Bayesian Approaches for Damage Identification of Historic Masonry Towers

The damage identification process of the Torre Grossa in San Gimignano (Italy) was investigated through model updating procedures based on data fusion. This approach provides the opportunity to account for different sources of uncertainty that are involved in the knowledge process and quantify their propagation. In this regard, Bayesian inference is frequently employed to compute the effects of parameter uncertainty, observation errors, and model inadequacy. Its application to historical constructions gained increasing interest as support to the identification of reliable computational models for historic masonry buildings, with the aim of evaluating their structural behaviour and performance under exceptional loads. In this scenario, the Torre Grossa represents an interesting case study due to the availability of different sources of experimental data from both static and dynamic Structural Health Monitoring (SHM) systems, as well as the detailed survey of the cracking pattern. This medieval tower is the tallest of the towers preserved in the historic centre of San Gimignano. It shows a square cross-section (measuring 9.5 x 9.5 meters) and an overall height of about 60 m. The sustaining walls of the tower are multi-leaf stone masonry walls with the internal layer constituted by brick masonry and external faces made by stone masonry. The internal core of the multi-leaf walls is composed of heterogeneous material (remainders brick tied by a poor mortar). Moreover, the tower is incorporated into the neighbouring building, the Town Hall, up to the height of 20 m. The paper shows the results in terms of structural model identification of the Torre Grossa by exploiting the use of different sources of data in a Bayesian perspective in order to investigate the use of this procedure as a tool for damage detection