A Bayesian network model for fire development and occupant response within dwellings

The provision of fire and rescue services is hugely complex due to the sheer number of different fire scenarios which can develop. Not only are there diverse types of locations, for example dwellings, public buildings, factories, etc., but there are also different circumstances within each type of location. This study focuses on dwellings, where variations arise in terms of geographical location, fire safety arrangements, characteristics of occupants, activities of occupants, among others. As for the occurrence of fire itself, each incident will be unique in terms of time of day, type of fire, state of occupants, fire cues, etc. What all these variations signify is that the potential magnitude of the next fire event and its consequences are generally unpredictable. Because of complicated scenarios, unpredictability of outcomes, and high frequency of incidents, Fire and Rescue Services have to be both capable and flexible in operation; however finding the optimal way of providing emergency cover and minimizing risk is a complicated task in a changing world. This study aims to contribute towards this task by assessing the case for dwelling fires in the UK The concept of probabilistic modeling under uncertainty within the context of fire and rescue through the application of the Bayesian Network (BN) technique is presented in this paper. BNs are capable of dealing with uncertainty in data, a common issue within fire incidents, and can be adapted to represent various fire scenarios. A model has been built to represent fire development within dwellings from the point of ignition through to extinguishment. The model is broken down into four parts; this paper presents parts I and II which deal with “initial fire development” and “occupant response and further fire development” respectively.