International Journal of Technology 12(3) 485-494 (2021)

Received July 2020 / Revised October 2020 / Accepted April 2021

International Journal of Technology

http://ijtech.eng.ui.ac.id

Evaluation of Face Support Pressure Prediction for Earth Pressure Balance

(EPB) Tunnelling using Analytical and 3-Dimensional Finite Element
Modelling

Fahmi Aldiamar1*, Masyhur Irsyam2, Bigman Hutapea2, Endra Susila2, Desyanti1

1Institute of Road Engineering, Ministry of Public Works, Jl A H Nasution No 264, Bandung 40294, Indonesia
2Bandung Institute of Technology, Jl Ganesa No 10, Bandung 40132, Indonesia

Abstract. A shield tunnelling technique is usually selected using earth pressure balance or slurry
methods for tunnel construction in urban areas with soft and saturated ground. Although shield
tunnelling has many advantages, incorrect determination of face pressure could cause ground
surface settlement or lifting during tunnel construction. Numerous approaches for determining face
support pressure have been published internationally, but a suitability evaluation based on local
ground conditions in Indonesia has not been conducted yet. The completion of Mass Rapid Transit
Jakarta (MRTJ) tunnel construction project using the earth pressure balance method, along with its
adequate data, has become a sample case of the effectiveness of each method to determine face
support pressure. The study discussed in this paper aimed to determine the linear relationship
between the calculated value and the actual measurement of face support pressure and to identify
which method most closely represents the actual condition according to the MRTJ case study. An
analytical approach using the limit equilibrium method and the numerical approach using Plaxis 3D
were conducted, followed by statistical evaluation in the terms of coefficient of variation. The result
shows that the limit equilibrium method is effective in predicting the mean value face support
pressure and the upper and lower perimeters for tunnel construction, while the overall face support
pressure result using the shell model of the finite element method are lower than the actual
measured values. The result probably indicates the balance state condition at the tunnel face, and
the additional 80 kPa after the second phase of excavation could indicate the need for greater
pressure for tunnel boring machine movement.

Keywords: Face support pressure; Finite element method; Mass rapid transit; Tunnels

1. Introduction
In urban areas, tunnel construction through soft and saturated ground requires special
caution because unsuitable construction methods can disturb the surrounding
infrastructures causing them to collapse. Under such conditions, shield tunnelling is usually
selected using the earth pressure balance (EPB) method or slurry method. Both methods
reduce the disturbance at the tunnel face and around the excavation area using a tunnel
boring machine (TBM). Although shield tunnelling has many advantages, incorrect
determination of face pressure could cause ground surface settlement or lifting during
tunnel construction. Numerous analytical, empirical, and numerical approaches for

*Corresponding author’s email: [email protected], Tel.: +62-22-7834487; Fax.: +62-22-7834487

doi: 10.14716/ijtech.v12i3.4158
486 Evaluation of Face Support Pressure Prediction for Earth Pressure Balance (EPB) Tunnelling
using Analytical and 3-Dimensional Finite Element Modelling

determining face pressure have been published in international journals or in technical

guidelines, but a suitability evaluation according to local ground conditions in Indonesia
has not yet been conducted. The Mass Rapid Transit Jakarta (MRTJ) tunnel construction
project was recently completed using the EPB method, and face pressure data according to
earth pressure gauges, along with soil profile and the results from in situ and laboratory
testing, are available as a sample case demonstrating the effectiveness of the limit
equilibrium method and the 3-dimensional (3D) finite element method to determine the
face support pressure. The study discussed in this paper aimed to determine the linear
relationship between the calculated value and the actual measurement of face support
pressure and to identify which method most closely represents the actual condition
according to the MRTJ case study.
The EPB method is based on equilibrium between soil pressure and water pressure
with jacking force applied on the cutterhead. A screw conveyor has the ability to adjust or
control the face pressure during an excavation. For a tunnel constructed below the ground
water level, the length of the screw conveyor must be designed to withstand hydrostatic
pressure and transform water pressure into atmospheric pressure. An illustration of the
EPB machine used in the MRTJ project is shown in Figure 1a, and the position of the
pressure gauge instrumentation is shown in Figure 1b. pressure gauges were installed to
measure the soil pressure exerted on the cutterhead and to inform the machine operator as
to whether the estimated pressures were still safe.

