Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

Middleton Hoult

Download as pdf or txt
Download as pdf or txt
You are on page 1of 19

Wireless Structural Health Monitoring of Bridges: Present and

Future

Neil A. Hoult1 and Campbell R. Middleton2

Abstract

As bridge infrastructure around the world ages the effects of material deterioration cause
reductions in their load carrying capacity while at the same time increasing live loads
mean that they are often exposed to stresses for which they were not designed. Yet there
are limited resources available to ensure that these bridges are fit for purpose, meaning
that new approaches to bridge maintenance are required that optimize both their service
life as well as the amount spent on maintenance. A potential tool that could support such
a maintenance plan is the use of wireless sensor networks (WSNs). WSNs have
advantages over conventional wired monitoring systems in terms of installation time and
potentially cost. In order to gauge the potential of these systems two WSNs were installed
on the Humber Bridge network. A relative humidity and temperature monitoring system
was installed in the anchorage of the main suspension, where the main cables of the
bridge are attached to the foundation, to ensure that a maximum relative humidity
threshold is not crossed. A second WSN which monitors aspects of deterioration on a
reinforced concrete approach bridge was also installed. Though both systems have
provided useful data to the owners there are still challenges that must be overcome in
terms of monitoring corrosion of steel, measuring live loading and data management
before WSNs can become a truly useful tool for bridge managers.

CE Database Subject Headings: Bridges, Bridge inspection, Bridge maintenance,


Monitoring, Corrosion, Weighing devices

1
Research Associate, Dept. of Engineering, Univ. of Cambridge, Trumpington St., Cambridge, CB2 1PZ,
UK. E-mail: nah25@cam.ac.uk
2
University Senior Lecturer, Dept. of Engineering, Univ. of Cambridge, Trumpington St., Cambridge, CB2
1PZ, UK. E-mail: crm11@cam.ac.uk

1
Introduction

Bridge infrastructure around the world is increasingly required to support higher loading
both in terms of frequency and magnitude. At the same time this infrastructure is ageing
resulting in increased levels of deterioration and maintenance. Unfortunately these
increased requirements are juxtaposed against limited government budgets to support the
maintenance and renewal of bridge infrastructure. As such new bridge maintenance
strategies are required so that bridge service lives and maintenance programs can be
optimized. In order to develop these optimized strategies, engineers require critical data
about bridge performance, delivered in real-time if required. One such way to obtain this
data is through the use of a monitoring system.

Traditional structural monitoring systems have consisted of sensor nodes that are wired
into a central data acquisition unit. These systems have the advantage that by using wires
large quantities of data can be transmitted reliably and sensors that require significant
amounts of power can be supported. However the use of wires also means that
installation times and overall system costs can be prohibitively high for maintenance
programs with limited budgets. Recent technological developments have made the use of
Wireless Sensor Networks (WSNs) a more appealing option. These networks eliminate
the cost associated with the installation of cabling, allow for sensors to be placed in
difficult to access locations and can be easily expanded to accommodate changing data
requirements.. However the drawbacks of these networks are three-fold. First, wireless
data transmission has less bandwidth and is not as robust when compared to wired data
transmission and so concessions must be made in order to ensure the integrity of the data.
Second, if the network is to be truly wireless the sensors must also be battery powered
which places further constraints on the monitoring and transmission rates. Finally,
although a number of sensors have been developed for use in WSNs, these sensors do not
necessarily provide the data that engineers and managers require to make informed
decisions about bridge infrastructure management.

2
In light of this background this paper will introduce two WSNs that have been installed
on the Humber Bridge system in the UK, outlining what data was required and how it
was monitored, what some of the issues were with these networks and also how these
networks have performed since installation. Although these networks have been
successful in terms of providing useful information to the operators, they are still limited
in terms of the data they can provide. As such this paper will also briefly explore some of
the developments that are required in order for WSNs, and structural monitoring in
general, to become a more powerful and useful tool for cost-effective bridge management
in the future.

