Fire Safety Science Digital Archive

Methods for estimating the time required for building evacuations range from simple calculations to highly complex computer-based models. Some of these more complex models attempt to represent the pre-movement behaviour of occupants, as well as their movement behaviours and physical movement characteristics, often using detailed rule-based algorithms, or artificial intelligence techniques. The quantification of such behaviours can be obscure and the models complex to use. Other models may ignore pre-movement behaviours. GridFlow has been developed to represent individual occupants in building spaces on a grid network, moving to exits through escape routes using an x,y coordinate and distance map method. Walking speeds, effects of crowding, obstacle avoidance, merging flows and flows through restrictions such as corridors, stairs and doorways are represented. Physical movement methods are therefore similar to those used in other models such as Crisp, Exodus, Simulex and others. GridFlow differs from other models in that premovement time and pre-movement-travel interactions are considered central to evacuation time modelling. In GridFlow, the pre-movement time is handled in the form of distributions obtained by direct measurements during monitored evacuations. The model therefore encompasses crucial elements of the evacuation process, without requiring detailed behavioural simulations. This is considered to be as informative as the most sophisticated models, using simple, transparent and easily verifiable behavioural inputs, derived from empirical data or specified and justified by the user.