Engineering Structures 28 (2006) 648659

www.elsevier.com/locate/engstruct

Fiber Bragg grating sensors for structural health monitoring of Tsing Ma

bridge: Background and experimental observation
T.H.T. Chan a, , L. Yu a,d , H.Y. Tam b , Y.Q. Ni a , S.Y. Liu b , W.H. Chung b , L.K. Cheng c
a Department of Civil and Structural Engineering, The Hong Kong Polytechnic University, HungHom, Kowloon, Hong Kong, PR China
b Photonics Research Centre, Department of Electrical Engineering, The Hong Kong Polytechnic University, HungHom, Kowloon, Hong Kong, PR China
c TNO (TPD), Delft, The Netherlands
d Blasting and Vibration Department, Changjiang River Scientific Research Institute, 23 Huangpu Street, Wuhan, Hubei 430010, PR China

Received 16 February 2005; received in revised form 15 September 2005; accepted 16 September 2005
Available online 27 October 2005

Abstract
The rapid expansion of the optical fiber telecommunication industry due to the explosion of the Internet has substantially driven down the cost
of optical components, making fiber optic sensors more economically viable. In addition, the rapid development of fiber-optic sensors, particularly
the fiber Bragg grating (FBG) sensors offers many advantages and capability that could not be achieved otherwise. In the past few years, fiber
Bragg grating sensors have attracted a lot of interest and they are being used in numerous applications. This paper describes the FBG sensors
developed for structural health monitoring, and were installed on Hong Kongs landmark Tsing Ma bridge (TMB), which is the world longest
(1377 m) suspension bridge that carried both railway and regular road traffic. Forty FBG sensors divided into three arrays were installed on the
hanger cable, rocker bearing and truss girders of the TMB. The objectives of the study are to investigate the feasibility of using the developed
FBG sensors for structural health monitoring, via monitoring the strain of different parts of the TMB under both the railway and highway loads
as well as comparing the FBG sensors performance with the conventional structural health monitoring system Wind and Structural Health
Monitoring System (WASHMS) that has been operating at TMB since the bridges commissioning in May 1997. The experimental observations
in this project show that the results using FBG sensors were in excellent agreement with those acquired by WASHMS.
c 2005 Elsevier Ltd. All rights reserved.

Keywords: Structural health monitoring; Fiber Bragg grating (FBG); Tsing Ma bridge (TMB); Strain

1. Introduction
Fiber photosensitivity is the main phenomenon involved
in writing Bragg gratings into the core of a fiber, which
was first demonstrated by K.O. Hill et al. in 1978 at
Canadian Communications Research Centre (CRC), Ottawa,
Ont., Canada [1]. However, pioneering work at the United
Technology Research Centre in fabricating fiber Bragg gratings
in a fiber core through its side, which was a significant
milestone for in-fiber Bragg grating (FBG) sensors, was
published eleven years later by G. Meltz et al. in 1989
[2]. This side-writing technique creates a Bragg grating
It is hereby stated that content of this paper has not been published
elsewhere and it has not been submitted for publication elsewhere.
Corresponding author. Tel.: +852 2766 6061; fax: +852 2334 6389.
E-mail address: cetommy@polyu.edu.hk (T.H.T. Chan).

c 2005 Elsevier Ltd. All rights reserved.

0141-0296/$ - see front matter 
doi:10.1016/j.engstruct.2005.09.018

directly in the fiber core using a holographic interferometer

illuminated with a coherent ultraviolet (UV) source. Since
then, many methods have been found to increase the refractive
index by improving both the UV exposure method, and the
photosensitivity of the fiber core [3]. Transverse holographic
methods are particularly useful for producing the modulated
UV since they can easily create FBG sensors with both the
desired spectral response and at any position along the fiber
[4]. In 1993, an advanced FBG production technique was
reported [5,6], which involved the use of an optical phase mask
to generate interference fringes. Similar techniques for FBG
production are used in most current FBG sensor work. Earlier
research on FBG sensors concentrated on the fundamentals of
the reliability of these sensors in terms of their strain sensitivity,
repeatability, signal demodulation techniques, and the general
characterization of their performance [79]. These studies have
shown encouraging results, underlining the potential of FBG

