Tunnel boring machine positioning automation
in tunnel construction
Xuesong Shen 1, Ming Lu 1*, Siri Fernando 2, and Simaan M. AbouRizk 1
1 Department of Civil and Environmental Engineering, University of Alberta, Edmonton,
Alberta, Canada
2 Design and Construction, Drainage Services, City of Edmonton, Edmonton,
Alberta, Canada
* Corresponding author (mlu6@ualberta.ca)
Purpo Tunnel construction using a tunnel boring machine (TBM) entails preci machine positioning and guidance in the underground space. In contrast to traditional lar-bad machine guidance solutions, the propod rearch aims to develop an automation alternative to facilitate TBMguidance and as-built tunnel alignment survey during tunnelling op-erations. Method A fully automated system is
propod, in which a robotic total station is employed to automate the continuous process of TBM -tracking and positioning in the 3D underground working space. ZigBeebad wireless n-sor networks are applied for wireless data communication inside the tunnel. A camera is mounted on the telescope of the total station to capture online operational videos. Real-time survey data are thus acquired, procesd and displayed on a tablet PC on the fly, resulting in: (i) TBM’s preci coordinates in the underground space; (ii) three-axis body rotations of the TBM; (iii) tunnelling chainage progress; and (iv) line and grade deviations of the tunnel alignment. Results & Dis-cussion For proof-of-concept, a prototype TBM-positioning automation system has been developed in-hou for labora-tory testing. The accuracy testing was conducted by the automation system and a specialist surveyor independently. The differences between the two ts of surveying results were less than 2mm, which sufficiently validated the high accuracy of the automation solution. In April 2012, the prototype will be field tested on a 2.4 m diameter and 1,040 m long drainage tunnel project in Edmonton, Canada.
Keywords: automation, tunnel construction, TBM, machine control and guidance.
张曜
I NTRODUCTION
For operators in the tunnel construction field, steer-ing a tunnel boring machine (TBM) is like driving a vehicle in complete darkness. The current practice for TBM guidance largely relies on traditional lar guidance systems which project a lar point onto a target board fixed on a TBM. Limitations of the prac-tice, however, potentially contribute to the high risks in executing tunneling projects, such as out-of-tolerance alignment deviations, project delay and budget overrun. Particularly, unforeen under-ground obstacles and variable geologic conditions further complicate tunnel alignment control. It is not unusual that TBM operators and site managers are caught by surpri with excessive out-of-tolerance tunnel alignment errors1. It may take weeks or longer time to determine the exact alignment deviations by survey specialists. Sometimes, the TBM can be trapped in the underground space, requiring consid-erable time, cost and effort for recovery; in worst-ca scenarios, the TBM has to be abandoned in the ground due to prohibitively high cost of rescuing it. This rearch aims to develop an automation system for TBM positioning and guidance. A fully automated solution is propod, in which a robotic total station is employed to automate the continuous process of TBM tracking and positioning in the three dimen-sional (3D) underground working space. ZigBee-bad wireless nsor networks are applied for wire-less data communication inside the tunnel. Real-time survey data are thus acquired and procesd on the fly, resulting in: (1) tunneling chainage progress; (2) line and grade deviations of the tunnel alignment; (3) three-axis body rotations of the T
BM; and (4) preci coordinates of any invisible points on TBM in the underground space. Further, the solution provides multiple role-bad ur interfaces and lends real-time, relevant assistance to TBM operators, tunnel surveyors, and project managers in making critical decisions.
The remainder of this paper is organized as follows: First, the pros and cons of the traditional lar guid-ance system are evaluated. We then illuminate sys-tem design of the propod automation solution, followed by a tunnel estimation ca to contrast improved and current work process. Main findings and the practical applicability of the rearch are summarized in conclusions.
T RADITIONAL LASER GUIDANCE SYSTEM
Lar guidance systems have predominated in tun-neling applications for many decades. Generally, a
lar station is firmly fixed inside the tunnel, project-ing a lar point onto a lar target board mounted on the TBM, as shown in Figure 1. Bad on the offts of the lar spot on the target board, the TBM operator infers the current line and grade tunnel alignment deviations.
Fig.1. Guiding lar beam inside tunnel
TBM’s three-axis orientations in the underground space are crucial to machine steering control. Cou-pled with the traditional lar system, a two-axis bubble leveler is commonly installed on the TBM to gauge its rotation angles of pitch and roll in vertical planes. Meanwhile, the advancing direction of the TBM (yaw in the horizontal plane) can be determined through installing a transparent front target along工作文章
点点滴滴
with the rear lar target board (e Figure 2).
