Following a seismic event, the serviceability of buildings, bridges and other infrastructure is critical. However, determining the safety of these structures post-disaster is challenging and time consuming. Most damage assessment is done through visual inspection, which can miss structural damage that is hidden from view, behind architectural elements or in hard to access areas. To improve the efficiency of post-earthquake damage assessment, low-cost sensors that could monitor structures and send alerts of significant damage would be invaluable. One specific example of critical damage would be that of steel fracture, which significantly impacts the safety of steel and reinforced concrete structures. Fractures are highly energetic, creating a distinctive gunshot-like sound. Microphones are already used to detect gun fire, and a similar methodology could be employed to detect and locate fractures in buildings. This project explored this possibility. A database of fracture sounds from experimental tests of structural components was built from a variety of sources, including an online research repository maintained by the National Hazards Engineering Research Infrastructure (NSF-NHERI). Specific features of the audio signals, for example Mel-frequency and linear predictive coefficients, were used to train machine learning algorithms  to classify these sounds and detect fractures. Physical experiments were also conducted to record rebar fractures using an array of low-cost microphones. The placement of the microphones and the difference in arrival times were used to estimate the location of the fractures, which were compared to the true location. This research constitutes the first step in the development of a robust acoustical monitoring strategy to aid in efficiently making decisions to restrict service to compromised structures following an earthquake. This concept is becoming increasingly viable as the availability of inexpensive instruments increases. As sensors improve, this approach could become the prevailing method for post-event assessment.

This research project will focus on the measurement of the influence of the fibers on the strength of fiber-reinforced concrete (FRC), examining both the experimental method used and the theoretical background needed to extract the salient material properties. Here, the primary interest is in the tension strength. The stress-strain relationship of most engineering materials is determined with direct compression and tension tests. However, this method proves to be unsuitable for testing the tension strength of FRC due to various factors. In this research project, we will determine the stress-strain relationship of FRC using a flexural beam test. The beam test is considered more reliable because the load can be controlled better than in the direct tension test, and stress concentrations and eccentricities can more easily be avoided. However, the experimental results must be combined with theory to extract relevant information. In structural analysis and design, it is common practice to begin with a known stress-strain relationship and the dimensions of a beam section and integrate to determine curvature for a sequence of moments. Our approach is the reverse of this process. Using a system of differential equations relating strains, moments, and axial stresses, and with strain measurements from the top and bottom midspans of the beam, we intend to inversely develop the stress-strain relationship of FRC in compression and tension through differentiation of a polynomial regression. We expect results to indicate that fiber-reinforced concrete has a higher residual strength than what is currently accepted and that our testing procedure will yield more accurate and valuable results than traditional tests. These findings could change the way cementitious materials are tested and improve efficiency in the built environment thereby decreasing carbon emissions. This presentation will highlight the steps, challenges, results, and implications of our project.