Lehigh University researchers have been on the forefront of field instrumentation and the testing of bridges, buildings and other structures for over 50 years. They have developed a vast database and considerable experience in testing and evaluating highway and rail structures.
Field testing and evaluation is a particularly challenging area of structural research and is the only method to accurately determine the true behavior of a structure in real time under actual loads. Field measurements can connect and validate laboratory testing and computer modeling to the behavior of real structures.
Accurate interpretations of structural behavior require accurate data. However, it is much more difficult to collect data several hundred feet in the air over a river, in the wind and cold, than in a laboratory. Many variables must be addressed prior to making an accurate interpretation of the measurements, and a thorough understating of data acquisition systems, sensors and structural behavior is essential. Therefore, it’s important to have a data acquisition system that can consistently perform under adverse conditions.
Lehigh University researchers have found the Campbell Scientific CR9000 measurement and control systems to be robust and capable of withstanding harsh field environments. They are the workhorses of the university’s infrastructure monitoring program.
Monitoring the Williamsburg Bridge
One recently completed field-testing project involved onsite measurements and long-term remote monitoring of New York City’s Williamsburg Bridge, which is currently undergoing a major rehabilitation. A significant portion of this project involves replacing a concrete-filled grid deck on both the south and north inner and outer roadways with a steel orthotropic deck.
Along with comprehensive laboratory testing, a team conducted an in-depth field instrumentation and testing program on the south outer roadway. This study investigated in-situ stress ranges and helped to better characterize the behavior of the complex structural system and its relationship to the laboratory response.
During controlled load tests, data were collected using the measurement and control systems from 82 strain gages at sampling rates up to 200 Hz. All data were temporarily stored on PCMCIA cards installed in the datalogger before being copied to the laptop at the end of each test for processing and backup.
A second portion of the program consisted of nearly seven months of remote monitoring. Data were collected using the same measurement and control systems. Program upload and data download were achieved with two modems configured by Lehigh University’s Advanced Technology for Large Structural Systems (ATLSS) researchers. One modem was placed at the site, the other in an office at the ATLSS laboratory. The data were downloaded to a desktop PC every one to 14 days.
Access to the structure was very difficult. Constant traffic and the remote location of the logger posed a problem to the monitoring program, since convenient, regular access was simply not possible. The measurement and control systems permitted Lehigh researchers to collect a considerable amount of data remotely, eliminating repeat trips to the bridge.
Recording and analyzing data
To minimize the volume of data collected, time histories were not recorded continuously. Rather, the system was programmed to begin recording when the stresses induced by live loads exceeded predetermined triggers. The appropriate magnitude of the triggers was determined apriori from the controlled load tests.
For example, assume it was determined that heavy trucks produced a peak stress of 6 ksi at a given location. Software triggers were then set at 5.9 ksi for these gages. If the stress exceeded that value, the time history was recorded.
Data were also recorded prior to the trigger event for a specified amount of time, say five seconds (i.e., a 5-second buffer was maintained). The system continued to record for an additional specified time period—again, say five seconds—and then stopped recording. These channels monitored, as triggered time histories were automatically re-zeroed on the whole and half hour using a digital balance algorithm. The appropriate length for the buffer and total recording time were determined on site. This technique provided sufficient time-dependent data and a method of verification for high stress range events to ensure that spurious signals were excluded from the stress-range histograms.
Stress-range histograms were developed using the rainflow cycle counting method contained in the system’s instruction set. The stress-range histograms were generated continuously and did not operate on triggers, thus all cycles were counted. In addition, the histogram was updated every 10 minutes, meaning data loss would be minimized in the event of a power failure. The stress range bins were divided into 0.5-ksi intervals. The results of the research will be incorporated in the AASHTO LRFD bridge design specifications.