Proceedings of the 1999 Winter Simulation Conference

P. A. Farrington, H. B. Nembhard, D. T. Sturrock, and G. W. Evans, eds.

SPECIAL PURPOSE SIMULATION TEMPLATE FOR UTILITY TUNNEL CONSTRUCTION

Simaan M. AbouRizk K.C. Er

Janaka Y. Ruwanpura Siri Fernando

Department of Civil and Environmental Engineering, City of Edmonton

220 Civil/Electrical Engineering Building Asset Management and Public Works
University of Alberta Drainage Services, Design and Construction
Edmonton AB CANADA T6G 2G7 14323 – 115 Avenue
Edmonton AB CANADA T5M 3B8

ABSTRACT Simphony. The template is based on the special purpose

simulation concepts described in AbouRizk and Hajjar
Utility construction projects have great opportunities for (1998).
simulation applications in construction. This paper
describes the special purpose tunneling simulation template 2 TUNNEL CONSTRUCTION USING TBM
developed based on the tunneling operations performed at
the City of Edmonton Public Works Department for Tunneling projects using shielded TBM involve the
shielded tunnel boring machines. The tunneling operations following activities:
are described, then the tunnel template and its components
are illustrated. The results generated from the template 1. excavation and support of the working shaft,
using the historical data to test the template and to analyze 2. excavation and support of undercut, and tail
the potential construction processes are presented. Future tunnel,
embellishments to the tunneling template are briefly 3. excavation of the tunnel,
described. 4. disposal of dirt from the tunnel face,
5. hoisting the dirt to the ground level,
1 INTRODUCTION 6. lining the tunnel,
7. extending the services and rail tracks,
Today’s city centers are struggling with increasing traffic 8. excavation and support of the removal shaft.
congestion, reduced surface vacancy and increased demand
for infrastructure. As a result, urban developers are Except for activities (1), (2) and (8) of the above, these
becoming increasingly interested in the tunneling are repetitive tasks in a tunnel construction cycle. The
alternative. Tunnels can be used to serve a variety of purpose of the pre-planning is to ensure that minimum
functions including subways, utility corridors and sewer waste occurs during the entire construction process. The
lines. In general, the term “tunneling” can be used to goal of eliminating the wait time is to optimize the
describe a wide range of underground excavation resources and the construction process. The advancement
operations. The City of Edmonton (CE) Public Works of the tunneling operation depends on the progress of the
Department has carried out numerous tunneling projects tunneling activities between the working shaft and the
over the last five decades. The CE started developing its tunnel face. In order to optimize the tunneling operation,
tunneling expertise in the early part of 1950's beginning all activities should be coordinated to minimize the delay
with hand tunneling. In 1965, a non-shielded tunneling occurring at either end.
boring machine (TBM) was used (also called as "spider Excavation on long tunnels is typically done with
mole"). After ten years, the CE purchased its first shielded shielded mechanical moles. The shield provides temporary
TBM and later purchased four more. The use of TBM support for the front sections of the tunnel and a safe
boosted the ability to construct TBM tunnels with working area for the crew. There are two types of tunnel
excavation as large as 6.8m diameter. boring machines: open-face and closed-face shielded
The paper explains the modeling and analysis of the machines. Open face machines are generally employed in
tunneling process for shielded Tunnel Boring Machines competent soils with reasonable stability. In conditions of
using the Special Purpose Tunnel Template developed with runny soils such as silt or sand, a closed-face-shielded

