Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
··································································································································································································································
Abstract
This study presents a Ground Lining Interaction (GLI) model to consider the deterioration of primary supports and realistic ground
conditions in evaluations of accurate New Austrian Tunnel Method (NATM) tunnel loads acting on concrete linings. Several
parametric studies of the major parameters that influence tunnel loads are performed and the results are compared with Terzaghi’s
formula. Numerical results demonstrate that the proposed GLI model is capable of reducing excessive tunnel loads by up to 30%.
Based on the obtained results, a simplified GLI formula that can be used in engineering practice is proposed for NATM tunnel loads
on concrete linings.
Keywords: NATM tunnel, tunnel loads, terzaghi’s formula, ground lining interaction, concrete lining, numerical analysis
··································································································································································································································
*Member, Professor, Dept. of Civil and Environmental Engineering, Yonsei University, Seoul 120-749, Korea (Corresponding Author, E-mail: soj9081@yon-
sei.ac.kr)
**Ph.D., Dept. of Civil and Environmental Engineering, Yonsei University, Seoul 120-749, Korea (E-mail: ychan009@yonsei.ac.kr)
***Member, Ph.D. Candidate, Dept. of Civil and Environmental Engineering, Yonsei University, Seoul 120-749, Korea (E-mail: kimkimym@yonsei.ac.kr)
****Ph.D. Candidate, Dept. of Civil and Environmental Engineering, Yonsei University, Seoul 120-749, Korea (E-mail: dohyun0403@yonsei.ac.kr)
− 672 −
Sang-Seom Jeong, Young-Chul Han, Yong-Min Kim, and Do-Hyun Kim
and deep tunnels are shown in Eqs. (3) and (4), respectively. 3. Ground Lining Interaction (GLI) Model
–K tan φ 2H
-------
γB
Proof = ------------------ ⎛ 1 – e ⎞
B
(3) The actual tunnel load acting on concrete lining is the domi-
2K tan φ ⎝ ⎠
nant design factor for NATM tunnels embedded in weathered
γB soil and rock. Although the design of concrete lining considers
Proof = ------------------ (4)
2K tan φ any uncertainties, a new model that can rationally evaluate tun-
nel loads is necessary for an economical design. The primary
B φ
B = 2 -----1 + H1tan ⎛ 45 – ---⎞ (5) purpose of this study is to evaluate the actual tunnel load and to
2 ⎝ 2⎠
preserve and improve the potential of the ground to support itself
In these equations, γ is the unit weight of the ground (kN/m3), B to the greatest extent possible. This task can be accomplished by
is the ground relaxation load breadth (m), K is the in-situ stress exploiting an arching effect in the surrounding ground around the
ratio, φ is the ground friction angle (o), H is the soil depth above tunnel. To achieve the goal of designing a practical and safe
the tunnel (m), B1 is the tunnel width (m), and H1 is the tunnel tunnel, a GLI model is proposed using a finite difference method
height (m). Terzaghi’s method is based on the assumption that (FDM) with FLAC 2D (Fast Lagrangian Analysis of Continua).
ground relaxation is inevitable and requires steel supports. The GLI model can simulate the load transfer to the secondary
However, the use of Terzaghi’s formula for shallow tunnels supports (e.g., concrete lining) caused by the deterioration of the
such as subway systems often leads to overestimation of the tun- primary supports. These unstable states result in the disturbance
nel load (Korea Rail Network Authority, 2010; Chang et al., 2003). of the ground-tunnel equilibrium. In this state, the concrete lining
To deal with this problem special form of Eq. 6 is suggested to co-supports the tunnel load with the ground self-weight and redis-
reduce an excessive tunnel load by considering the effect of soil tributes the load caused by additional ground deformation. Even-
cohesion. tually, these unstable states will reach a new state of equilibrium.
