Perspective on structural health monitoring of bridge scour

This paper celebrates A. S. E. Ackerman’s centennial publication on an apparatus for the monitoring of piles by providing a perspective on instruments used to monitor bridge pile scours. A short review of early works provides the reader with a historical perspective on the development and application of bridge scour monitoring devices. After, a discussion on contemporary measurement techniques reveals how these early devices have evolved, and how vibration-based monitoring techniques have gained significant attention. Lastly, thoughts on future needs for these structural health monitoring solutions are shared, and include remarks on the required characteristics to construct the next generation of high-performance bridge scour measurement device and monitoring systems.


Introduction
A century ago, A. S. E. Ackermann published an article in the Journal of Scientific Instruments, now Measurement Science & Technology, on an autographic apparatus for measuring and recording the movement of a driven pile [1].The apparatus consisted of a weight held by a spring suspended from a hook, which hook was attached rigidly to the pile, and a pencil suitably attached to the weight could be used to trace and record the movement of the pile on paper.Since this landmark contribution, technologies to monitor performance of piles, either during driving or post-installation, have considerably evolved, Original Content from this work may be used under the terms of the Creative Commons Attribution 4.0 licence.Any further distribution of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.especially given the important advances in measurement technologies, for instance in strain and acceleration sensing as well as various nondestructive evaluation techniques [15,18].As an example, networks of fiber optic sensors are regularly deployed to monitor strain along a driven pile and indirectly residual strain and forces [6], and accelerometers and other vibration-based techniques are utilized to conduct dynamic pile evaluation and testing [2,12].
This paper celebrates Ackermann's centennial publication on a pile monitoring device by providing a perspective on instruments used to monitor bridge pile scours.This perspective is conducted by highlighting technological progresses leading to modern bridge scour measurement devices, and offering insights on future needs to produce the next generation instruments.Bridge scour is a common cause of bridge failure [9].For example, there has been 100 bridge failures cased by 54 flood events recorded in Great Britain up to 2013 [24], and there were more than 20 000 scour-critical bridges  in the United States until 2009 [35].Figure 1 is a picture of the pier scour-induced collapse of the 2008 Houfeng Bridge, crossing the Dajia River in Taiwan.
Scour at a bridge pier is caused by local erosion provoked by a downward flow induced at the upstream end of the bridge pier, and amplified by horseshoe vortices developed due to the separation of flow [32] (see figure 2).The reader interested in learning more about the science behind bridge scour is referred to work from Pizarro et al [31] that discusses the scouring process, including types of scour, sediment particle motion, and methods to estimate scour depth.Invariably, the on-time evaluation of this problem affecting soil-structure interaction is critical in ensuring structural integrity and user safety.When detected, one can implement appropriate bridge scour remediation work based on prognosis [4].The most popular technique to conduct bridge scour evaluation is through visual inspections.Yet, visual inspections may fail at being conducted timely, require divers, and is difficult to conduct during extreme events such as flooding where scouring is more frequent and problematic.Consequently, there has been efforts in developing technologies capable of automating the inspection process, known as structural health monitoring (SHM) solutions.
This perspective paper focuses on bridge scour measurement devices targeting such SHM applications, and the rest of the article is organized as follows.The next section provides a short historical timeline on the development and application of bridge scour monitoring devices.The subsequent section discusses contemporary measurement techniques.The last section provides thoughts on future needs.

