Sensor Placement Optimization for Early Damage Detection in Gerber Girder Bridges

Bridge structural health monitoring (SHM) is a continuous process that involves the use of various sensors, data analysis, and assessment techniques to evaluate the condition, safety, and performance level of a bridge structure over its operational lifespan. To identify anomalous deviations from the expected behaviour and issue alerts in case of potential risks, sensors should be strategically placed. In this sense, high-density sensor networks can guarantee better coverage and improved data accuracy for damage detection. Yet, when dealing with large-size systems, the definition of a minimum scalable number of sensors is critical to meet budget constraints. Determining the sensor locations that provide the most relevant and valuable information while optimizing SHM-related resources and cost is not straightforward as infrastructure systems evolve over time, hence their monitoring needs may change. This work addresses the problem of optimal sensor placement for SHM of existing bridges by proposing a simple greedy algorithm based on a sequential backward selection process to optimize the sensor layout of a Gerber girder bridge in view of future SHM activities. To this end, a physics-informed model representative of the real bridge structure is employed to account for different structural conditions and potential damage scenarios aimed at identifying the best non-redundant sensor deployment that would ensure the safe monitoring of the bridge in the long run.