948
 AbouRizk, Ruwanpura, Er, and Fernando

machine is used. An important property of TBMs is their construction methods were also added as alternatives for
excavation rate, which is dependent on the soil conditions future developments.
and the TBM horsepower. Another important property is
the stroke length, which determines how often the TBM 1. TBM close face – 2.3 meter diameter,
will need to be re-set. 2. material handing – two trains with muck cars
The dirt handling process involves the transportation and belt conveyor at tunnel face,
and disposal of spoil from the tunnel face to the shaft 3. crane at shaft,
where it is lifted to the surface. Spoil can be hauled 4. liner installation: pre-cast concrete segments,
horizontally using trains and/or belt conveyors. The 5. single tracking layout inside the tunnel and
selection criterion depends on the tunnel site conditions. two tracks at undercut.
Belt conveyers have the advantage of providing a
continuous spoil removal system. However, they typically 2.1 Repetitive Construction Processes at Tunnel Face
require excessive maintenance. Train haulage is energy-
efficient and is compatible with most excavating and All activities at the tunnel face are repetitive. Figure 1
loading methods and is adaptable to almost all sizes of shows the bar chart schedule for the repetitive activities at
tunnels. Trains can also be fitted with special cars capable the tunnel face. No wait time is shown in the bar chart
of transporting laborers and support liners. Depending on schedule (for illustrative purposes) and it assumes that the
the tunnel diameter, a single or double-track system can be muck cars are available to commence the next cycle.
used. In most cases, a track switching system is utilized at However, the whole schedule depends on the travel time of
the undercut to allow multiple trains to share a single track. the train and the other repetitive operations at the shaft.
The working shaft is utilized to remove the spoil and to The second train could arrive at any time. In order to
transport the construction materials and personnel. The dirt have an optimized situation; the second train must arrive at
can be hoisted with a skip, a clamshell bucket, or a gantry the tunnel face when lining installation is done. If the
crane. Clamshells are typically used in shallow tunnels. In second train is late, the TBM has to wait until it arrives.
medium depth tunnels (10 m to 20 m), gantries or cranes Therefore, there are two waiting factors that determine the
are more economical and in deep tunnels (>30 m), a cage production and the tunnel advance rate; train waiting time
or skip is used with a head frame for hoisting the muck. for TBM or TBM waiting time for train.
The two major types of tunnel support systems consist
of either (1) rib-and-lagging or (2) concrete segments. Rib- 2.2 Repetitive Construction Processes at Shaft
and-lagging method has a record of high performance in a
variety of ground conditions. During installation, lagging The activities at the shaft also involve repetitive operations.
is wedged circumferentially between rib and soil. The rib- Figure 2 shows the bar chart schedule for the activities that
and-lagging support system acts as the primary lining occur at the shaft. The processes at the shaft depend on the
system. A secondary layer made of cast-in-place concrete travel time of the train and the operations at tunnel face
is placed when tunneling excavation is finished. Pre-cast plus the efficiency of the hoisting operations at shaft. It is
concrete segment lining is the alternative to rib-and- also assumed that there is no waiting time for any of the
lagging, which act as both the primary and final lining resources when scheduling the bar chart. The second train
systems. Each segment is designed as a compact structural could arrive at any time. In order to have an optimized
unit and thus requires the least amount of handling during situation; the second train must arrive at the shaft when the
erection. The full ring typically consists of four identical first train is finished loading the liners and unloading the
segments. It is partially installed inside the shield of the dirt If the second train is delayed, the crane waits for the
TBM. The ring is expanded tightly against the soil as it train.
leaves the shield. Metal spacers are inserted in the gap There are many other basic factors that affect
created by the ring expansion to maintain its structural production. A few of those factors are TBM penetration,
integrity. The gap is subsequently filled with concrete and train time or speed, and encountering difficult soil
the joints are patched with cement mortar. conditions. Simulation allows for the modeling of a given
In order to design the Simphony Template, the system on the computer in order to allow the user to
following resources and specifications were added to the experiment with alternate scenarios and compare the
initial template and tested using the historical records results. The test analysis based on the Simphony template
available at CE. The other specifications, resources and and the assumptions are presented in Sections 4 and 5.

949
 Special Purpose Simulation Template for Utility Tunnel Construction

Activity Duration
Boring
Reset TBM
Liner Installation

Unload Liners (if required)

Load Dirt into Muck Cars
Extend Tracks and Services

First Train Arrives

First Train Leaves *
Second Train Arrives

Solid = First Cycle Shaded = Next Cycle

* The leaving of the train depends on the capacity of muck cars. Refer to Section 4.1:Modelling Parameters
and Processes for details.

Figure 1: Repetitive Construction Processes at Tunnel Face

Activity Duration
Hoist muck car to Ground Level
Dump Dirt, Clean and Lower the Car
Load Material Car with Liners *
Lower the Material Car to Undercut *

First Train Arrives

First Train Leaves
Second TrainArrives

Solid = First Cycle Shaded = Next Cycle

* These activities can be performed before muck cars are unloaded, if time permits.

