Numerical Simulation of Construction Disturbances during Bidirectional Advancement of Undersea Large-Diameter Dual-Line Shield Tunneling

Abstract

Relying on the Mawan undersea large-diameter, dual-line, mud–water-balanced shield tunnel project and focusing on the characteristics of the tunnel, such as the complex geological conditions at the expected intersection location and the existence of a superimposed perturbation or secondary perturbation effect, theoretical calculations and three-dimensional numerical simulations were used to reveal the ground disturbance situation of the large-diameter, two-lane mud–water shield when it is propelled under various working conditions. The working conditions were set for the dynamic intersection of the left and right lines, with stopping and moving as the two modes, and a traversing simulation was carried out under three conditions related to the strata. The results show that the surface settlement curve for the two-lane construction became a “W”-shaped bimodal curve due to the superposition effect; the dynamic intersection construction greatly disturbed the ground layer and there was a plastic zone expanding outward at a small angle above the tunnel, with shear damage in the soil layer and tensile damage in the rock layer. A “one line stops, and another advances” intersection can reduce the impact of disturbance; the surface settlement value after the completion of the advancement was smaller than the dual-line intersection. The surrounding rock stress and displacement under the advancement of a single shield machine did not change to a great degree, there was no obvious change in the surface settlement above the tunnel, and the effect of the secondary disturbance was small.

1. Introduction

With the rapid development of coastal cities around the world, underwater tunnels have become an important part of cross-sea transportation. Under the soft soil or weak rock environment of the seabed, the shield method is one of the main modern tunnel-boring techniques and, increasingly, undersea shield tunnels have been built or are being planned [1]. When traversing complex strata, because of the large volume of work, high technical difficulty, and hidden nature, the construction of undersea tunnels involves risks, such as destabilization of the excavation surface, difficulty in determining the construction parameters, and unknown stratum disturbance [2,3,4,5]. This risk is further exacerbated when a tunnel’s diameter is enlarged and dual-lane advancement construction increases the degree of disturbance to the strata [6,7,8]. Therefore, it is of great significance to accurately predict the impact on the surrounding environment during undersea shield tunnel advancement and to control the disturbance effect of shield construction on the strata.

For ground disturbance caused by shield construction, the main research methods include an empirical method based on Peck’s formula, numerical analysis, model testing, and actual measurement analysis. Peck [9] proposed a formula for estimating lateral ground settlement based on the statistical results of specific engineering data. He believed that ground settlement is caused by soil loss and that the volume of soil loss is equal to that of the ground settlement trough. Later, scholars [10,11] revised the estimation based on different geological conditions and empirical data and extended the empirical formula to the calculation of ground settlement under two-lane tunnels with large-diameter shields.

Many scholars have studied the problem of tunneling through various strata using numerical simulations and model tests [12,13,14,15,16]. In terms of model studies, Kawamata [17] conducted several shaking table tests on shield tunnels passing through soft and moderately stiff strata. The results showed that the stress state of the tunnel was significantly optimized due to the design of the vibration damping layer. In a numerical simulation study, Xue et al. [18] used numerical simulation software to explore the safety impacts of different construction schemes, such as the step method, center diaphragm wall method (CD method), and cross diaphragm wall method (CRD), on the safety of undersea tunnels that cross fault fracture zones, providing theoretical guidance for on-site construction. Concerning uneven strength distributions in dangerous working conditions, Li et al. [19] used finite element software to analyze in detail the interaction between the rock seepage effect and the secondary stress field of tunnels to obtain the distribution of the tunnel plastic zone of the Qingdao undersea tunnel’s surrounding rock fragmentation. Wang et al. [20] studied the influence of the shield diameter, equivalent layer thickness, excavation surface support pressure, and other factors on the settlement of undersea tunnels from the perspective of fluid–solid coupling and verified the accuracy of the model. Dias, et al. [21] monitored two cross tunnel sections of underground work for a subway in Lyon, France. By comparison with measurements of other projects, it appeared that the face instability and the annular gap identified after the shield release were the main sources of short-term settlements. Yuan et al. [22] analyzed the vertical deformation of the soil surrounding four adjacent tunnels due to new tunnel construction in the context of a shield tunnel in soft soil in Shanghai. It was shown that the vertical displacement of the tunnels was significantly affected by the soil properties. Zeng et al. [23] focused on the endings of shield tunnels and investigated the instability and damage of end soils in loose and weak strata and, using numerical simulation, the authors found that the reinforcement effect increases with the increase in thickness of the added solid, but the change rate becomes smaller and smaller. To investigate the shutdown problem of shield tunneling machines, Hasanpour et al. [24] utilized finite-difference software to simulate the excavation process in unfavorable strata, and the results help to realistically assess the effect of unfavorable geological conditions on shield blockage.