Fig. 2. Lar target boards mounted on the TBM: (a) transparent front target, (b) rear target
One of the major limitations associated with the tradi-tional lar guidance system lies in relatively low accuracy and reliability due mainly to three factors, namely: (1) potential manual errors in initializing or calibrating the lar beam’s alignment, (2) dispersion and refraction of the lar beam over a long dis-tance, and (3) difficulty to receive lar’s projection becau of excessive TBM deviations 2. Typically, the maximum application distance for the lar guidance system is around 200 m. Besides, the lar beam’s alignment toned to be frequently calibrated by spe-cialist surveyors (at least once every other day). As a result, the tunneling productivity can be considerably
undermined by operation and maintenance of the lar guiding system.
In order to facilitate tunnel alignment control, com-mercial companies have developed advanced TBM guidance systems by integrating sophisticated me-chanical, optical and electromagnetic subsystems 3. Tight space constraints and harsh work conditions in the tunnel may not satisfy system installation re-quirements. On the down side, the high complexity in system design may compromi system reliability while considerably increasing the system’s price and consumption cos
t, including system maintenance and technical rvice 4.
TBM POSITIONING AUTOMATION SYSTEM
The propod TBM positioning solution combines four functions: (1) TBM tracking automation through surveying-computing integration; (2) wireless data communication enabled by ZigBee-bad wireless nsor networks; (3) “virtual lar target board” pro-gram for TBM guidance; and (4) real-time visualiza-tion of tunnel construction in a 3D environment.
TBM tracking automation
The system employs a robotic total station to realize an automated, continuous process of TBM tracking and spatial data collection inside the tunnel, as illus-trated in Figure 3. TBM’s coordinates as well as its line and grade deviations from the as-designed tun-nel alignment are computed in real time. By u of a limited quantity of tracking targets fixed on the TBM, the three-axis orientations of the TBM in the under-ground working space are computed by applying innovative “point-to-angle” algorithms, without the need of using any gauges (such as levelers, gyro-scopes, inclinometers and compass)1,4
.
Fig.3. Automated target tracking for TBM positioning and orientations computing
Wireless data communication
Wireless nsor networks are purpo-deployed in the system design, enabling on-site data communi-cation between key components of the TBM tracking system, namely, the total station, a control laptop computer in the underground tunnel, as well as a monitoring computer on the surface.
ZigBee-bad wireless nsor networks are de-
ployed in this system. In general, the emerging wire-less nsor networks technology provides a smart, cost-effective and energy-efficient network infrastruc-ture, which consists of a group of intelligent nsor nodes that can wirelessly communicate with one another. ZigBee reprents a global specification for wireless nsor networks bad on the IEEE 802.15.4 standard5. Typically, the battery life of a ZigBee nsor node is around veral months, which can be further extended to years under the “sleep” operation tting (analogous to tting a computer to the sleep mode).
In the field implementation of the propod solution, a control laptop computer is placed adjacent to the steering panel of the TBM. One ZigBee wireless node is linked with the robotic total station through a rial data cable, the other ZigBee node with the USB interface is plugged in the control laptop, as shown in Figure 4. Real-time surveying results are transmitted to the computer, while remote control commands are forwarded to the total station through
the same wireless data communication channel. Fig. 4. Wireless nsor networks for data communica-
tion between the robotic total station and control laptop Virtual lar target board program
A unique interface of the software system is a “virtual lar target board”, which is displayed in the control computer to guide the TBM. Four fundamental mod-ules are integrated in the program, including (1) total station communication module (TSCM): this module handles wireless communication between the control laptop computer and the total station; TSCM controls the total station operations by executing prepro-grammed point tracking and surveying commands, and translates the feedback from the total station in its “machine language” for further computing; (2) tracking and positioning computing module (TPCM): this module forms the core of the whole system and it computes TBM’s positions and attitudes from the coordinate data of the surveyed points. The compu-ting results are pasd over to the data publishing module; (3) analytical data publishing module (ADPM): the purpo of this module is to connect a data producer (for example TPCM) to a data con-sumer (for example the ur interface module). It stores all analytical results in a queue and propa-gates any updates to all subscribers; (4) role bad ur interface module (RUIM): urs of different roles have different ur interfaces and each ur interface has its own policy to render d
ata.