(a) EPB tunneling machine (b) Position of the earth pressure gauges
Figure 1 Illustration of an EPB tunnel machine and the position of the earth pressure gauges
(SOWJ, 2015)

2. Methods
2.1. Face Support Pressure Calculation
According to the German Tunneling Committee (ITA-AITES, 2016), several methods
can be used to determine face support pressure. Since they are simple and practical, limit
equilibrium and limit state methods are usually used in the design phase based on the
assumption that the ground stress distribution, which is influenced by the elevation of the
ground water level and the unit weight of the soil, could indicate the need for supporting
face pressure to prevent collapse.
In the limit equilibrium method, the failure mechanism basically refers to the support
force divided by the acting force on the sliding wedge mechanism. The support force and
soil shear strength act as stabilizing forces, while the sliding wedge weight and the load
from the prism are defined as destabilizing forces. The equilibrium conditions are
formulated on the sliding surface (Figure 2) in perpendicular and parallel directions.
Aldiamar et al. 487

where Ere : support force due to earth pressure, G : weight of the wedge, Pv : vertical load from the soil prism, T : shear
force on the vertical slip surface,  : sliding angle, ’ : friction angle of the soil, c : cohesion of the soil, D : e shield diameter,
Q : shear force on the inclined surface,  : unit weight of the soil

Figure 2 Forces on the sliding wedge mechanism (Anagnostou, 2012)

The required support pressure can be calculated using Equation 1. The equilibrium
condition is based on variations in the sliding angle of the wedge ( ), which determines the
support force and which has an impact on the highest support face pressure (Emax,re).
𝐷2
(𝐺+𝑃𝑣 ).(sin(𝜗)−cos(𝜗).tan(𝜑 ′ 2 ))−2.𝑇−𝑐′2
sin⁡(𝜗)
𝐸𝑟𝑒 (𝜗) = (1)
sin(𝜗).tan(𝜑 ′ 2 )+cos(𝜗)

The vertical load force from the soil prism on the wedge ( 𝑃𝑣 ) is calculated using
Equation 2 by multiplying the area on the top of the wedge by the vertical effective stress
acting on the wedge.
𝐷
𝑃𝑣 = 𝐴. 𝜎𝑣 (𝑡𝑐𝑟𝑜𝑤𝑛 ) = 𝐷. tan 𝜗 . 𝜎𝑣 (𝑡𝑐𝑟𝑜𝑤𝑛 ) (2)
𝑐𝑟𝑖𝑡

where 𝜎𝑣 (𝑡𝑐𝑟𝑜𝑤𝑛 ) is the vertical surcharge from the prism on the wedge, 𝐴 is the cross-
sectional area of the silo (m2), 𝜗𝑐𝑟𝑖𝑡 is the sliding angle of the wedge in the highest support
face pressure.
Vertical stress at elevation z can be calculated considering the soil arch, which is taking
into account the total weight of the soil above the tunnel if the overburden is less than twice
the tunnel diameter (Equation 3) and using Equation 4 if the overburden is otherwise. The
coefficients of lateral earth pressure from recent studies i.e. Kirsch and Kolymbas (2005),
and Girmscheid (2008) are shown in Table 1.
𝜎𝑣 (𝑧) = 𝛾1,𝑎𝑣 . 𝑧 + 𝜎𝑠 for 𝑡𝑐𝑟𝑜𝑤𝑛 ≤ 2. 𝐷 (3)
𝐴
.𝛾1,𝑎𝑣 −𝑐 ′ 1 𝑈 ′ 𝑈 ′
𝜎𝑣 (𝑧) = 𝐾𝑈 .tan(𝜑′ ) (1 − 𝑒 −𝐴.𝐾1.z.tan(𝜑 1) ) + 𝜎𝑠 . 𝑒 −𝐴.𝐾1.𝑧.𝑡𝑎𝑛(𝜑 1) for 𝑡𝑐𝑟𝑜𝑤𝑛 > 2𝐷 (4)
1 1

