See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/44083063

FIERAsystem Theory Documentation - Smoke Movement Model

Article · January 2002

DOI: 10.4224/20386250

CITATIONS READS
3 45

4 authors, including:

Ahmed Kashef Noureddine Benichou

National Research Council Canada National Research Council Canada
126 PUBLICATIONS   509 CITATIONS    225 PUBLICATIONS   768 CITATIONS   

SEE PROFILE SEE PROFILE

George Hadjisophocleous
Carleton University
214 PUBLICATIONS   1,147 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Mid-Rise wood construction project View project

NEWBuilDS View project

All content following this page was uploaded by George Hadjisophocleous on 22 September 2014.

The user has requested enhancement of the downloaded file.

NRC Publications Archive (NPArC)
Archives des publications du CNRC (NPArC)

FIERAsystem Theory Documentation - Smoke Movement Model

Fu, Z.; Kashef, A.; B�nichou, N.; Hadjisophocleous, G. V.

Web page / page Web

http://nparc.cisti-icist.nrc-cnrc.gc.ca/npsi/ctrl?lang=en
http://nparc.cisti-icist.nrc-cnrc.gc.ca/npsi/ctrl?lang=fr

Access and use of this website and the material on it are subject to the Terms and Conditions set forth at
http://nparc.cisti-icist.nrc-cnrc.gc.ca/npsi/jsp/nparc_cp.jsp?lang=en
READ THESE TERMS AND CONDITIONS CAREFULLY BEFORE USING THIS WEBSITE.

L’accès à ce site Web et l’utilisation de son contenu sont assujettis aux conditions présentées dans le site
http://nparc.cisti-icist.nrc-cnrc.gc.ca/npsi/jsp/nparc_cp.jsp?lang=fr
LISEZ CES CONDITIONS ATTENTIVEMENT AVANT D’UTILISER CE SITE WEB.

Contact us / Contactez nous: nparc.cisti@nrc-cnrc.gc.ca.

Smoke Movement Model

Fu, Z.; Kashef, A.; Bénichou, N.;

Hadjisophocleous, G.

IR-835

http://irc.nrc-cnrc.gc.ca/ircpubs
FIERAsystem Theory Documentation

SMOKE MOVEMENT MODEL

Zhuman Fu, Ahmed Kashef, Noureddine Benichou,

and George Hadjisophocleous

Internal Report No.

Date of Issue: March 2002

This is an internal report of the Institute for Research in

Construction. Although not intended for general distribution, it
may be cited as a reference in other publications.
 THEORY OF SMOKE MOVEMENT MODEL

Zhuman Fu, Ahmed Kashef, Noureddine Benichou and George

Hadjisophocleous

TABLE OF CONTENTS

TABLE OF CONTENTS...............................................................................................i
SUMMARY.................................................................................................................... 1
NOMENCLATURE ...................................................................................................... 2
1. INTRODUCTION..................................................................................................... 7
1.1 Assumptions and Limitations................................................................................ 8
2. DERIVATION OF GOVERNING EQUATIONS ............................................... 10
3. COMBUSTION AND CHEMISTRY.................................................................... 15
3.1 Heat Release Rate................................................................................................ 15
3.2 Combustion Chemistry........................................................................................ 17
3.3 Calculation of ( )O or ( )F ............................................................................ 20
4. FLUID FLOW ......................................................................................................... 22
4.1 Plume Entrainment.............................................................................................. 22
4.2 Door/Window Vent Flow.................................................................................... 23
4.3 Ceiling Vent Flow ............................................................................................... 27
4.4 Mechanical Ventilation ....................................................................................... 29
4.5 Stair Shaft Flow................................................................................................... 31
4.6 Species Concentration ......................................................................................... 33
5. HEAT TRANSFER ................................................................................................. 34
5.1 Conduction Heat Transfer ................................................................................... 34
5.2 Convection Heat Transfer in Non-Fire Rooms ................................................... 34
5.3 Ceiling Jet-Induced Heat Transfer ...................................................................... 36
5.4 Convection Heat Transfer in Fire Rooms ........................................................... 38
5.5 Radiation Heat Transfer ...................................................................................... 38
5.5.1 Derivation of Equations ............................................................................... 38
5.5.2 Calculation of View Factors......................................................................... 41
5.5.3 Gas Emissivity.............................................................................................. 42
6. NUMERICAL METHOD....................................................................................... 44
6.1 Introduction ......................................................................................................... 44
6.2 Numerical Properties of the ODE System........................................................... 46
6.2.1 Stiff Equations.............................................................................................. 47
6.2.2 BDF Method................................................................................................. 47