To summarize, the current results on shield tunneling through strata are mainly related to small-section shield construction and tunneling through the strata of a single condition via single-line boring or two lines with the propulsion in the same direction. Few scholars pay attention to phased advancement and there is a lack of research on the effect of construction methods on the disturbance to the undersea shield tunnel strata. In this study, combined with the Mawan Cross-Sea Passage (Moon Bay Avenue–Cross-River Expressway) project, three kinds of traversing strata were set up for the complex environment of the expected intersection of its two-lane shield tunnels: fully and strongly weathered mixed-granite strata; medium and slightly weathered mixed-granite strata; and soft upper and hard lower strata. Combining theoretical calculations and numerical simulation software (FLAC3D 5.0), the impact of the shield structure on the strata with the two construction methods of dynamic intersection and “one stop, one movement” was investigated. By comparing and analyzing the results of the numerical modeling under each working condition, the traversing method with the smallest disturbance was obtained and the results are of reference value to the actual project.

2. Engineering Profile and Settlement Calculations with Empirical Formulas

2.1. Engineering Profile

The Mawan Cross-Sea Channel (Moon Bay Avenue–Cross-River Expressway) project is located in the Shenzhen Qianhai Co-operation Zone, Mawan Area, and Baoan District, Dachan Bay Port Area. The route starts from the Mawan Avenue and Moon Bay Avenue intersection of the Mawan Port Area in Qianhai and ends at the Baoan Dachan Bay Area Yanjiang Expressway Dachan Bay Tollbooth and the Jinwan Avenue–Xixiang Avenue intersection. The shield section plan is shown in Figure 1; the total length of the left line is 2060 m and the right line is 2063 m and it uses two 15.43 m diameter slurry Tunnel Boring Machines (slurry TBM). A cross-section of the project is shown in Figure 2. The tunnel is expected to continuously pass through a full cross-section of soft and weak strata (ooze, clay, sandy soil, and fully and strongly weathered mixed rocks), a full cross-section of rock strata (medium and slightly weathered mixed rocks), and soft upper and hard lower strata (soft and weak strata and rocks), which makes for a very complex geological environment. According to the construction plan, the left and right lines of the Mawan Shield Tunnel are advancing toward each other. Combining the construction period, organization of the construction, shield speed, and other factors, it is expected that the two shield tunnels will meet and pass through the position before and after K2 + 900 in the sea area section. The overburden thickness of the meeting section is approximately 30 m and the clear distance of the tunnel is about 21 m. The tunnel body involves, from top to bottom, layers of clay, medium sand, sandy clay, strongly weathered granite, and slightly weathered mixed granite.

At the expected intersection location, the geological conditions are highly variable, which may involve fully weathered strata, soft upper and hard lower mixed strata, and medium and slightly weathered hard rock strata. In the case that the shield excavation diameter is greater than 15 m, there will be perturbation superposition and secondary perturbation effects in the intersection area, and the perturbation by the shield of the surrounding strata will be more complicated, which has certain construction risks. Therefore, this study used three kinds of strata for the numerical simulation: soft upper and hard lower strata; a fully, strongly weathered mixed-granite stratum; and a medium, slightly weathered mixed-granite stratum. To explore the optimal sequence for the shield’s construction, two kinds of intersection methods were applied to the three types of strata: dynamic intersecting construction; one line in which the machine is stopped and passes through the intersection, while the other line advances. The surface settlement, deformation, and stress of the surrounding rock caused by the two-lane shield tunnel with the different intersection methods in the different strata were investigated.

The test was to study the surface settlement caused by the construction of dual-lane shield tunnels in different strata and using various intersection methods, as well as the deformation of the surrounding rock and the change in the stress state, and to analyze the construction risks of this shield tunnel. Combined with certain theoretical calculations, the surface settlement law during shield tunnel advancement was analyzed, and it was mutually verified by the numerical simulation results, increasing the credibility of the simulation.

