Face Stability Analysis of Shield Tunnels in Homogeneous Soil Overlaid by Multilayered Cohesive-Frictional Soils

In order to better interpret failure features of the failure of soil in front of tunnel face, a new three-dimensional failure mechanism is proposed to analyze the limit support pressure of the tunnel face in multilayered cohesive-frictional soils. The new failure mechanism is composed of two truncated cones that represent the shear failure band and a distributed force acting on the truncated cones that represents the pressure arch effect. By introducing the concept of Terzaghi earth pressure theory, approximation of limit support pressures is calculated using the limit analysis methods. Then the limit support pressures obtained from the new failure mechanism and the existing approaches are compared, which show that the results obtained from the newmechanism in this paper provide relatively satisfactory results.


Introduction
Nowadays, tunnel constructions in urban shallow soft ground are more frequently being carried out using closed shields, which are basically divided into compressed-air shields, slurry shields, and earth pressure balanced shields.By supporting the excavation face and preventing the seepage flow to the face, closed shields control the surface settlement and limit the risk of tunnel face failure through the continuous support of the face during excavation.However, lack of sufficient face support leads to the instability of tunnel face.In extreme cases, the collapse propagates up to the ground surface creating a surface depression.Researchers have developed experimental, numerical, and analytical approaches to investigate the tunnel face stability and to determine the limit support pressure of the tunnel face in cohesive-frictional soils.
From the results obtained from experimental tests and numerical simulations, it is shown that the failure of soil in front of tunnel face shows two main features, that is, shear failure band in the lower part and pressure arch effect in the upper part.Referring to the remarks claimed by Sloan [32] and considering the specific problems of tunnel face stability analysis, the main characteristics of the limit equilibrium and limit analysis to assess the stability of tunnel face in cohesive-frictional soils are summarized as follows.As a tradition stability analysis method, the limit equilibrium methods are widely used in the practical engineering to  analyze the stability of tunnel face.Due to its simplicity, a multitude of prediction models have been developed to solve the complex situations of tunnel face, for example, seepage flow conditions and bolt reinforcement.To reflect the feature of pressure arch effect, most of failure mechanisms based on limit equilibrium methods adopted Terzaghi earth pressure theory.However, limit equilibrium methods neglect the stress-strain relationship of the soil, which is a condition that must be satisfied for a complete solution stated by Chen [33].Compared to the limit equilibrium methods, the limit analysis methods have a strictly theoretic basis.
Based on the limit analysis methods, Zhang et al. [34] proposed a three-dimensional failure mechanism, which is composed of four truncated cones and a distributed force acting on those truncated cones.Their failure mechanism combines the generality of earth pressure theory with the rigour of limit analysis.And the results show that their improved treatments are effective.However, their works are limited to the case of homogeneous soils.Practical experience shows that the stability of the face is often a problem in heterogeneous soils (Ibrahim et al. [9]; Senent and Jimenez [11]).The shear strength parameters of soils in different strata are different from each other.In this paper, for multilayered cohesive-frictional soils, a new three-dimensional failure mechanism is proposed to determine the limit support pressure of the tunnel face using limit analysis methods.The new failure mechanism is composed of two truncated cones on which a distributed force acts.The distributed force could be calculated by using Terzaghi earth theory.Then, the limit support pressures obtained from the new failure mechanism and the existing approaches are compared.

Limit Analysis of the Tunnel Face Stability
2.1.The New Failure Mechanism.The tunnel face stability to be analyzed in this paper is idealized as shown in Figure 1.By considering that the tunnel is a rigid circular cylinder of diameter  driven under a depth of cover , a surcharge   is applied on the ground surface, and   is the uniform support pressure on the tunnel face.
Referring to the concept proposed by Zhang et al. [34], a new prediction model is established to analyze face stability of shallow circular tunnels in multilayered cohesivefrictional soils.The new failure mechanism is composed of two truncated cones on which a distributed force acts.The two truncated cones were firstly proposed by Leca and Dormieux [26] to investigate the tunnel face stability in homogenous soil.Since this paper considers a multilayered cohesive-frictional soil (i.e., the crossed layer is assumed to be homogeneous whereas the cover soil is layered), two truncated cones proposed by Leca and Dormieux [26] are modified and only the part in crossed layer of the two truncated cones is preserved.The new failure mechanism considers the effect of the cover layers as a distributed force according to the Terzaghi earth theory.The same treatments were adopted in the works of Broere [18]; Kirsch and Kolymbas [19]; Anagnostou [20]; and Anagnostou and Perazzelli [21], which are based on the limit equilibrium method.

Geometric Properties.
The intersections of two truncated cones, tunnel face, and horizontal plane crossing the top of the tunnel face are two ellipses and a circle and are called Σ 1 , Σ 1,2 , and Σ 2 .The intersection of the first truncated cone (adjacent to the tunnel face) with the circular tunnel face is an ellipse Σ 1 , with semiaxis lengths of  1 and  1 and with area of  1 that are calculated as follows (cf. Figure 1): where  defines the opening angles of the two truncated rigid cones that are equal to 2 and  is the angle between the axis of the first truncated rigid cone which is adjacent and the horizontal.
The area of the contact elliptical surface Σ 1,2 between two truncated cones is ellipse with semiaxis lengths of  12 and  12 and with area of  12 is described as follows: The intersection of the second truncated cone with the horizontal plane across the top of the tunnel face is a circle Σ 2 with radius of  2 and with area of  2 as follows: In addition, the lateral surfaces and volumes of the two truncated cones are as follows:

