Fluctuations in Pedestrian Evacuation Times: Going One Step Beyond the Exit Capacity Paradigm for Bottlenecks

For safety reasons, it is important that the designs of buildings and public facilities comply with the guidelines compiled in building codes. The latter are often premised on the concept of exit capacity, i.e., the mean pedestrian flow rate through a bottleneck (at congestion). Here, we argue that one should duly take into account the evacuation time fluctuations when devising these guidelines. This is particularly true when the narrowing is abrupt and the crowd may behave competitively. We suggest a simple way to assess the extent of (part of) these fluctuations on the basis of the statistics of time gaps between successive escapes through the considered bottleneck, which in practice could be garnered by analysing recordings of future real evacuations or, perhaps, realistic drills (in the limits of what is ethically possible). We briefly present a test of the proposed strategy using a cellular automaton model and confirm its validity under some conditions, but also disclose some of its limitations. In particular, it may severely underestimate fluctuations in the presence of strong correlations in the pedestrians’ behaviours (while still performing better than only the mean capacity).