Probabilistic mechanics-based loss scenarios for school buildings in Basel (Switzerland)
Michel et al
Developing earthquake scenarios for cities in areas with a moderate seismicity is a challenge due to the limited amount of available data, which is a source of large uncertainties. This concerns both the seismic hazard, for which only recordings for small earthquakes are available and the unknown earthquake resistance of the majority of structures not designed for seismic loading. The goal of the present study is to develop coherent probabilistic mechanics-based scenarios for a mid-size building stock including a comprehensive analysis of the uncertainties. As an application, a loss assessment for the school buildings of the city of Basel is performed for different scenarios of historical significance, such as the 1356 event, and from the deaggregation of the Swiss Probabilistic Seismic Hazard Model of 2015. The hazard part of the computations (i.e. ground motion estimation) is based on this model, a regional microzonation and recordings of small earthquakes on a dense strong motion network to compute site-amplification factors. The school buildings, which are mainly unreinforced masonry or reinforced concrete shear wall buildings, have been classified according to a specifically developed taxonomy. Fragility curves have been developed using non-linear static procedures and subsequently, vulnerability curves in terms of human and financial losses are proposed. The computations have been run with the OpenQuake engine, carefully propagating all the recognized uncertainties. Scenarios before and after retrofitting measures show their impact on the earthquake safety. A sensitivity analysis shows that the largest uncertainties come from the ground motion prediction although an improvement of all parts of the model is necessary to decrease the uncertainties. Although improved data and models are still necessary to be developed, probabilistic mechanics-based models outperform the capabilities of deterministic and/or empirical models for retrieving realistic earthquake loss distributions.