanical and Deformation Characteristics of the Secondary Lining

Analysis of on-site monitoring data, as shown in Figure 5 through its time-series curve depicting contact pressure between the secondary lining and initial support, reveals a distinct phased evolution within the cement-rich sandstone tunnel.

This evolution progresses through three characteristic phases: an initial rapid increase corresponding to the period of accelerated concrete strength development post-placement of the secondary lining; a subsequent slow increase phase reflecting stress redistribution adjustments between the initial and secondary support layers; and finally, stabilization into a dynamic equilibrium state. Spatial analysis further indicates that peak contact pressure is most pronounced at the right Arch Shoulder of the secondary lining, followed by the Crown region. Crucially, the overall contact pressure values remain significantly lower than the direct pressure exerted by the surrounding rock on the initial support. This finding indicates that the secondary lining primarily bears deformation pressure transmitted from the initial support, rather than the direct loosening pressure of the surrounding rock.

The strain-time curve of the secondary lining concrete, as captured in Figure 6, exhibits an oblique ‘L’ shape, indicating structural compression. Strain development comprises two distinct stages: an initial phase of uniform growth corresponding to the slow increase in contact pressure between initial and secondary support, followed by a stage of declining deformation rate culminating in progressive stabilization. This progression reflects gradual attainment of mechanical equilibrium within the rock-support system. Notably, slight rebound phenomena at isolated monitoring points suggest elastic recovery from local stress relaxation or reduced rock creep rates. Furthermore, the data presented in Figure 6 reveal an asymmetric stress state in the Invert reinforcement—manifested through coexisting tensile and compressive stresses—highlighting spatial anisotropy in load transfer under water-rich conditions. This substantiates the necessity of designing secondary lining reinforcement to resist non-uniform tensile stresses and mitigate cracking. The findings demonstrate that the mechanical behavior of the secondary lining is significantly governed by surrounding rock creep and groundwater effects. Contact pressure evolution confirms deformation pressure transmission from initial support as the primary loading mechanism, requiring approximately 20 days for stabilization. The pressure peak during formwork stripping coincides with the most critical structural safety condition. This characteristic curve further underscores the sustained influence of long-term rock mass creep on support deformation under overall compression. Local rebound phenomena and asymmetric reinforcement stresses reflect the complexity of stress redistribution in water-rich soft rock. Consequently, reinforcement design optimization for non-uniform tensile resistance and enhanced invert closure procedures are essential to improve differential deformation resistance.

In water-rich red-bed formations, the secondary lining serves dual functions as a structural safety reserve and long-term stability maintainer by accommodating rock creep and initial support deformation. Structural design thus requires stringent consideration of formwork stripping timing, localized tensile strength.

4. Discussion

4.1. Mechanisms of Surrounding Rock Deterioration Induced by Water-Rock Interaction

The deterioration of surrounding rock in water-rich red-bed soft rock tunnels is primarily governed by evolving water-rock interaction dynamics. It is important to note that while the hydrogeological conditions and water inflow were characterized (as shown in Table 1), the wireless remote monitoring system deployed in this study focused on the mechanical responses of the rock and support structures (i.e., stress, strain, displacement, and contact pressure) rather than direct measurement of in situ pore water pressure or real-time water inflow. Consequently, the influence of groundwater and its temporal fluctuations discussed herein is inferred indirectly from the mechanical degradation, time-dependent deformation behavior, and the documented hydrogeological setting. The fundamental mechanism involves the interrelation of three factors: physical degradation of rock structures by groundwater, deterioration of mechanical parameters, and stress field redistribution. This interplay induces anomalous responses in support structure load-bearing characteristics. Physical softening and mechanical degradation act as primary catalysts for instability in Neogene mudstone-sandstone sequences exhibiting muddy weak-cementation characteristics.

As Figure 7 schematically represents, hydration of hydrophilic minerals (e.g., montmorillonite, illite) directly causes intergranular bonding strength reduction and significant decreases in rock mass integrity coefficient. Previous molecular dynamics simulations Zhu et al. [57] quantify this process, utilizing the CLAYFF force field and a representative montmorillonite-water system under periodic boundary conditions. The simulations, conducted at 300 K and 1 atm, demonstrated a 1.8 Å interlayer expansion upon saturation, corresponding to a 42% reduction in macroscopic elastic modulus. Consequently, rock strength substantially declines (sandstone saturated UCS: 2.43 MPa; sandy mudstone: 0.16 MPa) with distinct strain-softening behavior characterized by progressive residual strength deterioration at elevated moisture contents.

