search

UMD     FPE



Probabilistic Models

The probabilistic category of fire models (also known as the stochastic) contains those models that ?do not make direct use of the physical and chemical principles involved in fires, but make statistical predictions about the transition from one stage of fire growth to another.? *

Probabilistic models can be further broken down into three subcategories: Network models, statistical models and simulation models.

1. Network models assign a probability to a specific event. A network model is good when you know your actions and outcomes, but you want to see which of them are the most probable. Network models also are very easy to read and relatively simple to construct.
Inputs: Probabilities of steps leading up to a final event
Outputs: Probability of final event
Types of Models: Decision trees, event trees and fault trees

2. Statistical models also use probabilities, but statistical models describe random variables by a probability distribution. Statistical models do not need a computer, but a spreadsheet is very helpful.
Inputs: Probability distributions of events
Outputs: Probability of final event
Types of Models:

3. Simulation models take an event and run the event as many times as a different combination of conditions is available. The most common type of simulation model deal with the evacuation of humans.
     Inputs: Evacuation routs, human sizes, probabilities of actions/reactions
     Outputs: Egress times, people flow, probabilities of exit selection
     Types of Models: Simulex, EXITT, EVACS, Allsafe

Much more information can be found in the NFPA and SFPE handbooks on these types of models.

* From Karlsson & Quintiere ?Enclosure Fire Dynamics? CRC Press 2000 pg 255 to 305

<< Previous      Next >>

[ Return to Research Home ]

 

   
Back to top      
FPE Home Clark School Home UMD Home