Experimental Study on the Soil–Structure Responses Induced by Tunnelling in Limited Space

Abstract

As the economy develops rapidly, newly built tunnels in congested urban areas are becoming more and more common. Based on similarity theory, this study conducted a laboratory model test to investigate the soil–structure responses induced by the construction of a new tunnel in limited space which was formed by the existing underground infrastructures. The soil movement, convergence of the tunnel and distribution of the soil stress were explored. The test results revealed that the limited space could have a significant shielding effect on the soil movement. The maximum ground subsidence and the range of ground settlement induced by the construction of the tunnel in limited space were 60% and 40%, respectively, smaller than those induced by the construction of a tunnel at a “greenfield” site. The deformation of the tunnel was also restrained distinctly. Moreover, a soil settlement trough appeared below the bottom slab of the underground structure. The findings can shed some light on the surrounding responses induced by tunnelling in limited space in soft clay and can offer the valuable guidance for the similar projects.

1. Introduction

Due to the sharp growth of the economy all over the world, the demands for tunnels in large cities are greatly increasing [1,2]. Many studies have reported that metro systems could bring multiple benefits to urban development but can also induce new engineering challenges and risks [3,4]. As the shield tunnelling approach has flexibility and could cause little negative influence on the environment, this construction technique has been widely adopted in many large cities.

It is widely recognized that the construction of new tunnels in non-confined spaces could induce great soil stress release [5,6,7,8,9], which may have inevitably significant impact on the ground settlement and the tunnel. Based on many case histories, Peck (1969) [10] and Schmidt (1969) [11] first proposed an innovative pattern for ground settlement and used Gaussian shapes to represent this pattern. Subsequently, many scholars conducted long-term studies to investigate the ground settlement and tunnel deformation caused by tunnelling. Many scholars [12,13,14,15,16] used the field data to provide valuable references for the validation of the proposed prediction methods. Based on the reported findings [17,18], tunnelling-induced soil movements are radial displacements towards the tunnel cavity and longitudinal displacements towards the advancing tunnel heading. Additionally, ground loss was the key factor that affected the range and maximum ground settlement [19]. Heravi et al. (2011) [20] investigated the factors affecting ground surface settlement due to shallow tunnel excavation and indicated that the buried depth and diameter of the tunnel could have a significant impact on the distribution of ground settlement. As the tunnel could deform distinctly during the construction process, the tunnel responses were also investigated extensively. When tunnels were excavated in non-confined spaces, Xue et al. (2020) [21] pointed out that the tunnel will be compressed vertically. Through field monitoring, model tests and numerical analysis [22,23], the ground loss ratio and the shape of the tunnel could have a direct influence on the tunnel deformation and the internal forces in the linings of the tunnel. Based on the model test and the numerical analysis, many investigators [24,25] concluded that the maximum positive bending moment of the tunnel was likely to appear at the tunnel crown and the maximum negative bending moment was at the waist of the tunnel.

In actuality, as the underground spaces in urban areas are becoming more and more congested, new tunnels will be inevitably constructed near the adjacent existing underground infrastructures. Accordingly, the responses of the surrounding environment induced by close-proximity tunnelling have become a major concern of society [3,26]. Many researchers have pointed out that the presence of the existing underground structures could have a non-negligible influence on the soil movement and deformation of tunnels [27,28]. Liu et al. (2012) [29] compared numerical simulation results and in situ monitoring data and concluded that the existing metro stations and surrounding buildings would affect the ground settlement pattern and that new tunnels would also experience uneven deformation. Guo et al. (2022) [30] conducted a model test to study tunnel–pile interaction. The results showed that existing piles would affect the range of the ground settlement to some extent. Liao et al. (2009) [31] focused on the shield tunneling and environment protection in Shanghai soft ground. The results revealed that existing buildings with large stiffness could greatly affect the distribution of the soil displacement and the tunnel deformation. By conducting model tests on close-proximity tunnelling, Fu et al. (2022) [32] also verified that the presence of the existing underground infrastructures could restrain the deformation of the tunnel to some extent and that there was some difference between the measured ground settlement pattern and the conventional soil settlement trough.

