Technical Approach and Understanding
The whisker-inspired flow sensors will be constructed using the magnetostrictive Galfenol wire developed at the University of Maryland. One end of a Galfenol whisker is fixed and the other end is free to deflect under fluid flow-induced drag forces as seen in the figure below. The bending-induced stress near the fixed end of the whisker produces local changes the magnetic domain orientation, which is accompanied by a global change in the magnetic flux density in the whisker, demonstrated in the figure below. A giant magnetoresistance (GMR) sensor at the fixed end of whisker is used to detect this change of magnetic field and to convert it into an electrical signal that, once amplified, can be easily transmitted as needed. A permanent magnet must be placed in contact with the whisker to align magnetic domains along the whisker when it is not deformed. Different strength magnetics can be used at different locations, as long as a proper magnetic field strength is achieved at the root of the whisker to ensure internal magnetic dipole rotation in response to bending of the whisker without saturating the GMR sensor. The sensor configuration is very simple to assemble from five main components: the whisker, a GMR sensor, a clamping fixture, a small permanent magnet and a small low power operational amplifier.
Theory behind magnetostrictive flow sensor
Application to bridge piers
The objective of this project is to test and deploy an embedded array of sensors located on or near the outer surface of the bridge foundations, at varying heights, that can determine the sediment depth and profile around the foundation in real time. The sensor array will be composed of bio-inspired, whisker-shaped magnetostrictive flow sensors that are highly rugged and detect water flow by bending, see the figures below. Those sensors located above the sediment level will be free to move with the current flow and will yield dynamic flow measurements. Those sensors located below the sediment line will be trapped and will return only static measurements. Knowledge of sensor depth will help to determine the sediment level in real time. An automated data acquisition base station will monitor sensor signals from above the water line, differentiating between static and dynamic sensor readings, estimating the sediment and water line elevations, monitoring for sensor failure, and sending scour alerts to relevant authorities to be visualized using the decision support engine.
|(a) Overview of bio-inspired scour sensor array concept for bridge piers|
|(b) Installation concept for monitoring the leading and trailing edges||(c) monitoring side scour|
Wireless smart scour-sensing posts for abutment, culvert, and bank monitoring
To make installation of scour sensor elements as practical as possible, modular sensor units, known as “smart scour-sensing posts” will be used in this project. These modular units shown in the figure on the left below consist of several magnetostrictive transducer elements, embedded data collection and interrogation electronics, a wireless communication interface, and a long-lived battery pack packaged in a galvanized steel post that can be driven into the ground near abutments, culverts, and in embankments. These devices are wireless to eliminate the need to run signal cables in the vicinity of the bridge. They will utilize low-power wireless signal networks (e.g., Zigbee) to communicate with local base stations that will aggregate the data from multiple on-site posts and send it via cellular data link to remote users. Low-power components and use of sleep mode will be employed to extend the battery life with the aim of achieving a battery life of ten years. An illustration of the proposed installation approach for monitoring of a bridge abutment is provided in figure on the right.
|Smart scour-sensing post.||Installation of posts at bridge abutments.|
Culvert monitoring: Scour at culverts induced by exiting flow patterns can be detrimental and threatening to the culvert stability. Riprap placement at the outfall can retard scour but has been shown to also fail under high-flow conditions. Scour occurring below the culvert foundation can result in culvert collapse and subsequent bridge or road failure. To detect this scour, smart scour-sensing posts will be placed on the downstream side of the culverts as close to the mouth of the culvert as possible, shown in the figure below. A nearby base station will collect the data from any posts within range (up to 700 m) and transmit that data to remote managers.
Monitoring of bank stability: Riverbank erosion is prevalent in rivers in which a river is not in equilibrium with its watershed due to watershed development or extreme rainfall events. Sensors will be located vertically in the riverbank at various lateral locations to detect loss of the embankment depicted below. The depth and number of these posts will be determined based on input from DOT agencies. A nearby base station will collect the data from any posts within range (up to 700 m) and transmit that data to remote managers.
Monitoring of culverts.
Monitoring of bank stability/erosion.