2.2. Empirical Calculations Based on the Peck Model

2.2.1. Presentation of Empirical Calculations

Peck observed the phenomenon of surface settlement and pioneered the concept of ground loss, summarizing the results of a large amount of measured data. Under the assumption that the soil is not drained, he argued that the volume of settling tank is equal to the volume of ground loss and he assumed that the lateral distribution of the surface settlement trough is normally distributed.

The formulas for estimating lateral ground settlement are as follows [9]:

$$\begin{array}{c}S\left(x\right)=S_{\max }\exp \left(-\frac{x^2}{2j^2}\right)\end{array}$$

(1)

$$\begin{array}{c}{S}_{max}=\frac{V_{loss}}{j\sqrt{2\pi }}=\frac{\pi R^2\eta }{j\sqrt{2\pi }}\end{array}$$

(2)

In Equations (1) and (2), x is the horizontal distance of the tunnel axis; S(x) is the ground settlement at position x; S_max is the maximum ground settlement above the tunnel axis; V_loss is the amount of soil loss per unit length; R is the radius of the tunnel excavation; η is the rate of soil loss; and j is the coefficient of the width of the ground settlement tank. Peck’s formula has been widely used in engineering due to its simplicity and practicality.

On the basis of different geological conditions and the complexity of geotechnical engineering, many scholars have revised and improved Peck’s formula, extending Peck’s empirical formula to two-lane tunnels by combining it with the superposition principle. One of the more mature ones is the hypergeometric method proposed by Suwansawat [25]. This method takes into account the effect of the construction of the advance tunnel on the backward tunnel when calculating the ground settlement due to the backward tunnel. It is assumed that the values of η and j for the backward tunnels are different from those of the forward tunnels but the ground settlement curve due to the backward tunnels is still symmetrically distributed. Because the settlement caused by the first tunnel is different from that caused by the later tunnel, the total settlement curve is not symmetrical after superposition. Take the example of the right-side tunnel excavated first, as follows:

$$\begin{array}{c}S\left(x\right)=\frac{\pi R^2{\eta }_f}{j_f\sqrt{2\pi }}\exp \left(-\frac{{\left(x-0.5L\right)}^2}{{2j_f}^2}\right)+\frac{\pi R^2{\eta }_s}{j_s\sqrt{2\pi }}\exp \left(-\frac{{\left(x-0.5L\right)}^2}{{2j_s}^2}\right)\end{array}$$

(3)

In Equation (3), j_f and η_f are the width coefficient of the ground surface settlement tank and the soil loss rate of the first tunnel, respectively, and j_s and η_s are the width coefficient of the ground settlement tank and the soil loss rate of the later tunnel, respectively. The method takes into account the effect of the tunnel spacing, L, and the asymmetry of the settlement curve, but the four parameters need to be determined, and it does not specify the values for j_s and η_s.

The above settlements based on Peck’s settlement theory and its modified formulas are two-dimensional settlement solutions after soil stabilization, but the soil settlement induced by actual two-lane parallel shield tunneling is a three-dimensional solution. Wei [26] expanded from two dimensions to three dimensions by superimposing the three-dimensional soil settlement induced by the first tunnel over the backward tunnel, as follows:

$$\begin{gathered} \mathrm{S}\left(x,y,z\right)=\frac{S_{max,f}}{{\left(1-\frac{z}{h}\right)}^n}\left[1-\frac{y}{\sqrt{y^2+h^2}}\right]\exp \left(-\frac{{\left(x-0.5L\right)}^2}{{2j_f}^2{\left(1-\frac{z}{h}\right)}^{2n}}\right)+ \\ \begin{array}{c}\frac{S_{max,s}}{{\left(1-\frac{z}{h}\right)}^n}\left[1-\frac{y}{\sqrt{y^2+h^2}}\right]\exp \left(-\frac{{\left(x-0.5L\right)}^2}{{2j_s}^2{\left(1-\frac{z}{h}\right)}^{2n}}\right)\end{array} \end{gathered}$$

(4)

In Equation (4), $S_{max,f}=\frac{\pi R^2{\eta }_f}{j_f\sqrt{2\pi }}$; j_f and η_f are the same as in Equation (3). When z = 0, Equation (4) becomes the surface settlement calculation formula. Wei Gang’s method was used in combination with measured data on more than 10 tunnels in China to carry out the inversion and verification of the parameter values; the error between the empirical formula and actual value was small and the effect was good. The Wuhan Yangtze River Tunnel and Mawan Tunnel are both underwater tunnels under similar geological conditions, which can be used as references. Therefore, Equation (4) was used to calculate the ground settlement.

