Lestuzzi, PierinoFürst, Valentin2020-01-232020-01-232020-01-232020-01-23https://infoscience.epfl.ch/handle/20.500.14299/164813Large-scale seismic risk assessment requires knowledge of the vulnerability of buildings. Under the action of an earthquake, buildings with different properties also behave differently. Since it is not possible to analyze each building individually, it is necessary to use types that group together buildings with similar behavior. For each of these types, parameters characterizing their vulnerability have been established. In this project, a data mining method, the Association Rules Learning (ARL), was applied to the building stock of the canton of Basel-Stadt in Switzerland to determine the type of each building. This data mining method helps for finding correlation between some specific building attributes and known types in a learning set. The three attributes considered are construction period, number of stories and roof shape and can be automatically obtained from public databases. For seismic vulnerability assessment at large scale, both methods developed within the framework of the European RISK-EU project were used. The level 1 method (LM1) is an empirical method whose calculation of damage probabilities is based on the observation of past earthquakes. The level 2 method (LM2) is a mechanical method because it requires describing the mechanical behavior of buildings by capacity curves (force at the base as a function of lateral displacement) to infer target displacements and then damage probabilities. An optimization of the N2 method, used in LM2 for the determination of the displacement demand, was also considered. The empirical and mechanical methods were firstly applied while considering the European classification. Then, they were applied by considering an update of this European typological classification adapted to the Swiss building stock, which will be called the Hybrid typology (EUCH). The results obtained from the application of these methods on two different typologies were analyzed and several seismic vulnerability maps were proposed. Finally, the influence of the consideration of new types developed for in-row buildings typical of Basel was investigated by considering the district of Iselin.Seismic vulnerability – Moderate hazard – Existing building – Association Rule Learning – European typology – Swiss buildings – LM1 – LM2 - N2 Opt – In-Row buildings - Basestudent work::master thesis