Download as pdf or txt
Download as pdf or txt
You are on page 1of 35

Accepted manuscript doi:

10.1680/jgeen.22.00051
Accepted manuscript
As a service to our authors and readers, we are putting peer-reviewed accepted manuscripts
(AM) online, in the Ahead of Print section of each journal web page, shortly after acceptance.

Disclaimer
The AM is yet to be copyedited and formatted in journal house style but can still be read and
referenced by quoting its unique reference number, the digital object identifier (DOI). Once
the AM has been typeset, an ‘uncorrected proof’ PDF will replace the ‘accepted manuscript’
PDF. These formatted articles may still be corrected by the authors. During the Production
process, errors may be discovered which could affect the content, and all legal disclaimers
that apply to the journal relate to these versions also.

Version of record
The final edited article will be published in PDF and HTML and will contain all author
corrections and is considered the version of record. Authors wishing to reference an article
published Ahead of Print should quote its DOI. When an issue becomes available, queuing
Ahead of Print articles will move to that issue’s Table of Contents. When the article is
published in a journal issue, the full reference should be cited in addition to the DOI.

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051
Submitted: 09 March 2022

Published online in ‘accepted manuscript’ format: 10 June 2022

Manuscript title: Design of Deep Rock Tunnels combining the Hyperstatic Reaction and

Convergence Confinement methods

Authors: Dianchun Du1, Daniel Dias2 and Ngocanh Do3

Affiliations: 1School of Civil Engineering, Southeast University, Nanjing, China; 23SR lab,

Grenoble Alpes University, Grenoble, France and 3Department of Underground and Mining

Construction, Faculty of Civil Engineering, Hanoi University of Mining and Geology, Hanoi,

Vietnam

Corresponding author: Dianchun Du, School of Civil Engineering, Southeast University,

Nanjing 211189, China.

E-mail: dudianchun@seu.edu.cn

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051
Abstract

Tunnel lining is usually installed with delay behind the tunnel face excavation, which requires that a

displacement has to be considered at the tunnel boundary before the support structure installation. This delayed

installation comes with a release and redistribution of stress in the rock mass. Internal tunnel lining forces

significantly rely on the stress relaxation process taking place in the rock mass surrounding the excavated

tunnel. One of the difficulties when designing tunnel supports is, therefore, to analyze the rock-support

interaction considering the tunnel lining convergence caused by the stress redistribution. In this study, a simple

and effective calculation process based on the combination of two methods, the Hyperstatic Reaction Method

(HRM) and Convergence Confinement Method (CCM), is presented to analyze the interaction of rock mass and

support structure. The rock mass is assumed to obey the Hoek-Brown criterion. The stress release is also taken

into consideration in the present method. The present method is validated by comparing results of the HRM in

terms of tunnel lining forces against the analytical ones. Thereafter, the effect of the stress release coefficient, of

the tunnel depth and of the Hoek-Brown criterion parameters (the Geological Strength Index (GSI) and σci), on

the lining internal forces are presented and discussed.

Keywords: Hoek-Brown failure criterion; Tunnel design; Lining forces; Convergence Confinement Method;

Hyperstatic Reaction Method

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051
1. Introduction

Calculating the tunnel lining forces due to an excavation is key to an efficient tunnel design.

Many different tunnel design methods are usually used in practical engineering, such as

empirical, numerical and analytical methods (AFTES, 2011).

The empirical design methods, e.g. Bieniawski’s RMR (1989) and Barton’s Q system (1975),

are based on engineering experience. They usually refer to the geomechanical

characterization. Although those methods can be applied to tunnel designs in hard rock, they

have obvious shortcomings in weak rock.

In analytical design methods (Erdmann, 1983; Barton, 2002), the external loads acting the

surrounding rock mass should be evaluated. Rock loads are approximated using experiential

solutions (Svoboda and Hilar, 2015; Cao et al., 2018; Huang et al., 2021; Zhang et al., 2021)

or rock mass classifications (Osgoui and Unal, 2009). The disadvantage of the above external

load concepts is that they do not consider the rock-support interaction.

Numerical methods are becoming popular as they can consider complex tunnel geometries,

rock mass profiles and rock-support interaction (Jin et al., 2018; Mishra et al., 2018; Zhao et

al., 2018; Cui et al., 2019; Lei and Wei, 2020; Rehman et al., 2020, Zhang et al., 2020). The

numerical methods could be divided into two-dimensional (2D) and three-dimensional (3D)

methods. The 3D numerical analyses can be able to predict the deformations and stress

redistribution induced by tunnelling. However, simplified 2D plane strain simulations are

often adopted to replace the 3D ones to save on computation time. The design of the tunnel

support should consider the convergence that occurs ahead of the support installation, as a

result of the stress-release of the rock mass. In the present work, an analysis combining two

methods, the Hyperstatic Reaction Method (HRM) and Convergence Confinement Method

(CCM), is implemented to analyze the rock-support interaction considering the stress release.

Both methods are able to consider the rock mass-support interaction.

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051
The CCM (Panet and Guénot, 1982; Carranza-Torres and Fairhurst, 2000; Panet et al., 2001;

Oreste, 2003; González-Nicieza et al., 2008; Almog et al., 2015; Taghizadeh et al., 2020)

was primarily considered for axisymmetric analyses. A fictitious radial pressure σ is assumed

to be applied to the excavation walls considering the stress release coefficient. It should be

noted that Carranza-Torres and Fairhurst (2000) just focused on the practical application of

the CCM, in which the CCM was used to estimate the magnitude of loads that the rock mass

will transmit to supports installed behind the tunnel face.

The HRM is efficient (Oreste, 2007; Du et al., 2018) for tunnel design with a very short

calculation time (less than 3 seconds on an Intel CPU 2GHz PC). The rock-support

interaction is simulated using springs in the HRM. The HRM requires that the active loads

from the surrounding rock to be defined. In addition, the Hoek-Brown failure criterion has

been widely used as a failure criterion to define the stress conditions of rock mass. Therefore,

considering the stress release supplied by the CCM, the HRM is adopted here to evaluate the

internal lining forces of deep-buried tunnels excavated in rock described by the Hoek-Brown

criterion.

