In Italy, a quarter of the building stock dates to years before 1946 (source ISTAT, 2011). The awareness of the vulnerability of strategic structures such as hospitals is of primary importance as local governments have the necessity of a global overview of seismic performance of these facilities to decide how to best allocate funds for restorations. From this point of view, the fragility curves of a building represent a powerful tool that allows to assess the probability of exceeding a damage threshold (or a limit state). Such information is useful for the assessment of the structure, and for investigating the possibility to maintain functions during and immediately after earthquakes.

In this paper the case study of an hospital facilities is presented. The building is a seven storeys reinforced concrete frame with dimensions in plane of approximately 120x34 meters. The structure is developed with frames in the longitudinal direction (the largest one), only connecting beams in the transversal direction and eight stair/elevator cores, made of weakly reinforced walls. The building was built in the 1970s and was designed only for vertical loads. The structure was analysed and verified by means of non-linear static analyses (Pushover) and incremental non-linear dynamic analyses (IDA). Furthermore, the fragility curves for four different damage levels (null, slight, moderate and severe) and two different seismic intensity measures were obtained: the peak ground acceleration and the spectral pseudo-acceleration in correspondence of the first mode of the structure. The first represents a very intuitive and practical parameter to use, the second represents a choice widely used in literature.

As expectable, the results obtained showed a completely different behaviour between the two directions, with the transverse one as the weakest, due to the high shear fragility of the walls. Nonetheless, the structure has a ductile behaviour in the longitudinal direction. Furthermore, the evaluation of the fragility curves highlighted that the choice of the intensity measure still requires a deeper evaluation. The variation of the first mode period due to the shear behaviour of the star/lift cores is too relevant, and the spectral acceleration showed to be not a representative intensity measure for this case study.