Early work
In 1961, four decades after the work published by Ackermann, Richardson et al [33] proposed to utilize a sonic depth sounder to conduct SHM of bridge scours.While its application to bridge scour was demonstrated only in a laboratory environment, the authors envisioned that the device could be modified into a portable version by increasing the maximum depth measurement capability and decreasing the device's costs.In 1975, Norman [29] published a study on bridge scours conducted on a group of bridges located in Alaska.The field investigation, constituting one of the first field application of bridge scour measurement devices if not the first, was conducted using a fathometer (a type of sonar) along with transducers mounted on a bridge pier and other being hand-held.
In 1986, a review of bridge scour measurement techniques was reported in [36].The publication mentioned a conductivity sensor developed by researchers in Arizona.The device consisted of electrodes dispersed within a tube driven into the soil.The conductivity between electrodes was measured along the tube to distinguish between water and sand.The review also reported on a heat-flow measurement device consisting of heaters and temperature sensors located along the tube.The heaters were actuated for a short period of time, and temperature gradients measured to differentiate between soil and water.The review also reported on fathometers and concluded that these devices held a great short-term promise for SHM of scours, and the authors proposed and evaluated a portable sensor mounted on a remotely controlled boat.Results showed the feasibility of a portable fathometer, and recommended that ground-penetrating radar (GPR) be evaluated as a technique for monitoring bridge scours.This was done by Gorin and Haeni in 1989 [17], where the authors reported successful measurements in water less than 7.6 m (25 ft) deep.
The 1990's were marked by a considerable evolution in research and applications of bridge scour SHM devices.For example, Mason and Sheppard [27] deployed an acoustic scour-depth monitoring system over 16 bridge piers, with the notable feature that the system transmitted data through satellite, thus permitting off-site data acquisition and processing.The authors reported challenges relative to the harsh marine environment.Hayes and Drummond [19] developed two scour monitoring systems, one based on a fathometer and the other on an electrically conductive probe.Both systems were applied and tested in the field.Results from the investigation showed that the fathometer performed well provided proper post-processing of the data, and that the electrically conductive probe had limited applicability during high flows and was vulnerable to debris.In a National Cooperative Highway Research Program (NCHRP) report, Lagasse [23] evaluated the performance of various bridge scour monitoring devices.The report concluded that sliding collar-and sonarbased devices were particularly promising in terms of accuracy (±0.3 m), dependability, and durability over a broad range of engineer demands.The authors recommended future areas of research, including piezoelectric sensor driven rods, and local and remote telemetry.Yankielun and Zabilansky studied the use of time-domain reflectometry [38].The authors reported continuous capabilities, and also an accuracy of the order of ±0.05 m.Forde et al [13] conducted further work on the use of GPR in shallow water by moving GPR equipment manually using a small boat, and discussed the requirements in antenna frequencies versus river depth.

Contemporary measurement techniques
Since 2000, a number of contemporary measurement techniques for bridge scour monitoring were proposed in the form of electronically advanced versions of those aforementioned.Most of these advances are attributed to improvements in electronic hardware, for example leading to the integration of micro-electro-mechanical systems in scour measurement devices [25].As a result, sonar-based (e.g.fathometers) and radar-based (e.g.GPRs) are still utilized, but are now commonly capable of remote data transmission to improve site accessibility.Remote data collection and transmission became a standard in the industry.For example, buried mechanical devices that are engineered to float-out when a scour reached an acceptable level are now capable of transmitting information when floating.The interested reader is directed to [5] for a report on the evaluation of such remote monitoring techniques.Other devices borrow on previously developed principles, but use modern transducers to improve robustness and accuracy of measurements.This is the case for some modern versions of the driven rods that now measure vibrations or strain along the rod using fiber-Bragg grating and other transducers [14,40] to extract the natural frequency of the rod, which in turn relates to the buried depth.
Examples of innovative devices developed in the last decade include the bio-inspired whiskers proposed by Swartz et al [37].The sensor consists of a magnetostrictive flow sensor array is designed to send a dynamic signal when exposed to water flow.Figure 3(a) is a picture of the sensor showing the magnetostrictive whisker flow sensors connected to the data acquisition and processing board (DAQ) powered by a rechargeable battery pack and transmitting data through a zigbee antenna.The sensor is maintained in place using concrete as a ballast.There is also Chen et al [7] who developed a smart rock that consists of a group of magnetic elements designed to roll down at the bottom of the scour hole.Figure 3(b) is a picture of a smart rock consisting of a magnet embedded in concrete.The organization of these rocks influence a measurable magnetic field that can be mapped to the scour.Such magnetic field has been measured using magnetometers installed on a truck and recently using unmanned aerial vehicles [41].
An important modern development in bridge scour measurement techniques arose from significant enthusiasm in leveraging the Fourier kingdom and structural dynamics that gave raise to the utilization on inclinometers/tiltmeters and accelerometers.These devices are typically installed on the side of the piers above the water line, and provide time series data to be mapped to structural behaviors.The system is analogous to that of a driven rod, except that the rod is the pile itself.A critical challenge associated with these devices is that signals are not quantities that can be clearly mapped to bridge scour  and usually require physical knowledge (e.g. a finite element model) of the bridge [32].Nevertheless, these techniques hold important advantages over traditional depth-measuring instruments in that they are relatively less expensive, can be installed easily without needing to dive, and can be used continuously even during a natural hazard.These advantages are likely responsible for the surge in contemporary research interest.Figure 4 is a schematic of the principal bridge scour techniques discussed in this perspective paper.
One of the critical challenges in instrumentation for bridge scour detection is in the mapping of instrument signals to actionable information.Often, the detection and quantification of bridge scours is not sufficient, as soil-structure issues may have already become too grave and would require serious economic resources to mitigate.It follows that the utilized instruments would preferably allow the timely discovery of scours or, even better, the prediction of scours over a sufficient horizon facilitating condition-based maintenance actions.This need for actionable data is often overlooked in researches that focus on sensor development.Notable works on forecast of bridge scour using sensor data include a time-based warning systems based on probabilistic analysis [4] and a deep learning machine learning model [39] for predicting scour progression.The interested reader is also referred to [10] for a review on bridge scour predictive methods.

