A Bayesian Probabilistic Framework for Building Models for Structural Health Monitoring of Structures Subject to Environmental Variability

Managing aging engineering structures requires damage identification, capacity reassessment, and prediction of remaining service life. Data from structural health monitoring (SHM) systems can be utilized to detect and characterize potential damage. However, environmental and operational variations impair the identification of damages from SHM data. Motivated by this, we introduce a Bayesian probabilistic framework for building models and identifying damage in monitored structures subject to environmental variability. The novelty of our work lies (a) in explicitly considering the effect of environmental influences and potential structural damages in the modeling to enable more accurate damage identification and (b) in proposing a methodological workflow for model-based structural health monitoring that leverages model class selection for model building and damage identification. The framework is applied to a progressively damaged reinforced concrete beam subject to temperature variations in a climate chamber. Based on deflections and inclinations measured during diagnostic load tests of the undamaged structure, the most appropriate modeling approach for describing the temperature-dependent behavior of the undamaged beam is identified. In the damaged state, damage is characterized based on the identified model parameters. The location and extent of the identified damage are consistent with the cracks observed in the laboratory. A numerical study with synthetic data is used to validate the parameter identification. The known true parameters lie within the 90% highest density intervals of the posterior distributions of the model parameters, suggesting that this approach is reliable for parameter identification. Our results indicate that the proposed framework can answer the question of damage identification under environmental variations. These findings show a way forward in integrating SHM data into the management of infrastructures.