Overview

Use of statistical data to choose likely fire scenarios can be challenging. Judgment is involved in selecting the categories that define alternative fire scenarios.

First, the dimensions and factors used to code and describe historical fires in available databases are not always the same dimensions and factors required as input data for a fire hazard analysis model. A translation is required, and every translation involves an assumed or demonstrated relationship between fire experience data characteristics and desired input data characteristics. These relationships must be substantiated as much as possible.

For example, it might not be valid to assume that all arson fires are fast or severe (i.e., have a steeply rising heat release rate or a high peak rate of heat release). Most arson fires do not involve the use of iccelerants and are started by juveniles rather than individuals sophisticated in fire ignition. For fire scenario purposes in a fire hazard analysis, an arson fire might behave like an ordinary trash fire. By contrast, it might be valid to assume that all fires starting with the ignition of flammable liquids, whether accidental or deliberate, show the heat release rate curves associated with those products.

Second, when selecting likely fire scenarios, the analyst might identify fire scenarios that collectively account for a large share, preferably most, of the relevant fires. However, fire hazard analysis calculations require input data with so many characteristics defined that no one fire scenario can be expected to account for more than a tiny percentage...