Wireless Sensor Networks at the Humber Bridge

The Humber Bridge, as seen in Fig. 1, is the fifth longest suspension bridge in the world
with a central span of 1.41 km. The main suspension bridge is accessed via a series of
smaller reinforced concrete bridges, which while not as visually impressive as the main
bridge are just as critical as they provide the only access points to the main bridge. As
part of a larger project investigating the use of WSNs to monitor Civil Infrastructure, two
WSNs have been installed at the Humber Bridge. In each case the objective of installing
the network was to provide the Bridge Master and his staff with long-term data that
would allow them to better understand critical performance areas of their structures. The
first WSN was installed in the north anchorage of the suspension bridge to provide
environmental data on demand while the second system was installed on a near-by access
bridge to monitor changes in structural and environmental conditions.

Anchorage Environmental Monitoring System

The main suspension cables of the Humber Bridge are 680 mm in diameter and each one
consists of 14,948 5 mm diameter high strength steel wires. The forces in the cables are
resisted by anchoring the cables into massive concrete foundations, which is done in four
large chambers with one chamber located at each end of the two main cables. In order to
connect the cables to the foundation, they have to be split into individual wire bundles,

3
called strands, which are then wound around steel anchors as illustrated in Fig. 2. This
connection detail means that the protective wrap that is used when the strands are
bundled together to form the main cable cannot be used within the anchorage chamber
and the individual steel wires are exposed to the surrounding air. Research (Nakamura
and Suzumura 2005) has shown that if steel wires are subjected to relative humidity (RH)
levels above 60 % their susceptibility to corrosion increases significantly. As such in
these chambers where the steel wires are exposed to the surrounding air there is the
potential for them to corrode if the RH level is not properly controlled. To ensure that the
critical RH threshold of 60 % is not crossed, a dehumidification system has been installed
in each anchorage. This dehumidification system is set to turn on automatically if the RH
in any of the chambers exceeds approximately 45 % (a conservative threshold has been
chosen to help ensure that corrosion does not occur). A wired RH monitoring system has
been installed in each of the chambers, but the data from this system can only be accessed
at wall-mounted display units within the anchorage chambers themselves. Thus to better
evaluate the performance of the dehumidification system a web-based monitoring system
was sought to provide access to this data from anywhere in the world.

In order to provide the bridge managers with the required access to the data in this critical
area of the bridge, a WSN was installed in the north-west and north-east anchorage
chambers. The network consists of a gateway, 10 nodes with RH and temperature
sensors, one inclination sensor node and one data relay node. The nodes are Crossbow
MICAz motes. Fig. 3 illustrates the layout of the nodes in the anchorage chambers. Three
of the wireless sensors have been placed next to the existing wired sensors for data
verification purposes. The gateway then transmits the data using an ADSL Internet
connection back to the central server at the University of Cambridge.

When the network was first installed in July 2007, it was found that the nodes in the west
anchorage chamber were unable to communicate with the nodes in the east anchorage
chamber or the gateway. This problem was the result of a loss of transmission strength
between the west and east chambers caused by the 15 m long corridor that connected the
two chambers. This problem was overcome by using 5 dB gain external antennas in place

4
of the original MICAz antennas on the gateway and on a special relay node (3) that was
added to the network and installed at the other end of the corridor to the gateway. This
problem is an excellent illustration of one of the main issues for WSN installation, which
is network connectivity. Until the network is installed it is virtually impossible to predict
where areas of fading (significant drops in radio transmission strength) will occur. In
many civil engineering applications, for example train tunnels or highway bridges, site
possession times are very short so there may be very limited time available to setup and
test a WSN. Thus the development of tools which enable wireless signal strength to be
predicted for any given network configuration is a particularly important topic that
requires further research if WSNs are to be adopted more widely by infrastructure
managers.

Overall the RH and temperature monitoring system in the anchorage has performed well
since it was installed 10 months ago. In fact, as illustrated in Fig. 4, the data has provided
the Bridge Master with useful feedback about the operation of his dehumidification units
and identified the potential for significantly reducing operating costs of the
dehumidifying system. One can see from the graph in Fig. 4 that the RH for nodes 2 and
8 located in the north-east chamber varies over a much smaller range than the RH for
nodes 1 and 4, located in the north-west chamber. This indicates that the dehumidification
unit in the east chamber turns on and off far more frequently than the unit in the west
chamber. In fact the unit in the west chamber is potentially more efficient as its total on
time during a day is approximately 105 minutes versus 300 minutes for the unit in the
east chamber, a difference of a factor of three. Until this monitoring system was installed,
the bridge manager was unaware of this difference in the operation of the two units and
the potential to reduce the energy used by changing the settings on the dehumidification
unit in the east chamber to match the unit in the west chamber. The power consumption
of the dehumidifying units in each chamber is 7.82 kW meaning that the resulting power
savings in using the west chamber dehumidification strategy in both chambers could be
9.3 MWh per year. This has the potential to reduce the amount of carbon dioxide
produced in powering the system by 4 tonnes per annum based on the assumption that

