Stability of a Rock Tunnel Passing through Talus-Like Formations: A Case Study in Southwestern China

Tayi tunnel is one of the component tunnels in the Jian-Ge-Yuan Highway Project located in Yunnan Province, southeast of China. It mainly passes through talus-like formations comprised of rock blocks of diverse sizes and weak interlayers with clayey soils with diﬀerent fractions. Such a special composition leads to the loose and fractured structure of talus-like formations, which is highly sensitive to the excavation perturbation. Therefore, Tayi tunnel has become the controlled pot of the whole highway project as the construction speed has to be slowed down to reduce the deformation of surrounding talus-like rock mass. To better understand the tunnel-induced ground response and the interaction between the surrounding rock mass and tunnel lining, a comprehensive in situ monitoring program was set up. The in situ monitoring contents included the surrounding rock pressure on the primary lining, the primary lining deformation, and the stress of steel arches. Based on the monitoring data, the temporal and the long-term spatial characteristics of mechanical behavior of surrounding rock mass and lining structure due to the excavation process were analyzed and discussed. It is found that the excavation of lower benches released the surrounding rock pressure around upper benches, resulting in the decrease of the surrounding rock pressure on the primary lining and the stress of steel arches. In addition, the monitoring data revealed that the primary lining sustained bias pressure from the surrounding rock mass, which thereby caused unsymmetrical deformation of the primary lining, in accordance with the monitored displacement data. A dynamically adaptive support system was implemented to strengthen the bearing capacity of the lining system especially in the region of an extremely weak rock mass. After such treatment, the deformation of the primary lining has been well controlled and the construction speed has been considerably enhanced.


Introduction
Southwestern China has a complex terrain and complicated geological conditions with crisscrossed mountains and rivers, inducing complex geotechnical formations such as round gravel and mudstone mixture [1], layered phyllite strata [2], and rock-soil mixtures [3]. e talus-like rock masses, widely distributed in Yunnan Province in western China, are a special kind of geotechnical mixture, which is distinguishingly different from the common rocks, soils, or rock-soil mixture. e talus-like rock masses are mainly distributed in the elurium, colluvium, and diluvial layers of Quaternary System. e main components of the talus-like rock masses are rock blocks with diverse sizes with interlayers filled with weak rock or clayey soils with different fractions, which can be regarded as special weak rock masses. Different from the traditional rock-soil mixture, the talus-like formations are a special kind of geotechnical mixture and have very complicated compositions with a wide range of grain diameters. e recent studies about geotechnical mixture are mainly focused on the traditional talus development [4,5] and the rock-soil mixture [6][7][8][9][10] including in situ survey and laboratory test. Coli et al. [11] carried out an in situ shear test on shale-limestone chaotic complex bimrock and calculated its strength parameters based on a limit equilibrium analysis. Afterwards, the possible correlations between the scarce direct information from the in situ shear test data and the large indirect information from the synthetic image parameters were presented based on geostatistics [12]. Xu et al. [13] obtained the rock proportion and the granular distribution of soil-rock mixture using a combined method with large-scale direct shear test and digital image analysis, which was then implemented to investigate the mixture's shear strength characteristics and failure mechanism. Kalender et al. [14] developed a preliminary empirical strength criterion for determining the strength parameters of an unwelded rock-soil mixture by considering the mechanical behavior of the boundaries between matrix and blocks based on in situ and lab tests data. Afifipour and Moarefvand [15] conducted a series of uniaxial compression tests on artificial bimrocks models with high rock block proportions to investigate the bimrocks' mechanical behavior. In addition to the physical tests, the numerical simulation is also a mighty approach to investigate the mechanical and failure mechanisms of rocksoil mixture. Tsesarsky et al. [16] presented the influences of volumetric block proportion, block shape, and block orientation on the elastic moduli using finite element simulation. Xu et al. [17] developed a new multi-circle representation method of random polygonal blocks to generate the mesostructure model of a rock-soil mixture and implemented the model into simulating the mechanical behaviors of rock-soil mixture. Meng et al. [18] presented a numerical study on the elastic property of a rock-soil mixture using random mesostructure generation and found that the elastic modulus decreases gradually with an increase in the model size. Khorasani et al. [19] derived an equation linking the safety factor of bimslope (a slope formed by bimrocks) with volumetric block proportion using physical tests and corresponding numerical simulations, making it possible to evaluate the range of safety factors based on volumetric block proportion. Although there are series of studies on rock-soil mixtures and talus formations, there are still no guides or criteria especially for the talus-like formations, which brings uncertainty and hidden danger to the supporting structure design of tunnels and slopes excavated in such type of rock mass. is paper investigates the stability of a tunnel passing through talus-like rock masses via site survey and field monitoring. Firstly, the paper describes the basic information about the project case. In the project, the in situ monitoring items included the pressure on the primary lining from the surrounding rock mass, the stress of steel arches, and the deformation of the primary lining. Afterwards, the influence of construction activities on the taluslike ground and the stability characteristics of the tunnel in talus-like rock masses are analyzed based on in situ monitoring. Finally, a dynamically adaptive support system is presented based on the mechanical characteristics of taluslike formations.

