Bayesian structural parameter identification from ambient vibration in cultural heritage buildings: the case of the San Jerónimo monastery in Granada, Spain

Abstract

The deterioration of Cultural Heritage assets caused by the natural hazards is a pressing issue in many countries. Therefore, reliable models based on the large-scale structural response of the assets is key to assess their resilience. However, reliable models such as large and detailed Finite Element (FE) models, require a large number of data and input parameters. This paper proposes a Bayesian learning approach to identify the main parameters of a FE model with quantified uncertainty based on ambient vibration data. As a novelty when compared with other Bayesian structural parameter identification methods from ambient vibration data, here the likelihood function is formulated in a principled way considering information from both frequencies and modes using a probabilistic version of the Modal Assurance Criterion for the modes. This method is embedded into a parameterised computational model to automate the simulation process, and a real case study for a sixteenth century heritage building in Granada (Spain) is presented. The results show the suitability and effectiveness of the proposed Bayesian approach in identifying the most plausible values of the uncertain model parameters in a rigorous probabilistic way, but also in obtaining the modelled frequencies and the modal assurance criterion values with quantified uncertainty.