where 𝜎𝑣 (𝑧) is the vertical stress at elevation z (kN/m2), 𝛾1,𝑎𝑣 is the average soil unit
weight in the overburden area (kN/m3), z is the vertical elevation from the surface (m),
𝑡𝑐𝑟𝑜𝑤𝑛 is the overburden height (m), 𝜎𝑠 is the surcharge on the surface (traffic load)
(kN/m2), 𝑈 is the circumference length of the silo (m), 𝐾1 is the coefficient of lateral earth
pressure within the silo, k0 is the coefficient of earth pressure at rest, ka is the coefficient of
active lateral earth pressure, kp is the coefficient of passive lateral earth pressure.
488 Evaluation of Face Support Pressure Prediction for Earth Pressure Balance (EPB) Tunnelling
using Analytical and 3-Dimensional Finite Element Modelling

Table 1 Coefficients of lateral earth pressure (𝐾1 )

Studies Coefficients of lateral earth pressure
Kirsch and Kolymbas (2005) K1 = k0 = 1-sin (𝜗′1 )
Girmscheid (2008) ka < K1 < kp, K = 1 recommended

The weight of the wedge (G) is calculated using Equation 5, while the shear resistance
force is calculated according to friction force and cohesion force (Equation 6).
1 𝐷3
𝐺 = 2 . tan(𝜗 . 𝛾2,𝑎𝑣 (5)
𝑐𝑟𝑖𝑡 )

𝑇 = 𝑇𝑅 + 𝑇𝑐 (6)
where D is the tunnel diameter (m), 𝛾2,𝑎𝑣 is the average soil unit weight in the tunnel face
area (kN/m3), 𝑇 is the shear resistance force on the vertical triangular plane of the wedge
(kN), 𝑇𝑅 is the friction shear resistance force (kN), 𝑇𝑐 is the cohesion shear resistance force
(kN).
Girmscheid (2008), DIN 4126 (2013), and Anagnostou and Kovari (1994) proposed
that the same stress was simultaneously present next to the top level of the wedge, next to
the vertical plane of the wedge, and at the bottom of the wedge, which can be calculated
using Equation 7. Kirsch and Kolymbas (2005) proposed that the vertical stress at the top
of the wedge was equal to the stress next to the wedge and linearly increases along the
vertical plane according to the unit weight of the soil, which can be calculated using
Equation 8.
𝐷 2 .𝜎 (𝑡) 𝐷 3 .𝛾
𝑇𝑅,1 = 𝑡𝑎𝑛(𝜑 ′1 ). 𝐾2 . (3.tan⁡(𝜗𝑣 + 6.𝑡𝑎𝑛(𝜗2 ) (7)
𝑐𝑟𝑖𝑡 ) 𝑐𝑟𝑖𝑡 )

𝐷 2 .𝜎 (𝑡) 𝐷 3 .𝛾
𝑇𝑅,2 = 𝑡𝑎𝑛(𝜑 ′ 2 ). 𝐾2 . (2.tan⁡(𝜗𝑣 + 6.𝑡𝑎𝑛(𝜗2 ) (8)
𝑐𝑟𝑖𝑡 ) 𝑐𝑟𝑖𝑡 )

𝑘0 +𝑘𝑎
𝐾2 = 2
(9)
2.2. Numerical Modelling
Numerical modelling using the finite element method is usually used to evaluate the
soil structure interaction between the TBM and the surrounding soil. The German
Tunnelling Committee (ITA-AITES, 2016) recommended 3D numerical modelling with
stage construction to evaluate the tunnel excavation phase (Figure 3). Sometimes, 2-
dimensional (2D) numerical modelling is used to evaluate face stabilization that is
perpendicular to the longitudinal excavation direction. Nevertheless, the 2D model cannot
give an accurate result because it does not evaluate the tunnel shape effect. In the study
discussed in this paper, an evaluation was conducted using shell elements before the tunnel
excavation began to determine the acting face pressure at the shell element. Comparative
evaluations according to the actual face pressure were then conducted to see the
effectiveness of this concept in the design phase.
Aldiamar et al. 489