The originality of the present work is that it permits development of a new method to

estimate the internal tunnel lining forces using an effective and efficient way considering the

convergence and stress release of the excavated rock mass supported by the tunnel lining. The

support structural forces calculated by the proposed method have shown good accordance

with the analytical ones, allowing validation of the present method. Thereafter, the effects of

the stress release coefficient λ, the tunnel depth H, the parameters from the Hoek-Brown

criterion, e.g. 𝜎𝑐𝑖 , and Geological Strength Index (GSI), on the lining internal forces are

shown and discussed. The presented method is able to calculate internal forces at any radial

point of the tunnel cross-section, which provides a useful tool for tunnel design. Based on the

presented method, tunnel designers can obtain the lining forces required in a plane strain

analysis in a simplified and effective manner.

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051
2. Hoek-Brown strength failure criterion

The Hoek-Brown strength failure criterion was introduced by Hoek et al. (1995) to estimate

the rock mass strength, which can be expressed as:


𝜎
𝜎1 = 𝜎3 + 𝜎𝑐𝑖 (𝑚𝑏 𝜎 3 + 𝑠)𝑎 (1)
𝑐𝑖

where 𝜎1 , 𝜎3 means the maximum and minimum principal stresses, respectively; 𝜎𝑐𝑖

represents the unconfined compression strength of the rock mass (0.01MPa≤ σ𝑐𝑖 ≤200MPa);

𝑚𝑏 , s, and a are the material constants of the rock mass, given by:
𝐺𝑆𝐼−100
𝑚𝑏 = 𝑚𝑖 exp⁡( 28−14𝐷 ) (2)

𝐺𝑆𝐼−100
𝑠 = exp⁡( ) (3)
9−3𝐷

1 1 𝐺𝑆𝐼 20
𝑎 = 2 + 6 [exp (− ) − exp (− 3 )] (4)
15

where 𝑚𝑖 (5 ≤ 𝑚𝑖 ≤ 35 ) is a material constant of rock mass; 𝐷 is a disturbance factor

(0~1.0). Hoek and Brown (2019) suggested that 𝐷 = 0 can be considered if the excavation is

undertaken by a tunnel boring machine. This manuscript mainly focusses on the study of

deep-buried tunnels excavated by tunnel boring machines, therefore, 𝐷 = 0 is taken into

consideration in the following analysis. GSI ( 5 ≤ 𝐺𝑆𝐼 ≤ 100 ) denotes the Geological

Strength Index which is typically used to evaluate the deformability and strength of rock

mass. Cai et al. (2004) suggested that the values of GSI could be evaluated on the basis of the

geological description of the rock mass.

The Hoek-Brown strength failure criterion described in Eq. (1) can be transformed into a

failure envelope according to Carranza-Torres and Fairhurst (2000). The transformation


(1−𝑎)/𝑎 1/𝑎
involves dividing the stress magnitudes by⁡𝜎𝑐𝑖 𝑚𝑏 and then adding the term 𝑠/𝑚𝑏 .

The scaled principal maximal and minimal stresses 𝑆1 ⁡and 𝑆3 can be defined as follows:
𝜎1 1/𝑎
𝑆1 = (1−𝑎)/𝑎 + 𝑠/𝑚𝑏 (5)
𝜎𝑐𝑖 𝑚𝑏

𝜎3 1/𝑎
𝑆3 = (1−𝑎)/𝑎 + 𝑠/𝑚𝑏 (6)
𝜎𝑐𝑖 𝑚𝑏

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051
Replacing the principal maximal and minimal stresses 𝜎1 and 𝜎3 by the scaled stresses in Eqs.

(5) and (6), one obtains the Hoek-Brown failure criterion in the following form:

𝑆1 = 𝑆3 + 𝜇𝑆3𝑎 (7)

where the parameter 𝜇 is


(2𝑎−1)/𝑎
𝜇 = 𝑚𝑏 (8)

For the case of 𝑎 = 0.5 , Eq. (7) becomes the form mentioned by Londe (1988). This

transformation was applied by Carranza-Torres and Fairhurst (2000) to the design of tunnels

in rock mass satisfying the Hoek Brown strength failure criterion. Eq. (7) leads to an

important simplification of the mechanical analyses compared to Eq. (1). In Eq. (7),

parameters 𝜎𝑐𝑖 , s and 𝑚𝑏 are included in the parameters 𝑆1 ⁡and 𝑆3 .

The rock mass modulus⁡𝐸𝑟𝑚 (MPa) could be given (Hoek et al., 2002) by:
𝐺𝑆𝐼−10
𝐷 𝜎
𝑐𝑖
𝐸𝑟𝑚 = 1000(1 − 2 )√100 ∙ 10( 40
)
if σ𝑐𝑖 ≤ 100 MPa (9)
𝐺𝑆𝐼−10
𝐷
𝐸𝑟𝑚 = 1000(1 − 2 ) ∙ 10( 40
)
if σ𝑐𝑖 > 100 MPa (10)

Note that for elastic-plastic analyses, the rock-mass shear modulus (MPa) is more often used

than 𝐸𝑟𝑚 (MPa). The 𝐺𝑟𝑚 can be evaluated as follows:


𝑟𝑚 𝐸
𝐺𝑟𝑚 = 2(1+𝑣 (11)
𝑟𝑚 )

where 𝑣𝑟𝑚 ⁡ is the Poisson’s ratio of the rock mass ( 0.1 ≤ 𝑣𝑟𝑚 ≤ 0.3 ). In the following

analysis, the case of 𝑣𝑟𝑚 = 0.3 is considered (Hoek and Brown, 1980).