Summarizing the available research, the excavation of new tunnels in congested urban areas is challenging and the existing adjacent infrastructures could pose a great threat to safe construction. However, to date, there is still a lack of studies about soil displacement and the surrounding responses induced by the excavation of tunnels in the limited space which was formed by the existing underground structures. Based on this background, this paper conducted a series of physical experiments to study the responses caused by tunnelling in the limited space which was formed by underground structures and diaphragm walls. Soil movement, changes of soil stress and tunnel convergence were deeply investigated in this study. Our findings may shed some light on the surrounding responses induced by close-proximity tunnelling and could offer some guidance to similar projects.

2. Project Description and Geological Condition

2.1. Project Overview

As one of the biggest cities in China, Shanghai ha experienced significant development in recent decades. Accordingly, the city is becoming crowded and the underground spaces in the downtown areas of Shanghai are becoming more and more congested. In order to relieve the heavy traffic pressure in Shanghai, a new tunnel was planned to be excavated near the adjacent existing underground structures due to the congested underground environment. Figure 1 illustrates the typical cross-section view of the project. The existing underground structure was constructed several years ago and served as the underpass. The top of the existing underground structure was about 2.5 m to 4.5 m below the ground surface (BGS) and the height of this underpass was 8.25 m. The existing diaphragm walls were on both sides of the underpass. The toe of the diaphragm wall extended to a depth of 28.5 m~30 m BGS. According to Figure 1, this new tunnel was planned to be constructed in the narrow space which was enveloped by the existing underground structure and the diaphragm walls. The new tunnel and the existing underground structures were vertically stacked and their overlapping length was about 6 km. Since the overlapping length between the tunnel and the underground structure was very long, the safety of the surroundings during the period of tunnelling became a major concern.

2.2. Geological Condition

Shanghai is located on the east side of China and the bedrock in Shanghai lies under thick Quaternary sediments. Hence, the main strata in Shanghai are soft silty clay and sandy clay [33]. Prior to the excavation of the tunnel, comprehensive in situ tests and laboratory experiments (e.g., cone penetration tests, boreholes, triaxial tests and oedometer tests) were conducted to investigate the surrounding and subsurface soil properties across the site. The typical soil profile along with the measured soil properties are illustrated in Figure 2. As shown in Figure 1 and Figure 2, the uppermost layer is the artificial fill (2.6 m). Next, a firm silty clay (3.4 m) lies beneath this. Under the firm silty clay, a very soft muddy clay extends to a depth of 21.8 m BGS. The fourth layer is clay and silty clay (8.5 m). The third and fourth strata contain organics, humus and sludge which could cause significant reduction of soil strength. Through comprehensive laboratory tests, these two strata were shown to feature low strength, high compressibility and large water content. As the aforementioned two strata are susceptible to disturbance, these weak soils could pose a great threat to the safety of the project. Below the clay and silty layer, a dense silty sand layer with a thickness of about 34 m was encountered. Directly beneath the dense silty sand is very dense silty sand which extends to the termination depth of in situ tests at 80 m BGS. The level of the long-term phreatic water is 0.5 to 1.7 m BGS which fluctuates slightly with seasonal changes. The confined aquifer with the piezometric head of 3.8 m BGS is located at 33–80 m BGS. Generally, the sensitive thick soft soils above the sand strata in Shanghai were characterized as poor (e.g., large void ratio, low undrained shear strength). According to the geotechnical parameters shown in Figure 2, the Shanghai soft clay will deform drastically and fail quickly once it is subjected to disturbances. In addition, the soil properties gradually become better as the depth increases. More detailed geological and hydrological conditions in Shanghai can be found in Tan et al. (2019) [34] and Tan et al. (2015) [35]. From Figure 1 and Figure 2, it can be seen that the new tunnel mainly passed through the very soft muddy clay which featured weak properties. Additionally, both the underpass and the existing diaphragm wall were located in the upper 30 m of thick soft soil. Considering the challenging field geological conditions and the congested space for tunnelling, the challenge and uncertainty of the construction was magnified significantly.

