Fire Extinguishing Performance Assessment

Study Notes

Time elapsed between the starting and extinguishing of a band fire is a key performance metric.
This performance is measured with respect to the probability (P2) that the duration is lower than 60 minutes.
Another performance metric is the probability (P3) of either fire extinction in due time (60 minutes) or the cooling system working.

PNs models are used to compute probabilistic quantities related to the described scenario.
The syntax of PNs models includes:
- # to indicate the marking of a place (e.g., #i is the marking of place i).
- jets to indicate the number of tokens.
- !! to introduce a list of variable assignments.
- ?? to specify a list of conditions that must be verified for the transition to be valid.
- drc d to indicate Dirac's law of duration d.
- @ (k) (e1, e2, …, en) to indicate a k out-of n logic (where ei are Boolean expressions).

The interaction between PNs models is achieved through variables (guards and assertions).
Examples of PNs models are presented, including condensate LOC, gas detector 1 (GD1) behavior, and gas detection system.

Safety bounder establishment has an efficiency of 90% and is delayed by 3 minutes.
Ignition sources prevention has an efficiency of 91% and is also delayed by 3 minutes.
Foam film establishment depends on water pumps and emulsifier.

There are 5 water pumps (3 on-site and 2 off-site).
The starting of water pumps can be performed in three ways:
- Instantaneously if the ESD push button works (probability of failure = 0.044) and the pumps work (with probability of failure for each pump = 0.019).
- After 3.50 minutes if the whole operational team (team 3) is available.
- After 7 minutes if the team is not available.

Monte Carlo simulation is used to assess performance parameters.
The principle of Monte Carlo simulation is to realize a great number of histories and perform classical statistics on the results.
Uncertainty assessment is achieved by means of Monte Carlo sampling, among other approaches.
Monte Carlo sampling method is used to handle uncertainty in model results.