Shell element to
predict face pressure

a. 3D Numerical Modelling

b. Side view
Figure 3 Numerical modelling schemes for the 3D model

Since plates or a shell model in the 3D finite element method (Plaxis, 2017) can only
display normal stress, shear stress, and bending moments according to the local axis, as
seen in Figure 4, circular pressure can be calculated using Equation 10.
𝑤
𝑀𝑟 = ⁡ 16 [𝑎2 (1 + ) − 𝑟 2 (3 + )] (10)
where 𝑀𝑟 is the radial moment (kN/m), 𝑎 is the slab radius (m),  is the Poisson’s ratio, 𝑟
is the any section distance r from the origin (m), 𝑤 is the intensity of uniformly distributed
load (kN/m2).

Figure 4 Circular slab for the uniformly distributed load model (Ramchandra and Gehlot, 2018)

2.3. Soil Stratification and Parameters

According to the boring logs for tunnel alignment that were compiled in 2010, 2016,
and 2017, soil stratifications in the research area are defined in accordance with ASTM
D2487-11 (2011), as shown in Figure 5. Tunnel construction was predicted to be going
through hard silty sand and below the ground water level. The soil parameters discussed in
this paper are based on laboratory testing results that were optimized using a soil test
facility. The empirical correlations presented in this paper refer to a publication by Lim et
al. (2010), i.e., the compression index, , is equal to 𝐶𝑐 /2.303 and the swelling index, K, is
equal to 𝐶𝑠 /2.303 where the ratio of 𝐶𝑠 /𝐶𝑐 or K/ is about 0.09–0.15. The pure Poisson’s
ratio can be assumed to be 0.2, as suggested in the Plaxis manual (Plaxis, 2017). Calvello
and Finno (2004) suggested stiffness parameters for the Hardening Soil Model (HSM) as
490 Evaluation of Face Support Pressure Prediction for Earth Pressure Balance (EPB) Tunnelling
using Analytical and 3-Dimensional Finite Element Modelling
𝑟𝑒𝑓 𝑟𝑒𝑓 𝑟𝑒𝑓 𝑟𝑒𝑓
𝐸50 =1/3 𝐸𝑢𝑟 and 𝐸𝑜𝑒𝑑 =0.7 𝐸50 . HSM and the Soft Soil Model (SSM) are the soil
constitutive models used in the 3D finite element model, as seen in Table 2 and Table 3,
respectively. The Over-Consolidation Ratio (OCR) is defined as the highest stress divided
by the current stress. Soil that is currently experiencing its highest stress is said to be
normally consolidated and has an OCR of 1. In numerical modelling, it is necessary to
consider the coefficient of strength reduction in the interface (Rinter) in order to correctly
simulate moving contact between the structure and the soil. In this paper, Rinter = 0.6 is used
in the soil modelling.

Figure 5 Soil stratification in relation to the tunnel alignment

Table 2 Hardening Soil Model (HSM) parameters

Unload/ Oedometer
Secant
Depth N-SPT  c’ ’ reload stiffness,
Name Description OCR stiffness,
(m) (mean) (kN/m3) (kPa) (o) stiffness, Eoed (kPa)
E50 (kPa)
Eur (kPa)
Layer 1 0.00-4.60 Stiff silty clay 5 15.48 11 28.2 4.7 22,000 66,000 15,400
Stiff clayey
Layer 2 4.60-9.20 6 15.10 11 28.2 1.8 18,000 54,000 12,600
silt
Very stiff
Layer 3 9.20-12.87 11 15.05 13.4 27.7 1.2 19,000 57,000 13,300
silty sand
Hard silty
Layer 4 12.87-50.00 45 17.00 1 30 1 35,000 105,000 24,500
sand

Table 3 Soft Soil Model (SSM) parameters

N-SPT  c’ ’
Name Depth (m) Description OCR  K
(mean) (kN/m3) (kPa) (o)

Layer 1 0.00-4.60 Stiff silty clay 5 15.48 11 28.2 4.7 0.0233 0.0047
Layer 2 4.60-9.20 Stiff clayey silt 6 15.10 11 28.2 1.8 0.0210 0.0021
Layer 3 9.20-12.87 Very stiff silty sand 11 15.05 13.4 27.7 1.2 0.0370 0.0037
Layer 4 12.87-50.00 Hard silty sand 45 17.00 1 30 1 0.00046 0.000046