Figure 2: Repetitive Construction Processes at Shaft

3 SIMPHONY PLATFORM FOR SPECIAL representations, navigation schemes within the environment,
PURPOSE SIMULATION. creation of model specifications and reporting are completed
in a format native to the domain itself.”
Simphony is a simulation platform for building general and Intended application domain, simulation theory and
special purpose simulation models. It is a Microsoft Windows object oriented programming are the three key ingredients
based computer system developed with the objective of of developing a special purpose simulation tool. The
providing a standard, consistent and intelligent environment successful design and implementation of a special purpose
for both the development as well as the utilization of the simulation depends on flexibility of the modeling
construction of Special Purpose Simulation (SPS) tools environment and on user-friendliness.
(Hajjar and AbouRizk 1999). AbouRizk and Hajjar (1998) In this respect Simphony fulfills the requirements to build
also defined SPS as “a computer-based environment built to successful special purpose simulation tools. Its object oriented
enable a practitioner who is knowledgeable in a given application framework provides a structured approach to build
domain, but not necessarily in simulation, to model a project any simulation template with ease, which comprises graphical,
within that domain in a manner where symbolic hierarchical, modular and integrated modeling techniques.

950
 AbouRizk, Ruwanpura, Er, and Fernando

Figure 3: Simphony Special Purpose Tunnel Simulation Template: Inputs, Processes and Outputs

951
 Special Purpose Simulation Template for Utility Tunnel Construction

4 SPECIAL PURPOSE SIMULATION TEMPLATE leaves as soon as its cars are filled, irrespective of whether
FOR TUNNEL CONSTRUCTION the TBM has bored one meter (smaller setup). The TBM is
stopped until a new train arrives at the tunnel face. If the
The Tunneling Template illustrated in Figure 3 explains in Train is partially filled when TBM achieves the required
detail the modeling elements used in the template, its meter, the train waits until TBM installs the liners for the
inputs, outputs and statistics. In this discrete event bored meter. The rest of the muck cars are filled in the next
simulation model, the input parameters can be global to the boring cycle. This material handling was an embellishment
system or individual to the modeling element. For to the preliminary model designed in order to validate the
example, the TBM element determines the operations at the model with the proposed sewer tunnel project at Mill
tunnel face based on the shift length parameter inputted at Creek, Edmonton. The train is the main entity of the
the parent level of the model. The Liner Installation Time model.
is a parameter within the TBM element, which can be The shaft-undercut element obtains the inputs for
available to the TBM element or any other element in the unloading time of dirt car from undercut to shaft and the
model. There is one parent “Tunnel” element in the model loading time of liners from shaft to undercut. The loading
with eight child elements. The children elements are: of liners is not a continuous activity for each train. If liners
are required, then this event is triggered. The crane is one
1. Muck Cars. of the resources of the model.
2. Shaft and Undercut. The track at undercut, intersection and breakout Track
3. Track Layout at Undercut. check various operations in the model and maneuver the
4. Track Intersection between Undercut and the traffic of the trains. If the first train is inside the tunnel and
Breakout. the second train has finished unloading dirt, the second
5. Waiting Track in the Undercut for Second train travels to the intersection and backs up to the waiting
Train. track. When the train comes out of the tunnel and travels to
6. Breakout Track. undercut, the train at the waiting track leaves to the tunnel
7. Tunnel Section. face. The user is also given the option to input the length of
8. TBM Machine. the breakout.
The simulation model allows the user to add as many
4.1 Modeling Parameters and Processes tunnel sections to the model depending on the soil
properties. For example, if the tunnel to be bored is 170
The parent element of the model has some input meters long in Clay soil and there are 20 meters of Bed
parameters: the total tunnel length to be bored, TBM type Rock in the tunnel to be bored starting at the 50th meter of
(open face/ close face), total bored length, shift length per the tunnel, the user can add three segments of tunnel
crew, mobilization time at the commencement of each shift section as in Figure 4. The type of soil and the length of the
length (or day), buffer time (the time between closing time tunnel section are the input parameters. The simulation
of the shift and the end of shift) at the end of the day and model determines the tunnel penetration rate and the swell
the decision to stop work for lunch. At the end of the shift, factor of the soil based on the input.
the simulation model determines whether it is possible to
bore another stroke length of the tunnel or install the liners
after the tunnel portion has been bored within the buffer
time specified. If the work is stopped for lunch,
productivity is affected by the stoppage of work before
lunch and re-mobilization time before construction starts
after lunch. Figure 4: Tunnel Segments for different soil types
The muck car element has parameters related to train
and muck cars. The capacity of muck cars, the number of The TBM element has a few more inputs: TBM size,
muck cars and the speed of the train are the input liner installation method, TBM re-set time, unloading time
parameters. The tunnel boring operation changes of liners, and liner installation time. The simulation model
depending on the total capacity of the muck cars. determines the availability of time and the need to unload
According to CE tunnel operations, TBM only bores one liners before the next boring cycle begins. TBM is a main
meter of tunnel before installing the pre-cast segment resource within the model.
liners. If the total capacity of the muck cars is greater than
the bank volume of one meter of dirt (larger setup), the 4.2 Modeling Outputs and Statistics
loaded train leaves to shaft as soon as TBM finishes boring
one meter of tunnel. If the total capacity of muck cars is In addition to Simphony’s own graphical reports and
less than the bank volume of one meter of dirt, the train statistics, there are various designer-built graphical reports
and statistics in the tunneling template. The designer-built