2H
γ B – 2c- ⎛ 1 – e–K ------
Proof = -----------------
B
- tan φ
⎞ (6) 3.1 Load Transfer Mechanism by the Deterioration of Pri-
2K tan φ ⎝ ⎠
mary Supports
here, c denotes the ground cohesion (kPa). A plastic zone of the surrounding soil around a tunnel can be
Fig. 3. Ground Reaction Curve on Primary and Secondary Supports Fig. 4. Typical 2D Model for Numerical Analysis
loss may develop around tunnels for several reasons, including under a load distribution ratio of 0-0-100 when compared with
improper grouting and ground erosion. the results estimated under a load distribution ratio of 30-40-
In a numerical analysis, the initial equilibrium state is of great 30, which is a typical tunnel design. The 0-0-100 load
significance. The specified initial stress distributions should match distribution ratio is the state during ground-lining interaction in
a calculation based on the self-weight of the material. The mo- which the maximum load is acting on the secondary support
deling of the tunnel and installation process of the primary sup- member. Therefore, by setting the load distribution ratio to 0-0-
ports is essential to generate the initial stress and to capture a 100, it is possible to consider the largest tunnel load acting on
realistic construction stages; therefore, processes pertaining to the the concrete lining compared to any other load distribution
installation and elimination procedures of structures are presented situation, thus enabling a safe tunnel design.
in subsequent section.
The load distribution ratio is the relative load-bearing rate on 4.3 Analysis Procedure
the ground, on the primary support member, and on the To evaluate the tunnel loads caused by the GLI model concept,
concrete lining. When setting the load distribution ratio to 0-0- each analysis was conducted based on the same procedures as
100, the secondary support member and concrete lining resist those used with NATM tunnel construction from a practical engi-
the overall tunnel load. Using this procedure, evaluating neering. The analysis procedures, shown in Fig. 6, consist of eight
extreme loads acting on the concrete lining via numerical steps to set the load distribution ratio.
analyses is possible. Fig. 5 shows the result of a tunnel load Step 1) The 2D plain strain numerical mesh including the
tunnel geometry is generated with a boundary condition. The
initial ground stresses are then applied to the numerical
model.
Step 2) The elements of the tunnel above the spring line are
removed to simulate the procedure of the NATM tunnel
excavation. At this time, the load distribution ratio is set to
zero.
Step 3) The primary supports, including soft shotcrete and
rockbolts, are installed in the excavated arch area. The soft
shotcrete and rockbolts are modeled as beam and cable elements,
respectively.
Step 4) The soft shotcrete installed in the previous step is
replaced by hard shotcrete after increasing the elastic modulus by
approximately threefold, and the numerical model is then
executed for the first time. At this time, the load distribution ratio
is set to 100.
Step 5) The elements of the tunnel below the spring line are
Fig. 5. Comparison Results for Tunnel Load under Different Load also removed.
Distribution Ratio Step 6) The soft shotcrete and rockbolt are installed in the
Fig. 6. Numerical Procedure of NATM Tunnel Analysis for a GLI Model: (a) Step 1, (b) Step 2, (c) Step 3, (d) Step 4, (e) Step 5, (f) Step 6,
(g) Step 7, (h) Step 8
5. Parametric Study tunnel loads for rock grades one to three as calculated with the
GLI model are smaller than those in Terzaghi’s rock mass table
To obtain detailed information about tunnel loads, several pa- and rapidly increase for rock grades of four and lower. The
rametric numerical analyses were conducted while varying the tunnel load height for rocks of grades one to three is less than 0.5
major influencing parameters, in this case a tunnel depth ranging
from 20- to 80 m and an in-situ stress ratio ranging from 0.5- to
2.0. Fig. 7 shows a tunnel cross-section considered in a param-
etric study, which represents a typical NATM tunnel design in
South Korea.