2.2.2. Settlement Calculation

In this study, the shield was selected for analysis of the expected position of intersection in the sea section. The meeting’s form takes the right shield tunnel first and the left shield later, with the two lines of the shield advancing from opposite directions. The calculation profile was selected as follows:

(1)

It was assumed that the intersection will occur near K2 + 900, and the excavated section is in the fully, strongly weathered mixed-granite layer.

(2)

Assuming that the meeting is near K2 + 935, the excavation section will be in the soft upper and hard lower layers.

If the shield intersection is located in the middle, slightly weathered mixed-granite formation, many of the assumptions concerning the empirical formula are no longer applicable, with the results of the calculations diverging too much from the actual values, which is not discussed here. Combining the actual project and this research [27,28], the specific parameters were calculated, and the values are shown in

Table 1. Calculated parameters from the modified formula.

Section	Depth (m)	Spacing (m)	${\mathit{I}}_{\mathit{f}}$ (m)	${\mathit{I}}_{\mathit{s}}$ (m)	${\mathit{\eta }}_{\mathit{f}}$ (%)	${\mathit{\eta }}_{\mathit{s}}$ (%)	Diameter
K2 + 900	27.7	28.715	8.7	9.1	1.83	1.86	15.43
K2 + 935	35.5	28.715	9.0	9.2	1.56	1.62	15.43

.

The section for the calculations was taken as the x-axis in Figure 3, which connects the midpoints of two tunnels and (−50, 50) was taken for x to obtain the total lateral ground settlement curve. The settlement in Figure 4 shows the settlement pattern when the two tunnels advanced 100 m from opposite directions, and the overall ground settlement trend was highly similar under both strata. Because of the principle of superposition, the ground settlement groove of the two tunnels shifted from a normal distribution, “V”-type single peak to “W”-type dual peaks. The maximum settlement occurred directly above the tunnel, and the settlement in the soft upper layer and hard lower layers was 12.5 mm. The settlement in the soil layer was 15.6 mm, which was larger than the settlement in the soft upper layer and hard lower layer. At 40 m from the calculated section, the settlement increased rapidly and the ground was disturbed to a greater extent. The settlement in the area between the two tunnels decreased and showed extreme values, with a settlement of approximately 3 mm.

The modified empirical formula based on the Peck model is lacking in the applicable conditions and accuracy and it cannot intuitively and accurately reflect the superimposed and secondary disturbances of the strata in the intersection area of large-diameter shield tunnels advancing from opposite directions to detect the construction risk. Therefore, FLAC3D 5.0 simulation software was further adopted to establish a more realistic model and working conditions, as well as to produce more realistic results for the ground disturbance.

3. Numerical Simulation of Shield Construction

3.1. Computational Model and Parameters

Model size is as follows: the diameter for the tunnel’s excavation (D) was 15.43 m, according to the Saint Venant’s principle on the disturbance range in shield tunnel construction. The outside of the tunnel to the boundary of the model was 2D, the depth of the tunnel extended to the top of the silt layer of the sea, the bottom of the tunnel was 3D, and the length of the tunnel advancement in the simulation was set at 200 m. Therefore, the model’s size was 200 m × 120 m × 70 m, as illustrated in Figure 5.

Boundary conditions are as follows: the left and right boundaries restrict the horizontal degree of freedom, the lower boundary restricts the vertical degree of freedom, and the upper boundary takes the free boundary and applies a hydrostatic pressure of 5 m.

Structural relationships and parameters are as follows: solid units were used for the tunnel perimeter rock and lining. The ground soil layers included a silt layer; sandy soil layer; sandy clay layer; fully, strongly weathered granite; and moderately weathered granite, in that order. The ground material was calculated using the Mohr–Coulomb model, and the yield criterion for damage was the Mohr–Coulomb criterion. The deformation mode used was large strain deformation and the structural material was the linear elasticity principal structure relationship. According to the results of the geological field investigation and related mechanical experiments, as well as considering the size effect of the geotechnical body, each rock layer parameter adopted in the numerical calculation was reduced by 80% based on the actual values (as shown in

Table 2. Mechanical parameters of the soil layer.