T.H.T. Chan et al. / Engineering Structures 28 (2006) 648659

sensors for strain monitoring. The measurand versatility and

the unique advantages offered by FBG sensors have resulted
in their use in a wide range of sectors such as in advanced
aircraft and space vehicles [10,11], marine [12] and medical
science [13]. Recently, it has been demonstrated that FBG
sensors have great potential for a wide range of applications
where quasi-distributed measurements of physical parameters
such as strain, pressure, vibration, temperature, ultrasound, high
magnetic field and high-g acceleration are required [14,15].
Comprehensive reviews of recent progress in the applications
of FBG are available elsewhere in the literature [1618].
These research works have shown that FBG sensors
have several inherent advantages over conventional electrical
sensors such as small size, light weight, non-conductivity, fast
response, resistance to corrosion, higher temperature capability,
and immunity to electromagnetic noise and radio frequency
interference. The distinct advantages of FBG sensors over other
types of fiber optic sensors are their multiplexing capability and
wavelength-encoded measurand information. A single string of
optical fiber can accommodate up to many tens of FBG sensors.
The measurand information is encoded in the wavelength,
which is an absolute parameter and thus FBG sensor systems
are less susceptible to signal amplitude fluctuations. Another
attractive feature of FBG sensors is their inherent ability to
serve as both the sensing element and the signal transmission
medium which opens new possibilities in the field of reliable
remote structural health monitoring [19,20]. In this regard,
application of this new technology would have a significant
impact on the health and efficiency of civil infrastructure
systems [21]. However, the practical applications of this kind of
sensor to real civil engineering structures have not been widely
adopted although the progress of fiber optic health monitoring
is impressive [22,23].
Different types of fiber Bragg grating (FBG) sensors
for simultaneous strain and temperature measurement and
for temperature independent strain measurement have been
developed by The Hong Kong Polytechnic University
[2429]. Fiber-laser-based wavelength-division-multiplexed
(WDM) FBG sensor interrogation techniques and a broadband
light source-based multiplexed FBG sensor interrogation
system have also been devised. The system can perform both
static and dynamic strain measurement (up to a sampling rate
of 52 Hz), and allows one to read the measurement data at
locations tens of kilometers away from the monitoring site.
After a series of successful laboratory tests and verification,
the system has been installed in different parts of the Tsing
Ma bridge [30] and compared with the results obtained by
a sophisticated long-term monitoring system, known as Wind
And Structural Health Monitoring System (WASHMS), which
was devised and implemented by the Highways Department
of Hong Kong SAR Government to monitor the structural
health and conditions of the three cable-supported bridges
in Hong Kong, including the Tsing Ma bridge, the Ting
Kau bridge and the Kap Shui Mun bridge. This on-structure
instrumentation WASHMS system consists of a total of about
800 sensors of different types permanently installed on the
three bridges, including strain gauges, GPS position sensors,

649

accelerometers, level sensors, temperature sensors and weighin-motion sensors. Because the WASHMS was implemented
several years ago, it has not benefited from the newly developed
optical fiber sensor technology. In order to investigate the
feasibility of using the developed FBG sensors for structural
health monitoring, a field test was carried out in May 2003,
in which a number of such FBG sensors were installed on
the Tsing Ma bridge to conduct real time and full scale
measurements. The results were assessed and compared with
the conventional strain gauges obtained from the WASHMS.
This paper first introduces the background of fiber Bragg
grating sensor technology and then follows by the high-speed
demultiplexing/interrogation system for FBG sensor arrays,
FBG sensor fabrication and package units. In this paper, the
installation and experimental setup in the Tsing Ma bridge are
also described, and the preliminary observation results on the
structural health monitoring of the Tsing Ma bridge using the
FBG sensor units are presented, analyzed and compared with
the conventional strain gauges employed in the WASHMS of
the Tsing Ma bridge. Some conclusions are made, and further
measurements and discussions in detail will be reported in
another subsequent paper [31].
2. Background of fiber Bragg grating sensor technology
In the field of fiber-optic sensors (FOSs) the FBG sensors
are one of the most exciting developments in recent years. They
have a unique property and many advantages over other FOSs
due to their quasi-point sensing and multiplexing capability.
The most important advantage of an FBG sensor is that the
measurand is encoded directly in terms of the wavelength,
which is an absolute parameter and does not suffer from
disturbances of the light paths [32]. Hence, the output signal
is independent of the intensity of the source, and losses in
the connecting fibers and couplers. Furthermore, each of the
reflected signals will have a unique wavelength and can be
easily monitored; an array of wavelength-multiplexed FBG
sensors may thus be implemented for simultaneous multiple
measurements using a single fiber. A typical FBG is a periodic
perturbation of the refractive index in the fiber core as shown in
Fig. 1.
2.1. FBG-based sensing principle
A FBG is a periodic structure, which is written into a
segment of germanium-doped single-mode fiber in which a
periodic modulation of the core refractive index (RI) is formed
along the fiber length by exposure of the core to a spatial
pattern of UV light at 197 or 248 nm wavelengths. When
light within a fiber passes through a FBG, multiple Fresnel
reflections take place along the entire length of the grating due
to the variations in refractive index. Constructive interference
between the forward wave and the contra-propagating light
wave occurs when the wavelength of the propagating light in
the fiber doubles the grating pitch, i.e. the Bragg (or phase
matching) condition is satisfied. This leads to narrowband backreflection of light. The reflected wavelength is known as the

650

T.H.T. Chan et al. / Engineering Structures 28 (2006) 648659

Fig. 1. A periodic perturbation of the refractive index in the fiber core.

