Tunnel construction in karst strata with abundant water causes changes in the surrounding groundwater environment, which can easily trigger geological disasters such as mud and water inrush. How to accurately predict the groundwater ahead of tunnel excavation face is a highly challenging problem. In order to improve the detection accuracy of groundwater during the construction of a deep buried tunnel, the transmission and reflection process of seismic waves at the interface and the relationship between the reflection coefficient and seismic wave signal have been analyzed on the basis of a two-phase medium theory in this paper. The expression of seismic wave stress–response relationship associated with the instantaneous amplitude and instantaneous phase and frequency is established. Then the relationship between seismic wave attributes and groundwater seepage potential energy is derived by combining the fluid mechanics theory, which is used as the basis for the determination and identification of groundwater volume and classified, and a new technology of an advanced detection of groundwater by seismic waves is established. This method has been applied to the Zhanghuai Railway in China and quantitatively predicted the karst water and caves in the Tianqiaoshan Tunnel before excavation. The engineering test proves the reliability and advancement of this technology.
The relationship between seismic wave information and surrounding rock stress change is established.
The relationship between groundwater seepage and disturbance stress change variable of surrounding rock under the action of seismic wave is proposed.
The method of fine detection of groundwater by artificial seismic waves is proposed and verified.
The disturbance of newly constructed underground projects in water-rich areas has changed the equilibrium state of groundwater and caused soil deformation, leading to major geological disasters such as ground collapse, settlement deformation, mud, and water inrush. Therefore, groundwater prediction has always been a technical problem of concern to engineers (Band et al. 2021). Since 1960, many hydraulic engineers have carried out a lot of research on groundwater level prediction, and made much progress in numerical modeling, prediction analysis, or real case studies on related subject applications, forming different groundwater level prediction methods of physics, statistics, artificial intelligence (AI), and hybrid methods (Post & von Asmuth 2013; Fathian et al. 2016; Dalkilic & Gharehbaghi 2021). In recent years, with the increase in the use of deep learning models, the Gated Cycle Unit (GRU) and long short-term memory (LSTM) neural networks are used as temporary frameworks, providing a huge opportunity for the improvement of groundwater level prediction methods (Shen 2018). Combining this model with different data preprocessing models such as singular spectrum analysis (SSA), variational mode decomposition (VMD), empirical mode decomposition (EMD), and wavelet decomposition, and establishing a numerical analysis mixture model can effectively improve the accuracy of groundwater level prediction (Azari et al. 2021; Bahmani & Ouarda 2021; Wu et al. 2021). The groundwater prediction method based on intelligent analysis has been widely developed and applied in the field.
The difference between karst strata and other strata lies in the presence of karst caves and their different sizes and sources. Due to the influence of karst development, groundwater in karst strata has the characteristics of concentration into cavities and contingency. The disturbance of groundwater during tunnel construction is highly likely to induce mud and water inrush. Unlike predicting groundwater levels in other formations, predicting groundwater levels in karst formations is more complex and challenging for safe tunnel construction. Although Zhai et al. (2021) and Huang et al. (2021) analyzed the dynamic response of building structures under regular wave action through numerical simulation and model experiments, and obtained the relationship between pressure increase and liquid flow velocity, the effect of using numerical technology to predict groundwater during the construction process of karst tunnels is not as good as actual technology due to the complex and varying geological conditions. Therefore, in the process of tunnel construction, civil engineers have developed a variety of groundwater detection technologies.
The infrared water detection technology has simple principles and flexible operation, which was first applied to predict groundwater environmental changes in tunnel engineering (Wang et al. 2003a, 2003b; Wang et al. 2020). But the humidity and magnetic field inside the cave reduce the detection accuracy. On the basis of a detailed analysis of the principle of infrared water detection, although many engineers (Alimoradi et al. 2008; Lu & Chen 2010; Zhai et al. 2017) conducted on-site tests in multiple newly built railway tunnels, studied the causes of interference in infrared water detection technology, improved anti-interference methods, and improved the accuracy of infrared water detection through joint application with technologies such as geological radar and TSP, its application customers are still very few due to its detection distance within 10 m.
