Babak Moaveni - Civil and Environmental Engineering, Tufts University
Peter Moser - Department of Civil and Environmental Engineering, Tufts University
Iman Behmanesh - Department of Civil and Environmental Engineering, Tufts University
Continuously monitoring structural vibrations is becoming increasingly common as interest in structural health monitoring (SHM) grows, as equipment becomes more affordable, and as system and damage identification methods develop. We designed a vibration-based continuous monitoring system and deployed it on the Dowling Hall Footbridge at Tufts University in Medford, MA (Figure 1). The Department of Civil and Environmental Engineering at Tufts University educates engineering students to become leaders in addressing society's problems such as engineering for structural sustainability. The continuous monitoring system on the test-bed bridge provides both a live laboratory for research in SHM of civil infrastructure systems and a teaching tool for vibrations courses taught at Tufts School of Engineering at both the graduate and undergraduate levels.
The Dowling Hall Footbridge is well suited for a continuous monitoring system for several reasons. The bridge is flexible and significantly excited by pedestrian traffic and wind; vibrations are easily measured and can be felt by an observer standing on the bridge; and the bridge is exposed to a wide range of environmental conditions and is large enough to exhibit complex structural behavior. These conditions provide an opportunity for a realistic assessment of environmental effects.
This project consists of the following three phases:
- Initial testing and system design
- Instrumenting the bridge and extracting dynamic characteristics
- Developing an automated damage diagnosis framework based on measured dynamic characteristics of the bridge and environmental conditions
Phase 1: Initial Testing and System Design
Initial testing and system design took place in the spring and summer of 2009 beginning with several vibration tests. The objective of these tests was to estimate the dynamic characteristics of the bridge and to assess the feasibility of a continuous monitoring project. Figure 2 shows the natural frequencies and mode shapes of the bridge obtained from a preliminary test in April 2009. Knowledge of the frequencies, mode shapes, and expected vibration amplitudes helped us design a continuous monitoring system. We used these preliminary tests to decide the number and location of the sensors in the continuous monitoring system.
Phase 2: Instrumenting the Bridge and Extracting Modal Parameters
We installed the continuous monitoring system in the fall of 2009. The system includes an array of eight accelerometers and 10 thermocouples, a rugged and remotely operable data acquisition system, a reliable communication system, and fully automated modal analysis programs. A set of data is recorded once an hour or when triggered by large vibrations. The system generates thousands of data records. Because of this, developing programs that automate the transfer, processing, and analysis of data was crucial to the success of the project. The monitoring system has been running continuously since January 2010 and is still providing data. Figure 3 shows the location of accelerometers and thermocouples on the footbridge.
Eight Piezoelectronics model 393B04 uniaxial accelerometers were permanently mounted to the underside of the bridge, as shown in Figure 4. We found the current sensor network adequate for estimating the six lower vibration modes considered in this project. The monitoring system measures the air temperature, the steel frame temperature, the temperature of the heated concrete deck, and the temperature of the piers at several different locations.
An NI cRIO-9074 integrated chassis/controller is the core component of the data acquisition system. The cRIO-9074 features a field-programmable gate array (FPGA) that allows for customizable access to low-level chip functions. The ability to sample different sensors at different rates (acceleration versus temperature) and direct on-chip scaling of raw data to account for channel sensitivities are two attractive features of the FPGA. The cRIO-9074 also includes a 400 MHz processor, 128 MB of RAM, a 488 MB solid-state storage drive, eight slots for NI C Series modules, two Ethernet ports, and an operating temperature range from -20 to +55 °C. The cRIO-9074 uses the LabVIEW Real-Time OS. Once programmed, the device can run independently of a host computer, making it ideal for remote applications where reliability and autonomous operation are required.
