Analyzing Concrete Cracking Along Civil Infrastructure using Distributed Fiber Optic Sensing

Crack monitoring using distributed fiber optic sensing (DFOS) in civil engineering has evolved significantly during recent years. Scientific and commercial applications often focus on distributed strain sensing based on Rayleigh scattering due to its high spatial and strain resolution despite its limited sensing range. Brillouin sensing techniques are more suitable for monitoring large-scale civil infrastructure as they can provide measurements over numerous kilometers. The improved sensing range, however, results in spatial limitations and therefore, restricted suitability for strain-based crack monitoring. This paper presents an enhanced laboratory test series, in which the suitability of various Brillouin interrogators for concrete crack monitoring was evaluated. Five individual concrete specimens equipped with multiple installation setups were investigated under well-known laboratory conditions, where the resulting fiber optic strain sensing data could be related to the true crack width obtained by high-resolution distance transducers over the crack. An alternative method relates alterations in the raw Brillouin frequency spectrum (BFS) to local distortion events like cracks, which is however an extensive, time-consuming process. Artificial intelligence (AI) is therefore applied to the test data to identify BFS anomalies and relate them to the locally arising crack width. First results demonstrate that AI can be an efficient tool to optimize traditional DFOS monitoring strategies, but further optimization is required for reliable determination.