Figure 8 presents the spatial distribution characteristics of the rock loosening zone under water-rock coupling effects, with measurements obtained from two critical sections: the tunnel arch ring (Figure 8a) and the tunnel excavation face (Figure 8b). The arch ring profile (a) depicts the loosening zone distribution around the tunnel periphery post-excavation and support installation, highlighting the depth of rock mass deterioration surrounding the constructed tunnel structure. In contrast, the excavation face profile (b) illustrates the loosening zone development ahead of the tunnel face during the construction phase, revealing the preemptive weakening of rock mass prior to excavation. The depicted loosening zones were characterized through a combination of borehole televiewer logging, multi-point borehole extensometer measurements, and seismic velocity profiling conducted at the monitored sections.

Field data confirm that groundwater penetration along discontinuities accelerates rock mass disintegration, resulting in the formation of loosened zones with thicknesses ranging from 2.6 to 3.3 m. Notably, concentrated hydraulic action near working faces further exacerbates the development of these loosened zones, critically compromising the surrounding rock’s self-support capacity.

Groundwater-driven stress field redistribution represents an advanced stage of degradation. Pore water pressure from fissure flow reduces the effective stress of the surrounding rock, diminishing its yield strength. Concurrently, post-excavation groundwater redistribution induces principal stress axis rotation, leading to stress concentration in critical regions, as evidenced in Figure 9. Field monitoring data confirm these mechanisms: at Xijialiang Tunnel’s right arch shoulder, water pressure accumulation produced a maximum deformation of 1204 mm, measuring 2.3 times greater than displacement at the left arch shoulder. This exceptionally large deformation accumulated over a period of approximately 60 days following excavation, primarily during the creep stage under sustained pore water pressure. The deformation led to severe yielding of the steel arch at the right shoulder, localized cracking of the shotcrete layer, and necessitated remedial measures including additional grouting and installation of supplementary steel ribs to stabilize the section. This pronounced asymmetry underscores the hydro-mechanical coupling effect. To further elucidate the stress redistribution process inferred from Figure 9, a conceptual description of the forces acting on the tunnel is provided here. Initially, upon excavation, the surrounding rock experiences instantaneous stress relief, leading to the development of a loosened zone around the tunnel periphery. The primary forces include the in situ horizontal and vertical stresses, which are redistributed tangentially and radially, resulting in stress concentration at the arch shoulders and arch waist. Over time, under water-rich conditions, pore water pressure reduces the effective stress, weakening the rock mass and promoting progressive yielding. The stress redistribution follows a temporal sequence: early stress buildup at the arch springing and waist due to their susceptibility to asymmetric loading, followed by delayed transfer to the crown and invert. This process is exacerbated by groundwater infiltration, which not only softens the clay-rich cementation but also induces swelling pressures, further altering the force equilibrium. The continuous creep deformation drives a shift from initial elastic stress distribution to a prolonged plastic adjustment phase, ultimately leading to the observed stress asymmetry and structural response documented in Figure 9. However, it should be noted that direct monitoring of pore water pressure or groundwater inflow was not conducted in this study due to technical and operational constraints during tunnel construction. Instead, the influence of groundwater was inferred indirectly through the observed mechanical response of the surrounding rock and support structures, such as the saturation-induced reduction in uniaxial compressive strength (UCS > 45%) and elastic modulus (E > 60%), as well as the characteristic time-dependent deformation behavior under sustained seepage conditions. While this approach provides valuable insights into the hydro-mechanical coupling effects, the lack of direct hydraulic measurements represents a limitation in fully quantifying the transient pore pressure development and its spatial variability along the tunnel alignment. Future studies incorporating in situ piezometers or distributed fiber optic sensors for pore pressure monitoring would enhance the understanding of real-time water-rock interaction dynamics and further validate the proposed deformation mechanisms.

