ANALYSIS OF NATM TUNNEL RESPONSES

DUE TO EARTHQUAKE LOADING IN VARIOUS SOILS

Zaneta G. Adme

Home Institution: Dept. of Civil and Env. Engineering FAMU-FSU College of Engineering
2525 Pottsdamer St., Tallahassee, FL 32310

Host Institution: Tokyo University

Advisor: Makola M. Abdullah, Ph.D.

ABSTRACT

Tunnels play a large part in the redevelopment of urban areas. In almost all urbanized
areas it is hard to avoid building in close proximity to previously constructed building.
Many of these urban areas also lie in close proximity to water, in which weaker soils
prevail. Some of these areas, also, are prone to frequent earthquakes. It is necessary,
therefore, for these areas to implement standard tunneling techniques. The New
Australian Tunneling Method (NATM), which addresses the concerns listed above, is
becoming increasingly popular and has been adopted by some countries as the preferred
method of tunneling. The NATM is a method that incorporates the surrounding rock or
soil into a ring-like support for the structure. With the growing popularity of this method,
it is necessary to determine which soils are best suited for this method of tunneling. In
this project the soils of various cities will be analyzed to determine which of the soils
perform better with the NATM using a two dimensional model of the tunnel that is being
exposed to earthquake excitations.

KEYWORDS

New Australian Tunneling Method (NATM), earthquake, finite elements.

PROBLEM STUDIED

There are several reasons for utilizing tunnels. They can be used to connect land masses,
to bypass impeding geologic formation, or stability issues, and to reduce environmental
concerns. Most tunnels, however, are used to increase the flow of traffic. Fifty percent of
the world’s population live in urban areas and seventy percent of the population live in
earthquake prone areas (Merritt, et al. 1985). Initially, tunnels were designed with no
regard to seismic effects, but, recently, there has been enhanced awareness of seismic
hazards for underground structures (Merritt, et al. 1985).

There are two broad categories of earthquake effects of tunnels: ground shaking and
ground failure. When seismic waves propagate through the earth’s crust, the resulting
ground motions are considered ground shaking. There are two basic categories of ground
shaking. Body waves travel within the earth’s inner layers. These waves can be either
longitudinal P or transverse shear S waves. P waves move in a compressional motion
similar to the motion of a slinky, while the S waves move in a shear motion perpendicular
to the direction the wave is travelling. These waves can travel in any direction
underground. Surface waves travel along the earth’s surface in the same matter a ripple
would travel through water. These waves can either be Rayleigh or Love waves. Love
waves shake the surface side-to-side. Rayleigh waves move the surface of the earth
around in a circle, forward and down then back and up. This is the same as the motion in
an ocean wave (Merritt, et al. 1985). Any tunnel structure will be deformed as the ground
is deformed by the traveling waves.

Ground failure can include different types of ground instability. These can include
faulting, liquefaction, and tectonic uplift and subsidence. Faulting occurs when an
increase in stress causes rocks to break. Liquefaction is a phenomenon in which the
strength and stiffness of a soil is reduced by earthquake shaking or other rapid loading.
Tectonic uplift and subsidence is the upward and downward movement of the ground due
to plate movement. These phenomena have been responsible for tremendous amounts of
damage in historical earthquakes around the world. Each of these hazards could possibly
be detrimental to tunnel structures (Merritt, et al. 1985).

Many tunneling methods are in common use, and a suitable one is generally chosen
according to geology, tunnel dimensions, and other factors (Kirzhner and Rosenhouse
2000). The New Australian is a method in which, after a section of tunneling is
completed, shotcrete is applied to the surface of the tunnel and the surrounding rock or
soil becomes integrated into the support structure (Yang, 2002). Extreme care is taken
during excavation and immediate application of support media prevents unnecessary
loosening of soil. These tunnels use rounded tunnel shapes to prevent stress
concentrations in corners where most failure mechanisms start (Yamaji, 1998). These
tunnels, also, utilize thin linings to minimize bending moment. Observation of tunnel
behavior during construction is an important part of NATM. This optimizes working
procedures and support requirements (Yang, 2002). Many countries have adopted this
method as the primary method of construction.

OBJECTIVE

The objective of this project is to determine which soil types, when used in conjunction
with a tunnel completed using the New Australian Tunneling Method, perform better
when subjected to seismic excitation. For this purpose, soils from seven cities were
selected and a finite element model of the tunnel created for each case. Comparisons
based on the maximum displacement each soil experienced when subjected to an
earthquake were then made.
 RESEARCH APPROACH

The first step was to identify the physical problem. This included describing the physical
structure, identifying the source of dynamic excitation, and determining the expected
outcomes. The next step in the process was defining the inputs and then defining the
model based on the inputs. The last step was to find the solution of the model and review
the results of the project.