952
 AbouRizk, Ruwanpura, Er, and Fernando

outputs and statistics are reported to various elements of

the model. Figure 3 shows all the outputs and statistics
generated from the model. The Section 5 of the paper
explains in details some of the designer-built reports.
The utilization, waiting time, and queue length for
each resource are generated from Simphony. The crane
resource ensures an orderly and realistic representation of
the loading and unloading operations. The track resource
ensures that a maximum of one train is allowed to proceed
at a time and the TBM resource ensures that no excavation
can take place without a train. Figure 6: Tunnel Advance Rate (Larger Setup)
5 ANALYSIS OF SIMULATION RESULTS
5.2 Test # 2: Encountering Different Soil
The preliminary validation of the Tunnel Simulation Conditions in the Smaller Set Up
Template was performed using available historical data and
the experience of the personnel at the City of Edmonton. The rate of tunnel production varies depending on the soil
Table 1 depicts the historical data obtained to test the conditions. In this modeling scenario, Bed Rock is
template with the specifications of 2.3 meter diameter encountered for 20 m when the tunnel reaches 30 meters.
tunnel using LOVAT M100 TBM and two trains in good The soil for all the other areas is soft clay. The total tunnel
clay soil. The tunnel is supported with pre-cast segment length is 100 m. Figure 7 shows the tunnel advance rate for
liners. this scenario. The tunnel advance rate is considerably
reduced when boring in Bed Rock.
5.1 Test # 1: Comparison of Results by
Changing the Site Set-Up at Shaft

The availability of site space is a governing factor for

tunnel production. If the space for shaft, undercut and tail
tunnel is limited, the project manager prefers to use a
smaller setup which requires smaller muck cars that are Bed
Rock
easier to be hoisted using a small crane. The alternative is
to use larger capacity muck cars which are hoisted using a Clay Clay
comparatively larger crane. The proposed Mill Creek
tunnel intends to use a smaller set-up due to the same
problem. The tunnel template was used to test the output
results for both. Table 2 outlines the results obtained. Figure 7: Tunnel Advance Rate with different soil
Figure 5 and 6 illustrate the average Tunnel Advance Rate conditions
(m/hr) for both cases.
The simulation results show that the smaller setup 6 FUTURE ENHANCEMENTS
generates a less productive tunnel advance rate. The TBM
utilization is also reduced in the smaller set up. The crane This simulation template predicts the results based on the
utilization increased in smaller set up due to higher algorithms developed in the model, which are entirely
frequency of muck car handling. based on the actual construction processes. Following are
future embellishments to the model.

1. Extend the capability of the template to

model the construction of a tunnel using
hand-tunneling method and other sizes of
tunnel boring machines for both open face
and closed face.
2. Allow the users to pick compatible resources
for each tunneling activity and predict the
comparative results.
3. In the present model, different soil properties
Figure 5: Tunnel Advance Rate (Smaller Setup) can be added on an arbitrary basis. The future

953
 Special Purpose Simulation Template for Utility Tunnel Construction

Table 1: Historical Input Parameters

Input Parameter Value

Train Speed (kmph) Uniform (3.00, 4.00)
Unloading Dirt for M uck Car (1.9 m3) - M ins Triangular (3.00,3.50,4.00)
Unloading Dirt for M uck Car (0.75 m3) - M ins Triangular (2.75,3.00,3.50)
No. of M uck Cars - Small Capacity 6
No. of M uck Cars - Large Capacity 4
Loading of Liners from Shaft to Undercut - M ins Cons tant (4.00)
TBM Penetration Rate for Clay (m/hr) Uniform (4.00, 4.25)
TBM Res t Time - M ins Uniform (1.00, 3.00)
Unloading of Liners from Trains - M ins Cons tant (4.00)
Liner Ins tallation Time Triangular (15.00,16.00,18.00)
Shift Length - M ins 480
M obilization time at the commencement of each day - M ins 15
Buffer Time at the end of the day - M ins 15
Break for Lunch Yes