The main ur interface is designed for the TBM operator, which mimics a traditional lar target board the operator is familiar with, as shown in Fig-ure 5. This ur interface consists of (1) two perpen-dicular lines in the center of the screen and the crosshair indicating the as-designed alignment of the tunnel project; (2) two points onto the screen repre-nting the current positions of the two center points at the tail and the head ctions of the TBM, which are practically invisible in the underground space; (3) a square box which defines the TBM deviation toler-ance limits. If the two points are both enveloped inside the box, it means at the current moment the TBM’s alignment deviations are well controlled within the specified tolerances. The Euclidean distances from the tail/head point to the two perpendicular lines define accurate measures of line and grade devia-
鸟字笔顺
tions of the tunnel alignment.
13一15小年同性
Fig. 5. Interface of virtual lar target board program
3D visualization of tunnel construction
A ur-friendly 3D platform is provided in the system in order to visualize analytical results describing the TBM's real-time position state, the tunnel design and the construction progress. It aids project managers in making critical decisions on a near real-time basis. The tunnel design and tunneling process are visual-ized in three steps: (1) before the construction pha, relevant environment data, like ground topog-raphy, strata information, geotechnical parameters, and the as-designed tunnel alignment are modeled in the system; (2) during the construction pha, the system reads TBM real-time positioning data and animates the construction process. The difference between the as-designed alignment and the as-built alignment can be readily visualized through 3D com-puter graphics, thus allowing project managers and engineers to monitor what is happening underground in an intuitive VR environment; (3) after the construc-tion pha, the tunnel alignment control process and the as-built tunnel alignment can be reviewed, while the TBM operator’s experience can be captured for performance asssment and training purpos.
Figure 6 visualizes a simulated tunnel project. The
progress of tunnel construction is prented in the complicated underground space, where the different colors of as-built tunnel ctions indicate the quality of the tunnel alignment (green – within tolerance; red
– out of tolerance).
Fig. 6. Real-time 3D visualization of tunnel construction
P RODUCTIVITY AND COST PERFORMANCE ANALYSIS
In this ction, a ca study of evaluating productivi-ty and cost performances on tunnel construction by
u of the two alternative TBM guidance systems is
prented. Suppo a 1,000-meter-long tunnel is to
be built, total project time and direct construction
cost are estimated for the traditional lar system
and the propod automation system, respectively.
The tunnel crew consists of one Tunnel Supervisor,
one Tunnel Forman, one TBM Operator, one Crane
Truck Operator, two Tunnel Laborers (level II) and
four Tunnel Laborers (level I). A survey crew consists
of three surveyors, as given in Table 1. The tunnel
crew works 8 hours/shift, 1 shift a day, 5 days a
week. The survey crew works on the site by ap-pointment only. Bad on the u of the traditional
lar system on previously completed tunnel projects
and productivity analysis using historical data, the
average production rate is determined to be 5
m/shift, which factors in different types of delays in
connection with survey checking and alignment con-trol.
Table 1: Crew and equipment cost information
Quantity Job Hourly salary 6
1 Tunnel Supervisor $ 33.544
1 Tunnel Forman I $ 28.554
1 TBM Operator $ 26.915
1 Crane Truck Operator $ 25.536
2 Tunnel Laborer II $ 25.936
4 Tunnel Laborer I $ 25.148
3 Surveyor $ 24.573
Quantity Equipment Hourly rental
fee
1 TBM $ 315.000
1 Crane Truck $ 120.000
For the traditional lar system:
∙
“Routine survey”: In every 10 m TBM advances, a shutdown for 1.5 hours is necessary for rou-tine checking of lar alignment by the survey crew;
∙ “Relocation survey”: In every 200 m a shutdown
for 5 hours for lar station relocation and lar realignment by the survey crew is necessary, which includes any “routine survey” if needed; ∙ “Misalignment shutdown”: In every 800 m a
potential shutdown for 1 week is required to fix TBM misalignment issues by the tunnel crew. Note: the surveyors are only paid bad on the num-ber of hours they spend on survey checking and lar realignment rvices; while the equipment rental fee is charged bad on time of availability on the site. Meanwhile, during non-productive shutdown
periods, the tunnel crew would work on tunnel
maintenance while still being paid by their hourly
rates.
Given the average production rate of 5 m/shift, the
total project time by u of the lar system is de-termined as:
(1000/5)*8 = 1,600 h
The cycles as required for routine survey, relocation
survey and misalignment shutdown are determined
as below, respectively:
1000/10-1 = 99 cycles
1000/200-1 = 4 cycles
1000/800 ≈ 1 cycle
Considering the overlap between survey rvices
and misalignment shutdowns, durations for routine
survey, relocation survey and misalignment shut-down are as below, respectively:
(99-4-1)*1.5 = 141 h
(4-1)*5 = 15 h
1*5*8 = 40 h
The total shutdown time is 196 h. Therefore, actual
tunneling time is 1,600-196 = 1,404 h.