i
 6.2.3 Error Estimation and Automatic Control of Step Size and Order................ 48
6.3 Solution of the Constant Temperature Compartments........................................ 49
6.4 Solution of the Physical Sub-models .................................................................. 50
6.4.1 Conduction ................................................................................................... 50
6.4.2 Ceiling Jet Heat Transfer.............................................................................. 53
7. VERIFICATION EXAMPLES.............................................................................. 54
7.1 Standard Single Compartment ............................................................................ 54
Analysis of the Results.......................................................................................... 54
7.2 One-Compartment with Mechanical Ventilation ................................................ 56
Analysis of the Results.......................................................................................... 58
7.3 Two Compartments ............................................................................................. 61
Analysis of the Results.......................................................................................... 61
8. REFERENCES ........................................................................................................ 64

List of Tables
Table 2.1 A List of the Governing Ordinary Differential Equations ................................ 13
Table 7.1 Zone Model Input Data for Single Compartment Example .............................. 55
Table 7.2 Comparison of Model Prediction with Reference [59] ....................................... 56
Table 7.3 Basic Input Data of the Test Facility................................................................. 58
Table 7.4 Results of Test 1 with Mechanical Ventilation ................................................. 59
Table 7.5 Results of Test 2 with Mechanical Ventilation ................................................. 60
Table 7.6 Results of Test 3 with Mechanical Ventilation ................................................. 60
Table 7.7 Zone Model Input Data for Two-Compartment Example................................. 61

List of Figures
Figure 4.1 Basic vertical-vent (door/window) configuration............................................ 24
Figure 4.2 Possible neutral planes in an opening .............................................................. 25
Figure 4.3 Basic horizontal-vent (ceiling vent) configuration .......................................... 27
Figure 4.4 Schematic of the mechanical ventilation opening ........................................... 30
Figure 4.5 Schematic of an effective ceiling vent for a stair shaft.................................... 31
Figure 5.1 Schematic of the radiation geometry ............................................................... 39
Figure 5.2 A schematic of flame radiation to a rectangular surface ................................. 42
Figure 6.1 Schematic Chart of Numerical Calculation ..................................................... 45
Figure 7.1 Schematic of the Test Facility ......................................................................... 57
Figure 7.2 Predicted and Experimental Results for Pressure Difference Between the Burn
Room and Corridor Near the Ceiling ........................................................................ 62
Figure 7.3 Predicted and Experimental Results for Averaged Gas Temperature of Burn
Room (BR) and Corridor (CR).................................................................................. 62
Figure 7.4 Predicted and Experimental Results for the Interface Heights of the Burn
Room (BR) and Corridor (CR).................................................................................. 63

ii
SUMMARY

This report describes a two-zone model developed at the National Research Council of
Canada (NRC) to predict fire growth and smoke production and movement in a multi-
compartment building. The development of this model is primarily intended to help
evaluate the risk from fires in buildings. This report describes the theoretical physical
models, numerical methods, and also presents some verification examples.

The report describes the derivation of the whole set of the governing ordinary differential
equations (ODE). The selected independent governing variables for each compartment
are pressure, enthalpy of upper layer, and mass of upper and lower layers. The
implemented fire physical sub-models are described; including combustion and chemistry
model, fluid flow model and heat transfer model. The combustion and chemistry model
includes the calculation of heat release rate and species production and consumption
rates, while the fluid flow model includes the calculation of plume entrainment,
door/window type vent flow, ceiling vent flow, mechanical ventilation, stair shaft flow,
and species transport. The heat transfer model includes the calculation of heat
conduction, heat convection, ceiling jet-induced heat convection, and radiation heat
transfer.