Project Overview.
e Jian-Ge-Yuan Highway Project is located in the southern Yunnan Province, linking up the Jianshui County, Gejiu City, and Yuanyang County, as shown in Figure 1. e length of the project is about 124.5 km, with 100 bridges and 29 tunnels. e length of bridges and tunnels occupies 72% of the total length of the project. Tayi tunnel, one of the component tunnels in the Jian-Ge-Yuan Highway Project, is located at the junction of Jianshui County and Yuanyang County, as shown in Figure 1(b). ere are two lines in the tunnel: the left line is 2616 m long (from Z5K62 + 502 to Z5K65 + 118) and the right line is 2593 m long (from K62 + 489 to K65 + 082). e buried depth mainly ranges from 100 m to 250 m with a maximum depth of 297 m.

Geological Conditions. Tayi tunnel passes across Red
River fault zone, resulting in developed secondary structures and complex lithology due to the rupturing movement. e tunnel mainly passes through moderately to strongly weathered slate mixed up with siltstone, limestone, mudstone, and soil. Dense fractures and minor structures in the rock mass are heavily developed with sharply steep dip angles and disordered dip directions. e development of weak bands with irregular distribution brings up fragmented and loose rock masses with poor integrity, inducing a low ability of self-stabilization of the excavated tunnel.
Up to now, 10 boreholes have been conducted to explore the geological conditions along the tunnel route, the detailed information of which is listed in Table 1. e encountered rock mass was evaluated according to the Chinese standard for engineering classification of rock masses [20] and code for design of road tunnels [21]. In general, the rock mass can be divided into different categories ranging from grade I to grade V based on the hardness degree, intactness index, and the quantitatively basic quality index BQ. e BQ value is obtained as where R C is the uniaxial compressive strength of the rock mass and K V is the intactness index of the rock mass. e higher the grade is, the more fragmented and weathered the rock mass will be. In Tayi tunnel, grade IV rock mass accounts for about 30%, and grade V covers about 70% of the route. e revealed rock mass in the boreholes is shown in Figure 2, where the range of tunnel section indicates the values of depth of tunnel vault and invert, respectively. It should be noted that the given range of the tunnel section is usually larger than the actual excavation height of the tunnel ( Figure 2). e surrounding rock masses exhibit an obvious characteristic of hybrid formation composited by rock and weak interlayer filled with soil or other weaker rock as shown in Figure 2. Figure 3 gives some examples revealed during the tunnel excavation showing the complex formation. At chainage Z5K62 + 936 of the left line, the rock mass is composited by slate and weak interlayer filled with mudstone and has a highly loose and fragile structure with low strength (Figure 3(a)), which can be easily crushed by hand. Figures 3(b) and 3(c) show a type of fractured slate sandwiched by clay soil and a mixed rock mass formed by slate, limestone, and carbonaceous mudstone at Z5K63 + 966 and Z5K64 + 455, respectively. In the right line, twisted layers composited by slate and limestone, isolated rock blocks cut by weak mudstone interlayers, and limestone with carbonaceous mudstone interlayer were explored at K62 + 860, K64 + 530, and K64 + 570, as shown in e abovementioned geological mixture composited by rock and soil (or weaker rock) revealed in Tayi tunnel is quite different from the traditional rock-soil mixture (often called talus formation) as explored by previous studies [17,[22][23][24], which can be regarded as rock-in-soil talus. us, we tentatively call such a kind of rock masses composited by rock and soil as talus-like formations. ey are sensitive to the excavation perturbation, inducing that Tayi tunnel has become the controlled pot of the whole highway project as the construction speed has to be slowed down in order to reduce the deformation of surrounding talus-like ground.
e explored engineering problems in the tunnel include (1) large deformation of the primary lining (Figure 4

