A binary logistic regression model to evaluate backdraft phenomenon
Ryan Falkenstein-Smith, Thomas Cleary
A technique to predict backdraft phenomenon using a binary logistic regression model is presented. The model is established from time-averaged temperature, global and local equivalence ratios, and oxygen concentration measurements obtained in a series of backdraft experiments conducted at the National Fire Research Laboratory at the National Institute of Standards and Technology. The experiments utilized methane and propane fires of different sizes in a reduced-scale enclosure to create conditions conducive to a backdraft phenomenon. Time-averaged measurements estimated immediately before an anticipated backdraft were observed to vary with the duration of the total fuel flow time into the compartment. The established model's accuracy was found to improve with the inclusion of all time-averaged measurements as opposed to fewer components.
Suppression, Detection and Signaling Research and Applications Conference
and Cleary, T.
A binary logistic regression model to evaluate backdraft phenomenon, Suppression, Detection and Signaling Research and Applications Conference
(SUPDET 2022) , Atlanta, GA, US, [online], https://tsapps.nist.gov/publication/get_pdf.cfm?pub_id=935430
(Accessed December 11, 2023)