The numerical method used to solve the system of governing equations is presented.
Backward differentiation formulas (BDFs) method has been used to solve the stiff ODE
system. Some details about the ODE solver are presented, including the basic properties
of stiff equations, basic algorithm of the BDF method and the control mechanism of
precision criteria, step size and order. The method of solving the constant temperature
compartments is also presented. Besides, the method applied to solve the one-
dimensional conduction partial differential equation is described, including discrete
equations and Tri-Diagonal Matrix Algorithm (TDMA) method. The numerical method
of calculating ceiling jet-induced convection heat transfer is also given.

Three sets of experimental data for single and two-compartment fire tests with and
without mechanical ventilation are compared to the predictions of the model. The
comparison shows that the model predictions compare favorably with the experimental
data, especially the upper layer gas temperature, interface height, neutral plane height,
and vent flow rate. The model, however, over-predicts the floor incident radiant heat flux
and the lower wall surface temperature, and under-predicts the lower layer gas
temperature.

1
NOMENCLATURE

A Area
Ae i Effective orifice area on the ith floor
Aint Area of interface
AS Cross-sectional area of stair shaft
AV Vent area
Aw Surface area of a wall, ceiling or floor
Bi I Biot number of the inside surface of a wall
Bi O Biot number of the outside surface of a wall
C Specific heat
Cconv A coefficient in the convection heat transfer formula
CCV Coefficient of ceiling vent flow
CE1, CE2 Coefficients in McCaffrey’s plume entrainment correlation
CLOL Lower oxygen limiting coefficient
CP Specific heat at constant pressure
Csoot Volume fraction of soot in smoke layer
CV Specific heat at constant volume
Cvent Coefficient of door/window type vent flow
De Equivalent diameter
di Density of people on the stairs between the ith floor and (i-1)th floor
E Internal energy
Fkj Configuration factor from surface k to surface j for radiation
Fo Fourie number
FOP Fuel to oxygen ratio in fire plume
FOS Stoichiometric fuel to oxygen ratio
Fr Froude number
g Gravitational acceleration
Gr Grashof number
H Vertical distance between the fire source and ceiling
h Heat transfer coefficient
~
h Characteristic heat transfer coefficient
Hi Gas enthalpy of layer I
hi Height of stair shaft between the ith floor and (i-1)th floor
hIC, hOC Convection heat transfer coefficients of the inside and outside surface of a
wall, ceiling or floor
(∆H)F Heat of combustion per unit mass of fuel
(∆H)O Heat of combustion per unit mass of oxygen

2
∑ H& si Rate of net enthalpy gain of layer i by mass flow across its boundary
K Conductivity
k Friction pressure loss coefficient
L Thickness of a wall, ceiling or floor
Le Equivalent mean beam length of smoke layer for radiation
m Mass
m& AV Average vent mass flow rate
m BF Mass of the burned fuel
m BFmain Main part of the burned fuel
m BFmin Minimal part of the burned fuel
mC Mass of carbon in fuel
m CO Mass production of CO
m CO 2 Mass production of CO2
m CRP Mass production of carbon-related products (Soot+CO+CO2)
m& e Mass entrainment rate
m& Fan Specified mass exhaust rate of the fan
mH Mass of hydrogen in fuel
m H 2O Mass production of water
& max
m Maximum smoke exhaust rate from the upper layer
& ML
m Mass flow rate exhausted from lower layer
& MU
m Mass flow rate exhausted from upper layer
mO Mass of oxygen in fuel
m O2 Oxygen consumption from air in combustion
m PF Mass of fuel pyrolysis
m PFmain Mass production of the main part of fuel
m PFmin Mass production of the minimal part of fuel
m soot Mass production of soot
m tox1 Mass of the first kind of toxic species of mPFmin
m tox 2 Mass of the second kind of toxic species of mPFmin
m tox 3 Mass of the third kind of toxic species of mPFmin
m TUCF Mass of the total unburned combustible fuel
m UF Mass of the unburned fuel
m UFmain Main part of the unburned fuel
m UFmin Minimal part of the unburned fuel