Construction Design and Support
System. Tayi tunnel is constructed using the bench cut method with three steps as shown in Figure 5. e heights of the three benches are 3.98 m, 3.76 m, and 2.86 m from the top-down, respectively ( Figure 5(b)). Usually, the length of the upper bench is about 8 m to ensure enough space for the operation of excavation machinery. e excavated tunnel is designed to be supported by a compound primary lining composited by mortar anchors, steel meshes, I-beam steel arches ( Figure 5(c)), and shotcrete. After the installation of steel arches of each bench, two-foot reinforcement bolts (42 mm radius and 3.5 m length) are used on both sides to strengthen the bearing capacity of steel arches ( Figure 5(a)). As shown in Figure 5(b), the spacing along the excavation direction of steel arches is 80 cm and thus the longitudinal spacing of mortar anchors is set as 80 cm, while the circumambient spacing is 120 cm. According to the longitudinal spacing of steel arches, the length of each excavation step of the upper bench is set as 1.6 m and the excavation length of middle and lower benches is set as 2.4 m, i.e., two and three times of the spacing, respectively. e secondary lining is in the form of cast-in situ reinforced concrete with 45 cm thickness, of which the reinforcement design is shown in Figure 5(d). e reserved deformation between the primary and second linings is 10 cm. Before excavation of the upper bench, grouting tremies with 42 mm radius, 3.5 mm thickness, and 4.5 m length are used to pre-support the surrounding rock masses. e longitudinal spacing of the tremies is 80 cm to match the installation of steel arches and the circumambient is 30 cm. e inclination angle to pipe the tremies is set as 40°t o ensure the splicing length between two adjacent tremies is larger than 3.9 m. It should be noted that this system support is designed for the general grade IV∼V surrounding rock masses.

Insight from the In Situ Monitoring
To investigate the mechanical characteristics of the primary lining in talus-like rock masses, in situ monitoring was established in Tayi tunnel, including the pressure on the primary lining from surrounding rock, the stress of steel arches, and the deformation of the primary lining.

Monitoring Scheme.
e deformation of the primary lining was measured by the total station. ere were 5 monitoring points in each monitoring section as shown in Figure 6(a), which were mainly located on the upper bench.
e monitoring points were established after spraying the shotcrete of the primary lining. e primary   lining deformation was carried out every 5 m along the tunneling direction and measured twice a day. 7 monitoring points are designed to measure the pressure on the primary lining from surrounding rock and the stress of steel arches as shown in Figure 6(b). e pressure on the primary lining from surrounding rock was measured by vibrating wire pressure transducers, which were located on the outer surface of steel arches (Figures 6(b) and 7). e stress of steel arches was monitored by vibrating wire stress transducers, which were located in the junction of the web and flanges of the steel I-beam as shown in Figures 6(b) and 7. ese transducers were fixed after the location of steel arches and before spraying shotcrete. Two monitoring sections in the left line, i.e., Z5K64 + 500 and Z5K64 + 540, were chosen to obtain the surrounding rock pressure and steel arch stress, which were measured once a day. e measurement items are summarized in Table 2. It should be noted that the tunnel was excavated from a larger to a smaller value of chainage, i.e., from Z5K64 + 540 to Z5K64 + 500.