5
0.43 kg of CO2 is generated for every kWh of electricity used (National Energy
Foundation 2008).

The first sensor (node 10) ran out of battery power in mid-April 2008 approximately 10
months after the sensors were first deployed but the rest of the nodes continue to be fully
operational. There have been three hardware failures to this point. Both nodes 5 and 10
stopped transmitting sensor data. In the case of node 10 the sensor board was replaced
and the problem has not reoccurred. In the case of node 5 both the mote and the sensor
board have been replaced as the problem has occurred more than once. A problem also
occurred with the gateway whereby multiple data points from the same sensor were
recorded as having arrived at nearly identical times. Since the sensors are programmed to
send data back every three to five minutes, having data recorded from the same sensor at
intervals of less than one second indicated that there was a potential problem with the
gateway. Initially this was believed to be a hardware issue to do with the clock on the
gateway malfunctioning, however it was eventually discovered that it was a result of
having a database running on the gateway. The gateway was both writing to and
extracting values from a large database resulting in an overall slowdown in processor
performance. This in turn meant that the clock slowed down and the data values, which
are time stamped based on this clock, received incorrect time stamps. This problem was
overcome by removing the database from the gateway and instead writing the data to text
files, which requires far less processing power. However, all three component failures
illustrate the need for redundancy if the data being monitored is safety critical. Otherwise
there is the potential for data to be lost due to software and hardware failures, although
this problem is not unique to wireless systems.

Reinforced Concrete Bridge Monitoring System

A second WSN has been installed on the Ferriby Road Bridge, a reinforced concrete
approach bridge to the north of the main suspension bridge as illustrated in Fig. 5. This
network is designed to supplement the regular inspection program that this bridge
undergoes. During the bridges last visual inspection in the summer of 2002 it was noted

6
that cracks had formed on the soffit of the bridge and that the elastomeric bearings were
inclined transverse to the span of the bridge. Whilst neither of these issues required
immediate attention at the time of inspection both have the potential to become severe
issues over the course of time. However one of the problems with visual inspections is
consistency. For example, although a crack may be noted from inspection to inspection,
one inspectors interpretation of its width and extent may vary from anothers. As such
monitoring has two potential benefits in this area as it would allow for the variation in
deterioration to be tracked over time and provide a more objective way of tracking this
deterioration. However, in order to install a wired system to do this, it would require the
installation of both the sensors and the cabling on the soffit of the bridge, which could
result in extensive road closures and delay costs on top of the installation costs.

Instead a WSN was installed on this bridge that consisted of six sensor nodes in the first
instance. Three of these nodes measure inclination of the elastomeric bearings as well as
RH and temperature both inside and outside of the box housing the sensor. The other
three sensors measure the change in crack widths on the soffit of the bridge as well as RH
and temperature. The WSN was installed over the course of two days in March and April
2008. It is anticipated that with more experience installing the sensors this time could be
reduced significantly.

This network is also innovative in that it provides fully wireless long-term monitoring.
The sensor nodes are powered using long-life lithium batteries which should allow the
inclinometer and displacement sensors to operate for 1.5 years and 8 years respectively
between battery replacements. The gateway is also battery powered and transmits data by
wireless, unlike the gateway in the anchorage network which relies on mains power and
transmits data over an ADSL Internet connection. The gateway at the Ferriby Road
Bridge is powered from a 12 V battery which is recharged by a solar panel mounted on
the side of the bridge as indicated in Fig. 5. A mobile phone connection allows the data to
be downloaded to the central server, located at the University of Cambridge, wirelessly as
well. These two features mean that bridges in virtually any location in the UK can be
remotely monitored.

