Sie sind auf Seite 1von 30

THE INVESTIGATION OF SIMPLIFIED

EXCAVATION METHOD ON TUNNEL


BEHAVIOR

BY: ALEX YOON CHAN YIP


INTRODUCTION
• A tunnel excavation is clearly a 3D problem, considering the third dimension should intuitively lead to more
accurate predictions.
• Simplified procedures that allow us to consider 3D effects within a simplified 2D plane strain analysis are
favourable in geotechnical design.

Figure 1: 3D arch support and 2D FE-approximation with support pressure


INTRODUCTION
•• Available research studies are not conclusive in demonstration that the accuracy of predictions
achievable in practical applications of full 3D methods is higher than those of properly used 2D
analyses with indirect incorporation of 3D effects.
• 2D methods typically require specification of just a few or any one parameter “” which integrates
the influence of all of the aforementioned factors that need to be considered in 3D analyses. It is
not possible to find a direct empirical relationship between “” and all the modelling aspects that
enter into the 3D simulations.
• Thus in practical applications, the parameter “” is often merely estimated.
• It must pointed out that the if the aim of analysis is the evaluation of the tunnel face stability,
appropriate 3D modelling is required and irreplaceable. Evaluation of the tunnel face stability is
however outside the scope of this presentation.
STIFFNESS REDUCTION METHOD
(-METHOD)
• Calculation phases of stiffness reduction
• method:
(a) Initial stresses are computed with gravity
loading but with reduced core stiffness
(a.E) and
(b) When excavation is occur, support
pressure; a reduced support pressure is
equal to zero
 factor which ranging from 0 to 1 (0 = no
stiffness of material in the tunnel opening; 1
= same stiffness of material inside as around
the outside the tunnel opening zone).
STRESS REDUCTION METHOD
(-METHOD)
• Starting from initial stresses, the first
• calculation phase tunnel installation is
simulated by switching off ground
elements inside the tunnel and reducing
the initial ground pressure which acting on
the inside of the tunnel with a factor of 
which ranging 0<  <1.
• In the second calculation phase, the
shotcrete lining is installed and the
remaining load is distributed over the
lining and the ground.
CONVERGENCE CONFINEMENT METHOD
(λ-METHOD)
• Starting from initial stresses, the first
• calculation phase tunnel installation is
simulated by switching off ground
elements inside the tunnel and reducing
the initial ground pressure which acting on
the inside of the tunnel with a factor of 
Figure 2: Convergence Confinement Curves (After Panet & Guenot, 1982 and Hoek, 2007)
which ranging 0<  <1.
• In the second calculation phase, the
shotcrete lining is installed and the
remaining load is distributed over the
lining and the ground.

Figure3: Change in confinement loss on tunnel centreline


PARAMETRIC ANALYSIS
• A typical tunnel in Kuala Lumpur limestone formation.
• The presence of groundwater is considered explicitly in the models, assuming phreatic water
surface at the ground surface.
• The soil and rock behaviour is assumed to be governed by a linear elastic-perfectly plastic
constitutive relation based on the Mohr Coulomb failure criterion.
• The presences of karstic cavities are not modelled explicitly.
• Unconfined compressive strength of rock is assumed as 20MPa.
ROCK PROPERTIES

 In accessing the rock properties, Mohr-Coulomb material parameters are fitted


to the Hoek Brown failure curve by adopting RocLab software.
DESIGN OF SHIELD TUNNEL LINING
 The most significant potential loads acting on
underground structures are earth pressure and
water pressure.
 To estimate the actual working load acting on
the lining, calculations were carried out using
two kinds of combination of vertical load;
(i) Effective overburden earth pressure together
with water pressure;
(ii) Terzaghi’s effective loosening pressure
together with water pressure.
ANALYTICAL METHODS
 Two empirical solutions were adopted and
compared;
(i) Total Stress and Effective Stress Methods (after
JSCE, 2001)
(ii) Muir Wood (1975) and Curtin (1976) Methods
PARAMETRIC ANALYSIS
The analysis is done in 3 main headings; 1. Comparison of simplified excavation method:
• Stiffness Reduction Method (-Method)
a) Induced movement on tunnel lining
• Stress Reduction Method (-Method)
b) Induced forces on tunnel lining • Convergence Confinement Method (-Method)
c) N-M Interaction Chart 2. Compare 2D, 3D & Analytical Methods
CONVERGENCE CONFINEMENT METHOD
λ - METHOD

• A key assumption which require the


• calculation of both LDP and GRC curves to
be adjusted in order to incorporate the
effects of TBM pressure .