Bragg wavelength, B , and is given by

B = 2n

(1)

where n is the effective refraction index of the fiber core and

the period of the index modulation. Both n and depend
on temperature and strain, consequently the Bragg wavelength
is sensitive to both strain and temperature. A FBG is thus
an intrinsic fiber-optic sensor. Light that does not satisfy the
Bragg condition passes through the FBG with very low loss,
as shown in Fig. 1 [33]. The changes of index created in FBGs
are relatively permanent and FBGs are sensitive to a number
of physical parameters. Thus, by monitoring the resultant
changes in reflected wavelength, FBG sensors can be used in
a variety of sensing applications to measure physical quantities,
for example, strain, temperature, pressure, ultrasound, high
magnetic field, force and vibration. Each of the reflected signals
will have a unique wavelength and can be easily monitored, thus
achieving multiplexing of the outputs of multiple sensors using
a single fiber. However, the FBG central wavelength will vary
with the change of these parameters experienced by the fiber.
For an applied longitudinal strain change of , the wavelength
shift, BS , is given by
BS = B (1 ),

(2)

where is the photoelastic coefficient of the fiber, given by

=

n2
[12 (11 12 )],
2

(3)

where 11 and 12 are the components of the fiber-optic

strain tensor and the Poissons ratio. For the silica fiber,
the wavelengthstrain sensitivities of 800 nm and 1.55 m
FBG sensors have been measured as 0.64 pm/ and
1.15 pm/ respectively [7,34]. For a temperature change of
T , the corresponding wavelength shift,  BT , is given by
 BT = B ( + ),

(4)

where and are the thermal expansion coefficient and

thermo-optic coefficient of the fiber material respectively. For
the silica fiber, the wavelengthtemperature sensitivities of
800 nm and 1.55 m FBG sensors have been measured with
values of 6.8 pm/ C and 13 pm/ C respectively [7,34].
The refractive index and the grating period, and subsequently,
the Bragg wavelength vary with changes in axial strain  and
temperature T . The shift in the Bragg wavelength in response
to strain and temperature is given by
 B
= (1 ) + ( + )T.
B

(5)

2.2. FBG multiplexing technique

The primary advantage of FBG sensors is their capability
for multiplexing operation. In most practical applications, FBG
sensors need to be multiplexed in order to achieve quasidistributed measurements and potentially to compete with
conventional electrical or other types of optical sensors. A large
array of FBG sensors may be addressed by a single source

T.H.T. Chan et al. / Engineering Structures 28 (2006) 648659

and detector using one or a combination of the four standard

techniques: time-division multiplexing (TDM), spatial-division
multiplexing (SDM), frequency-division multiplexing (FDM)
and wavelength-division multiplexing (WDM). In principle, the
multiplexing techniques which are suitable for conventional
fiber-optic sensors can also be applied to FBG sensors [35].
Fig. 2 shows the schematic of a scanning filter FBG
interrogation system based on WDM techniques. A broadband
source is used to interrogate a chain of wavelength-multiplexed
FBG sensors. A quasi-distributed chain of FBG sensors would
then be multiplexed in the wavelength domain to separate
the individual sensor signals. WDM interrogation systems
typically require highly reflective FBG sensors each operating
in a distinct wavelength window. It is one of the most
straightforward techniques for FBG elements in which the FBG
sensors in the network are arranged such that their operational
regions do not overlap. The maximal number of FBG sensors
along a single fiber that could be interrogated by a WDM
interrogation unit depends on the optical bandwidth of the light
source, operation range of each FBG sensor and the tuning
range of the tunable filter. The typical number of FBG sensors
that can be dealt with by a WDM interrogation unit is about 40
and is mainly limited by the light sources optical bandwidth.
The FBG sensors would be assigned an operating wavelength
band wide enough to cover the measurand-induced wavelength
shift without overlap [36]. Commercial FBG interrogation
systems based on scanning FabryPerot (FP) filters are
available from several companies. Spatial-division multiplexing
(SDM) is a simple technique that involves the splitting of
the interrogation light, typically using a fused fiber-coupler,
to several strings of optical fiber. The reflected light from
the FBGs of each fiber is either routed to an equal number
of wavelength measurement units or time-shared by a single
wavelength measurement unit with the help of an optical
switch. The FBG-reflected signal from each string of fiber does
not interfere and thus identical FBGs (i.e. FBGs with the same
Bragg wavelength) can be used in each fiber. This technique
combined with the WDM technique can increase the number of
FBG sensing points significantly.
Fig. 3 shows the schematic diagram for a FDM interrogation
system [36]. A wavelength controlled laser source is used to
address a chain of identical FBG sensors. The source has
a periodically modulated chirp in frequency that is smaller
than the line width of the gratings. The reference signal from
the first coupler interferes with the reflected FBG signals.
If the FBG sensors operate at different frequencies, the
signals from each can be separated using switching electronics.
Successful demultiplexing relies on no overlap between the
FBG modulation frequencies.
Fig. 4 shows that schematic of a TDM interrogation system
where narrow optical pulses were launched into an optical fiber
containing many low-reflection FBG sensors with virtually
identical Bragg wavelengths. Light takes about 10 ns to
propagate one round trip along 1 m of optical fiber and
therefore the separation between adjacent sensors must be
greater than 1 m for interrogating pulses of 10 ns wide.
Individual sensors are distinguished by measuring the time of

