Much of Europe's railway infrastructure is approaching the end of its technical lifetime. However, there are clear environmental and financial incentives for prolonging the life span of infrastructure such as bridges. The key to safely extending a bridge's lifetime is monitoring. Traditional monitoring consists of inspections carried out at intervals of several years. Advanced Bridge Monitoring (ABM) allows both road and rail bridges to be monitored continuously: both in terms of their condition and the traffic they carry. Together with preventive maintenance and traditional inspection techniques, advanced monitoring techniques extend bridge life and avoid the disruption caused by unplanned maintenance.
A significant proportion of European railway bridges are more than 60 years old. Since they were built, the load models on which they were designed have changed. For much of this period, the projected loads were overestimates: we may expect many bridges to comfortably exceed their design lifetime. More recently however, increases in goods transport have caused projected loads to be adjusted upwards. Furthermore, changes in regulation and increasing international trade mean that the loads to which a bridge is subjected are more diverse than in the past. The uncertainty associated with future loads reinforces the need for measurement. A detailed and up to date knowledge of a bridge's condition, and its load history, will allow us to predict the future condition of the bridge with greater certainty and to anticipate and pre-empt safety issues.
Traditional monitoring is based on regular physical inspection of the structure for incipient cracks. Typically, the bridge is fully inspected every six years, with smaller investigations every two years. Advanced monitoring complements these inspections; it has three components:
Structural health monitoring is the long term, low frequency, observation of a set of representative characteristics of the bridge. It is appropriate for monitoring, for instance, tilting caused by soil slides; deformation caused by loading, and corrosion caused by the bridge's environment. The choice of properties to be measured will vary with the nature of the bridge. However, generally, deformation of the bridge at midspan is the most informative parameter. No matter which measurements are chosen, for this type of monitoring it is not the absolute value of a parameter, but its change over time, that is the critical factor. The challenge is therefore to be able to resolve an extremely small effect (of the order of milliradians in the case of deformation) from spurious long-term trends such as sensor drift.
Bridge weigh in motion, in contrast, is the high frequency measurement of the load applied to the bridge. The greatest danger for bridges is fatigue, i.e. cumulative damage caused by dynamic loads. Fatigue damage is a function of a load and the number of cycles for which it was applied. For railway bridges a fatigue event can be the arrival of the train—but it can also be the cyclic loading induced as each axle crosses the bridge. Because a train may consist of carriages from several countries—which follow different axle load rules—it may be necessary to identify a given axle as an abnormal load. A BWIM system must operate with sufficient speed to be able to resolve individual axles.
Since it runs continuously, taken over time, BWIM can produce very large data sets. However, this information is only useful if the correct analytical techniques are applied. The third component of an advanced monitoring system is its analytical toolbox. In the case of BWIM, a significant amount of pattern recognition and structural modelling is required to extract high resolution loading and fatigue information. SHM produces far more modest volumes of data, however, these too are meaningless without the correct analytical tools. For SHM, acquired data are combined with structural models in a stochastic framework to develop risk predictions. These allow the client to evaluate the likelihood of various scenarios. PSP has developed several software applications, running on a central server, that integrate seamlessly with our field measurement devices to produce rigorous risk analyses.
The actual hardware of an advanced bridge monitoring system can be split into two parts: a local device (or cluster of devices) mounted on, or near, the bridge; and a central server at a remote location. The local device captures data from multiple sensors, performs pre-processing and data caching before passing the data to a central server. A typical device has 10 sensor channels which, in the case of BWIM, are sampled at up to 3kHz. The devices must be sufficiently robust to operate in a hostile environment at location that may not be easily accessed. The units are designed for low maintenance and can be powered with photovoltaic cells.
For SHM applications, we typically use one local device per bridge. This will take a few measurement-sets a day and will report these to a server once a day. For BWIM applications we use one device per railway track—clusters of devices form a LAN. One of the devices is designated as a master: this collects data from the others for each train event and passes the data to the server in near-real-time.
The second physical component of the system, the server, carries most of the computational load. This has an internet connection with the local device and so may be placed at a convenient central location. The server has several tasks: most importantly the computationally demanding data analysis and data-driven simulation. However, it is also responsible for the preparation of reports and the archiving of data. Together the local device and the server allow high resolution data to be acquired from remote locations, and analyzed using complex numerical methods.
Railway bridges are chokepoints for any industrial society. It is vital that they either remain in operation, or at worst are withdrawn from service in a planned manner. Knowledge of a bridge's condition and its load history allow informed decisions to be made about the use of the bridge and its maintenance. This is particularly important in Europe with its relatively mature stock of bridges.
Advanced bridge monitoring is made possible by the availability of three enabling technologies. Reliable, low power, embedded systems make it feasible to acquire data from physically remote locations. Cheap computational power makes it possible to undertake complex analyses and simulations using the acquired data. Finally, the spread of internet technologies makes it possible to easily link these two technologies. The combination of these technologies means that it is now possible to deploy systems that dramatically improve both the bridge instrumentation and its analysis.