The 13% is not an error rate. This is the uncertainly of the input variables and the uncertainly associated with the measured variables. There are references that provide the details of how this data was generated. The plots provide a means to visually assess if the predictions are under-estimated or over-estimated. The 45 degree line is for no difference in the measured or predicted HGL. If the data points fall into the lines of uncertainty, then one can say that there is no different between the measured and predicted values.

No. These models are very different. The combustion and fire sciences deal with very complex and highly coupled fire phenomenon.The models are constructed to provide tractable solutions by making assumptions that allow a model to be developed or a numerical solution to be developed to solve a complex set of equations. These assumptions produce some limitations and computational expense.

Most of the correlations were developed at a time when measurement and computational power were very limited. Fast forward to today, and we have very powerful computers that can practically approximate the solution to a complex set of non-linear equations required to model the fire's interactions with compartments.

For example, the HGL is a single point measurement and prediction for the individual correlations. The zone model provides an average temperature of the upper layer, and the CFD models provide estimates of discrete points in the HGL depending on the number of grid cells.

Douglas J. Carpenter, MScFPE, CFEI, PE, FSFPE
Vice President & Principal Engineer
Combustion Science & Engineering, Inc.
8940 Old Annapolis Road, Suite L
Columbia, MD 21045
(410) 884-3266
(410) 884-3267 (fax)
www.csefire.com