3. Model and Test Descriptions

3.1. Similarity of the Model Test

Many researchers pointed out that model tests can be an irreplaceably effective way of investigating complex engineering problems. Therefore, laboratory model tests were widely adopted to conduct investigations about tunneling in a congested environment. In order to reproduce the actual geotechnical condition and to reflect the interaction between the tunnel and the existing underground structures, the models must satisfy the main requirements (e.g., the geometry and physical–mechanical characteristics) [32,36]. Based on the similarity theory and the well-documented model test cases, the similarity ratios of the main parameters in this study can be written as follows:

$$\left\{\begin{array}{l}{C}_{\sigma }=C_{\gamma }\times C_l\\ C_{\delta }=C_{\epsilon }\times C_l\\ C_{\sigma }=C_E\times C_{\epsilon }\\ C_{\epsilon }=C_{\phi }=C_{\mu }=1\end{array}\right.$$

(1)

where C represents the similarity ratio between the model and the prototype; and $\sigma$, $\gamma$, $E$, $l$, $\delta$, $\phi$, $\epsilon$ and $\mu$ refer to the stress, gravity, elastic modulus, length, deformation, friction angle, strain and Poisson ratio, respectively.

In light of the effect of the self-weight stress field in this study, the gravity in the model test was equal to that of the field condition. Therefore, the $C_{\gamma }$ was 1. Considering the operating conditions and the size of the experimental devices, the similarity ratio of the length, $C_L$, was set as 1:30.

In order to minimize the adverse boundary effect of the model tank, the dimension of the model tank was determined as 1800 mm × 600 mm × 1800 mm. As shown in Figure 3, the model tank was made of steel. To achieve better observation of the soil movement during the test process, a high-strength glass plate was set at the front side of the model tank. Hence, the ground settlement and the soil deformation during the whole process can be monitored well by a digital image acquisition system. In general, the size of the test tank met the experimental requirements in this study.

3.2. Preparation of the Model Soil

According to Figure 1 and Figure 2, the underground structure and the tunnel were mainly buried in the soft soil. If each layer at the site was remolded in the laboratory test, the cost and the difficulty would be greatly increased. Therefore, for the convenience of experimental operability, this study simplified the experimental conditions to a certain extent to ensure the operability of the test. The main strata adopted in this model test were 1 layer of remolded clay and 1 layer of sand which were based on the typical very soft muddy clay and dense silty sand in Shanghai. Since the sand layer was located at the bottom of the model box and served as the bearing stratum, Fujian standard sand was employed as the model sand in this study. According to Yang et al. (2017) [37], barite powder, clay, bentonite, washing powder and other materials can be used to reproduce model clay. This model clay could well reflect the characteristics of Shanghai soft soil. Prior to the model test, a series of laboratory tests which mixed the artificial materials (e.g., clay, barite powder, whiting powder and bentonite) were conducted to simulate the model clay. The final mass ratio for each component, i.e., barite powder, clay, bentonite, silicone oil, washing powder and water, was 4:1:1:4.7:3.6:4.9. The typical mechanical parameters for the prototype clay and model clay are listed in

Table 1. Typical mechanical parameters for prototype clay and model clay.

Type	E (MPa)	γ (kN/m3)	φ (°)	C (kPa)	Water Content (%)
Prototype clay	4.32	17.6	18.5	33	58.7
Model clay	0.143	17.3	18.3	1.105	53.6
Similarity ratio	30.00	1.02	1.01	29.86	1.10

. The view of model clay and the model sand are illustrated in Figure 4.