Equivalent Mohr-Coulomb Parameters

The friction angle 𝜑⁡and cohesive strength 𝑐⁡are usually used in practical tunnel design. The

friction angle, 𝜑 ′ , and equivalent cohesion, 𝑐 ′ , for the rock mass satisfying the Hoek-Brown

strength failure criterion could be expressed as follows (Hoek et al., 2002):


′ ](𝑠+m 𝜎 ′ )𝑎−1
σ𝑐𝑖 [(1+2𝑎)𝑠+(1−𝑎)m𝑏 𝜎3𝑛
𝑐′ = 𝑏 3𝑛
𝑎−1
(12)
′ )
(1+𝑎)(2+𝑎)√1+[6𝑎m𝑏 (𝑠+m𝑏 𝜎3𝑛 ]/[(1+𝑎)(2+𝑎)]

′ ) 𝑎−1
6𝑎m𝑏 (𝑠+m𝑏 𝜎3𝑛
𝜑 ′ = arcsin⁡[ ′ ) 𝑎−1 ] (13)
2(1+𝑎)(2+𝑎)+6𝑎m𝑏 (𝑠+m𝑏 𝜎3𝑛

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051
′ ′
where 𝜎3𝑛 = 𝜎3𝑚𝑎𝑥 /𝜎𝑐𝑖 . Based on Hoek et al. (2002), for deep-buried tunnels, the value of

𝜎3𝑚𝑎𝑥 ⁡can be obtained by:
′ −0.94
𝜎3𝑚𝑎𝑥 𝜎′
𝑐𝑚
′ = 0.47 ( 𝛾𝐻 ) (14)
𝜎𝑐𝑚

in which, 𝛾⁡is the unit weight of rock mass and H means the tunnel depth. If the horizontal

stresses are higher than the vertical ones, the horizontal stresses should be used in place of

𝛾𝐻. 𝜎𝑐𝑚 is the rock mass strength, which is defined by:
m
[m𝑏 +4𝑠−𝑎(m𝑏 −8𝑠)]( 𝑏 +𝑠)𝑎−1

𝜎𝑐𝑚 = σ𝑐𝑖 ∙ 4
(15)
2(1+𝑎)(2+𝑎)

In the following study, the equivalent Mohr-Coulomb parameters strength parameters

obtained from Eq. (12) and Eq. (13) are used to design deep circular tunnels excavated in the

rock mass satisfying the Hoek-Brown strength failure criterion.

3. CCM and HRM

3.1. Convergence-Confinement Method

As a 2D method, the CCM describes a procedure to estimate the active loads applied to the

tunnel lining considering its delayed installation after the tunnel excavation. The detailed

CCM was introduced by Carranza-Torres and Fairhurst (2000). Only a brief outline is

presented here.

To describe the problem, Fig. 1 shows a cylindrical tunnel subjected to a rock mass stress 𝜎𝑖 .

The problem is assumed to be 2D using plane strain conditions. Fig. 1b shows the cross-

section of a circular tunnel lining (external radius R and thickness⁡𝑡𝑐 ). The displacement

𝑢𝑟 ⁡represents the inward rock mass displacement corresponding to the external active loads

𝑝𝑠 . A reaction pressure (pi) of the support is developed and it contributes to act against the

rock mass movements. An equilibrium state of the tunnel-rock mass system is reached when

the active load ps is equal to the internal reaction pressure pi.

Fig. 2 shows three basic components of the CCM that are, (1) the Longitudinal Deformation

Profile (LDP); (2) Support Characteristic Curve (SCC) and (3) Rock mass Reaction Curve

(GRC).

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051
The support is usually installed some distance behind the tunnel face. The CCM assumes that

the support structures carry a proportion of the external loads and the remainder by the rock.

The vertical segment NK in Fig. 2 corresponds to the tunnel face normalized pressure at the

time of support installation. When the effect of the support is diminishes, system equilibrium

will be reached at point D. The load 𝑝𝑠𝐷 at point D shown in Fig. 2 is the design load. The aim

of the analysis using the CCM in the present work is to determine this final pressure 𝑝𝑠𝐷 .

Assuming the rock mass obeys the Hoek-Brown criterion, Eqs. (5-7) are taken into

consideration. The variables characterizing the strength of the rock mass include the intact

rock parameter 𝑚𝑖 , unconfined compressive strength σ𝑐𝑖 , rock mass parameters 𝑚𝑏 , s and a

introduced in Section 2.

According to the Londe transformation (1988) (Eqs. 5 and 6), the internal load 𝑝𝑖 ⁡ and

hydrostatic stress σ𝑖 can be calculated. The scaled internal load 𝑃𝑖 ⁡and scaled hydrostatic

stress 𝑆𝑖 are calculated as follows:


𝑝𝑖 1/𝑎
𝑃𝑖 = (1−𝑎)/𝑎 + s/m𝑏 (16)
σ𝑐𝑖 𝑚𝑏

σ𝑖 1/𝑎
𝑆𝑖 = (1−𝑎)/𝑎 + s/m𝑏 (17)
σ𝑐𝑖 𝑚𝑏

The scaled internal load 𝑃𝑖 is given by the following expression according to Carranza-Torres

and Fairhurst (2000):


1 2
𝑃𝑖 = 16 ∙ (1 − √1 + 16𝑆𝑖 ) (18)

Then the actual load 𝑝𝑖 can be obtained from Eq. (16):


(1−𝑎)/𝑎 s
𝑝𝑖 = σ𝑐𝑖 𝑚𝑏 [𝑃𝑖 − 1 ] (19)
m𝑎
𝑏

Clearly, the internal load 𝑝𝑖 ⁡ could be calculated by the hydrostatic stress 𝜎𝑖 . Tunnel

excavation causes a stress redistribution in the surrounding rock mass and convergence of the

tunnel linings result due to the loss of confinement. Therefore, the rock relaxation that occurs

during the excavation process decreases the hydrostatic stress, 𝜎𝑖 , below the initial rock mass

stress 𝜎0 , which is then given by:

𝜎𝑖 = (1 − 𝜆) · 𝜎0 (20)

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051
where 𝜎0 ⁡is the initial stress of rock mass and 𝜆 represents the stress release coefficient. The

most critical point in the CCM is to determine the value of stress release coefficient λ. The

initial vertical rock mass stress 𝜎𝑣0 ⁡can be calculated by 𝜎𝑣0 = 𝛾𝐻 (Hoek and Brown, 1980).

