Computer modelling

ENGR 403 – Fire Engineering

 Computer modelling
Suggested reading

z FEDG – Chapter 12, Fire Modelling with Computers

2
 Computer modelling

z A model can come in many shapes, sizes and styles

z A model is not the real world but merely a human construct to
help us better understand real world systems

3
 Computer modelling

z Deterministic – predict the outcome from a

specific set of input variables
z Stochastic – model that estimates the probability
distributions of potential outcomes by allowing for
random variation in one or more inputs over time
z Can combine a deterministic model with stochastic
selection of input variables

4
 Computer modelling
Stochastic models

Fire starts in
apartm ent Fire controlled sprinklers
1.000 0.903

No fire spreads through

Fire not controlled apartment fire separation
sprinklers to corridor
0.098 0.580

Fire spreads through No fire spread through

apartment fire separation corridor fire separation to
to cooridor one stairway Pressurization effective
0.420 0.580 0.675

Pressurization not
effective
0.325

Fire spread through

corridor fire separation to
one stairway Pressurization effective
0.420 0.675

Pressurization not
effective
0.325

5
 Computer modelling
Deterministic models
z Simple correlations
‰ Physics based correlations
‰ Simple curve fits

z Complex simultaneous equations

‰ Fire dynamics
z Zone models
z Computational Fluid Dynamics (CFD) models
‰ Human behaviour
z Node networks or detailed geometry
z Psychological and physiological assessment
‰ Structural fire (finite element)
z Heat transfer
z Mechanical performance

‰ Survey in 2003 found 168(!) computer fire models

6
 Fire dynamics models

7
 Computer modelling
Zone model assumptions
Zone Model
z Assumes two distinct homogeneous layers form

Hot upper layer

Plume Vent flow

Cool lower layer

z Implies local effects are unimportant or non-existent (at least

from a practical point of view)
‰ No spatially sensitive information
‰ No heating or cooling due to local flows such as heating and flows
due to the plume
‰ Uneven combustion due to varying oxygen levels throughout the
room not accounted for
8
 Computer modelling

z Fire specified by user (area, rate of heat release,

toxic gas yield etc.)
z Plume described by appropriate sub-model
z Geometry suited to being modelled as a rectangular
parallelepiped
0.50m10.00m
m

1.50m
1.50

6.00m
m

m
1.50

6.32

2.50m
m

Office
1.50

Main room
39 sq m

10.00m

9
 Computer modelling
Compartment fire data

2.40
15
31
2.10 60
300
1.80 3000-4200
Height Above Floor (m)

2-Zone
1.50

1.20

0.90

0.60

0.30

0.00
0 50 100 150 200 250 300
Temperature (C)

10
 Computer modelling

z B-RISK (formerly BRANZFIRE)

‰ Layer conditions (temperature, depth, gas species)
‰ Smoke flows through vents
‰ Smoke and heat detection activation
‰ Mechanical ventilation
‰ Glass fracture

11
 Computer modelling

FDS – Fire Dynamic Simulator

z Breaks the space into small
control volumes
z Solve the Navier-Stokes equations
for each volume
z Predicts the heat release rate
based on the amount of oxygen
consumed
z Provides detailed answers but can
have a high computational price
z Requires expert user

12
 Computer modelling

z FDS can deal with complex geometry

z Temperature, mass flows, gas species
in each individual computational cell
z Conduction heat transfer through
solids
z Radiation heat transfer
z Ignition of surfaces
z Mechanical smoke movement
z Detection and suppression systems

13
 Computer modelling
Grid dependence

300 mm grid 150 mm grid 25 mm grid

14
 Computer modelling

z FDS can plot many data

and display them as vectors,
surfaces etc

15
 Structural fire models

16
 Computer modelling

Load bearing
Construction Activity capacity

Room geometry
Fire Element geometry
Fuel load
Model Applied loads
Fire characteristics Structural
Model Mechanical properties

Fire thermal Thermal

exposure gradients

Element geometry
Heat transfer (from Welch and Lennon, 2001)

Model Thermal properties

Heat transfer coefficient

17
 Computer modelling

z FIRES-T3, THELMA – heat transfer only

z SAFIR, VULCAN – heat transfer and mechanical

performance

18
 Evacuation models

19
 Computer modelling

z Building space can be represented in two broad

ways
‰ Fine network – grid of equal size to represent a space
‰ Allows accurate representation of the geometry, but the
input can be tedious, with many of the programs requiring
CAD drawings

‰ Coarse network – represents each space as a node

‰ Nodes are connected via arcs, which represent the actual
connection of the building components
‰ Approach increases the speed of computation but does not
take into account overtaking or other local interaction
between individuals

20
 Computer modelling

z Occupants can be represented either globally or

individually
‰ Global perspective represents the occupants as a
homogeneous grouping of people, without considering
occupants might have different identities
‰ Individual perspective allows diversity of the
occupants. Assigns personal attributes to the occupants,
allowing individual movement and decision-making etc.

21
 Computer modelling

z Simulex, Exodus, STEPS, FDS-Evac (fine node,

individual perspective)
z EvacuatioNZ (coarse node, individual perspective)

22
 Computer modelling

z Applications
‰ Design calculations
‰ Post-incident analysis
‰ Development or extension of simple correlations
9.0 225
1200 2.5
8.0 200

7.0 without smoke control 175

Layer temperature

layer temperature (deg C)

6.0
with smoke control
150
Lining temperature
1000
layer height (m)

without smoke control

5.0 with smoke control 125
Bedroom layer depth 2.0
4.0 100

800

Temperature (˚C)
3.0 75

Layer depth (m)

2.0 50 1.5
1.0 25

0.0 0
600
0 300 600 900 1200 1500
time (s)
T17 with RHR x 4 at t = 1560 s onwards then 1.0
FAST rise to window ventilation limit at t = 1735 s
400 plus.

Compartment : Half window breaks at 1735 s and

fully breaks at flashover. 0.5
200
O2 contribution not considered.

0 0.0
0 300 600 900 1200 1500 1800 2100
Time (secs)

23
 Computer modelling

z General guidance
‰ The model must be shown to be valid for the calculation
being performed
‰ The user must have sufficient understanding of the model
and the phenomenon being modelled
‰ Details of model name, version, input files etc. need to be
provided as part of the documentation
‰ The sensitivity of the model to the inputs must be
investigated

See Section 12.2 in the FEDG for more details.

24