Deformation of Primary Lining
(1) Influence of Construction Activities on Primary Lining. Tayi tunnel was shut down from the noon of September 19 th in 2020 to the morning of September 24 th in 2020 due to scheduled maintenance of large-scale mechanical equipment. erefore, it is reasonable to regard the duration from September 20 th to 23 rd as downtime, when the chainage of the excavation face was Z5K64 + 576.6. e monitoring data from September 11 th to 28 th were selected to show the deformation of the primary lining before, during, and after the downtime. Taking Z5K64 + 600 as an example, Figure 8 illustrates the displacements of five monitoring points (as shown in Figure 6). e diurnal displacement of the primary lining during the downtime was smaller than the data before and after the downtime, especially from September 21 st to 23 rd . Figure 9 shows the diurnal displacement of point A at the tunnel vault in chainages from Z5K64 + 610 to Z5K64 + 580, which yields the same rule with Figure 8. It can be concluded that the deformation of primary lining caused by the construction activities, including face excavation, boring, and drilling pipes, is larger than the deformation caused by surrounding rock creep in the talus-like ground.
In addition, the construction activities have more impact on the closer sections (Z5K64 + 610 to 600) than the farside sections (Z5K64 + 595 to 580).
(2) Spatial-Temporal Evolution of Primary Lining Deformation. Figure 10 shows the time-history curve of deformation of primary lining at Z5K64 + 500 and Z5K64 + 540. All the monitoring points were measured on the first day when they were located, inducing that the horizontal axis indicated the number of monitoring days and begun from 1. e time-history curves of different monitoring points yield a similar evolution rule: the diurnal displacement increased rapidly soon after the excavation of the upper bench, then dropped down, and finally tended to be a stable small value about 31 days after the excavation of the upper bench; i.e., the total displacement reached a steady value. In general, the excavation of middle and lower benches influenced the deformation of the primary lining in the form of increasing the diurnal displacement at Z5K64 + 500. However, such a  phenomenon was not distinct at Z5K64 + 540. In terms of the spatial distribution (Figure 11), the primary lining bore an uneven and asymmetrical deformation at both sections. e maximum total displacement was located at the left or right spandrel of the primary lining while the minimum value happened to the springing. Figure 12, almost all the monitoring points had positive data; i.e., the force on the primary lining from the surrounding rock is pressure. In consideration of the deformation shown in Figure 10, the surrounding rock shrunk after the excavation, which accords with the rational analysis. It should be noted that points D and G at Z5K64 + 500 had small negative values after the excavation. e reason might be that there was a small gap or void between the primary lining and the surrounding rock induced by the concrete shrinkage, which would cause a small magnitude of tension on the surface of the transducers.

Advances in Civil Engineering
(1) Temporal Evolution of Surrounding Rock Pressure. e surrounding rock pressure on the primary lining increased rapidly in the first 5 days after the excavation of the upper bench as shown in Figures 12(a) and 12(c). Afterwards, the surrounding rock pressure decreased in the subsequent several days. Once the middle bench was excavated, the surrounding rock pressure around the upper bench dropped. Meanwhile, the pressure on the primary lining of the middle bench increased. Similarly, the surrounding rock pressure on the primary lining of the upper and middle benches dropped slightly after the installation of the primary lining of the lower bench. e reason inducing the decrease is that the excavation of lower benches can release some amount of surrounding rock pressure around upper benches. In addition, the excavation of the lower bench had a relatively bigger impact on the surrounding rock pressure on the primary lining of the middle bench than the one of the upper bench. e peak value of the surrounding rock pressure on the primary lining of upper bench appeared 4∼6 days after the excavation. As shown in Figures 10(a) and 10(c), the surrounding rock pressure tended to be stable about 35 days after the excavation of the upper bench. Moreover, the steady values decreased from top to bottom, revealing that the upper part of the primary lining bears more surrounding rock pressure than the lower part.
(2) Spatial Evolution of Surrounding Rock Pressure. e spatial distribution of the surrounding rock pressure in Figures 12(b) and 12(d) had an obvious characteristic of discreteness due to coupling factors such as surrounding rock properties, excavation patterns, supporting parameters, and in situ stress level. e peak and steady values at sections Z5K64 + 500 and Z5K64 + 540 were quite different. In general, the deformation at Z5K64 + 540 was larger than the one at Z5K64 + 500 as shown in Figure 12, inducing larger peak and steady values at Z5K64 + 540. According to the site survey, the dominating formations around the two sections were both strongly weathered slate while there was a small interlayer of soft carbonaceous mudstone around Z5K64 + 540, resulting in a worse ground condition at this section. erefore, the section of Z4K64 + 540 had a larger peak and steady values in surrounding rock pressure.
Influenced by the in situ stresses and uneven excavation of the tunnel periphery, the surrounding rock pressure showed obvious asymmetry. For the primary lining of the upper bench, the pressure on the right was larger than the left in both sections. However, the pressure on the primary lining of the middle and lower benches yielded an opposite rule. e spatial distribution of the surrounding rock pressure along the perimeter of the tunnel also had an obvious characteristic of discreteness. In general, the surrounding rock pressure around the upper part was larger than the lower part.    Advances in Civil Engineering illustrates the stress of steel arches at the section of Z5K64 + 500. Similar to the rules of surrounding rock pressure, the stress of steel arches in the upper bench increased rapidly soon after the excavation, then dropped, and finally tended to be stable. e stress of steel arches at middle and lower benches was less affected by the excavation than the upper bench, which first increased and then tended to be stable. ere is no clear descent stage of the data curve of middle and lower benches. e excavation of lower benches released the surrounding rock pressure on the primary lining, inducing the decrease of the force undergone by the primary lining. is phenomenon appeared in the form of the stress decline of the steel arches as shown in Figure 13(a). e stress of steel arches tended to be stable about 40 days after the excavation of the upper bench. Similar to the surrounding rock pressure, the stress of steel arches unevenly distributed along the tunnel periphery (Figure 13(b)).