The initial horizontal rock mass stress 𝜎ℎ0 ⁡depends on the coefficient at rest 𝐾0 , and can be

estimated by 𝜎ℎ0 = 𝐾0 · 𝜎𝑣0 (assuming no water pressure).

Numerous 2D numerical studies conducted by Dias and Kastner (2013), Do and Dias (2017),

indicating that the CCM could be able to correctly simulate the final surface settlements if the

stress relaxation coefficient is estimated when the support structure is installed. According to

Panet et al. (2001), the value of the stress relaxation coefficient 𝜆 can be determined by:
0.75∙𝑅 2
𝜆 = 1 − 0.75 (0.75∙𝑅+𝐿) (21)

This relation ship indicates that the coefficient 𝜆 varies from 0.25 (at the tunnel face, L=0) to

0.75. In the following analyses, values of 𝜆 in the range from 0.3 to 0.7 are considered. Once

a stress relaxation coefficient has been defined, the stress 𝜎𝑖 applied to the excavation walls

can be obtained according to Eq. (20). Replacing the obtained stress 𝜎𝑖 into Eq. (17), one

obtains 𝑆𝑖 . According to Eqs. (18) and (19), the value of the critical pressure 𝑝𝑖 could then be

obtained, which is also the final goal pressure 𝑝𝑠𝐷 (see in Fig. 2).

3.2. Hyperstatic Reaction Method

According to Oreste (2007) and Do et al. (2014), the HRM can be effectively used to design

the tunnel lining (Oreste, 2007; Do et al., 2014; Du et al., 2020). The tunnel lining is divided

into a finite number of linear elements (see Fig. 3) which interact with each other through

nodes in this method. It assumed that the elements interact with the rock mass with normal

springs and shear springs which are distributed over the nodes. The interaction of rock-lining

in the HRM is realized by the active loads and springs connected to the element nodes (𝑝𝑣𝑖

and 𝑝ℎ𝑖 shown in Fig. 3).

The analysis scheme used in the HRM is given in Fig. 3. Since the tunnel is assumed to be

deep, constant hydrostatic stresses are used in the analysis (Carranza-Torres and Fairhurst,

2000).

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051
Fig. 4 shows a beam-type element in HRM. Since the structure interacts with the rock mass

by springs, the stress characteristics of each element could be calculated after obtaining the

node displacements. The displacements can be calculated by means of defining the global

stiffness matrix K of the elements.



The 𝐾 could be assembled by the local stiffness matrix 𝑖 of element under the global

Cartesian coordinates, which is given as follows:


(1) (1)
11 12 0 ⁡0⁡ ⁡0 ⁡0⁡ ⁡0⁡
(1) (1) (2) (2)
21 22 + 22 23 ⁡0⁡ ⁡0 ⁡⁡0⁡ ⁡0⁡
(2) (2) (3) ⁡ ⁡ ⁡0 ⁡0⁡ ⁡0⁡
0 32 33 + 33
(3)
𝐾= 34 ⁡0⁡ ⁡0⁡ (22)
(3)
0⁡ 0⁡ ⁡ 43⁡ (3) (4)
44 + 44 ⁡ ⁡ ⁡0⁡
0⁡ 0⁡ ⁡0 (𝑛−1)
⁡ ⁡
(𝑛−1)𝑛
⁡ ⁡
0 0 0 0 0 (𝑛−1) (𝑛−1)
[ ⁡ 𝑛(𝑛−1) ⁡ 𝑛𝑛 ]
(𝑖) (𝑖) (𝑖) (𝑖)
where the terms 𝑖𝑖 , 𝑖(𝑖+1) , (𝑖+1)𝑖 and (𝑖+1)(𝑖+1) represent 3 3 sub-matrices of 𝑖 under

the global Cartesian coordinates (i=1, 2, 3…, 360). It should be noted that taking the

calculation accuracy and calculation time into account, the number of elements in the present

work is taken as 360. The 𝑖 is calculated as follows:

[ ]𝑖 = 𝜆𝑖 ∙ ̅̅̅̅
[ ]𝑖 ∙ 𝜆𝑖 (23)

where ̅𝑖 means the local stiffness matrix under local Cartesian coordinates. 𝜆𝑖 is the

transformation matrix. ̅ 𝑖 and 𝜆𝑖 are calculated as follows, respectively:


𝐸 𝐸
0 0 − ⁡ ⁡0⁡ ⁡0
𝐿𝑖 𝐿𝑖
12𝐸𝐼 6𝐸𝐼 12𝐸𝐼 6𝐸𝐼
0 0⁡ −
𝐿3𝑖 𝐿𝑖 𝐿3𝑖 𝐿𝑖
6𝐸𝐼 4𝐸𝐼 6𝐸𝐼 2𝐸𝐼
0 0 −
𝐿𝑖 𝐿𝑖 𝐿𝑖 𝐿𝑖
̅̅̅̅
[ ]𝑖 = (24)
𝐸 𝐸
− ⁡ ⁡0⁡ 0 0 0
𝐿𝑖 𝐿𝑖
12𝐸𝐼 6𝐸𝐼 12𝐸𝐼 6𝐸𝐼
⁡0 − − 0 −
𝐿3𝑖 𝐿𝑖 𝐿3𝑖 𝐿𝑖
6𝐸𝐼 2𝐸𝐼 6𝐸𝐼 4𝐸𝐼
⁡0 ⁡𝐿 ⁡ 0 −
[ 𝑖 𝐿𝑖 𝐿𝑖 𝐿𝑖 ]

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051
𝑐 𝑠 𝑖 𝑠 ⁡ 𝑖 0 ⁡0⁡ ⁡0⁡ ⁡0
−𝑠 ⁡ 𝑖 𝑐 𝑠 𝑖 0 ⁡0⁡ ⁡0⁡ ⁡0
𝜆𝑖 = 0 0 1 0 ⁡0⁡ ⁡0 (25)
⁡0⁡ ⁡0⁡ 0 𝑐 𝑠 𝑖 𝑠 ⁡ 𝑖 0
⁡0 ⁡0⁡ 0 −𝑠 ⁡ 𝑖 𝑐 𝑠 𝑖 0
[ ⁡0 ⁡0⁡ 0 0 0 1]
th
where⁡ 𝑖 means the inclination angle of the i element relative to the horizontal (Fig. 3).