The tunnel excavation phase in the 3D finite element method is simulated according to
the excavation rate in a real construction sequence, i.e., 1.5 m for each tunnel lining segment
during tunnel construction. The tunnel lining parameters used in the modelling are
presented in Table 4.
Aldiamar et al. 491

Table 4 Tunnel lining parameters

Tunnel inner Tunnel lining Material  E
Name 
diameter (m) thickness (m) model (kN/m3) (kPa)
Lining 0.60 0.25 Linear Elastic 24 33,000,000 0.15

2.4. Statistical Evaluation

Coefficient of variation (COV) is used to describe the variations in the evaluation results
from different populations (Liu, 2012). The bias value is used to measure and evaluate the
relative dispersion data, which is equal to the ratio between the standard deviation and the
mean (Equation 11) for the entire face pressure calculation dataset in comparison to the
actual measurement by the soil pressure gauge. A bias value greater than 1 indicates that
the calculated value is smaller than the actual measurement, while COV should be near 0.
𝜎
𝐶𝑂𝑉 = 𝜇 (11)

where 𝜎 is the standard deviation, and 𝜇 is the mean value.

3. Results and Discussion

3.1. Analytical and Finite Element Method Modelling Results
An analytical evaluation was conducted using the limit equilibrium trial and error
method to obtain the critical equilibrium condition in accordance with the variations in the
sliding angle of the wedge ( ), which determines the support force in the highest, lowest,
and mean value of support face pressure along the tunnel crown alignment, i.e., Emax crown,
Emin crown, and Emean crown (Figure 6). Face support pressure is influenced by the
overburden height, which is indicated by the face pressure variations along the tunnel
length. The influence of the arching effect was also evaluated to determine which of the
following equations give the closest result to the actual pressure. The COV for the
calculation without the arching effect is smaller than those for the calculation with the
arching effect (Figure 7). Those findings indicate that excluding the arching effect in the
calculation gives a better result than including it, as described in Section 2.4. The actual face
support pressure stress seems to meet the upper boundary of the maximum stress
according to the analytic calculation.

Figure 6 The arching effect evaluation

492 Evaluation of Face Support Pressure Prediction for Earth Pressure Balance (EPB) Tunnelling
using Analytical and 3-Dimensional Finite Element Modelling
300
With arching effect
250 : 1,34
: 0.11

Actual face pressure (kPa)

COV : 0,082
200

150

100
No arching effect
: 1,23
50
: 0,099
COV : 0,081
0
- 50 100 150 200 250 300
Emax prediction (kPa)

Figure 7 COV and bias evaluation between the limit equilibrium measurement and the actual
measurement

Numerical calculation using Plaxis 3D was conducted applying a construction stage to

each of the excavation phases, and the shell elements placed in front of the excavation face
were monitored for each step, as shown in Figure 8. The thickness of shell elements is fixed
at 0.45 m according to thickness of the cutterhead used in the MRTJ project and the shell
model parameters (Table 5).

Table 5 Shell model parameters

Name Shell thickness (m) Material model  (kN/m3) E (kPa) 

Lining 0.45 Isotropic elastic 78 2,000,000 0.15

Figure 8 Face pressure acting on the shell element

A comparison of the shell element pressure resulting from the bending moment finite
element modelling and the actual face support pressure is shown in Figure 9. The results
for the HSM and the SSM are close; unfortunately, the finite element method measurement
for face pressure was lower than the actual face pressure, which is also indicated by the bias
mean values and COV (Figure 10). However, at the early stage of tunnel construction, the
face pressure and actual pressure are perfectly matched when using the finite element
method, which probably indicates the balance state condition at the tunnel face. ZTV-ING
(2012) stated that calculations of the support pressure deviations for an EPB tunnel must
be considered to be +/- 30 kPa. After the second phase excavation of numerical modelling
Aldiamar et al. 493

for face pressure, in which +80 kPa is added, the face pressure measurement is similar to
the actual pressure measurement (Figure 10). The additional load could be different
depending on the elevation of the ground water level, the soil condition, and the overburden
pressure.