Numerous technical barriers are anticipated and must be investigated during the course of the project. This section will discuss these barriers and the planned approach for overcoming them in four broad categories: 1) those barriers associated with sensor protection; 2) those associated with signal generation; 3) those associated with signal processing and classification; and 4) those associated with flow of data/information and decision support. Each of these categories of challenges is discussed below.
Sensor Protection: Despite their inherent mechanical robustness, protection of the proposed magnetostrictive sensors from corrosion is an issue that must be addressed for use in their application. Galfenol is an iron-based alloy which has corrosion properties that are similar to those of steel. The flow sensor design must incorporate sufficient corrosion protection for the Galfenol whiskers to have a life expectancy that is matched to the life expectancy of the batteries that power the wireless data acquisition system.
Tata Steel (tatasteel.com) has a comprehensive publication to provide "guidance on the corrosion and protection of H-section (steel) universal bearing piles.” They note that fresh water can contain dissolved salts, gases or pollutants that are harmful to ferrous alloys (like Galfenol). And although corrosion loss from river water immersion is generally lower than for seawater, their report shows results from studies of H-section steel bearing pile corrosion that exhibited corrosion rates in fresh water of 0.02 – 0.05mm/side/year, and in corrosive soils of up to 0.015mm/side/year. With these corrosion rates, if made from an unprotected Galfenol strip with a thickness of 0.50 mm, the proposed whiskers would vanish several years (<5) after installation.
Canadian researchers have conducted the only known studies on the corrosion of Galfenol. Their studies suggest iron provides a good basis for initial Galfenol corrosion design considerations. They indicate a negligible corrosion rate for Galfenol (with gallium content that ranges from 15% to 27%) in de-aerated 3.5%NaCl solution at 25oC. They suggest that this because a protective film is formed on the alloy surface causing its potential (E) to fall below -0.79~-0.86, which is below the cathodic protection criterion for iron.
Compositions of Galfenol, AISI 1021 steel, and bearing steel.
|AISI 1021 steel||0.18-0.23%||0.60-0.90%||0.04%||0.05%||1.00-2.00%||balance|
Schematic diagram of corrosion-protective layers for Galfenol strip mounted on bridge pier in immersed conditions
This protective film resulted in a plateau potential region during the anodic polarization scan. Gallium oxide (Ga2O3) layer as a major component of the protective film formed like dense aluminum oxide (Al2O3) layer on aluminum, and the increase of gallium content in Galfenol is more effective for protecting on the surface. However, they observed pitting and crevice corrosion on Galfenol with gallium content greater than 18.4% after cyclic polarization scan. For comparison, AISI 1012 steel and 304 stainless steel tested in the same corrosive condition with Galfenol and their compositions are summarized in table above. The average corrosion rate of Galfenol in naturally aerated 3.5% NaCl solution is approximately four times lower than that of AISI 1012 steel, while it is higher than that of 304 stainless steel.
As an alternative, cathodic protection, epoxy coating and/or chromium electroplating can be used for protection. Cathodic protection can be easily applied to immersed structures. Aluminum or zinc alloy pieces are employed as a sacrificial anode. The anode will impart corrosion immunity by rendering the Galfenol whisker cathodic relative to the adjacent anodes for a sufficiently negative potential in electrochemistry. The effective life of the Galfenol whisker can be further increased by the use of an epoxy coating that covers the surface of the Galfenol whisker that is to water. In the case of static senor strip buried in riverbed composed of soils and sands, it also requires abrasion resistance. Alternatively, a thin chromium film deposited by electroplating can be applied between Galfenol strip and external epoxy layer to ensure the protection with abrasion resistance. We will need to investigate the impact of the stiffness of the chromium film and epoxy layers on the bending characteristics of the whiskers. An increase in whisker surface area may be required to increase pressure drag associated with a given flow past the whiskers to ensure suitable bending induced moment rotation occurs for a given flow rate and sensor sensitivity to water flows. The previous figure shows the schematic diagram of these anticipated corrosion-protection coatings and layers. The project team would also like to explore the use of magnetostrictive iron aluminum whiskers as alternative to Galfenol. Although iron-aluminum has less than half the sensitivity of Galfenol, it will cost less that Galfenol and has significantly better corrosion properties so may be a viable alternative alloy for the whisker sensor application.