The vector of nodal displacements S=[S1, S2, …, Sn]T can be obtained by the following

relation ship:

[𝐾] ∙ [𝑆] = [ ] (26)

where F=[F1, F2, …, Fn]T is the vector of the nodal forces applied to the lining under the

global coordinates. S=[S1, S2, …, Sn]T is the vector of the nodal displacements, which consists

of three displacements for each node.

After obtaining the displacement vector S under the global coordinates, the nodal

displacements under the local coordinates could be easily calculated. Thereafter, the nodal

stress characteristics under the local coordinates could be calculated Eq. (26).

The stiffness of shear springs (ks) and normal springs ( 𝑛) are respectively computed by the
normal apparent stiffness ( 𝑛) and shear apparent stiffness ( 𝑠 ). The shear stiffness 𝑠(𝑖) and

normal stiffness 𝑛(𝑖) ⁡are given by the formula:


(𝐿𝑖−1 +𝐿𝑖 ) 𝑝 𝑖𝑚 𝑝 𝑖𝑚 (𝐿𝑖−1 +𝐿𝑖 )
𝑠(𝑖) = 𝑠(𝑖) ∙[ ∙ 1] = ∙ (1 − 𝑝 )∙ (27)
2 𝑖 𝑖𝑚 + 0 ∙ 𝑖 2

(𝐿𝑖−1 +𝐿𝑖 ) 𝑝𝑛 𝑖𝑚 𝑝𝑛 𝑖𝑚 (𝐿𝑖−1 +𝐿𝑖 )


𝑛(𝑖) = 𝑛(𝑖) ∙[ ∙ 1] = ∙ (1 − 𝑝 ) ∙ (28)
2 𝑛𝑖 𝑛 𝑖𝑚 + 𝑛 0 ∙ 𝑛 𝑖 2

where is the apparent stiffness, which is given by the 𝑝⁄ ratio. The relationship (Oreste,

2007) between the support deformation and reaction pressure 𝑝 is adopted in the analysis

(Fig. 5):
𝑝 𝑖𝑚
𝑝 = 𝑝 𝑖𝑚 ∙ (1 − 𝑝 ) (29)
𝑖𝑚 + 0 ∙

where 0 and 𝑝 𝑖𝑚 are respectively the initial stiffness and the maximum reaction pressure of

the rock mass.

The maximum normal reaction pressure⁡𝑝𝑛 𝑖𝑚 can be evaluated by:


2∙𝑐∙𝑐 𝑠 1+𝑠𝑖𝑛 𝜎 +𝜎 𝑣
𝑝𝑛 𝑖𝑚 = + 1−𝑠𝑖𝑛 ∙ ∙ 1−𝑣 (30)
1−𝑠𝑖𝑛 2

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051
where 𝜑, 𝑐⁡and 𝑣𝑠 ⁡are respectively the friction angle, cohesion and Poisson’s ratio of the rock

mass. The active loads (𝑝𝑣𝑖 𝑝ℎ𝑖 ) are calculated by Eq. (20) in the analysis.

The initial normal stiffness is determined as follows (Do et al., 2014; Du et al., 2021):
𝐸
𝑛0 = 1+𝑣 ∙ (31)
𝑅

where is a dimensionless factor ( =2.2 is adopted in the calculation); 𝐸𝑠 means the

Young’s modulus of rock mass.

The active loads (𝑝𝑣𝑖 𝑝ℎ𝑖 ) used in HRM are calculated by means of the CCM (𝑝𝑖 ). In the

present analysis both methods are combined by the active loads which correspond to the

pressure 𝑝𝑖 calculated in CCM.

4. Validation with analytical solutions

The analytical solutions play an important role as they can be used to validate numerical

calculations. This section shows the comparison between the results of the present method

and an analytical solution.

Erdmann (1983) presented relatively simple analytical solutions, in which the maximum

normal force Nmax and the maximum bending moment Mmax of the tunnel lining are,

respectively, evaluated as follows:


𝛼 𝛽
𝛼∙𝑅 𝜎 +𝜎ℎ (1+ (1+ )+ )∙𝑅 𝜎 −𝜎ℎ
1 4(1+ )
𝑁𝑚𝑎𝑥 = 𝛼∙𝛽 · + 𝛼∙(3− ) 𝛽∙(5−6 ) 𝛼∙𝛽 · (32)
𝛼+ + 2 1+ (3−4 )(1+ + + 2
1+ 1 ) 4(3−4 )(1+ ) 1 (3−4 )(1+ )

𝛽
(1+ )∙𝑅 𝜎 −𝜎ℎ
(1+ )
𝑀𝑚𝑎𝑥 = 𝛼∙(3− ) 𝛽∙(5−6 ) 𝛼∙𝛽 · (33)
2+ + + 2
6(3−4 )(1+ ) 4(3−4 )(1+ ) 6(3−4 )(1+ )

where and are the relative stiffness, which can be given, respectively, as follows:
𝐸 𝑅3 𝐸𝑅
= , = (34)
𝐸𝐼 𝐸

Where Es means the rock mass elasticity modulus; and R is tunnel radius. The positive

direction of the internal forces used in the analysis is given in Fig. 6.