3
∑ m& e Net mass flow rate entering a layer by plume entrainment

∑ m& e, K
Net mass flow rate of species K entering a layer by plume entrainment

∑ m& si
Net mass gain rate of layer i by mass flow through its boundary

∑ m& V
Net mass flow rate entering a layer by vent flow

∑ m& V ,K
Net mass flow rate of species K entering a layer by vent flow
NS Number of species
Nu Nusselt number
P Pressure
Pb Pressure difference at the bottom position
PCO 2 Partial pressure of CO2
Pg Gas pressure
PH 2O Partial pressure of water vapour
Pr Prandtl number
Pt Pressure difference at the top position
P1(Z), P2(Z) Pressures of the two sides of a vertical-vent at height Z
∆PF Compartment floor pressure difference relative to ambient
∆P(Z) Pressure difference between the two sides of a vertical-vent at height Z
Q Heat release rate
Q′ Equivalent heat release rate considering upper layer effect on ceiling jet
Qc Convection heat release rate
Qceil, qceil Convection heat loss and flux to ceiling surface
Qconv, qconv Convection heat loss and flux to a surface of a wall, ceiling or floor
Q *eq Dimensionless heat release rate at the interface
QF Heat release rate directly radiated out of the fuel
Q *H Dimensionless heat release rate
qi Net radiation heat transfer rate of surface i
q IF Average radiant incident heat flux to a compartment floor
q IR Net radiation heat flux to the inside surface of a wall, ceiling or floor
QN Nominal heat release rate of the fuel
q OR Net radiation heat flux to the outside surface of a wall, ceiling or floor
∑Q si Net heat gain rate of layer i by heat transfer or combustion
R Gas constant
r Radial distance from the stagnation point on the ceiling surface
RC Radius of the circular equivalent ceiling surface
ReH Reynolds number
S& i Net energy gain rate of layer i as a source term
T Temperature
t Time

4
 t Time step size
Ta Ambient temperature
Tad Adiabatic ceiling jet temperature
Tceil Ceiling surface temperature
TFLR Compartment averaged floor interior surface temperature
Tg Gas layer temperature
TI Assumed interface temperature
Tw Surface temperature of a wall, ceiling or floor
TWL Compartment averaged lower wall interior surface temperature
TWU Compartment averaged upper wall interior surface temperature
T(x, 0) Initial temperature profile of a wall, ceiling or floor
V Volume
VEX, MX Maximum exchange volume flow rate
V& Specified volume exhaust rate of the fan
Fan
VH Volumetric flow rate from high to low-pressure side
VL Volumetric flow rate from low to high-pressure side
V& Volume flow rate exhausted from lower layer
ML

V& Volume flow rate exhausted from upper layer

MU

VN Net volumetric flow rate

x Spatial coordinate in heat conduction equation
x Space step size
YK Mass fraction of species K
YLOL Lower oxygen limiting mass fraction
YO 2 Oxygen mass fraction
Z Height
Zceil Ceiling height
Ze Plume entrainment height
Zext Mechanical ventilation opening extension
Zfire Height of burner’s surface
Zf1, Zf2 Floor elevation of the two compartments connected by a vertical-vent
ZI Height of interface of upper layer and lower layer
ZN Neutral plane elevation of a vertical-vent flow
ZS Thickness of smoke layer
Zsi Sill height
Zso Soffit height
Z12 Entrainment height of vent flow-induced plume

Greek Symbols
Γ Perimeter of stair shaft
ρ Mass density
β Tread coefficient of stair shaft