3.3. Preparation of Model Tunnel and the Model Underground Structure

Given the cost and difficulty of model fabrication, aluminum alloy was finally chosen to be the material for the model underpass and model diaphragm wall. In fact, the actual connection between the underpass and the diaphragm wall was rigid. Hence, in this study, the overlapping parts between the model diaphragm wall and the underpass were welded together tightly to ensure integrity. The detailed information of the model underpass is illustrated in

Table 2. Typical parameters for prototype underpass and model underpass.

Type	Thickness of the Top Slab (m)	Thickness of the Middle Slab (m)	Thickness of the Bottom Slab (m)	Thickness of the Side Wall (m)	Material
Prototype underpass	1	1.2	1.2	0.8	reinforced concrete
Model underpass	0.01	0.015	0.015	0.009	aluminum alloy

. The thickness of the model diaphragm wall was 0.015 m. The length of the model underground structures (i.e., model underpass and diaphragm) along the advancing direction of the tunnel was 0.6 m. Moreover, the picture of the model underground structures is presented in Figure 5.

As for the model tunnel, many scholars pointed out that polyvinyl chloride (PVC) tubes can be used to simulate tunnels in model tests. However, many researchers have indicated that the connection effect from the longitudinal and circumferential joints in the segment should be taken into consideration. Hence, the grooving method was widely employed to simulate the reduction of joint stiffness induced by bolts. Since the tunnel–soil–underground structure interaction was the main research topic in this study, the model tunnel was supposed to be assembled to reflect the actual condition to a great extent. According to the findings reported by Ritter et al. (2018) [38], 3D printing technology can be an extremely effective method for revealing the detailed features of a scaled model. Based on the aforementioned advantages, 3D printing technology was adopted in this model study. The model segments consisted of 5 standard segments (67.5°) and 1 top segment (22.5°), which was in accordance with the distribution of typical tunnel segments in Shanghai.

It is well known that the compression stiffness, tension stiffness and bending stiffness of the bolts are the most important characteristics of the joints between the segments. Therefore, according to the similarity theory described in Ye et al. (2014) [39], the final number of longitudinal bolts was selected to be 12 and the number of transverse bolts was chosen to be 6. The material used for the model bolts was iron and the diameter of the model bolts was 1.2 mm.

The length of the model tunnel was 0.6 m, which was equal to the length of the model underpass and diaphragm wall. The detailed photographs of the model tunnel are illustrated in Figure 6. The detailed information of the model tunnel is listed in

Table 3. The parameters of linings of prototype tunnel and model tunnel.

Type	Outer Diameter (m)	Inner Diameter (m)	Thickness of the Lining (m)	E (MPa)	Poisson Ratio	Material
prototype tunnel	6.6	6.0	0.3	3.45e4	0.2	concrete
model tunnel	0.22	0.2	0.01	953.67	0.2	modified organic plastics
similarity ratio	30	30	30	36.2	1	/

.

To simulate the excavation process and the volume loss induced by tunneling, the drainage method was employed in this study. The outer surface of the model tunnel was wrapped with a large water bag. Two rubber hoses were connected to the water bag. The volume loss during the excavation process of the tunnel was simulated and controlled by the discharging volume of water in the water bag. The view of the water bag and the drainage-controlling device during the model test are shown in Figure 7.

The two end sides of the model tunnel were sealed with a transparent membrane, which could prevent the clay or water from entering the inside of the tunnel during the process of the test. According to the findings of Wei et al. (2010) [40], the common volume loss rate during tunneling in China varies between 0.5% and 3%. To facilitate observation of the soil movement and the deformation of the models, the volume loss in this study was selected to be 2%.

3.4. Test Schemes and Monitoring Equipment

The detailed layout of the experimental devices in the model test is depicted in Figure 8. To obtain a better understanding of the influence induced by tunnelling in limited space, a model test for the excavation of the same tunnel at a “greenfield” site [41] was also conducted.