651

flight of signals returning to the interrogation unit. A single

optical pulse consists of many wavelength components but FBG
sensors reflect only the wavelength component that matches
their Bragg wavelength. Normally, the wavelength shift of the
sensor is determined by using a linear optical filter which
converts wavelength shift to optical power variation. TDM
systems utilize identical, low reflectivity FBG sensors (typical
4% reflection) all operating in the same wavelength window.
However, the signal-to-noise ratio of TDM systems is lower
than that of WDM systems and therefore the performance of
TDM systems is generally not as good as that of WDM systems.
The TDM approach is able to multiplex a large number of FBG
sensors without the need for wavelength selective components.
The wavelength measurement accuracies demonstrated by
WDM interrogation systems and TDM interrogation systems
are typically 1 pm and 10 pm, respectively. This is equivalent to
temperature and strain measurement accuracies of 0.1 C (1 C)
and 1 (10 ), respectively, for WDM (TDM) interrogation
units.
The FBG sensor technology has been undergoing rapid
development since its first demonstration by Morey in 1989
[7]. This is because the FBG sensors have many advantages
compared to other optical and electrical sensors. All the
aforementioned features make FBG sensors very suitable for
health monitoring for civil structures, such as bridges, tunnels,
and dams. FBG sensors can be employed in existing structures
by attaching the sensors onto their surfaces. New structures
can be monitored by embedding FBG sensors in them during
construction without affecting the structural integrity because
of the small size of optical fiber. The first field application of
FBG sensors for bridge health monitoring was demonstrated in
1995, by embedding the FBG sensors into the concrete girders
of a new road bridge in Calgary, Alberta, Canada [37].
3. High-speed demultiplexing/interrogation system for
FBG sensor arrays
In the past few years, the FBG sensor is demonstrated to
be very suitable for strain sensing. An important advantage
of a FBG sensor is that it can be easily surface-mounted
to or embedded in different structures. The FBG sensors are
lightweight, small and can even be embedded and integrated
in a composite structure. Furthermore, FBG sensors are
very suitable for simple wavelength multiplexing so that a
single detection unit can be used to interrogate many FBG
sensors in one fiber. A new demultiplexing/interrogation system
(DEMINSYS) for FBG sensor arrays has been developed by
TNO TPD [38]. This potentially low cost system combines a
high readout frequency (about 20 kHz) for all the FBG sensor
channels with absolute measurement, and has a detection limit
of less than 1 pm. The detection limit of the DEMINSYS for
dynamic signal is verified with a tunable laser (New Focus
6200). The wavelength of the tunable laser is modulated with
a frequency of 600 Hz and the amplitude is 0.6 pm (pp).
The spectrum of the DEMINSYS output signal showed that
the 600 Hz modulation can clearly be observed in the spectrum
and the amplitude of 0.6 pm (pp) is found to be about 25 dB

652

T.H.T. Chan et al. / Engineering Structures 28 (2006) 648659

Fig. 2. Schematic of a scanning filter FBG interrogation system based on WDM.

Fig. 3. Schematic of a FDM interrogation system.

Fig. 4. TDM interrogation system of multiple FBG sensors.

higher than the noise level. Therefore detection of a sub-pm

wavelength shift with a readout frequency of about 20 kHz is
feasible.

cause disturbances to the DEMINSYS. This can be verified by

using a reference FBG sensor.

The DEMINSYS is designed for Structural Health

Monitoring for aerospace applications, but is also suitable for
civil structures. Other specifications of the DEMINSYS are
shown in Table 1.

4. FBG sensor fabrication and package units

For this field experiment the maximum readout frequency

has been used for noise reduction by averaging to achieve subpm resolution. Since the strain gauge signal in the WASHMS
has a maximum detection frequency of about 20 Hz, the FBG
sensors can also provide information of high-frequency events.
During the measurements, the DEMINSYS is located in the
bridge close to the FBG sensors. Vibration of the bridge will

Standard telecommunication single mode fibers (CorningSMF28) were used to fabricate FBG sensors. In order not to
weaken the mechanical strength of the FBG, the outer coating
of the fiber was removed by soaking a short length in a
warm acid bath instead of using a mechanical stripper. After
the FBG inscription, this short un-coated FBG was annealed.
To facilitate the installation process while maintaining the
straightness of the FBG, the FBG was mounted on nitinol
(an acronym for Nickel Titanium Naval Ordnance Laboratory)
strips with thickness of 7.5 m (0.0029) and dimensions

T.H.T. Chan et al. / Engineering Structures 28 (2006) 648659

653

Table 1
Other specification of DEMINSYS
No. of input fibers
4

No. of FBG sensors

32 (8 per input fiber)

No. of fiber input

Resolution

FBG channel spacing

Wavelength range

Better than 1 pma (1 )

About 4 nm

15301565 nm

a Depending on sampling frequency.

Fig. 5. Packaged FBG strain gauges boxed in ABS enclosures with two 3 mm
fiber outlets (inset shows the FBG sensors epoxied on nitinol sheets in an oven
after baking at 80 C for 5 h).