Fig. 9. Comparison Results for Tunnel Load under Different Tunnel Depths: (a) 20 m, (b) 40 m, (c) 60 m, (d) 80 m
m, which is similar to the damage zone from the excavation blast situ stress ratio ranges from 0.5 to 4.0 and is significantly af-
(Lim, 1996; Lee et al., 2000), as shown in Fig. 8(b). fected by the joint condition and ground strength, particularly for
long tunnels. In this study, tunnel loads with different in-situ stress
5.2 Effect of Tunnel Depth ratios (0.5-2.0) for various rock conditions are compared.
When designing tunnels, it is important to consider the various When the vertical stress is two times greater than the horizontal
rock and ground conditions as well as the tunnel depth. The tun- stress (K0 = 0.5), the tunnel loads estimated by Terzaghi’s
nel depth is assumed to be 20 m in urban areas and 80 m in moun- method and by the GLI model are fairly similar (Fig. 10(a)).
tainous areas. Fig. 9 compares the results for tunnel loads with However, when the in-situ stress ratio (K0) is 2.0, the tunnel
varying tunnel depths. loads tend to decrease greatly for both Terzaghi’s method and the
The tunnel loads obtained by Terzaghi’s rock mass table pre- GLI model when compared to an in-situ stress ratio (K0) of 0.5.
sent no differences for varying tunnel depths. Thus, Terzaghi’s Thus, Ter- zaghi’s rock mass table cannot consider the effect of
rock mass table does not consider the tunnel depth. The tunnel the in-situ stress ratios, and Terzaghi’s formula cannot be used for
loads estimated by Terzaghi’s formula increase as the tunnel high grade rocks. However, the GLI model is not only
depth increases. The GLI model provides significantly smaller compatible with all rock grades but also affords results similar to
tunnel loads for shallow tunnel depths when compared to those those of Terzaghi’s rock mass table under the condition of a
by Terzaghi’s formula. However, as the tunnel depth increases, decreasing in-situ stress ratio.
the tunnel loads obtained with the GLI model in weathered rock
are similar to those with formulated using Terzaghi’s method. 6. Proposed Ground Lining Interaction Formula
5.3 Effect of the in-situ Stress Ratio The GLI model can reduce the tunnel load acting on concrete
According to numerous domestic ground investigations, the in- lining by considering the arching effect of the ground and by
formula, H is the soil depth above the tunnel (m), c is the ground
cohesion (kPa), φ is the friction angle (o), and E is the elastic
modulus (kPa).
Fig. 10. Comparison Results for Tunnel Load under Different in-situ Fig. 11 shows the tunnel load based on the GLI model and the
Ratios (H = 40 m): (a) Terzaghi’s Methods, (b) GLI Model
proposed formula when compared with Terzaghi’s formula. For
a tunnel depth of 40 m, the correlation factor is 0.994, as shown
reducing the tunnel load height which is linked to the magnitude in Fig. 11. In Fig. 12, the numerical analyses results using the GLI
of the tunnel load. This section presents a tunnel load formula model and the results estimated by the proposed formula are com-
based on the GLI model results from this study. pared. The correlation factor between the results is 0.977 for tun-
A simplified GLI equation for estimating tunnel loads is devel- nel depths of 20-80 m. Therefore, the tunnel load estimated by
oped by interpreting the results of numerical analyses. The pro- the proposed formula is nearly identical to that estimated using
posed equation is a function of the unit weight (γ), elastic mo- the GLI model. Because the proposed formula is based on the
dulus (E), cohesion (c), friction angle (φ), and tunnel depth (H). GLI model, the tunnel load estimated by the proposed equation
The proposed GLI formula is able to consider the elastic modulus is smaller than that using Terzaghi’s method. The ability to esti-
(E) of the ground that surrounds the tunnel. In contrast, Terzaghi’s mate the tunnel load regardless of the tunnel depth is another
formula does not consider the elastic modulus which is an im- advantage of the proposed formula.
portant engineering factor; for this reason, the elastic modulus
for the given rock and ground conditions must be considered when 7. Conclusions
estimating tunnel loads. The proposed GLI formula is written as
E In this study, a new tunnel load evaluation method (Ground
γ ( B + H ) – c ----------------- -
PGLI = --------------------------- ⎛ e
1000γ B⎞
(7) Lining Interaction model) was presented and discussed while
7 tan φ ⎝ ⎠
considering the rock grade, tunnel depth, and in-situ stress ratio.
here, PGLI is the vertical tunnel load (kPa), γ is the unit weight of Moreover, a new equation that evaluates the tunnel load acting
the ground (kN/m3), B is the same parameter in Terzaghi’s on the concrete lining is proposed using the results obtained from
the GLI model. From the findings of this study, the following cord, Issue 307.
conclusions are drawn. Duddeck, H. and Erdmann, J. (1985). “Structural design models for tun-
Based on the results of Terzaghi’s rock mass table, the tunnel nels in soft soil.” Underground Space, Vol. 9, pp. 246-259.