Bad on the information given in Table 1, the hourly
wages for the tunneling and survey crews are calcu-lated as $ 267.013 and $ 73.719, respectively. The
direct labor costs for the tunnel crew and the survey
crew are calculated as below, respectively:
1600*267.013 = $ 427,220.8
(141+15)* 73.719 = $ 11,500.164 Since the equipment rental fee is charged bad on time of availability on the site, the equipment rental time is estimated as: (1600/8)/5 = 40 weeks = 6,720 h The direct equipment rental cost is: 6720*(315+120) = $ 2,923,200 In total, the direct project cost is
$ 3,361,920.964. For the propod automation system: ∙ “Routine survey”: In every 50 m TBM advances, a shutdown for 1 hour is necessary for routine checking by the survey crew;
∙“Relocation survey”: In every 200 m a shutdown for 2 hours is required for total station relocation by the survey crew, which includes any routine survey if needed;
∙Shutdowns due to misalignment fixing are not required.
Assuming the same tunneling hours are required on the project when the automation system is applied, the actual tunneling duration is 1,404 h.
The cycles for routine and relocation surveys are:
1000/50-1 = 19 cycles
1000/200-1 = 4 cycles
日进斗金是什么意思Considering the overlap between survey rvices and shutdowns, the durations for routine and reloca-tion surveys are:
(19-4)*1 = 15 h
4*5 = 20 h
The total shutdown time is 35 h. Then, the total pro-ject time is 1,404+35 = 1,439 h.
As such, the average production rate using the au-tomation system is:
1000/1439*8 = 5.56 m/shift
The direct labor costs for the tunnel crew and the survey crew are:
1439*267.013 = $ 384,231.707
35* 73.719 = $ 2,580.165
The equipment rental time is estimated as:
(1439/8)/5 = 36 weeks = 6,048 h
The direct equipment rental cost is:
6048*(315+120) = $ 2,630,880 Therefore, the direct project cost is $ 3,017,691.872. Table 2 compar
es the total project times, the direct construction costs and the average production rates for the two TBM guidance systems. When the auto-mation system is applied to replace the traditional lar system, the contractor would save $ 344,229 on the direct construction cost. Meanwhile, it is esti-mated that 10.1% shorter project duration and 10.2% lower direct cost would be achieved, while the aver-age production rate would be incread by 11.2% to 5.56 m/shift.
Table 2: Productivity improvement and cost savings by u of the propod automation system
Lar system Automation
system
舞台灯光效果图Compari-
son
Project time 1600 h 1439 h
名诗句
-10.1%
Direct con-
struction cost $ 3,361,921
$ 3,017,692 -10.2%
Production
rate
5 m/shift 5.5
6 m/shift 11.2%
S YSTEM PROTOTYPING AND FIELD TESTING
A prototype of the propod automation system was developed in-hou at the University of Alberta. The automation prototype mainly consists of three mini tracking targets (model: CTS Leica Compatible Mini Prism 65-1500M), a robotic total station (model: Leica TCPR1203+) and three ZigB
ee wireless n-sor nodes (model: SENA ProBee ZS10 and ZU10), as shown in Figure 7. In clo collaboration with the Design and Construction Section of the City of Ed-monton, the new solution is scheduled to be imple-mented in a 2.4 m diameter and 1,040 m long drain-age tunnel project in Edmonton, Canada for field performance evaluation from the end of April 2012.
(a) (b)
Fig.7. Prototype of the TBM positioning automation system: (a) three tracking targets mounted on a 2.4 m diameter TBM, (b) robotic total station linked with a ZigBee wireless node
C ONCLUSIONS
Increasing demands for better underground infra-structure have spurred tunnel construction all over the world, within which the TBM tunneling method is the most commonly applied. The lack of effective TBM guidance solutions, however, potentially con-tributes to incread risks and uncertainties in tunnel construction
In this rearch, we have developed an automation solution for TBM positioning, which integrates auto-mation control mechanisms, innovative computing algorithms, and wireless network technologi
es. Meanwhile, the multiple role-bad ur interfaces lend substantial decision support for TBM operators, tunnel surveyors, and project managers to track the construction progress as well as visualize any tunnel alignment deviations on the fly.
The project estimation ca study indicates by adopting the propod automation system tunneling productivity would be improved by 11.2% against using the traditional lar system. The resulting pro-ject duration and the direct construction cost would both be reduced by about 10%. The realistic system performances and the productivity improvement will be further validated through conducting extensive field experiments of the automated TBM positioning system at Edmonton, Canada starting from the end of April 2012.
