Ali Saeidi - Centre d’études sur les ressources minérales, Université du Québec à Chicoutimi
Olivier Deck - Université de Lorraine, GeoRessources, UMR ۷۳۵۹, Nancy, F-۵۴۰۴۲, France
Thierry Verdel - Université de Lorraine, GeoRessources, UMR ۷۳۵۹, Nancy, F-۵۴۰۴۲, France

The extraction of ore and minerals by underground mining or other underground workings often causeground subsidence phenomena. In urban regions, these phenomena may induce small to severe damage tobuildings. We have developed vulnerability functions for determining damage to building in subsidenceregions. The methodology uses Monte Carlo simulations, and existing analytical methods based on the beamtheory for the evaluation of damage in the subsidence area. It allows taking into account uncertainties bothon the geometrical and mechanical parameters of buildings, and on the phenomena of soil structureinteraction for analytical methods. This paper focuses on uncertainties on soil-structure interaction. Thedetermination of damage with analytical methods requires values of the horizontal strain and the deflectiontransmitted to buildings. But the available geotechnical parameters are the horizontal ground strain and theground curvature; soil structure interaction parameters are then required to determine how these geotechnicalstrains are transmitted to buildings. The value of these later parameters are dependent on several factors,such as the soil and the building rigidity, the building type, the mine or tunnel depth, the localisation ofbuilding in a subsidence basin. All of these factors increase the uncertainties in building vulnerabilityfunctions, and must be considered in the development of these functions. We observed that low values ofground curvature coefficient (KΔ) lead to a flattening of the vulnerability curve (low vulnerability), and thatlow values of horizontal strain soil-structure interaction coefficient (K) lead to a shift of the vulnerabilitycurve to the right (reduced vulnerability). Finally, it also appears that the building vulnerability curves aremore sensitive to KΔ values than to K values.