7
As with the anchorage network, there were initial radio transmission problems that had to
be overcome before this WSN was fully operational. Fig. 6 shows the general layout of
the nodes in this work with the distance from the gateway to node 1 being approximately
40 m. After the initial deployment data was only consistently received from nodes 1, 4
and 7 with data being received intermittently from node 3. Nodes 5 and 6 were
inclinometer sensors that were attached to the elastomeric bearing above the column,
which meant that they were in a recess between the top of the column and the underside
of the bridge slab as illustrated in Fig. 7. It was felt that this recess was limiting the
transmission range of these sensors. In order to overcome this problem external antennas
were attached to nodes 5 and 6 that extended beyond this recess as demonstrated in Fig.
7. Node 2 was an LPDT node mounted on the underside of the slab approximately 6 m
away from the gateway. In this case it was preferable not to have to use an external so
that the sensor was as inconspicuous as possible in order to minimize the possibility of
vandalism. Instead an external 5 dB gain antenna was mounted on the gateway. As a
result of these changes data could be received from nodes 2, 5, and 6 as well as nodes 1, 4
and 7. However data was no longer being received by node 3. This was felt to be as a
result of changing the location of the gateway antenna, which caused node 3 to fall into a
region of fading. The solution to this issue has not yet been implemented but will be to
install an additional relay node as illustrated in Fig. 6 although due to the unpredictable
nature of radio transmission it is not possible to guarantee the success of this solution.
Both of these transmission issues illustrate one of the key obstacles to be overcome in
order for WSNs to gain widespread acceptance, and that is the ability to ensure network
connectivity. In this instance, as with the anchorage, network connectivity was developed
through a trial and error approach however on structures where possession times are
limited this approach would be less than ideal. The answer to this problem is not clear as
modelling can only suggest at the likelihood of fading and not indicate the exact locations
of it. Similarly onsite testing to locate areas of fading is influenced by the presence of
personnel and equipment meaning that a system works while the testing is taking place
may not work during everyday operation. Overcoming these transmission issues by
installing more relay nodes has cost implications in terms of equipment, delay and

8
personnel costs and so would need to be optimized in order to make the use of WSNs a
cost-effective option. This is an area that requires more investigation.

Future Sensing Requirements

Although these initial WSN deployments have shown promise, there remain several key
issues that need to be addressed. One such issue is that the sensor technology does not
currently exist to accurately measure corrosion in reinforced concrete structures or
suspension bridge cables, either using wired or wireless systems. Another concern in
terms of developing effective bridge monitoring systems is the need to be able to measure
live load on the structure. One must also give careful consideration to what data is
required in order to properly maintain the bridge inventory whilst avoiding the pitfall of
monitoring for the sake of monitoring. The following section outlines the significance of
each of these issues and what some of the key challenges to be overcome are in order for
them to be integrated into the monitoring systems of the future.

Corrosion Sensors

Deterioration caused by corrosion affects the capacity of all structures that have steel
elements by reducing the area of steel available to resist the applied loads. In order to
estimate the reduced structural capacity and remaining service life of the structure
engineers need to know the extent (the remaining area of steel), location and the rate of
corrosion. However these values are not easily estimated even if the corrosion is visible
and become especially difficult to quantify for structures where the load carrying steel
members are embedded, such as in reinforced concrete, or wrapped such as in suspension
bridges.

Researchers have proposed a wide variety of techniques for measuring corrosion in


reinforced concrete structures including half-cell, chloride ingress, eddy current, linear
polarization, acoustic emission and thermographic measurements. Unfortunately each of

9
these tests has limitations which call into question their ability to predict the three key
parameters of extent, location and rate. For example, half-cell and chloride ingress
measurements provide an indication of the likelihood of corrosion but do not actually
quantify the extent of corrosion and only provide information about a specific location on
the structure. Proponents claim that linear polarization provides some indication of the
rate of corrosion but again only at a specific location on the structure. The measured rate
from a linear polarization test is highly dependant on a number of factors including the
moisture and chloride content of the concrete as well as the temperature, which makes the
accuracy of the test questionable unless precise data exists. Acoustic emission and
thermographic monitoring offer the possibility of being able to monitor the entire
structure but can only indicate the location of corrosion meaning that other techniques are
required to then determine the extent and rate of corrosion.