The smart scour-sensing post concept requires that the magnetostrictive flow sensors be capable of surviving the installation process which will involve being driven into the soil near abutments, culverts, and in river banks. Flanges installed just below the sensors will protect their bases during driving. The course of action to be taken is to install them in a coiled condition utilizing a water-soluble tie to hold the entire whisker within the protective shadow of the base flange; that tie will dissolve after installation when water is present. This will be done by employing a 4-inch diameter “installation pipe” that threaded to the exterior of the base flange depicted in the fifth figure (mouse over to emphasize). A Rhino post-driver will be used to drive the base flange and this installation pipe to a desired depth in the soil/bank. The 3-inch diameter scour post with attached flow sensors will be inserted into the center of the installation pipe and a latching or threaded connection used to attach the scour post to the base flange. The outer installation pipe will then be removed while a vibrational force is imposed, allowing sediment to surround the scour post.
Signal Generation: Generation of appropriate signals under certain conditions is envisioned to be another potential barrier to successful implementation of this approach. For low-flow conditions, there may be little dynamic signal generated by the whisker sensor which may make unburied sensors appear to be buried sensors. The project team will perform extensive laboratory studies to quantify which conditions have sufficiently low enough flow levels to create false indications of the trapped condition to understand where these conditions may occur in the field. In addition, the geometric properties of the whiskers as well as their support conditions provide some avenues for alleviating this problem. By varying the whisker profile to increase fluid-structure instability and by altering the roughness around the base of the whisker, additional turbulence can be introduced into the system that will benefit free-condition detection.
Signal Processing and Classification: Several signal processing and classification barriers must be overcome for successful validation of this approach. To accomplish low-power operation of the remote sensing hardware it is important to take advantage of embedded processing techniques. Embedded processing techniques are important in low-power and wireless application as transmission of raw data represents the most energy intensive process for such systems. Local, embedded data processing allows sensors to transmit only results from the engineering algorithms rather than lengthy time-history records saving energy and extending battery life.
The most basic embedded processing task is flow detection. The system operates by differentiating between static and dynamic flow signals returned from the magnetostrictive whiskers. For fast moving and turbulent water bodies, such a distinction is relatively trivial to make from highly varying sensor data. For slow moving bodies exhibiting laminar flow around piers, application of sophisticated autonomous signal processing techniques can help to distinguish genuine perturbations from noise. The laboratory experimental study will help to highlight conditions under which flow detection becomes difficult. Here, advanced signal processing techniques will be used to differentiate between noise and very low signals based on the vibrational properties of the whisker sensors themselves. Differences in resonant frequencies between air-coupled and water-coupled whiskers will be employed to identify sensors that are above the water line.
The array of magnetostrictive whisker sensors is designed to provide sufficient measurement points to provide useful scour measurements in real time. The automated system will monitor the locations of sensors returning static and dynamic data and maintain a map of the estimated channel bed profile. The system will also note sensors that are topographic outliers: either dynamic sensors surrounded by static sensors, or static sensors surrounded by dynamic sensors. Such sensors may indicate unusual scour, impingement of whiskers by trees or other debris, or a sensor fault condition. Because some applications for these sensors require installation below the water line (e.g., bridge piers), the cost of installing the system becomes a concern. The team will work with state DOT partners to establish the optimal number and placement of sensors depending on the application and establish installation guidelines according to feedback received from these agencies.
In any long-term installation, sensor reliability and sensor failure are important problems. To reduce false alarms due to transducer failure, there has been significant progress made in the field of fault detection for permanently installed sensor arrays. The issue of sensor failure is particularly troublesome in systems such as the scour monitoring system in which static, or noise only, sensor signals are used as an important indicator (in this case, presence of sediment). Without embedded sensor fault detection algorithms, signals measured from damaged sensors may be easily interpreted as dynamic data potentially triggering false alarms, or worse, as static data, potentially missing hazardous scour events. Robust sensor failure detection algorithms will be installed to look for common failure modes including excessive noise, loss of signal, intermittent railing, and drift. Additional fault modes discovered during the course of the study will be added to the fault detection profile.