This hydro-mechanical coupling further drives multi-stage deformation evolution through rock rheological responses, establishing the foundation for time-dependent instability via sustained terrain deformation. Rheological effects under water-rock coupling constitute the intrinsic driver of long-term deformation in water-rich Red-Bed tunnel rock masses. Subject to pore water pressure, deformation in water-rich mudstone-sandstone sequences exhibits a characteristic four-stage evolution characterized in Figure 10: rapid deformation, slow deformation, relative stability, and creep. Crucially, creep-stage deformation constitutes over 40% of total displacement, with its rate increasing proportionally to pore water pressure elevation. This sustained deformation is driven by progressive expansion of the surrounding rock plastic zone into deeper strata, while distant-field stresses simultaneously compact the excavation-damaged zone, amplifying tunnel perimeter deformation pressure. These mechanisms establish a self-reinforcing cycle wherein increasing pore water pressure triggers rock strength deterioration, inducing intensified surrounding rock deformation and subsequent stress concentration around the tunnel periphery. The cascading response of support structures to these interactions reveals systemic hazards. Groundwater erosion degrades anchor-rock interface bonding, negating suspension and composite beam effects of full-length bonded rock bolts. Premature shotcrete closure during initial support installation obstructs drainage pathways, increasing rock mass density by 10–15% and forcing steel arches into premature yield. Under prolonged pore water effects, the secondary lining exhibits complex contact pressures (e.g., a peak pressure of 75 kPa at the Arch Shoulder) and undergoes extended stress redistribution cycles exceeding 20 days—a duration threefold longer than observed in non-water-rich tunnels. While existing creep constitutive models (e.g., the Nishihara model) describe time-dependent deformation, these models inadequately characterize the dynamic hydro-mechanical-chemical coupling prevalent in water-rich soft rocks. This limitation is particularly evident concerning the quantitative effects of recurrent montmorillonite hydration swelling on creep acceleration [58]. Collectively, water-rock interactions drive progressive rock mass deterioration through coupled mechanisms encompassing physical softening, mechanical degradation, and stress redistribution. This degradation process spans the continuum from initial integrity loss to ultimate support system failure. Furthermore, the interaction between erosion and creep plays a critical role in the long-term degradation of red-bed soft rock. Erosion of fine particles along fissures accelerates the permeability increase, which in turn facilitates further water infiltration and enhances creep deformation. To minimize such interactions, improved drainage measures and the use of erosion-resistant grouting materials are recommended.

4.2. Time-Dependent Deformation-Dominated Rock Mass Instability Mechanism

Rock mass instability in water-rich red-bed soft rock tunnels constitutes a gradual evolutionary process driven primarily by time-dependent deformation (creep and flow), rather than instantaneous failure. This conceptual framework posits that under sustained stress, progressive internal damage accumulation systematically degrades strength parameters, ultimately leading to failure. The governing mechanism operates through three interdependent dimensions: spatio-temporal variability in deformation rates, continuous stress redistribution adjustments, and progressive failure pathway development.

Deformation evolution exhibits a distinct four-stage temporal progression. Immediately post-excavation, rapid deformation occurs (1–3 days) with high deformation rates driven by elastic rebound and instantaneous unloading-induced plastic slippage. This transitions to a slow deformation phase (7–15 days) characterized by significantly reduced deformation rates and partial residual stress release through micro-fracture propagation. Subsequently, a relative stability stage emerges (15–30 days) featuring further rate reduction; this represents a transitional creep initiation phase rather than mechanical equilibrium. The process culminates in steady-state creep, where deformation increases linearly with time over months to years, accounting for >40% of total displacement. This long-term deformation fundamentally arises from the viscoelastic response of clay cementation in water-rich environments. Synergistic effects of internal particle sliding and pore water migration-including pore pressure dissipation–drive progressive rock strength reduction. When surrounding rock stresses remain between long-term strength and creep critical thresholds, deformation persists indefinitely without stabilization.

Based on the spatiotemporal deformation data captured in Figure 10, a conceptual model of the tunnel’s response can be delineated. The deformation is characterized by pronounced asymmetric convergence, predominantly manifesting as significant inward displacement at the arch waist and sidewalls (arch springing), accompanied by substantial crown settlement. This deformation pattern evolves temporally through the four distinct stages identified: initially, the excavation-induced stress redistribution causes rapid and largely symmetric deformation around the tunnel perimeter (sharp deformation stage). Subsequently, the influence of hydro-mechanical coupling promotes asymmetry, particularly as water-induced softening weakens the sidewall rock mass, leading to accelerated inward movement at the arch waist and springing during the slow and stable stages. The prolonged creep stage sees the continued and often non-uniform progression of these deformation components, driven by sustained pore water pressure, ultimately leading to the development of a characteristic “elliptical” or “eggshell” distortion of the tunnel cross-section over time. The spatial heterogeneity in deformation rates, notably the 2.3 times greater displacement at the right arch shoulder compared to the left (as documented in Section 4.1), underscores the critical role of local geological variations and stress redistribution in shaping the final deformation mode.