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051
Figures 7 and 8 show the results of Mmax and Nmax calculated by the analytical method and the

present method, respectively. Calculations are conducted assuming tunnel radius R=6.02 m,

lining thickness t=0.5 m, lining Young's modulus E=29.5 GPa, rock mass unit weight γ=25

kN/m3, σ𝑐𝑖 = 50 MPa, 𝑚𝑖 = 10, GSI=50, D=0, tunnel depth H=400 m. Stress release

coefficient values λ varying from 0.3 to 0.7 are considered.

From Fig. 7, it is found that the Mmax of both methods decreases with the increase of the stress

release coefficient λ. The results of the present method coincide with Erdmann’s method

(1983) for all examined stress release coefficients. The difference of the results in terms of

Mmax for both methods is lower than 1.5%. Fig. 8 gives the same conclusion regarding the

normal forces. The difference between the two methods in terms of normal forces is lower

than 1.4%.

Figure 8 presents the results of Nmax calculated by the present method are in excellent

agreement with analytical calculations, which indicates that the present method is validated

for the interaction evaluation between a Hoek-Brown rock mass and a tunnel support

considering the stress release due to the tunnel excavation.

5. Parametric impact and Discussion

Since the rock mass parameters effect on the lining performance was studied by Oreste

(2007), only the influence of the stress release coefficient λ, tunnel depth H, and the

parameters of Hoek-Brown failure criterion, e.g. 𝜎𝑐𝑖 and GSI, on the lining internal forces

(bending moment M and normal force N), are presented herein. The aim is to provide a

reference for the design of deep-buried tunnels excavated in Hoek-Brown rock mass.

5.1. Impact of λ

The impact of λ on the lining behavior is studied while the other parameters are kept constant

as shown in Section 4. Values of λ varying from 0.3 to 0.7 are considered. The influence of λ

on the bending moments and normal forces is shown in Fig. 9.

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051
Fig. 9 shows that the coefficient λ has a great impact on M and N. The M and N decrease with

the increase of λ. For example, the magnitude of M at the tunnel crown decreases from 22.17

kN.m/m to 5.48 kN.m/m when λ increases from 0.3 to 0.7, however, the magnitude of N

decreases from 0.98 MN/m to 0.04 MN/m. The reason is that the increase of λ leads to the

decrease of active loads applied to the tunnel lining, which causes a decrease of the lining

internal forces. The maximum absolute values of M appear at the invert arch, sidewall and

tunnel crown. However, the maximum absolute values of N only occur at the tunnel sidewall.

5.2. Impact of the tunnel depth H

The impact of H on M and N was examined by keeping the related calculation parameters

constant, as shown in Section 4, and considering a stress release coefficient value of λ =0.3.

Tunnel depth H is varied from 200m to 800m and its effect on the lining performance is

presented in Fig. 10. The internal forces are significantly influenced by the tunnel depth. This

influence increases with the increase of tunnel depth. Specifically, the values of M and N

increase as the tunnel depth increases. This is because the impact of tunnel depth is dual. On

the one hand, the increase of tunnel depth increases the load applied to the tunnel lining. On

the other hand, it reduces the horizontal stress coefficient K0. Therefore, the magnitude of M

at the tunnel crown increases from 17.19 kN.m/m to 77.13 kN.m/m when H increases from

200 to 800 m. Similarly, the magnitude of N increases from 0.44 MN/m to 2.38 MN/m. The

maximum absolute values of M appear at the tunnel crown, sidewall and invert. The

maximum absolute values of N are also located at the tunnel sidewall. This may be due to the

decrease of 𝐾0 ⁡with the increase of the tunnel depth.

5.3. Impact of unconfined compression strength of the rock mass 𝝈𝒄𝒊

In this section, the impact of 𝜎𝑐𝑖 (𝜎𝑐𝑖 =10, 25, 50, 75, 100 MPa) on M and N is shown. Clearly,

this effect is significant. The M and N decrease rapidly as 𝜎𝑐𝑖 increases. The magnitude of M

at the tunnel crown decreases from 106.98 kN.m/m to 5.66 kN.m/m with the increase of 𝜎𝑐𝑖

from 10 to 100 MPa. Similarly, the magnitude of N at the tunnel crown decreases from 5.58

MN/m to 0.11 MN/m. The maximum absolute N still occur at the tunnel sidewall. The

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051
absolute maximum absolute M corresponding to different 𝜎𝑐𝑖 appear at the tunnel crown,

sidewall and invert arch.

5.4. Impact of Geological Strength Index (GSI)

The effect of GSI on the tunnel lining performance is studied herein. Carranza-Torres and

Fairhurst (2000) pointed out that a disturbed rock mass will have a GSI close to 30. Since the

rock mass surrounding the excavated tunnel could not remain intact, herein GSI values from

30 to 60 are taken into consideration. Fig. 12 shows that the influence of GSI on the lining

internal forces decreases as the GSI increases. The M and N of tunnel lining decrease as the

GSI increases. Specifically, the magnitude of M at the tunnel crown decreases from 110.64

kN.m/m to 8.91 kN.m/m with the increase of GSI from 30 to 60. Similarly, the magnitude of

N at the tunnel crown decreases from 4.15 MN/m to 0.13 MN/m. This is because higher GSI

values represent lower discontinuities and better rock mass surface conditions. The integrity

and the self-stability of the excavated rock mass are therefore better.

One could also find that the maximum absolute M are present at the tunnel crown, sidewall

and invert. In contrast, the tunnel sidewall gives the maximum normal forces.

6. Conclusions

Based on two combined calculation methods, the CCM and the HRM, a calculation process is

presented to design deep-buried tunnels in rock mass obeying the Hoek-Brown failure

criterion. The present method is validated by comparison against existing analytical solutions.