Figure 9 Comparison of the actual face support pressure and prediction pressure based on
numerical models

Figure 10 Comparison of the actual face support pressure and prediction pressure based on
numerical models with + 80 kPa

4. Conclusions
Evaluation of the effectiveness of the limit equilibrium method to predict the mean
value face support pressure and the upper and lower perimeters for tunnel construction
yielded a good result; however, the initial mean pressure is relatively higher than the face
pressure measured when using a TBM pressure gauge. Overall, the face pressure results
using the shell model of the finite element method are lower than the actual pressure
measurements; yet, at the early stage of tunnel construction, the face pressure results
perfectly match the actual measured pressure at the tunnel face. This result probably
indicates the balance state condition at the tunnel face, and the addition of 80 kPa after the
second phase of excavation could indicate the need for greater face pressure to ensure TBM
movement.
494 Evaluation of Face Support Pressure Prediction for Earth Pressure Balance (EPB) Tunnelling
using Analytical and 3-Dimensional Finite Element Modelling

Acknowledgements
The author would like to acknowledge the Institute of Road Engineering, Ministry of
Public Works, the Research Group of Geotechnical Engineering, Department of Civil
Engineering, Faculty of Civil and Environmental Engineering, and the Institute for Research
and Community Services (LPPM) of Bandung Institute of Technology.

References
Anagnostou, G., Kovari, K., 1994. The Face Stability in Slurry-shield-driven Tunnels.
Tunnelling and Underground Space Technology, Volume 9(2), pp. 165–174
Anagnostou, G., 2012. The Contribution of Horizontal Arching to Tunnel Face Stability.
Geotechnik, Volume 35(1), pp. 34–44
American Society for Testing and Meterials (ASTM D2487-11), 2011. Standard Practice for
Classification of Soils for Engineering Purposes. ASTM International, West
Conshohocken, United States of America
Calvello, M., Finno, R.J., 2004. Selecting Parameters to Optimize in Model Calibration by
Inverse Analysis. Computer and Geotechnics, Volume 31, pp. 410–424
Deutsches Institut fur Normung (DIN 4126), 2013. Nachweis der Standsicherheit von
Schlitzwänden (Stability Analysis of Diaphragm Walls), Beuth Verlag GmbH, Berlin
Girmscheid, G., 2008. Baubetrieb und Bauverfahren im Tunnelbau (Construction Operation
and Construction Methods in Tunnel Construction). Ernst und Sohn Verlag, Berlin
International Tunneling Association-Association Internationale Des Tunnels Et De L Espace
Souterrain (ITA-AITES), 2016. Recommendations for Face Support Pressure Calculations
for Shield Tunneling in Soft Ground. German Tunneling Committee, Koln, Germany
Kirsch, A., Kolymbas, D., 2005. Theoretische Untersuchung zur Ortsbruststabilität
(Theoretical Investigation of Face Stability). Bautechnik, Volume 82(7), pp. 449–456
Lim, A., Ou, C-Y., Hsieh, P-G., 2010. Evaluation Clay of Constitutive Models for Analysis of
Deep Excavation under Undrained Conditions. Journal of GeoEngineering, Volume 5(1),
pp. 9–20
Liu, S., 2012. Confidence Interval Estimation for Coefficient of Variation. Master of Science
Thesis, Georgia State University, Atlanta, United States of America
Plaxis, 2017. Part 3: Plaxis Material Models Manual. Delf University of Technology & Plaxis
B.V., The Netherlands
Ramchandra, Gehlot, V., 2018. Limit State Design of Concrete Structures. Scientific
Publishers, Jodhpur, India
Shimizu-Obayashi-Wijaya Karya-Jaya Konstruksi (SOWJ), 2015. Jakarta MRT CP104/105
construction progress. Technical Meeting Material, Wijaya Karya Persero, Jakarta
Zusätzliche Technische Vertragsbedingungen und Richtlinien für Ingenieurbauten (ZTV-
ING) 2012. Teil 5 (Part 5): Tunnelbau (Tunnel Construction). Bundesanstalt für
Strassenwesen (Federal Highway Research Institute), Koln, Germany