of 6 110 mm, which were cleaned with high concentration

isopropanol to remove grease stains. Nitinol is a room
temperature super elastic metal which is corrosion resisting
and can withstand 8% elongation without deformation. The
FBGnitinol sheet combo was sandwiched and pressed together
using two Teflon sheets to minimize the thickness as well as
the evenness of epoxy between the FBG and the nitinol sheets.
Fig. 5 shows a packaged FBG strain gauge boxed in ABS
enclosures with two 3 mm fiber outlets. The inset of Fig. 5
shows the packaged FBG sensors epoxied on nitinol sheets in
an oven after baking at 80 C for 5 h.
In order to protect the packaged FBG sensor from moisture
and dust, weather proof ABS enclosures measuring 120
80 60 mm that complied with IP65 specifications were used.
These enclosures are similar to the WASHMS resistive strain
gauge enclosures in terms of functionality and appearance that
are in use by the Highways Department of Hong Kong SAR.
To attach the packaged FBG sensor to the ABS enclosure, a
rectangular opening was cut from the bottom of the enclosure
so that the sensor can be attached to the structure through this
opening. The sensor was then connected to more rigid 3 mm
single mode optical fiber cables and led out from the enclosure
through two stress relieving boots as shown in Figs. 5 and 6.
In the case where an extra FBG is needed for temperature
referencing inside the enclosure, this FBG was connected to
the strain sensing FBG before the 3 mm cable is connected.
This temperature referencing FBG must be free from stress to
avoid cross-sensitivity of strain and temperature [39]. Fig. 6
shows the inside of the enclosure. It can be seen that the FBG
is protected with a nitinol backing sheet which is attached to
the FBG with high strength epoxy. The FBG sensor and the
nitinol sheet are attached to the steel using a two-part epoxy

Fig. 6. Inside of the FBG sensor enclosure.

which is suitable for the application. The FBG is led out from
the enclosure by fusion splicing the FBG to two 3 mm optical
fiber cables that are much stronger than the 250 m bare fiber. In
order to maintain consistent glue thickness and pressure among
the sensors, a pressure plate which is supported with a spring at
the back is installed at the inner side of the enclosure lid. This
pressure plate exerts pressure evenly on the FBG sensor when
the lid is fixed with screws.
5. Installation and experimental setup in the Tsing Ma
bridge
5.1. WASHMS of the Tsing Ma bridge
The Tsing Ma bridge (TMB) is the longest suspension bridge
carrying both highway and railway traffic. The bridge has a
double deck configuration with the expressway on the upper
deck and the railway below. The upper deck carries a dual threelane carriageway and the sheltered lower deck contains two
tracks of railway and two single emergency roadways that allow
limited traffic to use the crossing during periods of very strong
wind. Structurally, the deck section of the Tsing Ma bridge is a
hybrid arrangement combining both truss and box forms [40].
Two longitudinal trusses to the full depth of the deck at 26
m centers act in conjunction with the steel orthotropic decks
of the upper and lower carriageways to provide the vertical
bending stiffness. Plan diagonal bracings at the upper and lower
levels enable the trusses to provide lateral bending stiffness.
Cross frames of Vierendeel form are provided at 4.5 m centers
with every fourth frame being supported from suspenders. The
bridge was commissioned on May 22, 1997. The monitoring
system, WASHMS, has been operating since then for the TMB.
The structural health monitoring system for the TMB comprises
sensors such as accelerometers, strain gauges, displacement

654

T.H.T. Chan et al. / Engineering Structures 28 (2006) 648659

Fig. 7. Strain gauge layout of the WASHMS installed on the Tsing Ma bridge.

transducers, level sensors, anemometers, temperature sensors

and weigh-in-motion sensors, installed permanently on the
bridges, and the data acquisition and processing system. 110
strain gauges were installed on bridge-deck sections to measure
the strain of members as shown in Fig. 7.
5.2. Installation and experimental setup
In this test, three different strategic locations: (1) hanger
cables, (2) rocker bearing, and (3) supporting structure on a
section (Chainage 23488) of lower deck respectively as shown
in Fig. 8, were chosen to install FBG sensors. Three photos are
taken after the FBG sensors are installed on these locations as
shown in Fig. 9, in which some FBG sensors were installed
nearby the resistive strain gauges in use such that comparisons
between conventional sensors and FBG sensors can be made. In
addition, strain-free FBG sensors were also used to measure the
temperature at different sensing points to provide temperature
compensation for the measurements.
The FBG DEMINSYS unit developed is a complete
system with the capability to perform multiple optical FBG
sensor measurements and provide a major advancement for
mechanical and temperature sensing applications. The unit
provides a rapid measurement rate of hundreds of FBG sensors
on several fibers. It includes an optical light source used to
illuminate the FBG sensors, and an optical detector to measure
the reflected optical signals on each sensor. An external portable
PC provides the on-line calibration data display/storage of the
FBG sensors under test. The system resolution and accuracy
are 1 pm and 10 pm respectively. The sampling rate of the
system is adjustable and can be increased up to 20 kHz. Fig. 10
briefly shows the experimental setup of the interrogation system
used to measure the wavelength of FBG sensors. The system
included a broadband light source operating around 1550 nm
and a wavelength detection module based on a sensitive chargecoupled device (CCD) array. Three strands of fibers, in which
up to 21 FBG sensors were serially connected, were installed
on the hanger cables, rocker bearing, supporting structure on a
section (Chainage 23488) of lower deck and suspender cables
respectively as shown in Figs. 8 and 9, respectively. To attach
the FBG sensors firmly on the structure, a small area of this

