Nondestructive evaluation (NDE) is a critical part of maintaining the structural health and safety of bridges. This paper applied the nondestructive testing technique of acoustic emission (AE) to test a bridge while it was load tested by the Florida Department of Transportation (FDoT). The bridge tested is a 182 ft long steel girder bridge with a concrete deck and four spans. It was built in 1951 and has recently been classified as structurally deficient. A four channel acoustic emission system was used to provide acoustic information from the different bridge members. The four channel system enabled live monitoring from sensors placed 40 ft above the ground on the steel bridge members. Classification neural network (NN), known as a Kohonen self organizing map (SOM), was used to classify the data into the three different AE sources. The test was also used to develop a new methodology for placing sensors and dealing with noise for future tests where the steel is experience fatigue cracking. While the bridge tested herein did not experience fatigue cracking, the ultimate goal of this research is to monitor for steel bridges for fatigue cracking and then predict remaining fatigue life using the captured AE data and back-propagation neural networks (BPNNs). The prediction was done on laboratory specimens by the authors with worst case of error of 20% for only first 25% of data [1-4].

Related Publications