• The relaxation at the location of lining


installation is obtained by constructing LDP
curve as proposed by Vlachopoulos and
Diederichs (2009) which was modified here
for the inclusion of TBM pressure.
SENSIBILITY ANALYSIS

Width, w (m) Height, h (m) Total Displacement, u (mm) • The sensibility analysis indicated the
150 100 21.72
200 100 21.71 geometry of the model only have minor
influences on the results. Model geometry
with width of 200m and height of 100m
Width, w (m)
200
Height, h (m)
75
Total Displacement, u (mm)
21.67
satisfied the criteria of Meissner (1996)
200 100 21.71 and Moller (2006).
• The appropriate mesh size was
Mesh Size Total Displacement, u (mm) implemented with the same model
Very coarse
Coarse
18.01
18.90
characteristics as used to determine the
Medium
Fine
21.71
22.60
optimum geometry size which the chosen
Very fine 22.60 model of 200m x 100m (width x height) as
mentioned.
• For the remainder of the numerical
analysis, a fine mesh size is chosen as
refined mesh would not produce much
more accurate results.
PARAMETRIC STUDIES
a) INDUCED MOVEMENT ON TUNNEL LINING

Induced displacement on Tunnel Lining  Different maximum value of relaxation


5.00 0.35
factor was achieved in all three ,  and λ
0.30
method in the numerical analysis as the
4.80
model could not attained equilibrium state.
0.25
 It is shown that at same relaxation factor, 
Total Displacement, u (mm)

4.60
λ - METHOD

Volume Loss, VL (%)


0.20 β - METHOD method shows the lowest displacement,
a - METHOD
4.40
VL (λ - METHOD) followed by λ method and  method.
0.15 VL (β - METHOD)
4.20
VL (a - METHOD)  λ method shows yielding behaviour before
0.10
soil body collapse.
4.00
0.05

3.80 0.00
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00

Relaxation Factor
PARAMETRIC STUDIES
b) INDUCED FORCES ON TUNNEL LINING

Induced Axial Load on Tunnel Lining  It is noted that an increase in relaxation


3100 factor will result in a reduction normal
3000
forces as presented. The higher the
relaxation factor, the lower the external
2900
loads act on the tunnel lining.
2800
 It is unexpected that the hoop forces
Axial Load, N (kN)

2700 λ - METHOD
β - METHOD
remain constant from relaxation factor
2600 a - METHOD ranging from 0 to 0.5 for case of  method
2500 which the behaviour is similar to the
2400
induced movement .
2300  For  method, the convergence only start
to occur when relaxation factor is more
2200
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 than 0.5 as recommended by Laabmayr &
Relaxation Factor Swoboda (1978).
  method shows the least convergence
followed by λ method and  method.
PARAMETRIC STUDIES
b) INDUCED FORCES ON TUNNEL LINING

Induced Bending Moment on Tunnel Lining  It could observed that greater bending
50.00
moment generated before soil body
45.00
collapse from λ method.
40.00  This phenomenon could be attributed to
35.00 the fact that stress relaxation on the tunnel
Beding Moment, M (kNm)

30.00
boundary has induced displacement on the
25.00
λ - METHOD
β - METHOD
surrounding ground where resulting
a - METHOD development of failure zone. The self-
20.00
weight of failure ground could be one of
15.00
the factors which leading to an increased in
10.00
induced bending moment on the tunnel
5.00 lining.
0.00
0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00

Relaxation Factor
PARAMETRIC STUDIES
c) N-M INTERACTION CHART

 - Method  - Method – Method

 According to the plotted N-M capacity diagram for C60 concrete with a nominal 0.13% reinforcement ratio, the overall
increase of confinement is actually contributing to stability.
 This clearly demonstrates that analysis of segmental tunnel lining require investigation of both upper and lower TBM
confinement ratios to ensure a robust design.
3D ANALYSIS OF SEGMENTAL LINING

 In the simplified model, a 3D simulation of


a tunnel supported by means of segmental
lining was modelled.
 The influence tunnelling process such as
the shield machine and the construction
loads (face support pressure and grouting
pressure) was taken into consideration.
 To avoid long lasting computations, the
step by step tunnel installation simulation
had simplified to only two phases. The first
phase is used to install a more or less
complete tunnel. To this end, soil elements
are switched off and lining elements are
switched on over a considerable tunnel
length to support the previous excavation.
The second phase is used to model a single
FACE SUPPORT PRESSURE
CALCULATIONS FOR SHIELD TUNNELLING
2500
 Face stability calculations in practice are
2000
typically performed using the limit
1500
Ere(ϕ)
equilibrium method.
1000