5
γ Ratio of specific heats
γC/CRP Mass ratio of the carbon in the CRP products to the total CRP products
γCO Mass ratio of CO production to carbon-related products (CO+CO2+Soot)
γsoot Mass ratio of soot production to carbon-related products
ε CO 2
Emittance of CO2
εD Emittance reduction resulting from spectral overlap
εg Gas total emittance
εH O 2
Emittance of water vapour
εj Emissivity of surface j
εsoot Emittance of soot
σ Stephen Boltsman constant
ν Kinematic viscosity
δkj Kronecker function
τkj Radiation transmittance from surface k to surface j
αkj Radiation absorptance from surface k to surface j
φ Equivalence ratio
χC Mass fraction of carbon in the fuel
χH Mass fraction of hydrogen in the fuel
χtox1 Mass ratio of the first kind of toxic species production to fuel pyrolysis
χtox2 Mass ratio of the second kind of toxic species production to fuel pyrolysis
χO Mass fraction of oxygen in the fuel
Π Dimensionless pressure difference
Ω& Net mass gain rate of layer i by mass flow through its boundary
i

Subscripts
U, L Upper, lower layer
H, L Higher or lower pressure side
FL Flooding condition
T, B Top, bottom sides
1, 2, F Upper surface, lower surface, flame in radiation formula

6
1. INTRODUCTION

Computational modeling of fire in a compartment can be achieved either using a zone

modeling approach or a field modeling method. In this study, the zone modeling
approach was used in which the gas within each compartment is generally divided into
one, or a few, control volumes (zones), and for each zone, the physical parameters such
as gas temperature and species concentrations are assumed to be spatially uniform. Then,
from the mass and energy conservation principle, as well as the ideal gas law, a set of
ordinary differential equations (ODE) is derived. In this type of model, the physical
details of the gas within a zone are not considered, however mass and energy transport
between zones has to be calculated by modeling the relevant fire sub-processes:
combustion and chemistry, fluid flow and heat transfer. [1-6]

Zone models may be grouped into two types based on the number of control volumes
(zones) in each compartment: one-zone models and two-zone models. One-zone models
are widely used in the analysis of post-flashover fires, as well as smoke movement in the
compartments remote from the fire room (network models). Two-zone models divide the
gas in a compartment into two distinct zones: an upper, higher temperature zone and a
lower, lower-temperature zone. These zones are a result of buoyancy-induced thermal
stratification. Two-zone models can be used to analyze pre-flashover fires, and some
models of this type have been developed. [7-9] A comprehensive review of existing fire
models can be found in Friedman. [7]

The one-zone modeling concept dates back to the work of Kawagoe, who developed a
single-zone approach for analyzing a post-flashover fire. [10] This approach was the basis
of the development of a series of single-zone post-flashover fire models. [11] The
application of the two-zone concept was pioneered by Thomas, Hinkley, Theobald, and
Simms, who constructed a steady state, two-layer model for calculating the flow of
smoke through roof vents.[11-13] Before the mid-1970s, however, fire modeling research
was focussed on post flashover fires; i.e., fully developed fires, using the one-zone
method.[14]

Quintiere has mentioned that the two-zone fire modeling of pre-flashover fires emerged
in the mid-1970s with the publication of a basis for the zone model approach by Fowkes
in his work with Emmons.[5, 15] Almost simultaneously, several zone computer models
were produced by Quintiere, Pape and Walterman and Mitler working with Emmons.[5, 16-
18]
In the late 1970s the first Harvard model and code was completed by Emmons and
Mitler, which might be the first comprehensive room fire model for one compartment.[18,
19]
However, the first true multi-compartment model of this type was formulated by
Tanaka.[1, 20] Zone modeling work continued, especially at NIST, where models such as
FIRST, FAST, CCFM and CFAST were developed.[21-23] Jones, Jones and Forney, and

7
Peacock and Forney developed the CFAST model based on FAST and CCFM, in which
the governing equation set is formulated to allow the actual physical phenomena to be
taken as source terms. This allows the greatest flexibility in modifying the terms that are
appropriate to the problem being solved.[1-3] In the CFAST model, the conservation
equations are solved in their original differential form, and the pressure is not assumed to
be steady state, nor the lower layer temperature to be at ambient conditions.