Stress of Steel
In the upper bench, the stress of steel arches at the tunnel vault was the largest and the right spandrel bore larger stress than the left spandrel. In the stable stage of the data curve, the steady stress decreased from upper benches to lower benches.

Mechanical Characteristics of Talus-Like Formations.
Generally, the mechanical characteristics of talus-like rock mass are similar to common weak rock masses [25][26][27][28][29][30] or layered rock masses [31][32][33]. However, there are still differences between the two types of rock masses. Firstly, the talus-like rock masses are more affected by the excavation activities than the common weak rock masses. With the bench cut method, the lower benches obviously release the     surrounding rock pressure around the upper benches, inducing the decline of the surrounding rock pressure on the primary lining and the stress of steel arches. After the installation of the primary lining of the lower benches, the surrounding rock pressure and the stress of steel arches increase slightly and finally tend to be stable. e increasement and speed of the stress of steel arches are larger than the ones of surrounding rock pressure. Secondly, the talus-like rock masses have more complex formations of rock blocks with wide-range sizes and soils with different fractions. erefore, it is extremely difficult to ensure the evenness of the tunnel periphery during mechanical excavation, which will induce uneven and asymmetrical contact between the surrounding rock mass and the primary lining. As a result, the surrounding rock pressure on the primary lining, the inner stress of the steel arches, and the deformation of the primary lining usually have a characteristic of discreteness and nonuniformity along the tunnel periphery. Accordingly, the evolution of the interaction between the surrounding rock and the primary lining can be concluded into three stages: (1) the deformed surrounding rock mass contacts with the primary lining and exerts sharp pressure on the primary lining, and the deformation of the surrounding rock mass is larger than the one of the primary lining; (2) the interactional pressure between the surrounding rock mass and the primary lining releases and decreases, and the deformation rate of the surrounding rock mass becomes smaller than the one of the primary lining; (3) the surrounding rock pressure tends to be stable and the deformations of the surrounding rock mass and the primary lining become coordinating.
In consideration of the complex formations and weakness of the talus-like formations, the deformation of surrounding rock masses should be emphatically disposed during the tunnel excavation. Termly advanced detection is recommended to obtain the information of rock mass properties ahead of the excavation face, such as the fragmentation degree, the lithological components, and water content. It is also suggested that the forepoling is established in time and changes momentarily if the advanced detection explores a worse rock mass. e invert part should be excavated and installed timely to enclose the primary lining into integrity to ensure the performance of the lining. Moreover, the secondary lining should be installed as soon as possible when the deformation of the primary lining tends to be stable to diminish the exposure duration of the primary lining.