The variation of the earth pressure during the tunnel construction was also a main research object in this study. Hence, prior to the beginning of the test, a series of micro earth pressure cells were put both inside and outside the confined space. The detailed locations of the micro earth pressure cells are illustrated in Figure 9.

The deformation of the tunnel during the excavation process was measured with an automatic tunnel convergence measuring device. The inner force of the model tunnel was measured with a high precision strain gauge. The measuring sections were selected at the rings 15, 30 and 45. Each measuring section was equipped with 8 strain gauges and the angle between the adjacent stain gauges was 45°. The locations of the strain gauges and the measuring sections of the inner force of the model tunnel are presented in Figure 10.

The ground settlements were measured with a high-accuracy displacement meter (the accuracy of the displacement meter was 0.01 mm). As the soil movement in the limited space was hard to monitor with traditional measuring instruments, this study adopted a non-contact measurement method with a digital image analyzing system (DIAS) [42,43]. The DIAS included a high-definition camera and an image-based recognition system. The photos of soil movement were taken during the whole process of the experiment. These photos were compared with the original state which was captured before the model test.

4. Results of the Model Test

The observed results in this model tests included the soil movement, changes of soil stress, tunnel convergence and distribution of bending moments of the tunnel, etc. It is worth noting that all the experimental data were converted to those of prototype [37] according to similarity ratios to make the results understandable.

4.1. Ground Settlement Induced by Tunnelling

The ground settlements induced by tunneling in the limited space and “greenfield” site are depicted in Figure 11. In Figure 11, $\eta$ represents the volume loss and the variations of $\eta$ represent the different stages of the model test. From Figure 11, it can be seen that there are some differences between the soil settlement pattern under these two circumstances. Similar to the findings reported by many scholars [10,11,14], when the new tunnel was constructed at the “greenfield” site, the ground settlement experienced a Gaussian distribution curve. However, when the tunnel was excavated in the limited space formed by an underpass and diaphragm wall (see Figure 1), a distinctive analogous flat-shaped ground settlement pattern was exhibited. When investigating the range of the ground settlement, as shown in Figure 11b, the analogous flat part of the settlement was −12 m to 12 m, which corresponded to the length of the underpass. Hence, it can be deduced that the underground structure experienced an overall subsidence. Moreover, the maximum ground settlement caused by the tunnelling in the limited space was 4.3 mm, which was 60% smaller than the result induced by tunneling at the “greenfield” site. In addition, the range of the ground settlement caused by the construction of the tunnel in the limited space was also much smaller (about 40%) than that obtained from tunnelling at the “greenfield” site. The results shown in Figure 11 indicated that the existing underground structure could restrain the ground settlement induced by the excavation of the new tunnel to some extent.

4.2. Convergence of the Tunnel

With the excavation of a tunnel, the tunnel will inevitably experience uneven deformation. Figure 12 depicts the typical pattern of tunnel convergence [44]. In Figure 12, the diameter of the tunnel at initial state is defined as D. When a tunnel is subjected to an external force, the tunnel will deform correspondingly [31,44]. The horizontal axis of the ellipse was defined as D_l while the vertical axis of the ellipse was defined as D_s. Finally, the ovality of the deformed tunnel, T, can be calculated as follows:

$$T=\frac{D_l-D_s}{D}\times 100\%$$

(2)

If T is positive, this means that the tunnel will be compressed vertically and the left and right waist section of the tunnel will bulge out horizontally. If T is negative, this means that the tunnel is compressed horizontally.

Figure 13 illustrates the convergence of the tunnel during the whole experiment process. According to Figure 13, as the T remained positive during the whole process, the tunnels were all compressed vertically and bulged out horizontally under the two circumstances. The measured pattern of the tunnel deformation was consistent with the reported research findings [44]. However, when tunneling in the limited space, the ovality of the final state of the tunnel was 0.310%, which was about 23% smaller than the result induced by the construction of the tunnel at the greenfield site.