Soil Layer	Thickness (m)	Specific Weight (kg/m3)	Young’s Modulus (kPa)	Poisson Ratio	Cohesion (kPa)	Friction Angle (◦)
Silt	5~6.5	1800	1.1 × 104	0.4	10	8
Sandy soil	4~7.5	2000	4.6 × 104	0.35	10	22
Sandy clay	2~7.5	2100	5.0 × 104	0.33	180	19.5
Fully weathered mixed granite	3.5~20	2420	1.2 × 105	0.25	700	20
Moderately weathered mixed granite	31.5~47.5	2500	2.84 × 107	0.27	4800	41.9
Shield pipe	0.65	2500	3.55 × 107	0.167	-	-

). By assigning different parameters to the strata, simulations of the excavation section with different strata were achieved.

3.2. Simulated Working Conditions

• Selection of the section:

  (1)

  It was assumed that the point of intersection was near position K2 + 900 and the excavation section was in the fully weathered mixed-granite layer (referred to as the soil layer intersection point below).

  (2)

  Assuming that the meeting is near K2 + 935, the excavation section is in the soft upper and hard lower layers (referred to as the topsoil and hard bottom layer intersection point below).

  (3)

  Assuming that the intersection will occur past K2 + 957, the excavation section is in the middle and slightly weathered mixed-granite layer (referred to as rock layer intersection point below).

• Determination of the intersection form:

  (1)

  Without stopping, the two line shield tunnels advance from opposite directions, with a dynamic intersection and passing.

  (2)

  One line shield tunnel stops and the other advances. The previously stopped shield tunnel resumes advancing after passing.

• Equivalence of layer loss.

The stratigraphic loss caused by the method of shield construction is closely related to the level of construction, and this calculation uses the stratigraphic dislocation caused by the difference in the length of time between the tunnel’s excavation and the tube sheet’s installation to simulate the stratigraphic loss under normal construction conditions.

3.3. Shield Tunneling Process Simulation and Monitoring Point Arrangement

The actual construction of a shield structure is a continuous process of cutter advancement, grouting, and installation of the pipe sheet, and its excavation can be approximated as a discontinuous process in the simulation. The specific excavation process was simulated using the stiffness transfer method, and the advancement of the shield was achieved with the life and death unit. When the shield advances, it disables the soil unit in front of the cutter’s plate, and the shield head unit changes from soil to shield shell, activating the lining pipe sheet unit at the end of the shield and inputting the circumferential stress in front of the pipe sheet to simulate the grouting pressure, as well as a reasonable amount of radial stress at the cutter plate’s surface to simulate the mud–water equilibrium pressure. With calculations using this model, it is assumed that the transverse connection between the lining pipe pieces and the longitudinal connection between the rings of the pipe pieces do not take into account the reduction in the overall stiffness of the lining’s structure, and the disturbed soil and grouting behind the wall of the lining pipe pieces are generalized to be homogeneous and equally thick iso-surrogate layers. The above assumptions and ideas can better reproduce the whole process of shield tunneling. A diagram of the shield propulsion is shown in Figure 6.

(1)

Parameters are assigned to each element, gravity and boundary conditions are imposed, and they are calculated to equilibrium to reproduce the initial ground stress;

(2)

The soil is excavated in front of the shield, the preset shield shell activated, and 100 kPa of equilibrium pressure applied to stabilize the palm surface;

(3)

The model is solved with certain steps to simulate the time difference between the tunnel’s excavation and tube sheet’s installation to achieve the emergence of ground loss;

(4)

The shield’s shell is removed and 300 kPa of normal stress is applied to the tunnel wall to simulate grouting pressure;

(5)

The shield’s tube sheet is activated, as well as a 0.1 m equivalent replacement layer. Then, the model is solved to static equilibrium.

The specific simulation conditions are as follows:

(1)

A simulation of single-line tunnel advancement is conducted to examine the effect of single-line construction on ground disturbance;

(2)

A simulation of two-lane tunnel construction is carried out in the phase direction at the same time, and the monitoring line is set at the calculated theoretical intersection;

(3)

A simulation of the one-stop and one-movement intersection form is performed, the monitoring line is set at the expected intersection position, and the stop–open distance of the two lines is set, using different intersection strata by setting the excavation progress.