In the case of suspension bridge cables, far fewer techniques are available to estimate the
key parameters engineers require. Currently in order to estimate the extent of corrosion
the protective wrapping must be removed from the cables, the strands are then wedge
apart before the individual wires are inspected visually. However, only a small
percentage of the wires can be inspected using this technique meaning that critical areas
of corrosion may not be uncovered. The rate of corrosion is currently monitored using
acoustic emission techniques where systems are installed to listen for the characteristic
noise of a steel wire breaking. While this approach may be appropriate for use with
suspension cables, there is a lack of research and testing that has been done to validate it.

As indicated by the preceding discussion there is a need for the structural engineering and
sensor development communities to reconsider how the extent, location and rate of
corrosion can be determined. Can MEMS sensors be developed that detect corrosion
products within concrete for example? And if so, can these sensors then be included in a
WSN that also allows other critical parameters to be measured thus providing the bridge
manager with a complete picture of the bridges performance? Or is the required tool a
scanning device of some kind that can be passed over the structure during an inspection
providing a complete picture of the condition of the steel elements? Recent work using

10
magnetic flux leakage scanning devices has shown promise for detecting flaws in
prestressed concrete structures (Mietz and Fischer 2007) and cable stayed structures
(Christen et al. 2003). However further research will be required to see if this technique
can be used to detect the level of corrosion as well.

Load Sensors

The two WSN systems presented in this paper track long-term trends relating to
environmental conditions and structural deterioration. However, there are also instances
when the aspect to be monitored is directly affected by the loading on the structure at the
time the sensor readings are taken. An example of this would be the force in the cables of
a cable-stayed bridge. In order for the measured force to be useful in an assessment of the
bridge, the engineer needs to know what the live loading (due to traffic, trains etc.) was
on the bridge at the time the forces were measured. Otherwise the engineer does not
know how the measured force relates to the capacity of the bridge. Thus there is a need to
develop accurate methods of determining the live loading on the bridge that can then be
correlated with the data from other sensors to allow the performance of the bridge to be
determined accurately. However, there are two challenges that must be overcome in order
to do this: first an accurate method of measuring the live load must be found and then a
method of integrating this technology into a WSN must be developed.

There exist several techniques to determine the weight of vehicles in motion using sensor
technology (European Commission 2001) including capacitance strip sensors, structures
instrumented with strain gauges and vehicle instrumentation (referred to as on-board
weigh-in-motion). Capacitance strip sensors are thin mats that measure the change in
capacitance as a vehicle passes over them which can then be related to the weight of the
vehicle. The data from these systems are difficult to interpret due to the complex dynamic
interaction between the suspension systems and the strip mounted on the road or bridge
surface. These effects can be reduced by using multiple mats and statistical correction
algorithms ((Stergioulas, Cebon and Macleod 2000).. Another possibility for measuring
the load is to instrument the structure itself. The CULWAY system has been shown to

11
provide accurate measurements of vehicle weights (Hood & Peters 1992)). This system
uses culverts instrumented with strain gauges to determine the weight of vehicles that
pass over them. The dynamic effects are minimized because there is no bump in the road
(as can occur with mats) to excite the vehicle and other dynamic effects are damped by
the mass of earth between the road surface and the instrumented culvert. Another
possibility is to obtain the weight of lorries from sensors mounted on the suspension
systems which allow weight data to be transmitted in real-time using wireless
transmission technologies. Such a system is already available on some commercial
vehicles (for example log-trucks in Australia) and its wider application for monitoring
heavy vehicle loads is currently being evaluated in New South Wales in Australia. This
is an extension to an existing wireless information system for monitoring the precise
location of very heavy trucks using GPS data transmitted back to a monitoring base
station. (http://www.rta.nsw.gov.au/heavyvehicles/iap/index.html) One can imagine a
system whereby vehicles passing over the bridge transmit their loading, speed and
location to the gateway which can then transmit this information back to the central
database along with other sensor data mounted on the bridge itself.