The transient electromagnetic method, as an electromagnetic induction method, has played an important role in the application of ground exploration of groundwater. It has been promoted and applied in the construction process of underground tunnels and has achieved some results since 2008 (Xue & Li 2008; Li et al. 2018b). One major advantage of transient electromagnetic observation of magnetic field response is the ability to conduct non-contact measurements, which makes it very convenient and fast to conduct three-component observations. Therefore, the three-dimensional (3D) forward modeling problem of transient electromagnetic method has gradually become a hot research topic for scholars. There are two main methods for 3D forward modeling of transient electromagnetic fields. One is to solve directly in the time-domain, and the algorithms used mainly include the finite-difference time-domain method (Sun et al. 2013, 2021), the time-domain finite element method (Um et al. 2010; Li et al. 2018a, 2018b), the integral equation method (Gunderson et al. 1986), and so on. The other is to first use the numerical simulation method to solve the frequency domain or Laplace domain, and then use the Fourier inverse transform or Laplace inverse transform to convert the results to the time-domain (Li et al. 2017). At present, the transient electromagnetic method for detecting groundwater using 3D inversion is widely used during tunnel construction.
The transient electromagnetic method mainly infers the existence of groundwater based on the electromagnetic response law of the geological environment. Its shortcomings mainly lie in the significant electromagnetic influence of the testing site, which cannot accurately determine the amount of groundwater. Affected by the environment inside the tunnel, the detection distance of transient electromagnetic method is generally 30–50 m.
In order to reduce the threat of groundwater to the safety of the project, many scholars have carried out a lot of theoretical and technical application research using resistivity methods in recent years. Li et al. (2017, 2018a, 2018b) from Shandong University, China, carried out theoretical and technical research on advanced water exploration using the resistivity method in combination with several tunnel projects in China. The inequality constraint representing the variation range of model parameters is introduced into the least squares linear inversion method as a priori information. This effectively improves the accuracy of inversion results, reduces the multiplicity of inversion problems, and establishes a technical system for 3D imaging and water volume estimation for advanced detection of water-bearing geological structures in tunnels. Cheng et al. (2020) established an ellipse expansion method for tunnel spatial observation that was based on a non-standard ellipse equation, realizing the identification and positioning of adverse geology in front of tunneling. A new method of tunnel seismic wave advanced detection imaging was developed based on ellipse expansion common reflection point superposition. Li et al. (2018b) applied the beamforming method to the inverse time migration imaging process and proposed a scanning stacking mode inverse time migration method and its calculation scheme, which aided in the refinement and quantification of seismic wave detection technology. On this basis, a pioneering breakthrough has been made in the quantitative detection of aquifer, and the quantitative identification with the detection distance of 1.5 times of the tunnel diameter (about 30 m) has been realized. The resistivity method is known as the 3D excitation method and become a necessary means of detecting groundwater in TBM construction tunnels.
The seismic wave method is widely used in petroleum exploration, mining exploration, and other fields due to its long distance and mature technology. Çakır & Coşkun (2022) cooperatively worked to image the near surface (<40-m) anomaly structures, i.e., cavity and ore body by the single-station Rayleigh surface wave group velocities and electrical resistivities are two data sets. The cavity is characterized by low seismic velocity and high electrical resistivity, while the ore body is characterized by high seismic velocity and low electrical resistivity. Levshin et al. (2018) and Çaklr (2019) consider fundamental mode group velocity dispersion of Rayleigh surface waves and electrical resistivities to solve the underground structural heterogeneities. They assume that the observed Rayleigh surface waves are obtained from a traditional common-shot gather, i.e., pattern of single-shot and linear geophone array. In deep karst soil layers, major mud and water inrush accidents occur from time to time due to insufficient safety distance between the tunnel excavation surface and the location of hidden groundwater. In order to improve the prediction distance of groundwater, Lou et al. (2018) and Man et al. (2020) have developed the 3D seismic wave geological prediction method and the prediction method by analyzing the propagation law of seismic wave in water-bearing stratum based on the seismic wave prediction of geological structures, which increased the prediction distance to 100–200 m. Lou et al. (2020) and Sun et al. (2020) have studied the detection direction of 3D seismic wave. On the basis of considering the characteristics of tunnel excavation, an efficient directional 3D seismic wave geological prediction method that can meet the requirements of tunnel construction space and excavation process is established which includes the observation system of directional detection, seismic wave excitation method, and data acquisition program. Some scholars (Kenda et al. 2020; Guleria et al. 2023) have achieved certain results in data statistical analysis, intelligent modeling, and other aspects, providing new analytical ideas for accurate prediction of groundwater.