Table 2: Statistics for Tunnel Simulation – Test # 1

Large Muck Cars Small Muck Cars

Tunnel Length 560 361
Days 45 45
Average Tunnel Adavance Rate (m/hr) 1.56 1.00
Average TBM Utilization 82.80% 52.30%
Average TBM Queue Length - Mins 0.14 0
Average TBM Waiting Time - Mins 1.9 0
Average Crane Utilization 39.65% 49.05%
Average Crane Queue Length - Mins 0 0.03
Average Crane Waiting Time - Mins 0 0.78

enhancement will be able to predict the aspect of the model. This would help the engi-
possibility of meeting difficult soils based on neers, planners and constructors to simulate the
a geotechnical analysis. The authors intend to operations using multiple runs and establish the
study the probability of meeting a difficult most cost-effective tunneling resources.
soil using Markovian Analysis and embellish
the template. Further this would allow an 7 CONCLUSION
accurate predicting of the tunnel penetration
rate for a composite and mixed soil strata An application of computer simulation for tunnel
when it bores horizontally. construction has been described using the historical data
4. The other possible problems during tunnel and the tunneling process obtained from the City of
operations will also be added to the template. Edmonton Public Works Department. The objective of the
These include breakdown of the TBM, analysis of the simulation model was to estimate the
maintenance of equipment, surveying, production of the tunnel advance rate and the optimization
adjustments for curvatures and slopes, of the resources. Further, the effect on tunnel advance rate
unforeseeable events such as water seepage due to the other variables has been analyzed. The results
and structural damages etc. show that the tunnel advance rate is inter dependent on
5. Based on the productivity and the utilization of various factors and activities in the tunneling operations.
the resources, a cost estimation module could The resources can be optimized if both cycles of tunnel
be added to the tunnel template to enhance the face and shaft operations are analyzed and adjusted to
project management and project planning achieve the maximum efficiency. This template will be

954
 AbouRizk, Ruwanpura, Er, and Fernando

tested again with actual data obtained from the proposed cover as well as various trenchless techniques. He has an in
sewer tunnel for Mill Creek at Edmonton to be depth knowledge and experience in shaft sinking and soft
commencing in the Fall of 1999. ground tunneling and has published several papers in
This study is another opportunity to show that various engineering magazines.
simulation can be successfully used in planning a
construction project. The future enhancements identified in SIRI FERNANDO is a Program Manager with City of
the paper will provide a more comprehensive template for Edmonton, Asset Management and Public Works,
tunnel simulation. The accuracy of the tunnel simulation Drainage Services. He Graduated from University of
template will assist the consultants and the contractors to Ceylon with a B.Sc. (Honours) Degree in Civil
develop more realistic approaches to meet the challenges Engineering in 1975 and completed a M.Eng in
they will face as construction progresses. Construction Engineering and Management at University
of Alberta in 1998. He has over 20 years experience in
REFERENCES design and construction of deep underground utility
structures, specializing in soft ground tunneling
AbouRizk, S.M., and D. Hajjar. 1998. A Framework for
Applying Simulation in the Construction Industry,
Canadian Journal of Civil Engineering, CSCE,
25(3):604-617.
Hajjar, D., and S. AbouRizk. 1999. Simphony: An
Environment for Building Special Purpose
Construction Simulation Tools. Submitted to the
Proceedings of the 1999 Winter Simulation
Conference, Phoenix, AZ.

AUTHOR BIOGRAPHIES

SIMMAN M. ABOURIZK is a Professor in the

Department of Civil and Environmental Engineering at the
University of Alberta. He currently holds the
NSERC/Alberta Construction Industry Chair in
Construction Engineering and Management. He received
his BSCE and MSCE in Civil Engineering from Georgia
Institute of Technology in 1984 and 1985 respectively; and
Ph.D. degree from Purdue University in 1990. His research
interests focus on the application of computer methods and
simulation techniques to the management of the
construction projects.

JANAKA Y. RUWANPURA is a Ph.D. candidate in the

Department of Civil and Environmental Engineering at the
University of Alberta. He earned his B.Sc. (Honours) from
the University of Moratuwa, Sri Lanka in 1992 and his
M.S. in Construction Management from Arizona State
University in 1995. His research focuses on applying
simulation to the construction project management
processes. He has received many local and international
scholarly awards in his career including a Fulbright
Scholarship.

K. C. ER is the Director of Design and Construction,

Drainage Services, Asset Management and Public Works
Department, The City Of Edmonton. He earned his B.Sc.
in Mechanical Engineering from the University of Alberta
in 1975. He has 24 years of experience in the construction
of drainage and water infrastructures using the cut and

955