Whilst both a strain gauge instrumented bridge and instrumented vehicles are viable
options for measuring the load on a bridge to correlate with the rest of the sensor data, it
does present challenges in terms of the development of WSN technology. For example,
the ideal situation in terms of assessment would be to have sensor readings for the critical
loading scenario. However, this would require a triggering mechanism whereby the
gateway sends out a message based on either a sensor reading or the data received from
an instrumented vehicle telling the rest of the sensors to start taking readings. Since it is
unlikely the sensors could be activated to take the critical reading at exactly the right
moment, they would instead need to take a window of readings. Depending on how often
the sensors were activated and their sampling rate, the energy required for the sensors and
transmission of data could be significant. In order to do this with a wireless system,
methods of improving energy use and battery life, such as power harvesting, would need
to be considered. Thus creating WSNs to be used to support bridge assessment will

12
require advances not only in weight measurement technologies but also in sensor and
network operation as well as energy performance.

Data Management

Currently many bridge managers have reservations about the practicality of extensive
monitoring, both wired and wireless, on the bridge network. They see the use of
monitoring systems as being limited to specific structures where problems have been
identified and which can be easily measured with the available sensor technology. Even
in these cases they are concerned about the usefulness of the data from these systems as
the quantity of data acquired can quickly become overwhelming thus making them
essentially useless. If structural health monitoring (SHM) systems, including wireless
SHM systems, are to be used more widely and become a useful tool in bridge
management, three key data management issues need to be addressed: data volume, data
quality and data visualization.

Given the current availability and cost of computer memory it is tempting simply to
record all the data collected by a SHM. However this should be avoided for two reasons.
Firstly, because most of this data is not critical to the operation of the bridge and so does
not need to be recorded. Secondly, because for a WSN transmitting all the data and not
just the critical data represents an inefficient use of power and bandwidth. As such careful
thought should be given to which measurements are actually useful for the evaluation and
assessment of the bridge, and the correct level of decimation performed on the data. The
responsibility of determining which data is critical lies with the structural engineering and
NDT communities. For example, for assessment purposes it maybe useful to have the full
strain load interaction diagram for a given structural element when a vehicle is crossing
the bridge, whereas a bridge manager may only be concerned if a certain threshold is
crossed. However, neither the assessor nor the manager is concerned with the strain data
gathered outside of this critical vehicle crossing window. Thus intelligent WSN systems
should be able to decimate the data in such a way so that only the critical data for
assessment and management is stored and presented.

13
Another consideration that affects the data volume is the choice of sensors and sensor
location. The development of microelectromechanical systems (MEMS) sensors has
created the opportunity for inexpensive monitoring of a variety of parameters and the
potential to install monitoring systems that might consist of thousands of sensors. In some
cases this may well be the correct approach, especially if the sensors are simply
monitoring whether a specific threshold is crossed. If, on the other hand, the sensors
produce large quantities of data, there is the possibility of data saturation whereby the
results presented to engineers are too extensive and complicated to be understood
properly. If this is the case it is possible that the results will simply be ignored. As such,
when developing a WSN, only critical parameters should be monitored rather than
monitoring other aspects of performance simply because it is possible.

Finally, in order to make the data as useful as possible, 3-D visualization techniques
should be developed to allow bridge managers and assessors to quickly and easily
understand the relevance of the data. If each sensor can be illustrated on a 3-D model of
the structure then it becomes more straightforward to understand the relevance of the
data. If the data has also been properly decimated it should be straightforward to present
the critical values, such as the minima or maxima, in a pop-up window over the sensor. A
3-D interface was developed for both the anchorage and the Ferriby Road Bridge WSNs
that worked in conjunction with Google Earth as illustrated in Fig. 6. This allowed the
Bridge Master to not only see the location of the sensors on the structures but also the
location of the sensor networks with respect to the surrounding infrastructure, which
could be especially useful if multiple WSNs are installed within the same network.

If the use of WSNs on bridges is to become pervasive, the concerns of bridge managers
with regards to the significance and use of the data must be addressed. As such further
research into WSN technology must give consideration to data volume, quality and
visualization. If these issues are dealt with properly than WSNs could become an
essential tool for use in a comprehensive bridge management system.