500

0
0 20 40 60 80 100
ϕ (degree)
1200
Support pressure at tunnel crown (kN/m2)

1000

800
OPERATIONAL RANGE
600

400

200

0
0 20 40 60 80 100 120
3D ANALYSIS OF SEGMENTAL LINING
 The total tunnel displacement of 3.227mm
is induced.
 Axial force acting on lining = 3044kN/m
 Induced bending moment = 35.06kNm/m
 The axial force on tunnel lining from
distance of from 0 to 48m is constant. It
noted that the tunnel lining is not carrying
full overburden load due to the soil arching.
 The result of total displacement of tunnel
and axial load acting on the lining is
relatable.
 For low TBM confining pressure, ground
deformation is maximized but giving low
lower bound lining forces whereas for high
TBM confining pressure gives lower bound
ground deformations but upper bound
lining forces.
REMARKS ON 2D SIMPLIFIED
EXCAVATION METHODS
On the basis of comparison with 3D numerical model, the better 2D deconfinement method was
estimated using the error function. Depending on what is the most important parameter need to be
monitored, different 2D deconfinement method should be adopted:
• When considering only the structural forces induced in the tunnel,  - Method should be chosen
as it gives the upper bound forces;
• When considering the vertical settlement develop on the soil surface,  - Method should be
chosen as it gives the highest tunnel displacement;
• When considering overall the behaviour of the soil and tunnel lining,  - Method is the better
one.
COMPARISON OF NUMERICAL AND
ANALYTICAL METHODS
Analytical Methods 2D Analysis
Induced Forces and 3D Analysis
Movements JSCE, 2001 MUIR WOOD, 1975 & CURTIS, 1976 λ - METHOD β - METHOD a - METHOD
Axial Force (kN) 2739.70 4458.60 2778 2750 2973 3044
Bending Moment (kNm) 46.63 31.24 31.13 31.07 45.61 35.06
Deformation (mm) - 3.06 4.15 4.20 4.28 3.227

λ - METHOD vs JSCE, 2001 λ - METHOD vs Muir Wood, 1975 & Curtis, 1976 λ - METHOD vs 3D Analysis

Axial Force (kN) 1.38% -37.7% -8.74%


Bending Moment (kNm) -49.78% -0.35% -11.21%
Deformation (mm) - 35.63% 28.61%

 The maximum induced forces and displacement on tunnel is comparable for all three 2D simplification
methods. As discussed, λ – METHOD reflect correct tunnel behaviour.
 λ – METHOD over predicted tunnel displacement.
 λ – METHOD under predicted tunnel lining forces.
 A rather good compatibility between λ – METHOD and 3D simulations. (Difference in magnitude is very less!)
CASE STUDY ON CL3 MRT SINGAPORE

 A fully numerical tunnel lining with


appropriate soil model which based on the
construction tunnel profile of Circle Line
Stage 3 (C852), Serangoon Interchange
Station was developed.
46m
 This case study is to validate the step by
step installation of tunnel simulation in 3D.
80m
 Comparison of tunnel lining forces:
63m
2D VS 3D

1. λ – METHOD (CCM) 1. VLM


25mbgl 2. VLM

λ – METHOD (CCM) IS THE MOST ACCURATE


SIMPLIFIED EXCAVATION METHOD?
CASE STUDY ON CL3 MRT SINGAPORE
COMPARISON OF 3D SIMULATIONS AND FIELD STUDY
0  A very good agreement between 3D
0 10 20 30 40 50 60 70 80 90
simulations and field study was achieved.
-10
 It is noted that tunnel movement
Induced Surface Setlement (mm)

-20
measured by BOTDR Distributed Fibre
-30
Optic Strain Sensing is lower than the
settlement marker G2407 and G2408
-40 which is incorrect.
-50  It is observed that settlement markers
measures quite amount of surface
-60
settlement was induced by tunnelling
-70 which is more than 1% Volume Loss.
Distance In Longtudinal Direction (m) However, the BOTDR Fibre Optic shows the
induced movement of tunnel is
G2407 G2408 G2419
R530 R540 BOTDR PLAXIS 3D 1% VL
approximately 1% Volume Loss.
BOTDR PLAXIS 3D 2% VL SM PLAXIS 3D 2% VL SM PLAXIS 3D 1% VL
 Overall, 3D simulations are able to capture
the behavior of induced movement by
CASE STUDY ON CL3 MRT SINGAPORE
SURFACE SETTLEMENT THROUGH
 For CCM, the relaxation factor,
0
0 10 20 30 40 50 60 70 • calibrated to ensure that the 2D andwas 3D
-5
analyses predicted as closely as possible by
-10 relating the tunnel wall deformation to the
Induced Surface Setlemen (mm)

actual physical location along tunnel axis.


