Strain Measurements Under Damage and Environmental Effects from a Real-Scale Masonry Testbed via Smart Brick Sensors
Description
Structural health monitoring (SHM) of masonry structures suffers from a critical scarcity of real-world strain data, particularly measurements capturing progressive damage under controlled yet realistic conditions. This dataset addresses that gap by providing strain time-series acquired through smart bricks: piezoresistive brick-like sensors, embedded directly within the fabric of the structure. Their seamless integration into the masonry allows for continuous, minimally invasive monitoring. The measurements were collected from a full-scale masonry prototype, a two-story building archetype constructed outdoors and left exposed to natural environmental conditions over multiple monitoring campaigns. As a result, the recorded strain signals carry the simultaneous imprint of two distinct sources of variation: structural damage and environmental effects (thermal cycles, humidity, seasonal drifts). This dual nature makes the dataset particularly suited for developing and benchmarking compensation strategies, novelty detection methods, and damage identification algorithms under realistic operating conditions. Three damage scenarios of increasing severity were induced on the structure. The first involved the sudden release of two central tie-rods (out of four originally installed). The second consisted of incremental static overloading applied to the roof slab. The third, and most severe, was a progressively induced differential foundation settlement. Each of the three accompanying CSV files corresponds to one damage scenario. All files share the same structure: the first two columns encode the timestamp (date and time), followed by thirteen measurement channels (SB1 to SB13), each representing the strain output of one smart brick. Each file contains approximately 720 time steps. For sensor placement within the structure and further details on sensor layout and damage scenarios, readers are referred to the linked publication, where a damage detection strategy based on linear cointegration is demonstrated using this dataset. Additional publications exploiting this dataset will be linked as they become available.
Files
Institutions
- University of PerugiaUmbria, Perugia