Future directions
Based on the SHM field's vision to the next generation measurement devices [22], it is foreseen that the next generation of high-performance bridge scour measurement devices will be required to combine the following financial, mechanical, electrical, and algorithmic characteristics, while maintaining prescribed measurement precision and accuracy requirements: Low cost: economic benefits of bridge scour monitoring will outweigh the cost of the sensor network including software and hardware acquisition, installation, and maintenance.Ruggedness: devices will be capable of surviving natural disasters such as flooding.Power: devices will be power-autonomous, ideally through energy harvesting, and will thus require intelligent power management.
Wire/data transmission: data will be transmitted wirelessly, with the use of wires minimized over the installed sensing system.Easy deployment: measurement systems will be plug-andplay, and deployable by minimally trained agents.Data management: devices will be pre-processing data to minimize the size of transmitting datasets and provide predigested information useful to infrastructure operations.Real-time capabilities: actionable information will be made readily available to infrastructure operators in real-time enabling punctual mitigation.Predictive capabilities: measured states will be combined with data-based or physics-informed algorithms to make predictions enabling condition-based maintenance decisions.
A promising path is that of accelerator-based solutions, because these instruments are readily available in highly rugged versions, their typical positioning make them less likely exposed to harsh events, and may be easily paired with power-autonomous wireless data acquisition systems [30].In addition, satellite-based techniques, such as interferometric synthetic aperture radar, are particularly promising at conducting remote scour detection due to their significant improvement in accuracy over the last decade [34].Advances in signal processing and autonomous underwater vehicles can also provide exciting opportunities, for example using underwater computer vision [16] to quantitatively evaluate scours [20].Furthermore, the fields of bio-inspiration [26], advanced materials [21], and robotics [3] may be harnessed to construct new sensing systems and strategies that can attain sensing performance levels never achieved before.
It is also envisioned that real-time monitoring and predictive capabilities will leverage advances in the collection and fusion of data from multiple devices and sources (Internet of Things or IoT) [11,28].For example, future monitoring strategies could combine sensor networks available on a bridge along with hydraulic sensors along the river and weather forecast.The integration of multivariate signals using cross-disciplinary techniques, including physical knowledge, will create new opportunities enabling faster and more accurate bridge scour characterization, predictions, and mitigation.

Figure 3 .
Figure 3. Two examples of recently developed SHM strategies for bridge scour monitoring: (a) bio-inspired whiskers consisting of a magnetostrictive flow sensor array (courtesy of R. Andrew Swartz); and (b) smart rock consisting of a magnet embedded in concrete [7].

Figure 4 .
Figure 4. Possible bridge scour measurement devices and their locations.