NRC has been conducting research work on fire risk evaluation of buildings which has
resulted in a comprehensive fire risk evaluation model FiRECAMTM for residential and
office buildings.[24-26] Currently, research on fire risk evaluation of light industrial
buildings is being undertaken, which is used to develop a system model called FIERA.[27]
As part of the FIERAsystem, the model described in this report will be used to calculate
smoke movement in industrial buildings, as well as to improve the calculation of smoke
movement in office and residential buildings.

Based on similar concepts used by CFAST, a two-zone model has been developed to
predict smoke and fire development in buildings. Both the model and interface are coded
using MS Visual Basic Version 5. This model uses a new form of the governing ordinary
differential equations and fire sub-models. The following sections present the governing
equation set for this model and describe the fire sub-models, including fire growth, fluid
flow, and heat transfer models.

1.1 Assumptions and Limitations

The developed smoke model uses mainly the principal laws of thermodynamics
(conservation of mass, energy, and momentum) to simulate the important time-dependent
phenomena involved in fires. However, in areas where knowledge is not available,
empirical correlations (when data are available) or even reasonable guesses were used.
The user should be aware of these limitations. Following is an overview of the
assumptions and limitations implicit with the current model:
• The model does not include a fire growth model. Rather, the fire is specified by the
user in terms of rates of energy and mass released by the burning objects. Such data
can be obtained from measurements taken in large or small-scale calorimeters. It is
the responsibility of the user to provide the appropriate rates, which depend on the of
problem under consideration.
• The entrainment coefficients are empirically determined values. These values are
mainly relevant for models up to a three-compartment case. For a larger number of
rooms, more data are needed. In these situations, the entraiment values may
introduce some effects on the fire plume or on hot gases flow from fire origin room to
the neighboring rooms.
• Each layer is assumed to be of uniform temperature, density, and species
concentrations. This assumption implies that the moment the hot gases enter a
compartment, the model calculates a hot gas layer that covers the entire ceiling. If the

8
hot gases entered a large room from one end, the model would predict that the
temperature at the far end would immediately increase. In other words the transient
time to fill the area is not considered. In real cases, there would be a delay time as the
hot gases travel across the large room.

9
2. DERIVATION OF GOVERNING EQUATIONS

Following the two-zone modeling concept, each compartment is divided into two zones.
For each zone, the mass, internal energy, enthalpy, density, temperature and volume are
denoted as mi, Ei, Hi, ρi, Ti and Vi, respectively, where i=U refers to the upper layer, and
i=L refers to the lower layer. The thermodynamic pressures for the upper and lower
layers, PU and PL are assumed to be identical and are denoted as P. Using thermodynamic
relations and definitions as well as the ideal gas law, the following equations are given:

mi
ρi = (2.1)
Vi

Ei = CV mi Ti (2.2)

H i = E i + PVi (2.3)

P = ρi RTi (2.4)

V = VU + VL (2.5)

CP and CV, specific heats at constant pressure and volume, are assumed to be constant for
the gas in the upper and lower layers, and the following relation exists:

γ = C P / CV , R = C P − CV (2.6)

where R is the gas constant. Applying Equations (2.1) through (2.5) to both the upper and
lower layers in a compartment results in 9 independent algebraic equations with 13
variables. To close the equation set, four additional independent equations are needed,
which can be obtained by applying mass and energy conservation principles to each zone.
The resulting equations are as follows:

Mass conservation:

dmi
= ∑ m& si (2.7)
dt

Energy conservation (First Law):[28]

dEi dV
+ P i = ∑ H& si + ∑ Qsi (2.8)
dt dt

10
where ∑ m& si is the net mass gain rate of layer i by mass flow across the boundary,
∑ H& si is the rate of the net enthalpy gain of layer i by mass flow across the boundary,
∑Q si is the rate of net heat gain of layer i by heat transfer or combustion.

Thus, the whole equation set has been closed. To facilitate the solution, the above system
of ODEs can be converted into other forms as given below.

Let us denote:

& = ∑ m&
Ω (2.9)
i si

S& i = ∑ H& si + ∑ Qsi (2.10)

& and S& are considered to be source terms.