protective paint was removed. Attention was paid during the

removal of paint to avoid damaging the steel beneath. Before
attaching the packaged FBG sensor, that area was cleaned with
clear water and isopropanol. Two types of adhesive glues were
used to attach the enclosure and FBG sensors on the steel. Fast
curing cyanoacrylate glue was applied on the outer rim of the
enclosure base and, on the inner rim of the enclosure and FBG
sensors, slow curing two-part weather proof epoxy was used.
A combination of the two glues at different positions provides
fast attachment of the sensor package in place while allowing
the much stronger epoxy bonding to cure slowly. The lid of
the enclosure is fixed with four captive screws and the joint is
finished with tongue and groove with neoprene gasket.
6. Measurement and observation based on FBG sensor
units
FBG sensors for strain measurement are excellent candidates
for structural health monitoring of the Tsing Ma bridge. A
field trial with the DEMINSYS system described in Section 3,
a FBG sensor array system, was carried out on Hong Kongs
landmark Tsing Ma bridge in 2003. The experiment was
carried out with a FBG interrogator based on a scanning
optical bandpass filter that provides a sampling speed of up to
70 sample/s as well as with a high-speed interrogation system
(DC to 20 kHz for all channels simultaneously) based on
CCD. Forty FBG sensors divided into 3 arrays were installed
on different parts of the bridge (hanger cable, rocker and truss
girders), as shown in Fig. 8. The goal of this field trial is to
monitor the strain of the different parts of the bridge under
railway load and highway load. Various measurements were
performed including an overnight measurement of about 20 h
with a sampling frequency of about 500 Hz. The measurement
results will reveal the dynamic responses of the Tsing Ma
bridge, which will be measured by the FBG sensors during train
passage. The results of the FBG sensor were also compared
with existing conventional strain gauges.
6.1. Temperature and strain measurement
The interrogation system was switched on continuously
for 24 h to monitor the structure and temperature change

T.H.T. Chan et al. / Engineering Structures 28 (2006) 648659

655

Fig. 8. Forty FBG sensors installed on the Tsing Ma bridge to measure temperature and strain at (1) hanger cable, (2) rocker bearing, and (3) truss girders of section
Chainage 23488.

Fig. 9. FBG sensors installed on (1) hanger cable, (2) rocker bearing, and (3) truss girders of section Chainage 23488.

of the bridge. Fig. 11 shows the wavelength variations of a

single sensor that incorporated both temperature referencing
and strain FBG sensors over a 10 min period. Fig. 11(a) and
(b), respectively, show the wavelength change of the strainfree temperature measuring FBG and the strain measuring
FBG which was affected by both thermal and strain effects.
Considering that the FBG is mounted on steel material which is
subjected to both thermal and mechanical effects [39], Eq. (5)

could be modified to express the wavelength change of the FBG

as:
 B
= (1 ) + (1 )(T ) + T
B

(6)

where B is the Bragg wavelength, is the strain, other

parameters have the same meanings as before. Therefore, the
actual strain response of the sensing point can be resolved by
offsetting the thermal effect using the temperature referencing

656

T.H.T. Chan et al. / Engineering Structures 28 (2006) 648659

Fig. 10. Experimental setup of FBG interrogation system.

FBG. Fig. 11(c) gives a comparison of the strain responses on

the sensing point measured by both the FBG sensor and the
conventional strain gauge over 10 min. At about 320, 420 and
560 s, three trains passed through the bridge. Meanwhile, three
overshoots were clearly shown by the interrogation system
and conventional strain measurements. The experimental data
show that the FBG results are in good agreement with the data
obtained by the installed WASHMS resistive strain gauges.
6.2. Hanger cable measurement
Six FBG sensors are mounted on the hanger cables. One
of the FBG sensors mounted on a hanger cable to the Ma
Wan tower was monitored with the high-speed DEMINSYS
interrogation system. Via another input fiber, an athermal
packaged FBG sensor in a vibration-isolated case is used
as reference. For the hanger cable measurement, the readout
frequency is set to 0.106 ms. The measured histories from the
two FBG sensors are shown in Fig. 12, where an arbitrary offset of 40 micro strain is applied to the reference FBG history
for display purposes: the upper time history is from the FBG
sensor on the hanger cable, the lower one from the reference
FBG sensor.
It shows that train passages at t = 307 s and t = 377 s can
clearly be detected by the FBG mounted on the hanger cable.
From the signals, train running directions can be deduced out
as in an opposite direction. In the middle of the measurement
history, the passage of heavy traffic can also be noticed. It
corresponds to the signal at t = 195 s.
6.3. Rocker bearing measurement
Various measurements with the FBG sensors mounted on the
rocker bearings (as shown in Figs. 8 and 9) were performed
including an overnight measurement of about 20 h with a
sample time of 2.1 ms; the data from strain gauges on the rocker
were also logged simultaneously. The corresponding history
results of a one-hour track are compared in Fig. 13, where the
upper time history is from one conventional strain gauge, and
the lower from one FBG sensor respectively.
In the Tsing Ma bridge construction, the rocker bearings are
used to support and hold the deck. Therefore, the loading on the

Fig. 11. Wavelength variations and comparison of strain responses measured

by FBG sensor (dot) and WASHMS strain gauge (line) respectively.

rocker is very complex and the strain depends strongly on the

position of the sensor. Despite the FBG sensors and the strain
gauges are not mounted on exactly the same location of the
rocker the signals of the 2 types of sensors, as shown in Fig. 13,
are found to be very similar and the train passages can clearly
be detected.

T.H.T. Chan et al. / Engineering Structures 28 (2006) 648659

657

Fig. 12. Hanger cable tension measurement history.

Fig. 13. Comparison between FBG (lower) and conventional (upper) strain gauges installed on rocker bearings.