The world cities selected for this project were:

• Agadir, Morocco
• Avezzano, Italy
• Chimbote, Peru
• Los Angeles, California
• Mexico City, Mexico
• Tangshan, China
• Tokyo, Japan

There were four basic criteria used to select the cities used for this project. The first
criterion was the earthquake history of the city. It was necessary to use cities that were
susceptible to large magnitude earthquakes. The next criterion was the population of the
city to evaluate the possibility of tunnel use, which is the third criterion. Tunnels are
more likely to be used in cities that have a large population. The last criterion used was
the variance in the soil types present in these areas. It was necessary to select cities with
contrasting soils to obtain a broad range of results.

There are currently15 different soil orders present in the world (Yamaji, 1998). Each city
used in this experiment was classified by soil order. That soil order was then classified by
the aggregate(s) associated with them as shown in Table 1.

Table 1. Soil order and soil types.

\
Location
DDDDDD Soil Order Soil Type
Agadir, Morocco Alfisols Low Plasticity Clay
Avezzano, Italy Ultisols Low Plasticity Silt
Chimbote, Peru Entisols Gravel-Sand Mixture
Los Angeles, California Mollisols Organic
Mexico City, Mexico Andisols Medium Plasticity Silt
Tangshan, China Inceptisols Sandy Gravel
Tokyo, Japan Oxisols High Plasticity Clay
The specific properties of each soil are listed in Table 2. This properties listed are needed
to in the calculations for each city.

Table 2. Soil Properties.

Internal
Mass Elastic
Poisson's Friction Cohesion
Soil Type Density Modulus
Ratio Angle (Pa)
(kg/m^3) (Pa)
(deg.)
Gravel Uniform 1600 4.00E+07 0.25 34 0
Sandy w/ few
2100 4.00E+07 0.25 35 0
fines
Sandy w/ silt or
2100 4.00E+07 0.25 35 1000
clay
Mixture of
2000 1.50E+07 0.25 38 3000
gravel and sand
Sand Uniform, fine 1600 1.50E+07 0.25 32 0

Uniform, coarse 1600 2.50E+07 0.25 34 0

Uniform, well-
1800 2.00E+07 0.25 33 0
graded
Silt Low plasticity 1750 4.00E+06 0.25 28 2000
Medium to high
1700 3.00E+06 0.25 25 3000
plasticity

Clay Low plasticity 1900 2.00E+06 0.28 24 6000

Medium
1800 1.00E+06 0.25 20 8000
plasticity
High plasticity 1650 6.00E+05 0.25 17 10000

Organic 1550 5.00E+05 0.25 20 7000

Rock Granite 2700 7.40E+10 0.25 51 5.51E+07

The physical structure used for this simulation consisted of a tunnel surrounded by two
ground layers, each 130 meters wide (Figure 1). The bottom layer was granite rock and
was 106 meters in height. The properties for this layer remained constant during each
trial The uppermost layer was a 24 meter deep soil layer that changed according to the
soil properties associated with each trial city. The tunnel was circular with a 22 meter
diameter and was buried 67 meters below the ground surface.

Finite element method was used in this project. This method is used to model and solve
complex two and three dimensional engineering problems. The Visual Finite Element
Analysis (VisualFEA) program incorporates powerful finite element processing software
with a user-friendly graphical interface that reduces the amount of time needed for
programming.
 Figure 1. Physical Structure.

First a two dimensional plane strain model was used. This is used in modeling three
dimensional structures that are uniform throughout their length, e.g. beams and cylinders.
The data was taken from a point between the rock and soil interface directly above the
highest point of the tunnel (Figure 2).

Figure 2. VisualFEA structure.

Fixed vertical ends were used to restrict the movement of the ends of the model. A
damper was placed at the bottom boundary to keep it from moving uncontrollably. For
this model Rayleigh damping of 0.05 was used to model the natural damping
characteristics of the rock and the soil.

The source of dynamic excitation used in this project was the acceleration record for the
1995 Kobe, Japan earthquake, as shown in Figure 3.

Figure 3. Acceleration record of the 1995 Kobe earthquake, Japan).

This earthquake data was used because this earthquake was the most devastating to civil
infrastructure in recorded history. The acceleration input from the earthquake was applied
to the bottom boundary.