Compared to other analytical methods, the proposed method is able to quickly calculate the

internal forces of structure along the whole tunnel lining considering the rock mass stress

release for deep-buried tunnels. Thereafter, the effect of different parameters, e.g. λ, H, 𝜎𝑐𝑖

and GSI, on the internal forces of tunnel lining in terms of bending moment M and normal

force N are shown and discussed. The conclusions are obtained:

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051
1. All studied parameters, e.g. λ, H, 𝜎𝑐𝑖 and GSI, have a significant influence on M and

N Specifically, the internal forces of tunnel lining along the whole tunnel decrease as

λ, 𝜎𝑐𝑖 and GSI increase, however, the internal forces increase as H increases, with

higher external pressure and lower rock quality, resulting in higher lining stresses.

2. The maximum absolute values of M appear at the tunnel crown, sidewall and invert

arch. Nevertheless, the maximum absolute values of N only occur at the tunnel

sidewall.

The present method can provide a reference for the design of deep-buried tunnels

excavated in rock mass.

Acknowledgements

The present research work was supported by the National Natural Science Foundation of

China (Grant No. 52108305), the Natural Science Foundation of Jiangsu Province (Grant No.

BK20210256), and the Jiangsu Provincial Double-Innovation Doctor Program (Grant No.

JSSCBS20210068), which are gratefully acknowledged. The authors would also like to thank

the reviewers and the editor and their comments have improved the quality of this

manuscript.

Notation

a material constant of the rock mass

𝑐′ equivalent cohesion for the rock mass

𝐷 a disturbance factor

EA normal stiffness of lining

EI bending stiffness of lining

𝐸𝑟𝑚 rock mass modulus⁡(MPa)

F=[F1, F2, …, Fn]T vector of the nodal forces

GSI Geological Strength Index

H tunnel depth

h initial node

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051
j final node
̅𝑖 local stiffness matrix under local Cartesian coordinates

kn stiffness of normal springs

ks stiffness of shear springs

Li length of element

M bending moment of lining

𝑚𝑏 material constant of the rock mass

Mmax maximum bending moment of lining

𝑚𝑖 a material constant of rock mass

N normal force of lining

Nmax maximum normal force of lining

phi horizontal active loads

𝑝𝑖 the internal pressure of the support

𝑝𝑖𝑐𝑟 critical load

𝑝𝑠 external active loads

pvi vertical active loads

R tunnel radius

s material constant of the rock mass

S=[S1, S2, …, Sn]T vector of the nodal displacements

u and v axial and transversal displacement

x and y local Cartesian coordinates

X and Y global Cartesian coordinates

and the relative stiffness

𝜎1 principal maximal stress

𝜎3 principal minimal stress

𝜎𝑖 ⁡ rock mass stress after excavation

𝜎𝑐𝑖 unconfined compression strength of the rock mass

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051
𝑣𝑟𝑚 Poisson’s ratio of rock mass

𝜑′ equivalent friction angle for the rock mass

𝛾⁡ unit weight of rock mass

θ rotation

λ stress release coefficient

𝜆𝑖 transformation matrix of element

References
AFTES (2011) Recommendations on the Convergence Confinement Method. AFTES.

Almog E, Mangione M, Cachia G (2015) Ground relaxation in segmental lining design using the Convergence-

Confinement Method. Underground Design and Construction Conference.

Barton N (1975) Classification of rock mass to distinguish self-supporting tunnels from those requiring support.

Proceeding of Society of Mining Engineers Meeting, Salt Lake City.

Barton N (2002) Some new Q-value correlations to assist in site characteristics and tunnel design. International

Journal of Rock Mechanics and Mining Sciences 39(2): 185–216.

Cai M, Kaiser PK, Uno H, Tasaka Y, Minami M (2004) Estimation of rock mass deformation modulus and

strength of jointed hard rock mass using the GSI system. International Journal of Rock Mechanics and

Mining Sciences 41(1): 3-19, https://doi.org/10.1016/S1365-1609(03)00025-X.

Cao C, Shi C, Lei M, Yang W and Liu J (2018) Squeezing failure of tunnels: a case study. Tunnelling and

Underground Space Technology 77(7): 188–203.

Carranza-Torres C, Fairhurst C (1999) The elasto-plastic response of under rock mass excavations in rock mass

that satisfy the Hoek–Brown failure criterion. International Journal of Rock Mechanics and Mining

Sciences 36 (6): 777-809, https://doi.org/10.1016/S0148-9062(99)00047-9.

Carranza-Torres C, Fairhurst C (2000) Application of the convergence-confinement method of tunnel design to

rock mass that satisfy the Hoek-Brown failure criterion. Tunnelling and Underground Space Technology

15(2): 187-213, https://doi.org/10.1016/S0886-7798(00)00046-8.

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051
Cui G, Wang X and Wang M (2019) Field tests on large deformation control measures of surrounding rock of

deep tunnels in fault zones with high geostress. Chinese Journal of Geotechnical Engineering 41(7): 1354–

1360 (in Chinese).

Dias D, Kastner R (2013) Movements caused by the excavation of tunnels using face pressurized shields—

Analysis of monitoring and numerical modelling results. Engineering Geology 152: 17–25,

https://doi.org/10.1016/j.enggeo.2012.10.002.

Do NA, Dias D, Oreste P, Djeran-Maigre I (2014) A new numerical approach to the hyperstatic reaction method

for segmental tunnel linings. International Journal for Numerical and Analytical Methods in

Geomechanics 38(15): 1617-1632, https://doi.org/10.1002/nag.2277.

Do NA, Dias D (2017) A comparison of 2D and 3D numerical simulations of tunnelling in soft soils.

Environmental Earth Sciences 76(3): 1-12, https://doi.org/10.1007/s12665-017-6425-z.

Du DC, Dias D, Do NA (2021) A Simplified Way to evaluate the effect of temperature on a circular tunnel.

Geotechnics 1: 385-401.

Du DC, Dias D, Do NA (2020) Lining performance optimization of sub-rectangular tunnels using the

Hyperstatic Reaction Method. Computers and Geotechnics 117(103279): 1-16.