FLAC 2D (2000). FLAC Ver 4.0 manual, Itasca Consulting Groups,
loads for every rock grade tend to be overestimated. Terzaghi’s
Minneapolis, Minnesota, USA.
formula can only be used for soft rocks and soils. However, the Goel, R. K., Jethwa, J. L., and Paithankar, A. G. (1995). “Indian experi-
GLI model can be adapted to all rock grades. This model can ences with Q and RMR systems.” Tunnelling and Underground
also reduce the tunnel load by up to 30% on the crown part of the Space Technology , Vol. 10, No. 1, pp. 97-109.
tunnel. Parker, H. W. (2004). Planning and site investigation in tunneling, 1o
Terzaghi’s rock mass table is inadequate when the effect of the Congresso Brasileiro de Túneis e Estruturas Subterrâneas, Seminário
tunnel depth must be considered. However, Terzaghi’s formula Internacional South American Tunnelling, Brazilian Association of
and the GLI model afford similar results for soft rock and soil for Soil Mechanics and Geotechnical Engineering, Brazil.
Iftimie, T. (1998). Overview and a new hypothesis on earth pressure
deep tunnel depths (deeper than 40 m). Because the GLI model
acting on circular tunnel lining, Tunnels and Metropolises, Negro Jr
provides reduced results for tunnel loads with shallow tunnel & Ferreira (Eds.), pp. 267-272.
depths (less than 40 m), adopting the GLI model for shallow Janssen, H. A. (1895). “Versucheüber getreidedruck in silozellen.” Zeit-
tunnels is more feasible. schrift des Vereins deutscher Ingenieure, XXXIX, Vol. 39, No. 35,
When estimating tunnel loads, Terzaghi’s rock mass table ig- pp.1045-1049 (in German).
nores the effect of the in-situ stress ratio. When the in-situ stress Jeong, S. S. and Seo, D. H. (2004). “Analysis of tieback walls using pro-
ratio is greater than 1.0, applying the GLI model rather than posed p-y curves for coupled soil springs.” Computer and Geotech-
Terzaghi’s formula is better because the results estimated by the nics, Vol. 31, No. 6, pp. 443-456.
Jiang, Y., Yoneda, H., and Tanabashi, Y. (2001). “Theoretical estimation
latter tend to be overestimated.
of loosening pressure on tunnels in soft rocks.” Tunneling and
The proposed tunnel load equation is based on the results esti- Underground Space Technology, Vol. 16, No. 2, pp. 99-105.
mated by the GLI model. This model reduces tunnel loads, which Kim, H. J. and Eisenstein, Z. (1998). “Prediction of lining loads from
prevents the overestimation of the actual tunnel loads acting on case histories.” World Tunnel Congress '98, Sao Paulo, Brazil, pp.
the concrete lining. Moreover, the proposed equation considers 299-304.
the effect of the ground’s elastic modulus. Kim, H. J. and Eisenstein, Z. (2006). “Prediction of tunnel lining loads
using correction factors.” Engineering Geology, Vol. 85, pp. 302-
312.
Acknowledgements Korea Rail Network Authority (2010). Seoul metropolitan high-speed
railway (Suseo~Pyoungtaek), Design of Roadbed and Optimization
This work was supported by the National Research Foundation of Concrete Lining.
of Korea (NRF) grant funded by the Korea government (MSIP) Koyama, Y., Okano, N., Shimizu, M., Fujiki, I., and Yoneshima, K.
(No. 2011-0030040). (1995). “In-situ measurement and consideration on shield tunnel in
diluvium deposit.” Proceedings of tunnel Engineering, JSCE, Vol. 7,
References pp. 385-390.