OUTCOMES

Since there are thirteen different soil types, the recorded data from uniform gravel and
medium to high plasticity silt are evaluated for comparison purpose.

Medium to High
Uniform Gravel Plasticity Silt

Disp.=0.5753 Disp.=0.2838

Figure 4. Graphical Results for Uniform Gravel and

Medium to High Plasticity Silt
Although the bands of color seem similar in their placement, their difference is relayed in
the scales of each soil. The scale for uniform gravel goes from 0-2.1406 cm and the scale
for medium to high plasticity silt only goes form 0-1.1526 cm.

The displacement results for all soils tested are listed in Table 3. For this project, it was
found that soils containing between ten and fifty percent fine particles performed better
under excitation than soils with smaller amounts of fine particles(<10%). Soils with the
greatest amounts of fines (>50%) performed unfavorably as compared to the other soils.
Soils that had a relatively low modulus of elasticity had a greater maximum displacement
than other soils in the study. The best performing soils were in the gravel category. These
soils performed better overall than sand, silt, or clay. The sand category performed the
worst overall, although uniform sand had one of the lowest maximum displacements in
all categories.

Table 3. Displacement results.

Maximum
Soil Type
Displacement (cm)
Gravel Uniform 0.5783
Sandy w/ few fines 0.5708
Sandy w/ silt or clay 0.02057
Mixture of gravel and sand 0.05604
Sand Uniform, fine 1.5550
Uniform, coarse 0.03947
Uniform, well-graded 1.1515
Silt Low plasticity 0.2129
Medium to high plasticity 0.28385
Clay Low plasticity 0.4304
Medium plasticity 0.85182
High plasticity 1.4148
Organic 1.7049

The results for each city are presented in Table 4. Tangshan, China recorded the smallest
maximum displacement for the point of interest. The maximum displacement was only
0.02057 cm. The location that recorded the highest maximum displacement was Tokyo,
Japan, 1.419 cm. Although Tokyo has the highest maximum displacement, these
displacements would only produce minor cracks in the tunnel structure (Yang, 2002).
Using the model representing a NATM tunnel, it is shown that tunnels constructed using
this tunneling method produces relatively small displacements under earthquake loadings
and could possibly considered as the primary method of construction in other projects.
Table 4. Displacement results for selected cities
Maximum Displacement
Location Soil Type
(cm)
Agadir, Morocco Low Plasticity Clay 0.4304

Avezzano, Italy Low Plasticity Silt 0.2129

Chimbote, Peru Gravel-Sand Mixture 0.05604

Los Angeles, California Organic 0.1704

Mexico City, Mexico Medium Plasticity Silt 0.2839

Tangshan, China Sandy Gravel w/ Silt 0.02057

Tokyo, Japan High Plasticity Clay 1.419

POSSIBLE FUTURE WORK

To increase the accuracy of this project, a three dimensional model with varying ground
layers can be used. A water table can be introduced to simulate the possibility of
liquefaction. Also, a structure that does not include a tunnel could be used for comparison
with each soil type.

ACKNOWLEDGEMENTS

I would like to thank MCEER, the REUJAT Program, the FAMU Undergraduate
Program, and the National Science Foundation. I would also like to thank my host
institution, the University of Tokyo, my host professor, Yozo Fujino, Ph.D., my home
advisor, Makola M. Abdullah, Ph. D., and graduate advisors, Terri R. Norton, M.S.C.E.
and Claudia M. D. Wilson, M.S.C.E.
References

Kirzhner, F. and Rosenhouse, G., 2000. Numerical analysis of tunnel dynamic response
to earth motions. Tunneling and Underground Space Technology, Volume 15, Issue
3, 249-258 .

Merritt, J. L., Monsees, J. E., Hendron, A. J., 1985. Seismic design of underground
structures: Proc of the 1985 Rapid Excavation and Tunneling Conference, New York.
International Journal of Rock Mechanics and Mining Science & Geomechanics
Abstracts, Volume 23, Issue 6, 255-282.

Yang, M. W. W., 2002. Principles of Tunnel Lining Design. AGS/IMM Technical

Meeting 2002 on “Underground Excavation in Urban Environments”.

Yamaji, H. Iwate-Ichinohe Tunnel on the Tohake New Trunk Line. JSCE, Volume 30,
Issue 10, 1435-1468 .

Intuition Software, 1999. VisualFEA-Visual Finite Element Analysis. Chonbuk, South

Korea: Intuition Software.