Four plume algorithms used by three zone type building fire simulators are evaluated against experiential data of Steckler and Nakaya. Significant differences in the room flow predictions are found with even the best performing plume algorithms predicting flows well below the measured values. Differences in plume behavior is attributed to (1) the background noise (turbulence) present when the data used in formulating the algorithms was collected, and (2) the inability of the plume algorithms to easily simulate the effect of plume blowing. The behavior of the McCaffrey plume in situations were the over-fire region dominates the plume flow is discussed.