It can be seen that most of the research work by mathematicians mainly focuses on how to improve the acquisition accuracy and interpretation technology of existing geophysical methods. Although great progress has been made in the prediction of geological structure and geological lithology, the accurate prediction of groundwater is still underexplored. Against this backdrop, this paper regards the geological body in front of the tunnel as a two-phase medium, investigates the changes in the stress field in the surrounding rock of the tunnel caused by the propagation of seismic waves, and then realizes the purpose of remote water exploration. Field tests were carried out, and promising results were achieved.
Constitutive relation of the wave theory
Relationship between surrounding rock stress and seismic wave
Here, denotes the stress–response, is the wave propagation time, represents the instantaneous amplitude, F(t) gives the instantaneous phase, and dF/dt represents the instantaneous frequency. The instantaneous amplitude is a measure of the intensity reflected that determines the change in the special rock stratum. Whenever there is a fault or fracture zone in the rock, the instantaneous amplitude changes evidently. The instantaneous phase is conducive to enhancing the co-phasal axes of reflection and can be implemented to display discontinuous faults. The instantaneous phase can be adopted as an index for identifying water-bearing structures as they cause phase changes. The instantaneous frequency is the change rate of the instantaneous phase, which reflects the change of different lithologies. and represent the averages of the instantaneous amplitude and instantaneous frequency of the whole reflecting layer, respectively. The stress–response characterizes the stress changes induced by the characteristics of the surrounding rocks in the process of seismic wave propagation which can reasonably assess the structural changes and unfavorable geological conditions inside the surrounding rocks.
Groundwater identification method based on the stress of surrounding rocks
Here, represents the trend in the flow of groundwater under the influence of fluctuating stress. The size of is related to the volume of groundwater when the surrounding rock conditions remain constant. The larger the volume of groundwater, the greater would be the flow trend under the influence of the fluctuating stress. Therefore, the combination of the stress–response of the surrounding rock and the wave velocity could identify the location and scale of groundwater.
3D SEISMIC WAVE DETECTION METHOD
3D observation system
Materials and methods
The loose rocks on the tunnel face were cleared, and a flat layout range of the observation system on the tunnel face was selected. The receiving point and shot point positions within the range with red paint were marked as shown in Figure 2.
A drill bit with a diameter of 50 mm was selected and drilled at the receiving point to form a geophone installation hole with an aperture of 55 mm and a depth of 50 cm.
The geophone in good condition was inserted into the hole smoothly. It is required that the geophone is perpendicular to the tunnel face and in close contact with the hole wall. It should be assured that the geophone cannot be loosened during the whole test process.
Multiple geophones were connected in series with the seismic wave collector, and the collector was adjusted to an appropriate position.
According to the order of geophones, the seismic wave was excited at the shot point position in the observation system by hammering to make it propagate to the surrounding rock in front of the tunnel excavation, and the reflected wave was collected by the receiver.
After confirming that the collected reflected wave information meets the test requirements, the data were saved and copied to the data processing computer.
A 3D observation system is established, considering the actual situation in the tunnel. The instantaneous amplitude and frequency of reflection in the same position are analyzed and extracted by professional software by using the reflection information of artificial seismic waves collected by the seismometer, thereby obtaining the stress–response of the surrounding rocks in front of the excavation face. The generalized relative flow velocity (trend) of liquid is calculated using Darcy's equation. According to the relative flow trend, the seismic wave reflection information (wave velocity), and stress–response of the rock, the geological conditions predicted can be quantitatively divided into six grades, namely the waterless state, water seeping, dripping, flowing in lines, streaming, and gushing.