Furthermore, the temporal evolution of deformation rate provides a quantitative basis for predicting support collaborative failure. As shown in Figure 10, the deformation rate during the creep stage exhibits a clear threshold beyond which the support system transitions from stable bearing to progressive failure. Based on field monitoring data from Xijialiang Tunnel, the critical deformation rate threshold for the surrounding rock is identified as approximately 2.5 mm/day. When the daily deformation rate exceeds this value for more than three consecutive days, the risk of support collaborative failure increases significantly. This criterion is derived from the correlation between accelerated creep and the loss of self-support capacity of the rock mass, which subsequently overloads the support system. The proposed deformation-rate-threshold-based criterion not only enhances early warning capability during construction but also provides a quantifiable index for optimizing support design in water-rich red-bed tunnels.

Progressive instability is primarily driven by time-dependent deformation-induced stress redistribution. Post-excavation, tangential stresses concentrate initially in regions including the Crown and Arch Shoulders.

As creep advances, stress migration toward deeper rock masses drives continuous plastic zone expansion. This process demonstrates significant spatial non-uniformity accompanied by principal stress axis rotation, inducing renewed shear deformation along existing discontinuities and further compromising rock mass integrity. Where localized stresses exceed rock mass residual strength, shear failures develop (e.g., longitudinal Arch Shoulder cracking). Stress redistribution from such failures accelerates adjacent zone deformation rates, establishing a progressive cycle where localized failure triggers stress transfer, subsequently inducing new failure zones, potentially culminating in global instability.

In water-rich environments, pore water permeation and water-rock chemical interactions (e.g., clay mineral hydration) significantly exacerbate time-dependent deformation effects. These processes not only accelerate creep but also progressively degrade rock mass microstructure, diminishing self-support capacity. This water-time synergy reduces critical instability timeframes and intensifies failure suddenness. Critically, delayed support system responses amplify instability risks: interfacial bond strength deterioration compromises anchorage reinforcement effectiveness; plastic deformation accumulation in steel arches correlates with creep rates; and secondary linings develop local stress concentrations under asymmetric deformation.

The observed stress asymmetry and bending effects in support elements highlight the importance of considering not only axial loads but also flexural and shear components in design. Future support systems could benefit from materials with enhanced corrosion resistance and microstructural stability to mitigate phase transformations under hydro-mechanical coupling.

4.3. Synergistic Failure Mechanism of Support Structures

Support system failure in water-rich red-bed soft rock tunnels originates from synergistic degradation of rock bolt, initial support, and secondary lining under combined hydrological and time-dependent deformation effects. This progressive failure sequence initiates with rock bolt system collapse, where water-rich environments and sustained rock creep degrade full-length bonded rock bolt interface performance, compromising suspension and composite beam effects. As Figure 11 depicts, rock bolt axial forces substantially exceeding yield strength, while rigid rock bolts limited ultimate elongation prevents accommodation of large creep deformations, inducing local tensile failure. Consequently, rock bolt holes further facilitate groundwater infiltration, exacerbating rock mass softening and establishing a feedback cycle: rock bolt failure accelerates surrounding rock deformation, thereby intensifying rock bolt stresses Field observations and monitoring data indicated that rigid bolts, with an ultimate elongation of less than 200 mm, typically sheared off at displacements exceeding 150–180 mm, whereas constant-resistance large-deformation bolts, capable of elongating up to 1000 mm, effectively accommodated creep-induced displacements beyond 300 mm without failure. When rock bolt lengths inadequately restrict expanding plastic zones, deep rock mass regulation capacity diminishes, subjecting initial support to intensified deformation pressure.

Premature shotcrete sealing creates water-seal effects that obstruct drainage, increasing rock mass density by 10–15% while reducing strength—exemplified by 120 cm maximum settlement at Xijialiang Tunnel section K5+747–K5+776. Although numerical manifold studies confirm support structure synergy can suppress compressive deformation propagation [59], water-rich conditions critically diminish this capacity through stress restructuring: steel arches enter yield-bearing states under sustained creep, exhibiting linear strain-time growth.