Two NI 9234 C Series 24-bit integrated electronics piezoelectric (IEPE) input modules measure the accelerometer channels. Each module monitors four acceleration channels and supports sampling rates of up to 51.2 kHz. Software-selectable options include IEPE signal conditioning with antialiasing filters and timebase export for tight synchronization between modules. Inputs are sampled digitally at 24-bit resolution. The modules have an operating temperature range of -40 to +70 °C. One 16-channel NI 9213 thermocouple input module monitors the temperature sensors. The module features up to 0.02° C temperature resolution, an autozero channel, a cold-junction compensator, and automatic voltage-temperature conversions for common thermocouples types. This module also has an operating temperature range of -40 to +70° C. Figure 5 shows the CompactRIO and the three NI modules located inside a weatherproof enclosure next to the footbridge.
We developed the monitoring program using the NI LabVIEW Real-Time and LabVIEW FPGA modules. We initially used the CompactRIO SHM reference architecture, which provided valuable start up assistance. The monitoring program continuously samples the acceleration channels at a 2,048 Hz sampling rate. Temperatures are recorded at a rate of one sample per second. A five-minute data sample is recorded to the storage drive of the cRIO-9074 once each hour, beginning at the top of the hour. The program also performs automatic triggering by continuously monitoring the one-second root-mean square (RMS) value of each acceleration channel and will record a five-minute sample if the values exceed 0.03 g. Sample recording can also be triggered manually.
In addition to data acquisition, the program performs file and memory management, automatic error recovery, and system status messaging. With the current settings, the cRIO-9074 storage drive can hold data from a 12-hour period. The program runs autonomously but can be accessed remotely. New data is retrieved hourly by a PC in the Civil and Environmental Engineering Department at Tufts University using an FTP synchronization utility. Communication with the cRIO-9074 occurs over the Tufts University wireless-G network. The cRIO-9074 controller connects to the campus network through a wireless bridge. The wireless bridge was installed inside the weatherproof enclosure (Figure 5) and required an external antenna. The Hawking HAO14SDP directional antenna features a 14 dB gain factor and all-weather construction.
Once the data is filtered and downsampled, an automated stochastic subspace identification method is used for continuous modal analysis of the footbridge. Automatic modal analysis was completed for each of the hourly measured records. The identified natural frequencies for a 16-week period beginning January 5, 2010 and ending April 25, 2010 are shown in Figure 6.
Phase 3: Development of a Damage Prognosis Framework
Note that the third phase of this project is still under investigation. As the first step in this phase, we’re removing the effects of changing environmental temperatures from the continuously identified modal parameters. Environmental conditions can have as large an effect on the modal parameters as significant structural damage, so these effects should be accounted for before applying damage identification methods. Figure 7 plots the identified natural frequencies of the six identified modes versus the temperature of the northern abutment. Correlation between natural frequencies and temperatures is evident: higher natural frequencies generally occur at lower temperatures. Several models have been proposed to represent the relationship between the natural frequencies and the measured temperatures. A fourth-order polynomial model was found to better fit the measured data than the other models used (bi-linear, ARX, quadratic, cubic, and so on). Comparisons between the identified and simulated natural frequencies of the six identified modes are displayed in Figure 8.
The system reliably continues to provide data for research in vibration-based SHM. Development of a probabilistic damage identification framework based on the identified modal parameters is the subject of ongoing research.
Moaveni, B., and Behmanesh, I. (2011). “Effects of changing environmental temperature on finite element model updating of the Dowling Hall Footbridge.” Proc. of the Eighth International Conference on Structural Dynamics (EURODYN 2011), Leuven, Belgium, July 4-6.
Moser, P., and Moaveni, B. (2011). “Design and deployment of a continuous monitoring system for the Dowling Hall Footbridge.” Experimental Techniques, in press (available online).
Moser, P., and Moaveni, B. (2011). “Environmental effects on the identified natural frequencies of the Dowling Hall Footbridge.” Mechanical Systems and Signal Processing, 25(7), 2336-2357.
Civil and Environmental Engineering, Tufts University