Monitoring

LONG-TERM BRIDGE MONITORING UTILISING WIRELESS MULTI-SENSOR SYSTEMS

Markus Rennen Dipl-Ing, Business Development Manager, Senceive

A closer look at wireless conditioning monitoring of aging infrastructure

Bridges connect people! Less poetically expressed, they help us to physically overcome or bypass obstacles. Bridges form the essential links in highways, railways and footways that allow the passage of traffic over obstructions, enabling everyday routines to be maintained. Compared to most other infrastructure, the failure of a bridge causes disproportionate disruption as well as social and economic damage.

Bridges are complex, high value structures and their construction and replacement can take years. Investment in bridges can only be justified by a long life of safe operation. It is not unusual for many bridges to be used well beyond their design life, both in terms of time and traffic loading, resulting in greater stress and accelerated deterioration. The engineers and owners responsible for them must decide how best to manage this deterioration, with options including replacement, strengthening, and doing nothing.

Replacement is usually viewed as too costly and too disruptive. Doing nothing brings considerable risk in terms of liability. Strengthening and upgrading are therefore more common, but both options require a robust understanding of structural condition and behaviour – which usually entails inspection, testing and monitoring.

Figure 1: High frequency mesh-based systems operate at afrequency of 2.4GHz and can support multiple sensor types in the same network. 

Inspections are required by law in many countries and these typically require significant investment in terms of time, expertise, traffic management and access equipment. In the UK, the Design Manual for Roads and Bridges (DMRB) outlines the requirements for two-yearly general inspections and six-yearly principal inspections. Robust inspections do not guarantee success however, with bridge failures still occurring in many countries – causing significant disruption and sometimes loss of life. In many cases, investigations in the aftermath show that a more effective surveys regime could have revealed early signs of defects and distress, potentially enabling effective intervention.

Where scheduled inspections identify potential issues, they often recommend follow-up surveys and testing, but these don’t always happen in a timely way due to the associated cost and resources. The growing availability of automated monitoring systems provides a welcome alternative that can resolve the balancing act between reasonable cost and mandatory diligence. Such systems deliver data at far greater frequency than manual surveys and without the subjectivity. Automated systems can monitor precise movement and changes in geometry as well as environmental and structural parameters such as temperature, strain, crack movement and groundwater behaviour.  Wireless systems can transmit data to a cloud platform, perform preanalysis and trigger automated alarms.

Monitoring of structures

The physical inspection is often accompanied by geodetic surveys, often using classic tacheometry, where total stations are employed to observe discrete, predefined points.

However, the data are collected expert analysis of results is indispensable, for example, to assess the engineering significance of trends and patterns in the data. Access to suitable tools is also essential. The physical inspection is often accompanied by geodetic surveys, often using classic tacheometry, where total stations are employed to observe discrete, predefined points.

Tacheometry can be automated to provide a relatively tight sequence of observations. This allows engineers to distinguish event-induced deformation from normal patterns of movement e.g. diurnal changes due to sun exposure or traffic loading. The total station is placed on a solid support and validates its position and orientation with every observation epoch by referencing itself to surrounding stable reference points. The monitoring points are discretely marked or signalled locations that will be repeatedly targeted.

In addition, total station observations require clear line of site between instrument and reflector; the presence of an inconveniently parked truck, a dirty prism or vigorous growth of vegetation can mean lost data. Vertical displacements can also be determined automatically and very accurately by hydrostatic levelling systems. Whilst effective, this technique entails a significant level of operational maintenance.

In addition, total station observations require clear line of site between instrument and reflector; the presence of an inconveniently parked truck, a dirty prism or vigorous growth of vegetation can mean lost data.

The use of tacheometry is therefore increasingly being accompanied or replaced by laser scanning. This provides a triangular irregular network (TIN) surface that, in comparison with results from an earlier scan, allows non-discrete analysis of structural deformation. Similar results can be achieved by unmanned aerial vehicle (UAV) based photogrammetry. A limitation of both techniques, however, is that they provide data that are limited to a single point in time and do not, therefore, support a tightly scheduled automated monitoring programme.

We are seeing growing adoption of fibre optic systems. These systems are inherently automated and require little maintenance. Despite the efficient and precise results and the relatively low price of the fibre optic cable, the complete system including the laser hardware and processing unit is extremely expensive and relatively challenging to install on existing structures. It is viewed by many experts as being more applicable as a structural health monitoring system built into new structures or those undergoing major upgrades because of the likelihood of a more generous budget at that stage.

Figure 2: Brücke Wittekindstrasse, a six-lane concrete bridge carrying the arterial B1 Highway over a four-lane road during load test.

Wireless condition monitoring

With the initial motivation in mind, i.e. extending the safe lifespan of existing bridges, wireless condition monitoring (WCM) offers a tool that allows the integration of a variety of autonomous long-life sensors, is easy to install and practically maintenance free. Wireless condition monitoring represents a specific variation of the Internet of Things (IoT). Wireless nodes collect data from internal or external sensors and transfer that data via a cellular gateway to a cloud-based platform.