Initial support failure constitutes a critical phase in synergistic deterioration. Steel arch and shotcrete stress distributions exhibit spatial asynchrony: at Arch Springing and Arch Waist regions, rapid stress escalation to yield plateaus occurs due to early surrounding rock softening. Conversely, Crown regions experience delayed stress response compounded by shotcrete-induced hydraulic sealing, enabling persistent rock strength degradation. This delayed steel arch stress progression accumulates plastic deformation, ultimately triggering weld seam tearing or buckling instability. Water-rich conditions further delay shotcrete strength development while increasing structural stiffness. Failure to achieve prompt support closure, particularly through delayed Invert construction, induces asymmetric deformation patterns characterized by sequential Crown settlement, sidewall compression, and exacerbated Invert bulging. This deformation hierarchy amplifies localized stress concentrations and induces secondary lining failure.

Secondary lining failure represents the terminal phase of synergistic degradation, with its stress state directly constrained by initial support effectiveness. Monitoring data confirms that the secondary lining primarily transmits deformation pressure from the initial support. Crucially, having inherited significant plastic deformation from the initial support, the secondary lining immediately assumes residual stresses and additional loading upon installation. Long-term rock creep increases contact pressure, while asymmetric deformation induces localized tensile-compressive stress alternation, initiating concrete cracking. When initial support fails, the secondary lining bears loads exceeding design capacity, ultimately causing structural collapse. This coordinated failure follows a defined mechanical progression: water-rich conditions and time-dependent deformation first compromise rock bolt functionality, subsequently accelerating plastic zone expansion, which induces premature initial support yielding. This disrupts load transfer, ultimately causing excessive secondary lining damage. Critically, water-rich environments accelerate this degradation by continuously weakening rock parameters and prolonging creep duration.

Comparisons with studies by Hong et al. [60] indicate that the failure sequence observed in this study is consistent with tunnels in similarly challenging hydrogeological conditions, though the role of water in accelerating interfacial degradation is more pronounced in red-bed formations.

4.4. Mechanical Behavior and Microstructural Considerations

The observed asymmetric deformations in steel arches are predominantly attributable to bending moments induced by non-uniform surrounding rock pressure. Spatial heterogeneity in stress distribution, particularly at the Arch Springing and Arch Waist, leads to differential yielding and localized buckling, confirming that bending components play a critical role in the structural response. Such asymmetry is further exacerbated by the viscoelastic behavior of the water-saturated rock mass, which promotes uneven load transfer and stress concentration around the tunnel periphery.

Regarding the potential influence of preliminary cold deformation on creep resistance, it is noted that while cold working can enhance yield strength through strain hardening, its efficacy in reducing creep rates in support structures remains limited under sustained hydro-mechanical coupling. The predominant mechanism of time-dependent deformation in water-rich red-bed soft rock is governed by pore water pressure and clay mineral hydration, rather than metallurgical processing of steel members. Thus, material-level interventions such as cold deformation may offer negligible improvement in long-term creep performance under such environmental conditions.

Microstructural degradation of the rock mass is highly influenced by mineralogical composition and diagenetic history, rather than crystalline system alone. Neogene cement-rich sandstones, characterized by high clay content and weak cementation, exhibit pronounced sensitivity to water-induced softening and disintegration. In contrast, more crystalline or well-cemented rock types may demonstrate greater resistance to hydro-mechanical degradation. This suggests that rock type—specifically the abundance of hydrophilic minerals and the degree of cementation—plays a decisive role in microstructural evolution and long-term stability, underscoring the need for lithology-specific support design strategies in tunneling through heterogeneous formations.

4.5. Limitations and Future Research Directions

While this study provides comprehensive insights into the time-dependent deformation and support failure mechanisms in water-rich red-bed soft rock tunnels, several limitations should be acknowledged. The findings are primarily derived from a single case study of the Xijialiang Tunnel, which may limit the generalizability to other geological settings with varying mineral compositions or hydrogeological conditions. Furthermore, the wireless monitoring system, though advanced, was deployed at a limited number of sections, which may not fully capture the spatial heterogeneity of rock-support interactions along the entire tunnel alignment. The laboratory tests and numerical simulations, while informative, did not fully incorporate the effects of temperature variations or chemical dissolution processes, which could further influence long-term rock behavior. Future research should focus on developing a fully coupled thermo-hydro-mechanical-chemical (THMC) model to better simulate the complex interactions in water-rich soft rock environments. Additionally, long-term monitoring of multiple tunnel sections under varying construction and hydrological conditions is recommended to validate and refine the proposed failure criteria and support design strategies.