14
Conclusions

As bridge infrastructure around the world continues to age and deteriorate it behoves the
engineering community to develop better ways to manage this infrastructure given the
limited resources that exist for maintaining and replacing it. One such tool that is
available as part of a total management strategy is the use of wireless sensor networks for
structural health monitoring. These networks have the potential to be cost-effective and
provide critical data for engineers and managers to make decisions about the condition of
individual structures as well as the overall bridge network. WSNs have been installed at
the Humber Bridge and on the adjacent Ferriby Road Bridge. A relative humidity and
temperature monitoring system installed in the north anchorage has been operating
successfully for 10 months. The network has provided managers with valuable insights
into the operation of the dehumidification system in this area and created the potential for
significant cost and energy savings. The second WSN was installed on a reinforced
concrete approach bridge and was designed to monitor deficiencies that have been noted
during visual inspections. Although this network has only been operational for a short
time, it has shown that is possible to use battery power not only for the sensors but for the
gateway as well, and to transmit the data back to a central server using a mobile phone
connection. There is still much work to be done if WSNs are to become an indispensable
tool for bridge managers and engineers. For example, techniques to measure corrosion in
reinforced concrete structures still need to be developed as only then can the capacity of
many of these structures be assessed accurately. Also, for the data from WSNs to be
useful in assessing the capacity of bridges, there has to be a method of measuring the load
on the bridge so that the sensor data can be properly correlated with the structural
performance. Finally, careful thought needs to be given to exactly what should be
measured on each specific bridge so that only the relevant data is captured. Work also
needs to be done to ensure that this data is processed and presented in a timely and
effective manner so that it supports the decision making process rather than detracting
from it. As such, though the present WSN developments are encouraging, much work
needs to be done to move towards a fully integrated future.

15
Acknowledgements

The authors wish to thank the EPSRC and the Humber Bridge Board for their generous
support. The authors also wish to thank Peter Hill, Tom Sanderson, Paul Fidler, Peter
Bennett, Ian Wassell and all those who made invaluable contributions to this research.

References

Christen, R., Bergamini,A., and Motavalli, M. (2003). Three-Dimensional Localization


of Defects in Stay Cables Using Magnetic Flux Leakage Methods. Journal of
Nondestructive Evaluation, 22 (3), 93-101.

European Commission (2001), Weigh-in-motion of axles and vehicles for Europe


(WAVE), prepared by Laboratoire Central des Ponts et Chaussees, April.(RTD project:
RO-96-SC,403)

Hood, R.G. and Peters, R.J. (1992). CULWAY in Western Australia.


Proceedings of Conference Australian Road Research Board, Nunawading (Aust), 16 (7),
199-216.

Mietz, J. and Fischer, J. (2007). Evaluation of NDT methods for detection of


prestressing steel damage at post-tensioned concrete structures. Materials and Corrosion,
58 (10), 789-794.

Nakamura, S. and Suzumura, K. (2005). Corrosion assessment of bridge cables. In


G.A.R. Parke and P. Disney (ed.), Proceedings of the 5th International Conference on
Bridge Managemnent, Guildford, 1, 28-36.

National Energy Foundation (2007). Simple Carbon Calculator.


<http://www.nef.org.uk/greencompany/co2calculator.htm> (Apr. 25, 2008).

Stergioulas, L.K. and Cebon, D. and Macleod, M.D. (2000). Static weight estimation
and system design for multiple-sensor weigh-in-motion. Proceedings of the Institution of
Mechanical Engineers, C: Journal of Mechanical Engineering Science, 214 (8). pp. 1019-
1036. ISSN 0954-4062.

16
Figures

Fig. 1 The Humber Bridge

Fig. 2 Connection of strands to concrete foundation

Gateway, 3 &11 9
5 10 5
1 & 10
7
Gateway
2 9000
6&7 8
4 8

E
15000

2 N
W
11
3
9000

4 6

1 9

(a) Elevation (b) Plan


Fig. 3 Node locations in north anchorage chambers

17
48
Mote 2 Mote 8
46

Relative Humidity (%) 44

42

40

38

36
Mote 1 Mote 4
34
00:00 04:48 09:36 14:24 19:12 00:00
Time (over 24 hours)

Fig. 4 RH data for west and east chambers

Fig. 5 The Ferriby Rd. bridge

Fig. 6 Node layout at the Ferriby Road Bridge

18
Fig. 7 Location of Nodes 5 and 6

Fig. 8 Anchorage WSN Google Earth model

19

You might also like