4.3. Distribution of the Bending Moment of the Tunnel

Figure 14 demonstrates the distribution law of the bending moment of the tunnel. According to the measured results, the distributions of the bending moment under the two conditions were similar. The bending moment kept increasing with the excavation of the tunnel. The maximum positive bending moment appeared at the tunnel crown and the maximum negative bending moment appeared at the waist of the tunnel. However, in general, the bending moments of the tunnel induced by tunnelling in the limited space were 35% smaller than those obtained from the excavation of the tunnel at the greenfield site. According to the results in Figure 13 and Figure 14, when the tunnel was constructed inside the limited space, the existing underground structure could effectively restrain the deformation of the tunnel and reduce the bending moment.

4.4. Distribution of the Soil Stress

It is well known that the excavation of a new tunnel will inevitably affect the surrounding soil. Hence, the change of the soil stress can also reveal the influence zone of the tunneling to a certain extent. Figure 15 presents the layout of the micro earth pressure cell at measuring section A. Figure 16 demonstrates the variation of the soil stress during the test. From Figure 16, the soil stress inside the limited space (i.e., A-1 to A-8) kept decreasing during the whole excavation process, which was because the excavation of the tunnel could induce the stress release of the surrounding soil and make the surrounding soil move towards the tunnel. On the contrary, the soil stress at A-9 to A-12 remained steady during the whole test, which revealed that the excavation of the tunnel inside the limited space cannot affect the soil outside the limited space. According to the results listed in Figure 11, Figure 13, Figure 14 and Figure 16, it can be reasonably concluded that the existing underground structure had a significant shielding effect on the soil movement induced by the tunneling.

4.5. Soil Movement in the Limited Space

Due to the DIAS, the movement of the soil inside the limited space was recorded uninterruptedly and the transverse soil settlement profile is illustrated in Figure 17. As shown in Figure 17, a soil settlement trough occurred below the base slab of the underpass and this soil settlement trough appeared right above the central line of the tunnel. Moreover, the settlement though in the limited space exhibited an obvious Gaussian distribution pattern. The results also verified that the underpass and diaphragm experienced an overall settlement, which was in a good agreement with the results concluded from Figure 10. The width of the settlement trough in the limited space was about 9 m, and this interspace between the base slab and the soil may have a significant negative effect on the deformation of the underground structure. As a result, soil reinforcement such as advance grouting was strongly recommended for this area to ensure the safety of the project.

5. Conclusions

Based on the elaborate laboratory model tests, this study deeply investigated the soil–structural responses induced by tunnel construction in a limited space. The following main conclusions can be drawn from the test results:

(1)

The limited space in this study was defined as an underground space surrounded by an underground structure. The limited space could have a significant shielding effect on the soil movement. During the tunnelling process, the soil near the tunnel could be disturbed drastically. On the contrary, the soil outside the limited space was hardly influenced by the excavation of the tunnel.

(2)

When the excavation was completed, the measured maximum ground subsidence and the range of ground settlement were about 60% and 40%, respectively, smaller than those induced by the construction of a tunnel at the “greenfield” site. The results revealed that an existing underground structure could effectively restrain the ground settlement caused by tunnel construction. The existing underground structure may also experience an overall settlement. Moreover, in the limited space, a soil settlement trough developed with the excavation of the tunnel.

(3)

The tunnel inside the limited space was vertically compressed during the whole test. However, when the excavation was completed, the final ovality of the deformed tunnel was about 23% smaller than the results caused by tunnelling at the greenfield site. In general, the bending moments of the tunnel induced by tunnelling in the limited space were 35% smaller than those obtained from the excavation of the tunnel at the greenfield site. Therefore, the presence of the underground structure can also effectively reduce the deformation of the new tunnel.