6.4. Truss girder measurement

average filter of 5000 points is shown in Fig. 15. Using such

filtering, sub-pm/ resolution can be achieved.

An array of 21 FBG sensors was mounted on different

locations of Chainage 23488 (as shown in Figs. 8 and 9). The
FBG sensors are placed close to the existing strain gauges for an
optimal comparison. In Fig. 14, the upper history from a FBG
sensor is compared with the signal of the lower corresponding
strain gauge. The sampling time of the FBG sensor is 0.0528
ms. A moving average filter of 10 points is applied to the data of
the FBG sensor and the detection bandwidth of the FBG sensor
is reduced to about 2 kHz. The FBG sensor signal has also an
arbitrary offset for display purposes. Although the sensors are
not located at exactly the same location, a great resemblance
has been found.
Zooming in the signal reveals that the noise in the FBG
sensor signal is mainly caused by a 13 Hz component and can
be reduced by appropriate filtering or placing the DEMINSYS
outside the bridge. The resolution is found to be about 1 pm and
can be reduced by further averaging. The effect of a moving

7. Conclusions
The background of fiber Bragg grating sensor technology was briefly introduced. The high-speed demultiplexing/interrogation system for FBG sensor arrays and the FBG
sensor package units were developed. A field trial with FBG
sensor arrays for measurement on a hanger cable, rocker bearings and truss girder of the Tsing Ma bridge was performed
successfully. The application of FBG sensors and interrogation
system to monitor the dynamic strain on Hong Kongs landmark
Tsing Ma bridge has been demonstrated. The FBG package
technique was proposed to apply for structural health monitoring applications. It can clearly and correctly detect the dynamic
strain responses of the bridge induced by the passage of trains
on the bridge. The measurement result of the interrogation system was in excellent agreement with those obtained by resistive strain gauge measurements. The FBG sensor system offers

658

T.H.T. Chan et al. / Engineering Structures 28 (2006) 648659

Fig. 14. Comparison between FBG (upper) and conventional strain gauge (lower) installed on a truss girder of section Chainage 23488 of the Tsing Ma bridge.

Fig. 15. Comparison between FBG (upper) and conventional strain gauge (lower) histories after filtering data in Fig. 14.

many advantages over traditional resistive strain gauges. These

include remote sensing, ease of installation, non-corrosive and
lower maintenance cost. This shows that FBG sensor technology is a good alternative for civil and structural dynamic strain
monitoring.
Acknowledgements
The support provided by the HK Highways Department,
the HK Research Grants Council, the Hong Kong Polytechnic
University Research Grants (with nos. G-YD20 and G-YX25)
and the National Natural Science Foundation of China (with
grant no. 50378009) is gratefully acknowledged.
References
[1] Hill KO, Fuji Y, Johnson DC, Kawasaki BS. Photosensitivity in optical
fiber waveguides: application to reflection fiber fabrication. Appl Phys
Lett 1978;32(10):6479.

[2] Meltz G, Morey W, Glenn W. Formation of Bragg grating in optical fiber

by the transverse holographic method. Opt Lett 1989;14(15):8235.
[3] Coll P. Holographically written Bragg gratings in photosensitive optical
fiber. Research project report. University of Sydney/OFTC; 1993.
[4] Morey WW, Ball G, Meltz G. Photo-induced Bragg gratings in optical
fibers. Optics and Photonics News. February, 1994. p. 814.
[5] Kersey AD, Morey WW. Multiplexed Bragg grating fiber-laser strainsensor system with mode-locked interrogation. Electron Lett 1993;1:
1124.
[6] Hill KO, Malo B, Bilodeau F, Johnson DC, Albert J. Bragg gratings
fabricated in monomode photosensitive optical fiber by exposure through
a phase mask. Appl Phys Lett 1993;62:10357.
[7] Morey WW, Meltz G, Glenn WH. Fiber optic Bragg grating sensors. SPIE
1989;1169:98107.
[8] Melle S, Liu K, Measures RM. Strain sensing using a fiber optic Bragg
grating. SPIE 1991;1588:25563.
[9] Huang S, LeBlanc M, Ohn MM, Measure RM. Bragg integrating
structural sensing. Appl Optim 1995;34:50039.
[10] Friebele EJ, Askins CG, Bosse AB, Kersey AD, Patrick HJ, Pogue WR
et al. Optical fiber sensors for spacecraft applications. Smart Mater Struct
1999;8:81338.