Du DC, Dias D, Do NA, Oreste PP (2018) Hyperstatic Reaction Method for the Design of U-Shaped Tunnel

Supports. International Journal of Geomechanics 18(6):04018030.

Erdmann J (1983) Vergleich ebener und Entwicklung räumlicher Berechnungsverfahren fürTunnel. Bericht

Institut für Statik, TU Braunschweig, 830–840.

González-Nicieza C, Álvarez-Vigil AE, Menéndez-Díaz A, González-Palacio C (2008) Influence of the depth

and shape of a tunnel in the application of the convergence–confinement method. Tunnelling and

Underground Space Technology 23(1): 25-37, https://doi.org/10.1016/j.tust.2006.12.001.

Hoek E, Brown ET (1980) Under rock mass Excavations in Rock. The Insitute of Mining and Metallurgy,

London.

Hoek E, Kaiser PK, Bawden WF (1995) Support of under rock mass excavations in hard rock. Rotterdam: A.A.

Balkema.

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051
Hoek E, Carranza-Torres C, Corkum B (2002) Hoek-Brown failure criterion-2002 edition. In: Proceedings of

the Fifth North American Rock Mechanics Symposium, Toronto, Canada, 1: 267-273.

Hoek E, Brown ET (2019) The Hoek–Brown failure criterion and GSI-2018 edition. Journal of Rock Mechanics

and Geotechnical Engineering 11(3): 445-463, https://doi.org/10.1016/j.jrmge.2018.0 8.001.

Huang D, Li XD, Song WC (2021) Precise grading of surrounding rock based on numerical calculation of

jointed rock mass. Proceedings of the Institution of Civil Engineers -Geotechnical Engineering 1: 1-12,

https://doi.org/1g/10.1680/jgeen.21.

Jin H, Yu K, Gong Q, Zhou S (2018) Load-carrying capability of shield tunnel damaged by shield shell

squeezing action during construction. Thin-Walled Structures 132(11): 69-78. https://doi.org/10

.1016/j.tws.2018.07.057.

Lei S and Wei Z (2020) Study on the mechanism of circumferential yielding support for soft rock tunnel with

large deformation. Rock and Soil Mechanics 41(3):1–8 (in Chinese).

Londe P (1988) Discussion on the determination of the shear stress failure in rock mass. ASCE Journal of

Geotechnical Engineering Division 14(3): 374-6.

Mishra S, Rao KS, Gupta NK, Kumar A (2018) Damage to shallow tunnels in different geomaterials under

static and dynamic loading. Thin-Walled Structures 126(5): 138-149,

https://doi.org/10.1016/j.tws.2017.11.051.

Oreste P (2003) A Procedure for Determining the Reaction Curve of Shotcrete Lining Considering Transient

Conditions. Rock Mechanics and Rock Engineering 36(3): 209–236.

Oreste P (2007) A numerical approach to the hyperstatic reaction method for the dimensioning of tunnel

supports. Tunnelling and Underground Space Technology 22: 185–205.

https://doi.org/10.1016/j.tust.2006.05.002.

Osgoui R, Unal E (2009) An empirical method for design of grouted bolts in rock tunnels based on the

geological strength index (GSI). Engineering Geology 107: 154–166,

https://doi.org/10.1016/j.enggeo.2009.05.003.

Panet M, Guénot A (1982) Analysis of convergence behind face of a tunnel. In Proceeding Tunnelling 82, IMM,

London, 197–203.

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051
Panet M, Givet P, Guilloux A, Duc J, Piraud J, Wong H (2001) The convergence–confinement method, Press

ENPC.

Rehman H, Naji AM, Ali W et al. (2020) Numerical evaluation of new Austrian tunneling method excavation

sequences: a case study. International Journal of Mining Science and Technology 30(3): 381–386.

Svoboda T, Hilar M (2015) Probabilistic analysis of tunnel loads using variance reduction. Proceedings of the

Institution of Civil Engineers-Geotechnical Engineering 168(4): 348-357. http://dx.doi.org/10.1680/geng.

Taghizadeh H, Zare S, Mazraehli M (2020) Analysis of rock load for tunnel lining design. Geotechnical and

Geological Engineering 38: 2989–3005, https://doi.org/10.1007/s10706-020-01202-y.

Zhao W, Zhao W, Zhang X and Yang C (2018) Construction technology of internal support ring arch

reinforcement for initial deformation of soft rock tunnel. Highway 63(3): 274–278.

Zhang J, Wang Z, Feng J et al. (2021) Deformation control for large-section tunnel construction in fractured

carbonaceous slate. Proceedings of the Institution of Civil Engineers–Geotechnical Engineering 1: 1-14,

https://doi.org/10.1680/jgeen.20.00212.

Figure captions
Figure 1. (a) Cross-section of the excavation; (b) Cross-section of a circular support

Figure 2. Three basic components in CCM: LDP, GRC and SCC

Figure 3. Analysis scheme of tunnel lining in the HRM under the global Cartesian coordinates (X and Y)

Figure 4. Beam-type elements under the local Cartesian coordinates (x and y)

Figure 5. Relationship between the support normal deformation and reaction pressure 𝑝 in Eq. (29)

Figure 6. Positive direction of the lining internal forces (M, N) in the analysis

Figure 7. Maximum bending moments Mmax obtained from the present method and the analytical method

considering different stress release coefficient λ

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051
Figure 8. Maximum normal forces Nmax obtained from the present method and the analytical method considering

different stress release coefficient λ

Figure 9. Effect of λ on M and N: (a) M; (b) N

Figure 10. Effect of H on M and N: (a) M; (b) N

Figure 11. Effect of 𝜎𝑐𝑖 on M and N: (a) M; (b) N

Figure 12. Effect of GSI on M and N: (a) M; (b) N

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.
Accepted manuscript doi:
10.1680/jgeen.22.00051

Downloaded by [ Southeast University] on [21/07/22]. Copyright © ICE Publishing, all rights reserved.

You might also like