The Distributed Force Acting on the Two Truncated
Cones.For cohesive-frictional soils, the results obtained from experimental tests and numerical simulations show that the failure of soil in front of tunnel face demonstrates two main features, that is, shear failure band in the lower part and pressure arch effect in the upper part.The existing researches [18][19][20][21] considered the effect of the cover layers as a distributed force and achieved satisfactory results, which is calculated using the Terzaghi earth pressure theory.In this paper, the distributed force acting on the two truncated cones is also calculated by Terzaghi earth pressure theory as shown in Figure 2.
Equation ( 6) is a differential equation for the vertical stress  V ().To solve the differential equation and consider the boundary condition, the result of vertical stress at any depth is determined as follows: Since the ground above the tunnel is heterogeneous and consists of  horizontal layers (cf. Figure 2), the vertical stress should be applied recursively for every layer from the top to down by treating the pressures of every overlying layer as a surface loading as follows: Then the final result of vertical stress distribution acting on the two truncated cones is obtained as follows: where   is the thickness of layer  and   ,   ,  0 , and   denote its unit weight and shear strength parameters.

Velocity Field.
The two truncated rigid cones are translated with velocities with different directions, which are collinear with the cones' axes and are at an angle  to the discontinuity surface.The velocity of first cone and the relative velocity between the two cones are described by the following equations: (10)

Limit Support Pressure.
To satisfy the stability conditions of the tunnel face according to the upper bound theorem, the following relation is considered: where   represents the power of the external loads and  V denotes the dissipation power.The power of the external loads,   , is the sum of three components:   , the power of the support pressure   ,   , the power of the vertical stress  V , and   , the power of the soil unit weight : where ( By equating the total rate of external work to the total rate of internal energy dissipation, as shown in (11), the pressure   at the tunnel face is obtained by ( 14) with (15) as follows: In (14),   ,   ,   , and   depend on the mechanical and geometrical characteristics , , and / and on the angular parameters of the failure mechanism .These parameters were obtained by maximizing   in ( 14) with respect to the angles .An upper bound solution can be found by numerically optimizing (14) with respect to the angle .

Comparisons
To validate the results obtained from the limit analysis developed in this paper, comparisons between the results of this work and existing approaches (Broere [18]; Tang et al. [30]; Senent and Jimenez [11]) were performed.Overall, the model proposed by Broere [18] is based on limit equilibrium methods while the models proposed by Tang et al. [30], Senent and Jimenez [11], and this paper are based on limit analysis methods, respectively.
Three sets of analyses were carried out by referring to the researches recently reported by Tang et al. [30] and Senent and Jimenez [11].And the detailed soil parameters of layers are shown in Table 1.The unit weight  is equal to 18 kN/m 3 for the layers in all cases (/ = 1.5).

Influence of the Variables of the Crossed Soil on Limit
Support Pressures.The first set of analyses described the cases of a single cover layer with constant strength parameters ( 1 and  1 ) and a single crossed layer with different strength parameters ( 0 and  0 and increments Δ 0 and Δ 0 ).Figures 3 and 4 show the curves of the limit support pressures with the variation of Δ 0 and Δ 0 , respectively.In Figure 3, the limit support pressures declined nonlinearly with the increase of Δ 0 from −5 ∘ to 15 ∘ .The decrease of the limit support pressures with Δ 0 from −5 ∘ and 0 ∘ was steeper than those with Δ 0 in the range between 0 ∘ and 15 ∘ , which shows that Δ 0 has a higher influence on the limit support pressures if the strength parameters of the crossed  soil were weaker than strength parameters of the cover soil.Moreover, the results from this paper and Senent and Jimenez [11] were between the results from Broere [18] (the highest solutions) and from Tang et al. [30] (the lowest solutions).The approach proposed by Broere [18] overestimated the limit support pressure compared to the results from other approaches.Figure 4 shows that the limit support pressures decreased linearly with the increase of Δ 0 .Similarly, the results from this paper and Senent and Jimenez [11] were almost equal and located between the results from Broere [18] (the highest solutions) and from Tang et al. [30] (the lowest solutions).It is noteworthy that rotational face collapse mechanism proposed by Senent and Jimenez [11] considers the whole face of the tunnel by employing a spatial discretization technique, which provides the results that outperform previous upper bound solutions.As mentioned above, the approximations of limit support pressures obtained from the new failure mechanism are not strictly an upper bound solution and the results from our paper are slightly less than the results from Senent and Jimenez [11] in most cases.The comparisons show that the results in this paper provide relatively satisfactory results.To compare Figures 3 and 4, it has also been found that the friction angle affected the support pressure effectively compared to the cohesion.

Influence of the Variables of
the Cover Soil on Limit Support Pressures 3.2.1.Single Cover Layer.The second set of analyses presented a reverse situation, which described the cases of a single cover layer with different strength parameters ( 1 and  1 and increments Δ 1 and Δ 1 ) and a single crossed layer with constant strength parameters ( 0 and  0 ).Figures 5 and 6 compared change rules of the limit support pressures with the variation of Δ 1 and Δ 1 , respectively.Figure 5 shows that the limit support pressures decreased nearly linearly with the increase of Δ 1 from −5 ∘ to 15 ∘ .And the results from this paper were slightly less than the results from Senent and Jimenez [11] and were much higher than the results from Tang et al. [30].
In Figure 6, the limit support pressures dropped linearly as Δ 1 increased.Similarly, the results from this paper were

Figure 2 :
Figure 2: Distributed force calculated by Terzaghi earth pressure theory.

Table 1 :
Soil parameters of layers.