Lee, I. M., Kim, S. K., Lee, D. W., Choi, J. W., Kim, D. H., and Kim, Y.
Barton, N., Lien, R., and Lunde, J. (1974). “Engineering classification U. (2000). “Optimization of tunnel blasting design by finite element
of rock masses for the design of tunnel support.” Rock Mechanics, method.” Journal of Korean Geotechnical Society (KGS), Vol. 16,
Vol. 6, No. 4. pp. 189-236. No. 5, pp. 63-74.
Bhasin, R. and Grimstad, E. (1996). “The use of stress-strength relation- Lim, S. B. (1996). An evaluation of dynamic damage adjacent to a
ship in the assessment of tunnel stability.” Tunneling and Under- blasthole in tunnel excavations, PhD Thesis, Kyounghee University,
ground Space Technology, Vol. 11, No. 1, pp. 93-98. Korea.
Chang, S., Lee, S., Suh, Y., Yun, K., Park, Y., and Kim, S. (2010). “A Mashimo, H. and Ishimura, T. (2003). “Evaluation of the load on shield
study on field change case of tunnel concrete lining designs using tunnel lining in gravel.” Tunneling and Underground Space Tech-
GLI (Ground Lining Interaction) model.” Journal of Korean Society nology, Vol. 18, Nos. 2-3, pp. 233-241.
for Rock Mechanics Tunnel & Underground Space, Vol. 20, No. 1, Negro J. A. (1988). Design of shallow tunnels in soft ground, PhD
pp. 58-64. Thesis, Department of Civil Engineering, University of Alberta,
Chang, S., Seo, S., and Lee, S. (2003). “Design of the secondary tunnel Edmonton, Canada.
lining using a ground primary support-secondary lining interaction O'Rourke, T. D. (1984). “Guidelines for tunnel lining design.” American
model.” Proc. of International Symposium on the Fusion Technology Society of Civil Engineers, pp. 82.
of Geosystem Engineering, Rock Engineering and Geophysical Ex- Park, J. J., Kim, D. H., and Jeong, S. S. (2012). “Analysis of tunnel
ploration, pp. 109-114. lining behavior under tunnel load.” Journal of the Korean
Craig, R. N. and Muir Wood, A. M. (1978). A review of tunnel lining Geotechnical Society, Vol. 28, No. 8, pp. 79-88.
practice in the United Kingdom, Supplementary Report, TRRL Rose, D. (1982). “Revising Terzaghi’s tunnel rock load coefficients.”
Suppl Rpt 335 Monograph, Transport and Road Research Laboratory, Proc. 23rd U.S. Symp. on Rock Mechanics, Berkeley, CA., AIME,
Berkshire England. New York, pp. 953-960.
Deere, D. U., Peck, R. B., Park, H. W., Monsees, J. E., and Schmidt, B. Seo, S. H., Chang, S. B., and Lee, S. D. (2002). “An analysis model of
(1970). “Design of tunnel support systems.” Highway Research Re- the secondary tunnel lining considering ground-primary support-
secondary lining interaction.” Journal of Korean Society for Rock Tunneling with Steel Supports, Proctor, R. V. and White, T. L.,
Mechanics, Vol. 12, No. 2, pp. 107-114. Commercial Shearing Co, pp. 17-99.
Singh, B., Jethwa, J. C., Dube, A. I., and Singh, B. (1992). “Correlation Whittaker, B. N. and Frith, R. C. (1990). “Tunnelling-design, stability and
between observed support pressure and rock mass quality.” Tunnel- construction.” The Institution of Mining and Metallurgy, London.
ling and Underground Space Technology, Vol. 7, No. 1, pp. 59-74. Wood, A. M. M. (1975). “The circular tunnel in elastic ground.” Geo-
Terzaghi, K. (1943). Theoretical soil mechanics, Wiley, New York technique, Vol. 25, No. 1, pp. 115-127.
Terzaghi, K. (1946). Rock defects and loads on tunnel support, Rock