There the raw data are processed into engineering parameters and made available for the user. Based on a qualified analysis of the data it is possible to characterise normal patterns of behaviour and distinguish them from anomalous events such as movement or cracking caused by forces including corrosion, traffic loading or disturbance from adjacent construction activity. If pre-defined threshold values are breached, then alarms will be triggered and transmitted to registered users via email or SMS.

The pre-stressed concrete bridge, built in 1957, consists of two separate parallel deck structures. It crosses the main road below diagonally and is supported mid-span by a row of concrete columns on the median strip. 

Today, the miniaturisation of sensors – such as micro electronic mechanical sensors (MEMS) – allows the design of compact nodes with extremely low power consumption which in turn leads to high durability1. Two families of communications platforms are available; mesh-based high frequency platforms and lower frequency long-range systems. High frequency mesh-based systems such as Senceive’s FlatMesh operate at a frequency of 2.4GHz and can support multiple sensor types in the same network (see Figure 1), as well as allowing remote users to change configuration settings such as reporting rate.

Most bridge monitoring projects use clusters of sensor nodes located within a relatively small area, with distances typically no more than a few tens of metres between them. Such distances can easily be covered using a 2.4GHz mesh platform. Occasional obstacles can be bypassed using repeater nodes. For more extended sites, lower frequency communication platforms are more applicable. Low power, long range (LoRa) based systems such as the Senceive GeoWAN platform operate at 8-900MHz and have a range of several kilometres. This extended range comes at the expense of reduced functionality and responsiveness and they offer fewer options for remote user-access. LoRa systems are therefore not often used in bridge and structural monitoring projects.

Figure 3: Installation of Senceive wireless senor nodes.

Many of the sensors themselves are not particularly new; strain gauges or potentiometric crack sensors and pressure cells, for example, have been widely used for many years. What is new is the way they function as part of a wireless node. The node determines the recording rate, provides both internal and external sensors with power and establishes the data communication with the gateway. The gateway provides short-term data storage, but most crucially, is the device that transfers data to the cloud server via cellular signal.

Highway Bridge, Dortmund

The Dortmund Roads Department (Tiefbauamt) is responsible for a range of highway structures across the north-west German municipality. One of these is the Brücke Wittekindstrasse, a six-lane concrete bridge carrying the arterial B1 Highway over a four-lane road. The pre-stressed concrete bridge was built in 1957 and consists of two separate parallel deck structures of roughly 31m length. It crosses the main road below diagonally and is supported mid-span by a row of concrete columns on the median strip (see Figure 2). Thus, the deck is formed of four discrete sections of 15 m length.

Whilst all concrete structures are expected to move to some extent, excessive deformation can cause hairline cracks that increase the risk of water and chloride ingress and corrosion of reinforcement or prestressing steel cables.

An engineering expert report confirmed the structural integrity of the bridge, but a decision was made to implement long-term monitoring in order to better understand its normal behaviour – such as movements due to diurnal or seasonal thermal variations or traffic load.

Whilst all concrete structures are expected to move to some extent, excessive deformation can cause hairline cracks that increase the risk of water and chloride ingress and corrosion of reinforcement or prestressing steel cables. Corrosion reduces the effective diameter of steel cables and reinforcement bars which therefore lose strength. Such deterioration is thought to have resulted in the recent partial collapse of the Carola Bridge in Dresden.

In cooperation with the municipality‘s surveying department (Vermessungs- und Katasteramt Dortmund) a monitoring plan was developed using wireless condition monitoring with the aim of securing safe operation throughout the bridge’s remaining lifespan.

A monitoring plan was developed using wireless condition monitoring with the aim of securing safe operation throughout the bridge’s remaining lifespan

A series of wireless Senceive tilt sensors and strain gauges were installed on the deck soffit along the projected line of prestressing cables by operatives working from a cherry picker access platform (see Figure 3).

A longitudinal constellation was chosen (see Figure 5 and Figure 7) that would make it possible to trace any vertical deflection. In addition, potentiometric crack sensors were installed to monitor movement of lateral cracks, and at two locations, temperature probes were installed in holes drilled approximately 800mm into the concrete.

Since installation in October 2023, the solar-powered gateway has been transmitting data from all the sensors to the cloud server at 30 minute intervals.

In-Situ Load Test

It was decided that six months into the monitoring programme, a load test would be conducted. The aims were to support future management of the bridge through better knowledge of its structural behaviour and to compare the effectiveness of different measurement techniques – laser scanning and wireless tilt nodes.

Figure 4: The 48t crane located on vital positions on the bridge in order to simulate significant traffic load.