Arrangement of the monitoring points is as follows: to analyze the effect of the ground disturbance in the intersection’s area and the force evolution characteristics of the tunnel structure, the tunnel measurement points were set up as shown in Figure 7. Because the meeting area is located under the sea, the layer dislocation caused by the construction disturbance will lead to layer delamination or fissure expansion, which will lead to serious fissure penetration and increase the water permeability, which are not conducive to safety during construction. Settlement indicators, especially around the tunnel and the ground surface, can reflect the degree of ground disturbance, i.e., the construction risk. Therefore, the monitoring points were located around the tunnel and near the surface above the tunnel.

4. Simulation Results and Analysis

4.1. Single-Line Shield Tunnel Advanced Impact Scope and Zoning

The shield tunnels were propelled separately from both ends to examine the disturbance of the ground layer, using single tunnel propulsion when they were not within a mutual influencing range, to judge the range of overrun influence when the shield was propelled and the amount of ground settlement when it was propelled normally. The surface monitoring point at the geometric center at the upper surface of the model was selected and, using a simulation, the settlement curve of the shield tunnel’s advancement to this point is shown in Figure 8.

From Figure 8, it can be seen that the trend in the settlement of the monitoring points with the shield excavation was similar among the three strata and, on the basis of the surface deformation rate, the range of disturbance can be classified into a micro-disturbed area, large disturbed area, and stable disturbed area. The micro-disturbed area in all three strata can be approximated to be within 40 m in front of the cutter plate, which is approximately 2.5 D. For −60 m to 20 m of the soil layer in Figure 8, it is a large disturbed area; the surface deformation increases rapidly and the deformation tends to balance after 60 m, which is the disturbed and stabilized area. The final settlement was 15.1 mm. The large disturbed area of the soft upper and hard lower layers was the distance range from 40 m to 10 m, and the surface was no longer deformed after 25 m, forming a 9.2 mm surface. The large disturbed area of the rock layer was further reduced, and the surface settlement was complete after 20 m, with a maximum settlement of 0.2 mm.

4.2. Stratigraphic Disturbance Analysis during the Dynamic Intersection

After the completion of the two-lane tunnel advancement from opposite directions, the distribution characteristics of the plastic zone in the different strata after being disturbed are shown in Figure 9. In the simulation, the surface settlement observation line was arranged symmetrically with the center line of the two tunnels, which was 100 m from the excavation.

4.2.1. Distribution Characteristics of Plastic Zone after Perturbation of Each Working Condition

When the dual-lane shield advanced through the tunnel, the rock and soil bodies around the tunnel under the soil conditions were significantly affected by the disturbance, forming a small plastic zone above the tunnel and expanding outward at a small angle, and the soil layer was damaged by the shear. The plastic zone was distributed symmetrically along the tunnel line and the center line of the tunnel, and the plasticity was the greatest in the area above the center of the line. There was a small area of tensile damage under the tunnel floor. After the excavation of the soft upper and hard lower strata, the soil body was disturbed and unloaded, and the distribution characteristics of the plastic zone were similar to that of the soil condition, with shear damage as the main damage, but the area of the plastic zone was reduced and the degree of plasticity above the center of the connecting line of the two tunnels was smaller than that of the soil condition. In the rock layer condition, the plastic damage zone caused by the construction disturbance was located in the surrounding rock bodies of the upper half of the tunnel, which was horizontally shaped, the degree of plasticity was smaller than that for the other conditions, and the surrounding rock was less disturbed.

4.2.2. Undersea Surface Settlement Curve

The surface settlement curves of each working condition can be drawn using the monitoring data from each point, as shown in Figure 10. Combining the surface settlement curves of the three conditions, the soil disturbance during the dynamic advancement of the dual line was quantitatively analyzed. It was observed that the overall settlement trends of the three ground settlement curves are similar, exhibiting a dual-peaked “W”-shaped curve, symmetrically distributed on the left and right. According to Peck’s settlement theory, the surface settlement of the single tunnel is a single-peaked normal distribution curve and, when the two tunnels are excavated, the amount of settlement is superimposed and the settlement troughs merge in the middle of the two tunnels, forming a dual-peaked or even multipeaked curve.

