| 
The collapse of the I-35W bridge in Minnesota in August focused a lot of attention on how we ensure the integrity of structures. The techniques for identifying structural problems are referred to as “structural health monitoring” (SHM). SHM today relies heavily on visual inspection. Only a handful of our bridges, buildings and other structures have sensors that track their performance over time. Much can be gained, however, with systematic instrumentation and installation of SHM systems. Such systems in principle provide diagnosis of the structure identifying that there is a problem, localizing the problem, and quantifying it. Furthermore, these systems should provide a prediction of the structure’s remaining life and make recommendation for follow-on actions. We often refer to structural monitoring systems as the “structural doctors.” Technological advances are making them much more practical.
A major obstacle to deployment of sensors in civil structures used to be the extensive amount of cabling needed for data and power transmission – a particularly difficult and costly task in existing structures. The Tsing Ma Bridge in Hong Kong required more than 26 km of cable for its instrumentation. The emergence of wireless sensor networks that are designed for low power consumption makes it possible today to install large numbers of sensors to track variety of structural functions. We developed the first wireless structural sensing node in 1996 and successfully tested it on the Alamosa Canyon Bridge in New Mexico. The low power consumption makes it possible for the system to operate on battery power over an extended period of time. In addition, new technologies that harvest power from structural vibrations or temperature changes of thin films are being developed that will make these networks fully autonomous.
Another difficulty with past systems is the extensive amount of data that needed to be interpreted by experts. With older systems, data typically are streamed to a central location where they are cleansed and analyzed. As an example, it took six months to two years after the 1989 Loma Prieta, California, earthquake to analyze the data that were collected from the handful of buildings and bridges that were instrumented at the time. Our current research has focused on developing damage detection algorithms that can be embedded at the sensor node serving as the diagnostic engine and as a data compression tool. These algorithms identify changes in the system at the location of the sensors. Only if damage is detected is an alert issued. The algorithms are becoming increasingly more robust, building on statistical pattern recognition methods developed for gene identifications. There is a considerable research effort underway to develop variety of damage diagnosis algorithms each intended for different type of structure or damage condition.
Such algorithms can be executed at the sensor location because small yet powerful microprocessors are available and can be combined with the sensors and the radios in a mini-system. We are also using 2 to 4 gigabyte storage devices to preserve the original data as well as analysis results. Forensic investigations often require that we review past data. Storing that data will enable us to perform more thorough investigations if such are needed. None of these capabilities were available to us only ten years ago.
Indeed much has changed in the past decade making it both technologically and economically feasible to install structural monitoring systems that can provide timely assessment of the state of the structure and issue warnings in case of impending failures. Traditional sensors are being replaced with small form micro-electromechanical (MEMS) sensors. In addition to the commonly used accelerometers, strain gages, temperature and humidity transducers, specialized sensors are continuously emerging. For example, research is being conducted on the development of corrosion sensors; piezoelectric and piezo-ceramic sensing skins for crack detection in steel; fiber-optic sensor cables for crack detection in concrete; carbon nanotube skins also for crack detection in steel members; and global positioning systems for large displacement measurements.
High resolution optical and infrared imaging techniques are being deployed for displacement and microcrack identification. Since structures are complex systems, combinations of sensors are typically needed to detect the different types of damage that can occur. A large number of sensors is desirable because changes in the structure can be located in a small area and those changes can be captured only if the sensors are close to the damage location. Excessive vibrations, such as those reported by drivers days before the collapse of the I-35W Bridge in Minneapolis, would have been captured by a handful of accelerometers triggering an alert as the vibrations became bigger than a critical limit. Hopefully in the future, more structure will carry such health monitoring systems.
|
|
|
Strategic Priorities