Furthermore, this study primarily relied on a comprehensive field monitoring campaign, which can be regarded as a large-scale natural experiment providing invaluable real-world data on the complex hydro-mechanical coupling phenomena. However, the inherent limitations of field studies, such as uncontrollable environmental variables and the difficulty in isolating single factors, necessitate complementary indoor simulation experiments for mechanistic verification and parametric analysis. Future work will involve designing sophisticated laboratory tests, including triaxial creep tests on saturated red-bed soft rock specimens under different pore water pressures and stress paths, as well as physical model tests of the support system. These controlled experiments aim to replicate the time-dependent deformation and support failure patterns observed in the field (e.g., the four-stage deformation law and the synergistic failure sequence). By comparing the results from these indoor simulations with the field monitoring data, the proposed mechanisms can be rigorously validated, and key parameters (such as the critical deformation rate threshold) can be further calibrated. This integrated approach, combining large-scale natural observation with reduced-scale physical simulation, will significantly enhance the robustness and generalizability of the findings, ultimately leading to more reliable predictive models and design guidelines for tunnels in water-rich red-bed soft rock.

5. Conclusions

This study, based on the Xijialiang water-rich red-bed Soft Rock Tunnel Project, employed an integrated methodology combining wireless remote monitoring, laboratory testing, and theoretical analysis. The investigation systematically elucidated the stress-deformation evolution mechanisms of the surrounding rock and the response characteristics of the support structure under water-rock coupling, leading to the proposal of key control measures. The principal conclusions are summarized as follows:

• In response to the first research objective, this study elucidated the four-stage time-dependent deformation and strength deterioration mechanism of the surrounding rock in water-rich red-bed strata under hydro-mechanical coupling. The deformation process exhibited a distinct four-stage evolution: sharp deformation, slow deformation, relative stability, and creep. The creep stage, governed primarily by pore-water pressure, accounted for over 40% of the total displacement. Groundwater infiltration induced hydration and expansion of clay minerals—notably montmorillonite and illite—leading to structural disintegration and the development of a loosened zone with a thickness ranging from 2.6 to 3.3 m. This process resulted in nonlinear deterioration of mechanical properties, including a reduction in saturated uniaxial compressive strength exceeding 45% and a decrease in elastic modulus surpassing 60%, accompanied by stress redistribution with concentrations notably at the arch shoulders and arch waist. The interaction between water and rock fundamentally drives the progressive degradation and sustained creep behavior of the surrounding rock.
• Addressing the second objective, the stress evolution and synergistic responses of support structures under water-rich conditions were quantitatively characterized. The rock bolt experienced severe degradation of interfacial bond strength, resulting in disordered axial force fluctuations and localized tensile failure. Traditional rigid rock bolts, with ultimate elongation capacities below 200 mm, proved ineffective for reinforcement mechanisms such as suspension or composite beam formation. This failure stemmed from insufficient embedment length, typically less than the loosened zone thickness, and an inability to accommodate large creep deformations. In contrast, constant-resistance large-deformation rock bolts, offering elongation capacities of 1000 mm or greater, demonstrated superior adaptability. Steel arches, subjected to elevated creep stresses, typically entered a state of plastic accumulation within 7–15 days post-installation. Stress distribution exhibited significant spatial heterogeneity: the Arch Springing and Arch Waist regions yielded early due to rapid stress escalation, while the Crown experienced delayed stress growth and eventual yielding. This was exacerbated by a “water-sealing effect,” where premature shotcrete closure obstructed drainage, increasing rock bulk density by 10–15%. The secondary lining primarily resisted deformation pressure transmitted from the initial support. Its contact pressure evolution was phased, peaking during formwork removal and gradually recovering under sustained rock creep, stabilizing approximately 20 days later. Concrete strain developed in an oblique ‘L’ shape under overall compression, though asymmetric deformation induced local tensile stresses, notably at the Invert. Furthermore, a deformation-rate-threshold-based criterion for support collaborative failure was established, providing a quantitative basis for early warning in engineering practice.
• Corresponding to the third objective, the progressive failure pathway of the support system was revealed, following a “local breakthrough–chain-reaction propagation–overall destabilization” sequence. Initial local failures, such as rock bolt fracture or steel arch yielding, induce stress redistribution that overloads adjacent components and ultimately leads to global instability. Furthermore, the spatiotemporal evolution characteristics of rock mass deformation rates offer quantitative criteria for predicting support failure, providing critical indicators for early warning in engineering practice.
• This research provides new insights into the multi-field coupling mechanism (water-rock-support) governing deformation and failure in water-rich red-bed tunnels. The revealed deformation patterns, failure pathways, quantitative failure criteria, and optimized support strategies provide a valuable reference and important basis for the safe and economical construction of deep-buried tunnels in similar geologies.