Accidental releases occurring in industrial platforms or during transportation of hazardous materials can entail the dispersion of toxic gas clouds. In case of such an event, the best protection strategy for people is to identify a shelter in a nearby building and stay in this room until the toxic cloud has finally been swept off. In addition to seeking refuge in an airtight room, this strategy called 'passive shelter-in-place' also includes closing all external openings and turning off all mechanical ventilation systems and openings. Following the AZF chemical accident (Toulouse, 2001, 30 deaths), a French law was adopted in 2003 that can compel public and private building owners to adopt such a shelter-in-place strategy. To prove that the shelter airtightness is sufficient and that the occupants will not be exposed to irreversible effects, simulations are required using for instance the modeling tool CONFINE. Originally developed by CETE de Lyon, this software is a pressure code able to model the infiltration of a pollutant inside a 3 zone - building (shelter, attic and rest of building). This paper aims at giving an overview of CONFINE (governing equations, modeling hypotheses...) and will illustrate its application on one example of shelter-in-place strategy for a public building. This paper will also present some unexpected results about the impact of wind velocity on shelter-in-place effectiveness. If a higher wind velocity results in a better dilution of the toxic gas outdoor, this situation does not necessarily lead to a lower concentration inside the room, and can conduce to more severe shelter airtightness.