To verify the reliability and practicability of the advanced water prediction technology based on the stress–response of surrounding rocks, a series of field tests and studies were made on the Tianqiaoshan Tunnel of Zhangjiajie-Jishou-Huaihua (Zhang-Ji-Huai) Railway. The test instruments included an SGD-SMT three-component seismometer and an SGD-SMH seismic wave recorder manufactured in Russia. A 10-pound hammer was selected as the artificial seismic source.
The newly constructed Zhang-Ji-Huai Railway is located on the north-and-south belt of Xiangxi, west of Hunan Province. It starts from Zhangjiajie City in the north, passes through Jishou City in Xiangxi Prefecture, and connects Huaihua City in the south. It connects the Qianjiang-Changde Railway, Jiaozuo-Liuzhou Railway, Chongqing-Huaihua Railway, Shanghai-Kunming High-speed Railway, Huaihua-Hengyang Railway, Chongqing-Hunan High-speed Railway, and the Tongren-Jishou High-speed Railway on its route. It covers major cities and prefectures from the western Hunan Province, passing through natural scenic spots such as Furong Town, Qinzhou Ancient City, and the Ancient Town of Fenghuang. The path along the railway consists of complex geology with developed karst and abundant groundwater, making tunnel construction difficult. Tianqiaoshan Tunnel is located at Morong Town, Guzhang County, which lies between Furong Town and the ancient city of Quanzhou. It starts from a mileage of DK104 + 231 and ends at DK111 + 139 having a total length of 6,908 m and a maximum buried depth of approximately 450 m, respectively. The tunnel area landforms consist of denuded low mountains located at the southeast foot of the middle section of the Wuling Mountains and have a humid subtropical monsoon climate. The tunnel site is located in the Xuefeng structural belt on the southeast margin of the Yangtze Block in the Qiangtang-Yangtze-South China plate. The exposed strata consist of the Quaternary System (Q), banded limestone of the Lower Cambrian Qingxudong Formation (€1q), the banded limestone interbedded with the mud of the Lower Cambrian Balang Formation (€1p), carbonaceous shale of the Lower Cambrian Niutitang Formation (€1n), limestone of the Upper Sinian Dengying Formation (Zbdn), dolomite of the Upper Sinian Doushantuo Formation (Zbd), and the pebbly sandstone of the Lower Sinian Nantuo Formation (Zan). The developed karst and abundant groundwater are present in this area, with the main geological hazards of karst, karst water, and shale gas.
Forecast example for DK107 + 037–DK106 + 937
On 2 June, 2019, a 3D seismic wave detection field test was carried out for the heading tunnel face at a mileage of DK107 + 037 on the exit side. According to the design principle of the 3D observation system combined with the actual on-site situation, the seismic source points and seismometers were arranged on the tunnel face. Eight seismometers were designed, which were divided into two rows, consisting of four in each row. The spacing between adjacent seismometers was set to 1.5–2.0 m, and the spacing between the two rows was 2.0 m. Ten artificial seismic source points were arranged on the tunnel face, which was evenly distributed on the left and right sides of each seismometer. The field test lasted for 18 min.
Forecast example of DK106 + 633–DK106 + 533
On 4 December, 2019, 3D field seismic wave detection was carried out for the tunnel face having a mileage of DK106 + 633 at the tunnel exit side. Based on the design principle of the 3D observation system and the actual on-site situation, the source points and seismometers were arranged on the face of the tunnel as well. Six seismometers were designed, which were divided into two rows, consisting of three in each row. The spacing between the adjacent seismometers was 2.0–2.5 m, and the spacing between rows was 2.0 m. Eight artificial source points were arranged on the face of the tunnel, distributed equidistantly on the left and right sides of each seismometer. The field test lasted for 15 min.
