Bridge safety goes wireless

Sitting at his desk in Lehigh’s ATLSS Center, Shamim Pakzad holds in his hand one key to the future safety of America’s bridges.
It is a wireless sensor unit and it looks like a computer chip. Small and rectangular, it contains five sensors that measure the phenomena that inflict stress on bridges: light, humidity, temperature and vibrations from passing traffic and wind. It also contains a small battery, circuitry for data collection, and a box-shaped “mote” fitted with a CPU (central processing unit) and a radio for wireless communication.
Just as the chip has revolutionized electronics, says Pakzad, the P.C. Rossin Assistant Professor of civil and environmental engineering, wireless sensors will change the way we protect bridges and the people who use them.
Wireless sensor networks, says Pakzad, will enable engineers to more accurately monitor a bridge’s ability to withstand the steady stresses of auto traffic and wind and the powerful, multidirectional loading of earthquakes. They will help locate and evaluate damaged portions of a bridge while saving engineers countless hours of time in physical inspections, often during inclement weather.
Because they are cheaper and easier to install and maintain than wired sensors, it is possible to fit a bridge with hundreds or thousands of wireless sensors, versus several dozen of the wired variety.
Three years ago, working with the University of California-Berkeley, Pakzad conducted a pilot project on San Francisco’s Golden Gate Bridge. The world-famous structure is fitted with a wired sensor network containing 90 sensors, almost all of them located on the piers of the bridge.
Pakzad’s group installed 64 wireless sensor units, or a total of 320 sensors, along the bridge’s main span and south tower, which sustains larger displacements due to wind and traffic than the pier. The project demonstrated that wireless sensor networks could work outside the lab; previously, the use of the sensors had been limited in scope.
While the wired sensor units on the Golden Gate Bridge cost thousands of dollars apiece, says Pakzad, each of the wireless prototypes installed by his group cost just $200. Mass-production and new design could drop that figure below $10.
This difference in cost, combined with the greater ease of installing and maintaining wireless sensor units, and comparable, if not superior performance, gives wireless sensor networks the potential to monitor bridges more cheaply and more thoroughly than wired units can.
But challenges remain. During the 2006 study of the Golden Gate Bridge, Pakzad’s group collected data at scheduled times to measure the bridge’s response to traffic. But no data was collected for the three earthquakes that struck San Francisco during the six-month period.
“The Golden Gate Bridge project lacked priority-based task management,” says Pakzad. “It had only scheduled task management. Its sensor network was not equipped to respond to earthquakes.”
Pakzad is working with Liang Cheng, associate professor of computer science and engineering at Lehigh, to enable the wireless sensor network to preempt a scheduled activity, such as monitoring traffic stresses, and switch to a more urgent task, such as monitoring responses to an earthquake or explosion.
“To develop priority-based task management, you have to equip hardware and software with new capabilities,” says Pakzad. “And data has to be processed differently. In ordinary times, there is no hurry to process data; it can wait until the base station has collected all the data.
“In the event of an earthquake, you need immediate analysis to tell you a bridge’s condition, whether it can be operated safely, whether emergency response teams can cross it to get to various parts of the city.”
Once a sensor network switches priorities, it must handle the rush of data and complex data-routing tasks brought about by an earthquake.
“A structure can vibrate very quickly during an earthquake,” says Pakzad. “You may need to take up to 500 samples per second. This produces a large volume of data that needs to be communicated through a wireless network.
“But a network is limited in the amount of data it can transfer, so data can quickly become unmanageable. We have to decide which data gets transmitted first, which does not need to be transmitted, and whether it is possible to send a summary of the data for later analysis.”
“Truly a multidisciplinary area of engineering
Pakzad is developing a software system that optimizes the performance of sensor networks by dealing with networking, routing and time synchronization. His goal is to enable wireless networks to identify and communicate the locations on a bridge that have sustained damage, either from traffic or from earthquakes or other events.
“In a large structure, the measurements taken by each sensor need to be synchronized very accurately through a large network. To measure the vibrations and accelerations at hundreds of points, sensors must measure everything at a precise time.
“During an earthquake, when you’re assessing a structure’s real-time response to multidirectional loading, if you lose some of your data, you need a protocol to ensure that communication, data collection and command dissemination are all reliable. The commands must be received by all sensors.”
Another part of Pakzad’s research deals with what he calls “translation of damage,” which occurs when engineers measure displacement, velocity and other structural responses to stress and then draw conclusions about a structure’s state of health or the extent of damage it has sustained. He is also writing an algorithm that translates data into performance parameters.
In addition to his collaboration with Cheng, Pakzad also works with professors Rick Blum, Miltos Hatalis and Marvin White in Lehigh’s electrical and computer engineering department.
“This is truly a multidisciplinary area of engineering, with a lot of electrical engineering, computer science and structural engineering,” says Pakzad.
Pakzad recently received a grant from the National Science Foundation’s (NSF) Sensors and Sensing Systems Program. His research is also supported by the Pennsylvania Infrastructure and Technology Alliance (PITA).
Pakzad is applying for additional funding from NSF and the National Institute of Standards and Technology to extend the use of wireless sensors into other areas of structural engineering, including early warning systems for progressive collapse.
--Kurt Pfitzer