Dynamically Adaptive Support System in Talus-Like
Formations. In consideration of the significant influence of the talus-like formations on the tunnel support system, a reasonable modification of the support system had been processed. Firstly, the talus-like formations, which had been classified into grade V rock mass, were divided into three subclasses according to the ground condition: (1) the extremely weak and fractured rock mass without water inflow as Vd-1; (2) the extremely weak and fractured rock mass with water inflow as Vd-2; (3) the extremely weak and fractured rock mass mixed with some amount of soil and water inflow as Vd-3. It should be noted that the item "Vd" came from the geological investigation report of Tayi tunnel and the letter "d" herein is just a notation showing that such a type of rock mass is weaker and more dangerous than grade V rock mass. Secondly, a dynamically adaptive support system was proposed to adapt to the varying conditions of talus-like formations as shown in Figure 14. e spacing of steel arches was changed to 60 cm and thus the excavation length of the upper bench was altered to 120 cm for grades Vd-1 and Vd-2 rock masses and 60 cm for grade Vd-3 rock mass, respectively. e excavation length of the middle and lower benches was set as 180 cm, i.e., three times of the steel arch spacing. In the dynamically adaptive support system, the pre-support scheme used pipe shed to prevent the deformation of the surrounding rock mass ahead of the excavation face. As for the primary lining, stronger structural elements were used to ensure that the deformed primary lining would not invade the design space of the secondary lining. Especially for grade Vd-3 rock mass, a double-layer primary lining (Figure 15) was adopted to prevent large deformation of surrounding rock mass. In addition, the reversed deformation between the linings and the secondary lining parameters were also adaptable according to the ground conditions. e detailed parameters of the dynamically adaptive support system are listed in Table 3. Every 6 m (ten times the excavation step) during the excavation, a combination method with the transient electromagnetic method, ground penetrating radar, and advance      14 Advances in Civil Engineering exploratory drilling was employed to detect the geological and hydrological conditions ahead of the excavation face to determine the subclass of the surrounding rock masses and the appropriate support parameters. As stated above, the talus-like formations are sensitive to the excavation activities and the construction speed of Tayi tunnel has to be slowed down to reduce the deformation of surrounding rock mass and prevent the rock collapse. With the original support system ( Figure 5) before August of 2020, the average excavation speed was less than 20 m per month in grade IV/V rock masses. e first-edition dynamically adaptive support system began to be used after August 2020 and the construction speed had been promoted to 30∼40 m per month in grade V rock mass. Afterwards, the dynamically adaptive support system was improved in January 2021 as shown in Figures 14 and 15. As a result, the construction speed had been enhanced to 70∼80 m per month in grade V rock mass (including the abovementioned grade Vd rock masses) and 160∼180 m per month in grade IV rock mass. e maximum deformation of the primary lining was reduced to less than ∼400 mm after the adoption of the dynamically adaptive support system.

Conclusion
is paper investigates the stability of Tayi tunnel passing through talus-like rock masses via site survey and in situ monitoring. In the project, the in situ monitoring items included the surrounding rock pressure on the primary lining, the stress of steel arches, and the deformation of the primary lining. e main conclusions are as follows.
(1) e deformation of the primary lining caused by the construction activities is larger than the deformation caused by surrounding rock creep in the talus-like ground. e construction activities have more impact on the closer sections than the farside sections. (2) With the bench cut method, the surrounding rock pressure and the stress of steel arches increase rapidly soon after the excavation of the upper bench, then slightly drop down, and finally tend to be stable. e lower benches obviously release the surrounding rock pressure around the upper benches, inducing the decline of the surrounding rock pressure on the primary lining and the stress of steel arches.
(3) In terms of spatial distribution, the monitoring data usually has a characteristic of discreteness and nonuniformity along the tunnel periphery due to the uneven and asymmetrical contact between the surrounding rock mass and the primary lining which is inherently dominated by the uneven excavation space caused by complex formations of talus-like rock masses. (4) It is suggested that the deformation of the surrounding rock masses should be emphatically disposed during the tunnel excavation in the taluslike ground. Termly advanced detection, flexible forepoling, timely closure of primary lining, and well-timed installation of secondary lining are recommended measures to reduce the deformation of the talus-like rock masses. (5) A dynamically adaptive support system was implemented to strengthen the bearing capacity of the lining system according to the geological conditions of talus-like formations. After such treatment, the deformation of the primary lining has been well controlled and the construction speed has been considerably enhanced.

Data Availability
e data used to support the findings of this study are included within the article.

Conflicts of Interest
e authors declare that they have no conflicts of interest.