DISCUSSION AND CONCLUSION
In view of the unique setting of the tunnel, based on the seismic wave theory of two-phase media, through the establishment of wave equation and mathematical derivation of the reflection interface, we obtained the mathematical expression of the stress–response of surrounding rock related to seismic wave information (instantaneous amplitude and instantaneous frequency), which can highlight the mechanical process of seismic wave propagation in different geological interfaces. Although some assumptions were selected for the reflection coefficient of seismic waves in the theoretical derivation process, we first proposed a new idea to analyze the properties of the propagation medium by studying the mechanical process of seismic waves at the reflection interface. Compared with the traditional methods (Çakır 2021; Liu & Cao 2022) of identifying seismic wave propagation medium changes by seismic wave velocity changes, the prediction results of the new method are more precise and accurate. It has a strong reference significance for the research of seismic wave fine detection.
Figures 4(c) and 8(c) show the stress gradient maps obtained after data processing. As shown in the figure, the stress gradient can characterize the integrity of the detected strata and rock masses. In this paper, the stress gradient is divided into 10 levels to describe the integrity of the rock mass, which characterizes the pore size present in the detected strata and the degree of fragmentation of the rock mass. In Figures 4(c) and 8(c), the porosity of the strata with a grade <4 is relatively small, and the integrity of the rock mass is good. The porosity of strata with a grade >6 is relatively high, and the rock mass is relatively fragmented. The pores in a formation with a level equal to 10 have reached their extreme value, indicating the presence of cavities.
Based on the mechanical process of seismic wave propagation at the interface, the relationship between groundwater seepage and seismic wave disturbance stress variation variables is established and a new groundwater prediction theory based on 3D seismic wave stress–response which related to the instantaneous amplitude and instantaneous frequency of seismic wave is proposed in this paper, and that can fully and completely reflect subtle changes of groundwater on seismic wave signals. The new method for groundwater prediction in this article sets the variation range of the groundwater flow trend parameter between 0 and 100, indicating the amount of groundwater. From Figure 4(d), it can be seen that when is less than 60, the water volume is large and there is water gushing in the formation, which affects the safety of tunnel construction. As shown in Figure 8(d), when is less than 40, the amount of water in the formation is small or anhydrous, which will not affect the construction safety of the tunnel. Compared with the methods of predicting formation water level (Band et al. 2021; Lin et al. 2022), electromagnetic method (Xue et al. 2021; Tang et al. 2022), and resistivity method, the method proposed in this paper can predict the location and size of groundwater within the range of 0–200 m, and has the advantages of long detection distance, strong reliability, and strong practicality.
The new 3D seismic wave detection method proposed in this paper is applicable to the axial detection along the tunnel during its construction in the deep stratum. The observation system of new method is ingeniously arranged in the excavation face and consists of multiple shot points and multiple survey lines. The obtained seismic wave signal has 3D characteristics and directionality, providing reliable seismic wave data for precise prediction of groundwater location and size. The observation systems of the existing seismic wave prediction methods (Zhang & Luo 2012; Dhang 2019) are all arranged on the side wall after tunnel excavation, and are composed of multiple shot points and two measuring points. The seismic wave data collected by them are two-dimensional data with many interference signals and the acquisition signal point is far away from the detection target, so they cannot predict the spatial position of the target, and the prediction results are prone to large errors.
On the basis of two-phase medium theory and fluid mechanics theory, the relationship between stress–response and fluid flow trend is preliminarily established, which makes it possible for seismic wave to predict groundwater. However, it is limited to obtain field test data, and more data samples need to be added to the regression relationship between seismic wave-induced stress–response and groundwater volume, so as to further improve the accuracy of groundwater prediction. In addition, the relationship between seismic wave-induced stress–response and surrounding rock integrity and the mechanism of its impact on groundwater potential energy change need to be further studied, which will be the focus of the next research.
This work was supported by the National Natural Science Foundation of China (Grant No. 41804145) and Key R&D projects of Hebei Province (Grant No. 19274207D) and Science and Technology Research Project of Colleges and Universities in Hebei Province (Grant No. BJ2021036).
DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.
CONFLICT OF INTEREST
The authors declare there is no conflict.