-15
 An overall agreement of 2D and 3D method
-20
is good as 2D method accurately estimate
-25
the settlement through shape.
 It is evident that the value of the
-30
settlement increases approaching the
-35 tunnel and the maximum value is at the
tunnel axis due to the symmetry
-40
conditions. Thus, the settlement decreases
Distance In Transverse Direction (m)
away from the center tunnel. This could be
2D 1% VL 3D 1% VL 2D CCM attributed to the fact that stress relaxation
limits the tunnel deformation to the
CASE STUDY ON CL3 MRT SINGAPORE
COMPARISON IN TUNNEL DISPLACEMENT
40  Both 2D and 3D adopting Volume Loss
Method shows a good agreement of tunnel
35
deformation although tunnel advancement
30
is not modelled explicitly in 2D (Wished-In-
Place).
Tunnel Displacement (mm)

25
 Convergence confinement method shows
20
induced highest displacement on tunnel.
 It is noted that 2D VLM shows lower tunnel
15
displacement compared to 2D CCM as no
10
relaxation was allowed in 2D VLM as the
tunnel is WIP.
5
 CCM shows higher tunnel displacement as
agreed in parametric studies.
0

2D CCM 2D VL 3D VL  The results is in a good agreement with


plotted transverse surface settlement as
CASE STUDY ON CL3 MRT SINGAPORE
COMPARISON IN AXIAL FORCE
0
 A very good agreement between CCM and
3D VLM was achieved.
-200
 This shows that CCM was able to allow
-400 tunnel to converge at a similar behavior as
3D. Hence, CCM was able to include tunnel
Tunnel Axial Force (kNm/m)

-600
arching effect in 2D analysis accurately.
-800  It is expected that 2D VLM shows the
-1000
highest axial force as WIP was adopted
thus no relaxation is allowed in tunnel.
-1200

-1400

-1600

Tunnel Location (m)

2D CCM 2D VL 3D VL
CASE STUDY ON CL3 MRT SINGAPORE
COMPARISON IN BENDING MOMENT
100  Both CCM and 3D VLM shows a similar
80
behavior of tunnel lining forces. The results
were still comparable.
60
 As expected that 2D VLM shows the
Tunnel Bending Moment (kNm/m)

40
highest induced bending moment on
20 tunnel lining as the tunnel lining is WIP
where no relaxation was allowed to occur.
0

-20

-40

-60

-80

-100

2D CCM 2D VL 3D VL
CASE STUDY ON CL3 MRT SINGAPORE
COMPARISON IN SHEAR FORCE
80  It was observed that both CCM and 3D
60
VLM shows similar behavior.
 The step by step installation of tunnel
40
which leads to realistic zig-zagging shear
Tunnel Shear Force (kNm/m)

20 force in the ring direction shows peak


shear stress generated.
0
 As expected that 2D VLM shows the
-20
highest induced shear force on tunnel
-40
lining as the tunnel lining is WIP where no
relaxation was allowed to occur.
-60

-80

Tunnel Location (m)

2D CCM 2D VL 3D VL
CONCLUDING REMARKS
• The result generated by λ method (CCM) is much more reasonable as sign of soil yielding failure was
shown as presented.
• It is believe that the ground will not fail immediately right after excavation as the excavated cavity could
support itself without any added structures which known as “Stand-up time”.
• The development of failure zone which occur around the tunnel could be highlighted by the development
of convergence which only indicated in λ method (CCM).
• This parametric analysis highlighted the fact that the importance of timing of support installation. The
support structure should not be installed too soon when the stresses imposed on the surrounding ground
is still high. However, the support structure should not be installed too late when the ground failure occur.
• In this parametric analysis, the tunnel lining should be installed before the stress relaxation coefficient
(relaxation factor) reached to a value of 0.6 which correspond to a volume loss of 0.27
• Overall a good agreement between λ method (CCM) and 3D step by step installation of tunnel was
attained.

Das könnte Ihnen auch gefallen