Are you an EPFL student looking for a semester project?
Work with us on data science and visualisation projects, and deploy your project as an app on top of Graph Search.
Among natural disasters, seismic activity which can cause earthquakes is a serious risk for human activities and most importantly their lives. Seismic risk assessment requires knowledge to evaluate existing buildings and their expected response against earthquakes. In a large scale assessment, it is not possible to analyze each one of them individually. Instead, each building is associated to a certain type to group them under similar behaviour. The literature already gives the parameters characterizing the vulnerability of these types. In the first part, this project focus on an advanced technique to determine the building type of the whole canton of Basel. By using data mining to have a list of features for the buildings and implementing a machine learning algorithm to interpret them, each building has a type attributed. A visual survey is also performed to validate the results from the machine learning method. To understand the taxonomy of buildings in Switzerland, the supervised algorithm called Random Forest (RF) was trained during another project with several swiss cities such as Neuchatel, Yverdon-Les-Bains and Solothurn. The results obtained by this method being close to the ones from the visual survey, this allows us to approve the first attempt of this methodology to try it on the whole country. In the second part, the risk assessment is performed using the European Risk-EU basis composed of two methods. Firstly, the empirical method (LM1) whose calculation of damage grade is based on past earthquakes. And secondly, the mechanical method (LM2) whose calculation of damage grade is based on the mechanical behaviour of buildings corresponding to their group. It consists in computing the lateral displacement in function of the lateral force at the base due to the seismic acceleration (those are called capacity curves). An optimization of the N2 method, used in the LM2 method to better compute the displacement demand was also considered. Several scenarios were considered to analyze the damage on the city of Basel and create seismic vulnerability maps.
Mario Paolone, Hamidreza Karami, Zhaoyang Wang, Pier Luigi Dragotti
,