As the shield tunnel moved forward, the layer loss increased and a settlement trough appeared above the tunnel. The settlement difference between the middle of the two tunnels and the upper part of the two tunnels became increasingly large, and a maximum difference of 10 mm occurred in the soil layer condition. When comparing the rock layer, soft upper and hard lower layers, and the soil layer conditions vertically, the scope of the over-disturbance of the shield gradually increased, and more obvious settlement curves appeared in the monitoring cross-section after 70 m of dual-line excavation in the soil layer and the soft upper and hard lower layers and after 90 m in the rock layer. The maximum settlement values of all three strata appeared above the centers of the two tunnels, with a maximum settlement of 34.6 mm in the soil layer and 22.5 mm in the soft upper and hard lower layers. The rock layer was the most stable, with only 1.02 mm. With the advance of the shield structure, the degree of ground disturbance increased and the surface settlement grooves continuously grew wider; the settlement grooves of the soil layer and the soft upper and hard lower layers were basically the same and the settlement grooves of the rock layer were narrow, which was caused by the better physical properties of these strata.

4.3. Stratigraphic Disturbance Analysis during the “One Line Stops, Another Advances” Intersection

A simulation of the “one line stops, another advances” intersection form was carried out and the two tunnels advanced simultaneously from opposite directions. According to Section 4.1, the over-disturbance ranges of the single line advancement among the three strata can be unified approximately as 40 m in front of the cutter plate. Combined with the tunnel size, the “one-stop, one-movement” intersection form was set as follows: the right tunnel stopped advancing when the two were 80 m apart, while the left tunnel continued advancing; through the intersection’s area, the right tunnel resumed advancing after the two lines were staggered head-on for 40 m. The monitoring points were arranged in the cross-section of the two lines. By analyzing the changes in the stress distribution of the tunnel’s surrounding rock and the vertical deformation of the surrounding soil, the secondary disturbance of the ground brought about by the backward tunnel advancement under the “one line stops, another advances” condition was investigated.

4.3.1. Stress Distribution in the Surrounding Rock of the Tunnel

The maximum principal stress clouds in the tunnel intersection area for the different strata, obtained via simulation, are shown in Figure 11. In the soil layer, as the advancement approached the monitoring surface, the compressive stress changed to tensile stress at the top and bottom of the tunnel, and concentrated stress appeared. When the left tunnel advanced to the monitoring surface, because of the secondary disturbance caused by the excavation, the tensile stress above the tunnel increased by 150 kPa. The overall stress field changes were not obvious and, after continuing to advance for 40 m, the stress at the monitoring surface increased by approximately 300 kPa, and the tensile stress at the top of the right tunnel changed back to compressive stress. The change in the stress was not large during the process of advancing, which had a small influence on the loading of the tube sheet. In the soft upper and hard lower strata, because of the difference between the physical and mechanical properties of the upper and lower strata, the stresses acted with the tube sheet after the tunnel’s excavation and a small amount of stress concentration appeared in the upper part of the tube sheet. The increase in the stress at the top plate during the construction of the left tunnel was 500 kPa and the increase in the stress during the construction of the right tunnel was 100 kPa. The construction disturbance lasted a long time and the monitored stresses increased slowly. Advancing in hard rock soil, the tube sheet was more uniformly stressed and the disturbance generated by the tunnel’s construction led to a growth in stress. The stress grew to approximately 100 kPa in the local area and the surrounding rock body had a better bearing capacity. The amount of stress acting on the tube sheet was smaller and the effect on the tube sheet was weaker.

4.3.2. Vertical Displacement in the Surrounding Rock of the Tunnel

Figure 12 shows the vertical deformation profiles at several locations during tunnel advancement under different working conditions. Intuitively, the excavation of the left tunnel in the soil layer caused an increase in the settlement displacement on the right side and the settlement range was enlarged. There was a settlement of 68 mm at the top of the left attempt. The surface settlement value of the intersection area in the soft upper and hard lower rock layers reached 10 mm and the maximum settlement value for the surrounding strata was approximately 24 mm. After the construction of the left tunnel, the range of the surface settlement increased, the settlement was tilted to the left side, and the range of the settlement influence for the whole section was more than 50 m. The maximum settlement value of the rock layer was in the upper area of the tunnel and the value of the settlement of the surrounding rock changed little during the construction of the left tunnel, which means that the influence of the construction disturbances by the adjacent tunnel was relatively small. The change in the surrounding rock settlement value during the construction of the left tunnel was small, indicating that the impact of the construction disturbance by the neighboring tunnel was small.