T.H.T. Chan et al. / Engineering Structures 28 (2006) 648659

[11] Foote PD. Fiber Bragg grating strain sensors for aerospace smart
structures. In: Proc to 2nd European conf on smart structures and
materials. 1994. p. 2903.
[12] Foote P, Roberts D. Carbon spars for superyachts and smart mast
technology. In: RINA Proc of the conf on the modern yacht. Paper no.
13. 1999.
[13] Fisher NE, Surowiec J, Webb DJ, Jackson DA, Gavrilov LR, Hand JW
et al. In-fiber Bragg gratings for ultrasonic medical applications. Meas
Sci Technol 1997;8:10504.
[14] Davis MA, Kersey AD. All-fiber Bragg grating strain-sensor demodulation technique using a wavelength division coupler. Electron Lett 1994;
30(1):757.
[15] Rao YJ. In-fiber Bragg grating sensors. Meas Sci Technol 1997;8:35575.
[16] Ferdinand P, Magne S, Dewynter-Marty V, Martinez C, Rougeault S,
Bugaud M. Applications of Bragg grating sensors in Europe. Optical
society of America technical digest series, vol. 16. 1997. p. 149.
[17] Rao YJ. Recent progress in applications of in-fiber Bragg grating sensors.
Opt Lasers Eng 1999;31:297324.
[18] Kuang KSC, Cantwell WJ. Use of conventional optical fibers and fiber
Bragg gratings for damage detection in advanced composite structures: a
review. Appl Mech Rev 2003;56(5):493513.
[19] Schulz WL, Udd E, Seim JM, McGill GE. Advanced fiber grating
strain sensor systems for bridges, structures and highways. In: Liu SC,
editor. Smart structures and materials 1998: Smart systems for bridges,
structures, and highways, SPIE 1998. vol. 3325. p. 21221.
[20] Sun BN, Ni YQ, Ko JM. Optical fiber sensor applications in civil
engineering. Research report. Hong Kong: The Hong Kong Polytechnic
University; 1999. p. 25.
[21] Bugaud M, Ferdinand P, Rougeault S, Dewynter-Marty V, Parniex P,
Lucas D. Health-monitoring of composite plastic waterworks lock gates
using in-fiber Bragg grating sensors. Smart Mater Struct 2000;9:3227.
[22] Vries MD, Bhatia V, DAlberto T, Arya V, Clause RO. Photo-induced
grating-based optical fiber sensors for structural analysis and control. Eng
Struct 1998;20(3):20510.
[23] Li HN, Li DS, Song GB. Recent applications of fiber optic sensors to
health monitoring in civil engineering. Eng Struct 2004;26:164757.
[24] Du WC, Tao XM, Tam HY. Fiber Bragg grating cavity sensor for
simultaneous measurement of strain and temperature. IEEE Photon
Technol Lett 1999;11(1):1057.
[25] Du WC, Tao XM, Tam HY. Temperature independent strain measurement
with a fiber grating tapered cavity sensor. IEEE Photon Technol Lett 1999;
11(5):5968.
[26] Guan BO, Tam HY, Tao XM, Dong XY. Simultaneous strain &

[27]

[28]

[29]

[30]

[31]

[32]
[33]
[34]

[35]
[36]
[37]

[38]

[39]
[40]

659

temperature measurement using a superstructure fiber Bragg grating.

IEEE Photon Technol Lett 2000;12(6):6757.
Guan BO, Tam HY, Ho SL, Michael Liu SY, Dong XY. Simultaneous
strain and temperature measurement using a single fiber Bragg grating.
Electron Lett 2000;36(12):10189.
Yu YL, Tam HY, Chung WH, Demokan MS. Fiber Bragg grating sensor
for simultaneous measurement of displacement and temperature. Opt Lett
2000;25(16):11413.
Guan BO, Tam HY, Chan HLW, Choy CL, Demokan MS. Discrimination
between strain and temperature with a single fiber Bragg grating. Microw
Opt Technol Lett 2002;33(35):2002.
Wong KY, Lau CK. Planning and implementation of the structural
health monitoring system for cable-supported bridges in Hong Kong. In:
Aktan AE, Gosselin SR, editors. Nondestructive evaluation of highways,
utilities, and pipelines IV, SPIE 2000; 3995: 26675.
Chan THT, Yu L, Tam HY, Ni YQ. Fiber Bragg Grating sensors for
structural health monitoring of Tsing Ma bridge: Measurement and
discussion. 2006 [submitted for publication].
Berkoff TA, Davis MA, Kersey AD. Fiber optic sensors for distributed
vibration monitoring. SPIE Vib Monit Control 1994;2264:14854.
Sipe JE, Poladian L, Martijn de Sterke C. Propagation through
nonuniform grating structures. JOSA-A 1994;11(4):1307.
Rao YJ, Ribeiro ABL, Jackson DA, Zhang L, Bennion I. Combined
spatial- and time-division-multiplexing scheme for fiber grating sensors
with drift-compensated phase-sensitive detection. Opt Lett 1995;20:
214951.
Kersey AD. In: Udd E, editor. Fiber-optic smart structures. New York:
Wiley; 1995. p. 40944.
Morey WW. Distributed fiber grating sensors. In: Proceedings of the 7th
international conference on optical fiber sensors. IREE; 1990. p. 2858.
Measures RM, Alavie AT, Maaskant R, Ohn M, Karr S, Huang S.
A structurally integrated Bragg grating laser sensing system for a carbon
fiber prestressed concrete highway bridge. Smart Mater Struct 1995;4(1):
2030.
Cheng LK. et al. Multi-channel FBG sensor system for static and dynamic
measurement up to 10 kHz. In: First European workshop on structural
health monitoring. p. 5607.
Kersey AD, Davis MA, Patrick HJ, LeBlanc M, Koo KP, Askins CG et al.
Fiber grating sensors. J Lightwave Technol 1997;15:144263.
Lau CK, Wong KY. Design, construction and monitoring of the three key
cable-supported bridges in Hong Kong. In: Lee PKK, editor. Structures
in the new millennium. Rotterdam (Netherlands): A.A. Balkema; 1997.
p. 10515.