Identification of seismic damage in masonry structures by two-step SSI and parametric inverse analysis

Masonry structures represent large part of the historical and cultural built heritage in Europe. The assessment of their structural behaviour is critical for the preservation and conservation of buildings, monuments and infrastructures especially in seismic areas. In particular, one relevant issue is the evaluation of damage and residual load-bearing capacity of historical masonry structures affected by earthquakes. Such evaluation is necessary to assess structural safety, in order to define posteriori actions such as evacuation, retrofitting or demolition. A successful technique aimed at better knowing the dynamical behaviour of a structure is Operational Modal Analysis (OMA), in which the modal properties of the system are identified based on vibration data collected when the structure is under operating conditions. In Structural Health Monitoring (SHM) analysis, any variation of such modal properties could be considered as an indicator of modifications induced in the structure, such as change in mass or stiffness due to environmental agents or development and propagation of damage. The identification of the location, the type and the severity of the damage entails the solution of a further inverse problem, in which modal data, possibly coupled to other sources of information, are exploited for the assessment of such damage. In this paper, a novel comprehensive procedure is proposed for the identification of damage in masonry structures after a seismic event. The procedure relies on the acquisition of operational vibration data before and after an earthquake. In a first phase, those data are analysed by means of Stochastic Subspace Identification (SSI) to estimate frequencies and modal shapes of the linear system in undamaged and damaged conditions respectively. Afterwards, such modal properties are used as input for the inverse problem of estimating material parameters for an approximate numerical model representing the structure. The proposed methodology is applied to a pseudo-experimental case study in which the acceleration data are provided as the output of a mesoscale numerical representation of a 3D masonry building. The same building is then modelled with a macroscale approach using a homogenous isotropic model, whose damaged and undamaged features are identified by means of inverse analysis against the SSI output. The results show the potential of the proposed approach, where simple and non-destructive measurements are coupled to advanced computational methodologies to provide a comprehensive framework for damage identification after seismic events.