This test was conducted in April 2024 and involved loading the structure with a 48t crane, provided by the Dortmund Fire Brigade.

Test locations were selected based on an assessment of a structural model which identified several points as vital for a characteristic analysis (see Figure 4).

Prior to placing the load, static laser scanning was carried out using three Zoller & Fröhlich phase-based terrestrial scanners to provide an initial model of the unloaded bridge.

The WCM system was switched to live mode for the duration of the load testing programme, with a 30-second data recording interval.

Working in this way, the effect of the increased loading could be observed in-situ to establish the point of maximum deformation.

Upon reaching this point, a second laser scan was conducted (see Figure 5) with the bridge under loaded condition. This was followed by a third scan following removal of the crane. This procedure was repeated for each crane position.

Results

Figure 5: Visualisation of load impact from tilt nodes.

The comparison of pre-load and loaded scan data was intended to provide detailed surface deformation data for the entire bridge deck slab. Analysing the bridge design, the civil engineering experts predicted 2-4mm deflection induced by the load.

The laser scan and simultaneous WCM were intended to validate each other’s results. The question was whether a) laser scanning would be sufficiently accurate to resolve these small values and b) whether tilt nodes and strain gauges would provide sufficiently representative results to determine realistic bridge deformation. Before each scan, the height of the scanner was meticulously observed by precise spirit levelling in order to avoid systematic effects.

The comparison of pre-load and loaded scan data was intended to provide detailed surface deformation data for the entire bridge deck slab.

Data processing and final analysis were carried out by the Dortmund municipality, with the process also conducted within the scope of a bachelor thesis at the geodetic department of the University of Applied Science Bochum/Germany. The report includes several key findings relating to the two measurement techniques.

Figure 6: WCM constellation and laser scanning during load test.

Firstly, the WCM live view showed the immediate (sub-minute) impact of the load as significant tilt change (see Figure 6).

As expected, two nodes on one side of the panel simultaneously pivoted in the direction of the loadpoint, while the other two tilted in the opposite direction. The strain gauge in the centre of the panel revealed a significant increase in strain.

In order to generate a deflection curve from the WCM data, the tilt values were interpreted as tangential deflection (see Figure 6, green graph).

This could be compared with a longitudinal profile section generated from the surface derived from the laser scan.

The results shown in Figure 7 demonstrate a strong correlation between the two datasets.

As expected, maximum deflection was induced when the load was placed on the outer lanes: this resulted in approximately 2mm vertical displacement in the centre of the panel (see Figure 7).

Both systems obtained representative results that coincided by fractions of a millimetre (see Figure 8).

These results were also obtained for the inner lanes where the deflections were smaller. Still the results coincided by the same ratio.

Conclusion

Laser scanning provides an extremely dense point coverage of the entire structure, similar in its accuracy to reflectorless total station observations.

The surface derived can be the basis of detailed structural analysis by generating a digital twin. However, the survey itself is conducted manually and requires some effort, both in terms of fieldwork and in processing.

Although advances in AI-supported processing have speeded-up the process it is still by no means automated or suitable for permanent monitoring. The wireless monitoring data, on the other hand, only represents discrete point locations and is therefore comparatively limited in terms of spatial coverage.

Figure 7: Visualisation of the surface deformation as determined by laser scan.

Figure 8: Comparison deformation derived by scan (blue graph) vs. WCM derived deflection (green graph).

With much more frequent (and potentially live) sampling, it is much more valuable in identifying temporal change.

With a node battery life in excess of 10 years it effectively offers permanent, maintenance free monitoring supported by automated processing and alarms.

Although advances in AI-supported processing have speeded-up the process it is still by no means automated or suitable for permanent monitoring.

Furthermore, a variety of geotechnical and structural sensors can be incorporated (strain, groundwater, crack movement, temperature etc.), exceeding the mere geometric analysis.

Within the scope of the presented monitoring task, both techniques revealed their respective strengths.
They should not be considered as competing, but instead supplementing each other.

Periodic laser scanning surveys can provide full spatial coverage and permanent WCM can provide continuous observation, albeit from a smaller sample of locations.

Thus, it can trigger alerts that can initiate more elaborate and detailed survey measures such as laser scanning, spirit levelling or visual inspection.

Used in combination, continuous, detailed, long term and yet cost-efficient monitoring of crucial infrastructure can be conducted, ensuring early warning and thus safe operation throughout the entire life span of the structure.

Markus Rennen Dipl-Ing, Business Development Manager, Senceive mrennen@senceive.com
senceive.com

Literature
Różański, Marcin: Geodätisches Monitoring mittels terrestrischer Laserscans im Zuge einer lastinduzierten Verformungsmessung an der B1-Brücke über die Wittekindstraße in Dortmund“ (2024), Bachelor Thesis, Faculty of Geodesy, Universoity of Applied Science Bochum.

---

(1) @30min data reporting rate >10 years autonomous operation.