An instrumentation system was implemented on the Powder Mill Bridge in Barre, Massachusetts. The instrumentation was designed as a research prototype for development of a structural health monitoring system. Strain data from permanently installed strain gauges was collected for truck events over a period of six months. Maximum strain values from heavy truck events were used to establish statistical distributions which describe behavior of the undamaged bridge under normal operating conditions. Prediction intervals were added to the statistical distributions. A calibrated finite element model of the bridge was used to simulate damage scenarios, including large scale and localized damage of the bridge deck. Analytical distributions of a damaged bridge model were plotted against experimental distributions of the undamaged bridge with prediction intervals. It was shown that, given 1500 future heavy truck events, all three damage scenarios studied would be detectable. |