The secondary disturbance brought by the continued advancement of the left tunnel was quantitatively studied after the stop of the right tunnel. According to the data from each monitoring point during the simulation, the excavation condition with the most significant secondary disturbance characteristics of the whole strongly weathered layer was selected for the study and the distributions of the settlement patterns for the left and right sides were plotted, as shown in Figure 13.

After the left tunnel advanced 30 m, the right tunnel was obviously disturbed by the secondary disturbance and the settlement above the right tunnel increased by 5.1 mm, which was smaller after approaching the area of the intersection. The influence gradually disappeared after being staggered by 20 m. The surface settlement near the left tunnel was small. After the left tunnel’s stoppage, the right tunnel’s advancement had no obvious effect on the settlement above the left tunnel. The surface settlement on the right side increased and the settlement was slightly larger than that for the left tunnel. The final settlements for the left and right tunnels were approximately 28.2 mm and 31.3 mm.

4.4. Comprehensive Analysis

Comparing the results of the theoretical calculations, the settlement trend was similar to the numerical simulation results, with the bimodal characteristics of a bilinear settlement curve, but the settlement values were quite different. This is due to the fact that there are less empirical data on existing shield tunnels with large diameters and complex strata, and it is difficult to approximate the actual situation by taking the values of the width coefficient of the settlement groove and the soil loss rate, which need to be further explored and analyzed.

The over-disturbance ranges for the single-line advancement among the three strata can be approximated and standardized as 40 m in front of the cutter plate, which is about 2.5 D. In addition, the stopping distance of the “one-stop and one-movement” intersection was thus set. According to the surface settlement curve, the disturbance range is classified into a micro-disturbance region, large disturbance region, and disturbance–stabilization region.

The “one-stop and one-movement” intersection form reduced the disturbance effect. Comparing Figure 13 with Figure 10, the final surface settlement caused by the “one-stop, one-movement” intersection decreased by 4 mm and, when the left tunnel advanced, the maximum change in stress among the three strata was approximately 0.5 MPa. The change in the tunnel’s peripheral settlement was no more than 20 mm and the change in the surface settlement was less than 5 mm, which show that the secondary disturbance caused by the advancement of the left tunnel had little effect on the tunnel’s peripheral rock mass. After the left tunnel stopped, the right tunnel advanced and, according to Figure 13b, the surface settlement value for the left tunnel was almost unaffected.

5. Conclusions

This study took the Mawan Cross-Sea Tunnel project as its background due to its large diameter; its layer complexity, involving fully and strongly weathered mixed-granite strata, soft upper and hard lower strata, and micro-weathered strata in the dynamic intersection; and its “one-stop and one-movement” intersection construction condition. According to the results of the three-dimensional numerical simulation, the main conclusions are as follows:

• The over-disturbance ranges of the single-line advancement among the three strata can be approximated and unified as 40 m in front of the cutter plate, and the amounts of ground settlement in the three strata were 15.1 mm, 9.2 mm, and 0.2 mm, respectively. The disturbance range was classified into micro-disturbance areas, large disturbance areas, and disturbance–stabilization areas based on the gradient of the ground settlement.
• When the two tunnels advanced from opposite directions toward the dynamic intersection, the soil above the tunnel in the soil layer extended outward along a certain angle to form a shear damage zone and the rock layer exhibited tensile damage. The dynamic intersection caused the tunnel settlement to increase, the settlement troughs of the two tunnels merged, and the surface settlement curve showed a bimodal shape. The over-disturbance range of the shield increased by approximately 10 m for the dynamic intersection. As the shield advanced, the ground was disturbed to a greater extent and the surface settlement groove continuously grew wider.
• The “one line stops, and another advances” intersection form can reduce the impact of disturbances. Under this condition, the change in the stress in the left tunnel was no more than 0.5 MPa, the increase in the stress in the right tunnel was less than 0.1 MPa, and the change in its surface settlement value was less than 5 mm. After the left tunnel stopped, the advancement of the right tunnel had almost no effect on the surface settlement above the left tunnel, and the change in the overall stress field and settlement was small.