Following the approach used in CFAST, Ω i i

From Equations (2.7) and (2.8) we have:

dmi &
= Ωi (2.11)
dt

dEi dV
+ P i = S&i (2.12)
dt dt

dP
Adding the Term Vi to the two sides of Equation (2.12), and considering Equation
dt
(2.3), we get:

dH i dP
= S&i + Vi (2.13)
dt dt

Adding the upper and lower layer versions of Equation (2.13), we have:

d (HU + H L ) dP
= ( S& L + S&U ) + V (2.14)
dt dt

From Equations (2.1), (2.2), (2.3), (2.4) and (2.6), we have:

CP
Hi = Vi P (2.15)
R

11
Adding the upper and lower layer versions of Equation (2.15), and substituting into
(2.14), we get:

dP γ − 1 &
= ( SU + S& L ) (2.16)
dt V

Differentiating equation (2.15), we have:

dH i C P dP dV
= (Vi +P i) (2.17)
dt R dt dt

Substitute Equation (2.17) into (2.13), we obtain:

dVi 1  dP 
=  (γ − 1) S& i − Vi  (2.18)
dt γP  dt 

From Equations (2.1), (2.11) and (2.18), and applying the quotient rule to
 dρ i d  mi  
  
 dt = dt  V   , we obtain:
  i 

dρ i 1  & & T ) + Vi dP 
=  − ( S i − C P Ω (2.19)
γ − 1 dt 
i i
dt C P TiVi 

 dT d  P 
Similarly, applying the quotient rule to  i =    , from Equations (2.4) and

 dt dt  Rρ i 
(2.19), we obtain:

dTi 1  & & T ) + V dP 

=  (Si − C P Ψ (2.20)
dt C P ρ iVi 
i i i
dt 

12
 Table 2.1 A List of the Governing Ordinary Differential Equations

Variables Equations
Mass dmi &
= Ωi
dt
Pressure dP γ − 1 &
= ( SU + S& L )
dt V
Volume dVi 1  dP 
=  (γ − 1) S& i − Vi 
dt γP  dt 
Density dρ i 1  & & T ) + Vi dP 
=  − ( S i − C P Ω
γ − 1 dt 
i i
dt C P TiVi 
Temperature dTi 1  & & T ) + V dP 
=  (Si − C P Ψ
dt C P ρ iVi 
i i i
dt 
Internal Energy dEi 1  & dP 
=  S i + Vi 
dt γ dt 
Enthalpy dH i dP
= S&i + Vi
dt dt

The complete list of equations derived is shown in Table 2.1, which includes equations of
mass, pressure, volume, density, temperature, internal energy and enthalpy. To close the
system of equations, we only need four independent differential equations. There are
many ways to select four independent governing variables from the above list. For
example, CFAST selects P, VU, TU and TL as its governing variables. In this model, the
equations of pressure, upper layer enthalpy, upper layer mass and lower layer mass are
used as follows:

dP γ − 1 &
= ( SU + S& L ) (2.21)
dt V

dH U dP
= S&U + VU (2.22)
dt dt

dmU &
=Ω U (2.23)
dt

13
dm L &
= ΩL (2.24)
dt

In the literature[1, 3, 29], the two terms on the right side of Equation (2.8) were combined as
one term, called enthalpy or total enthalpy. This could result in a little confusion between
this term and the real thermodynamic enthalpy term. Thus in this derivation, the equation
for thermodynamic layer enthalpy has been given as a complementary to the table
“Conservative Zone Modeling Differential Equations” appearing in references [1, 3, 29],
although Hi is directly proportional to Ei by in this case.

14
3. COMBUSTION AND CHEMISTRY

3.1 Heat Release Rate

The heat release rate of combustion can be obtained by:

Q N = m& PF (∆H ) F (3.1)

where QN is the nominal heat release rate,

(∆H ) F is the effective heat of combustion per unit mass of fuel which is obtained
from experimental data,
m& PF is the mass pyrolysis rate.

Following the oxygen consumption principle, the oxygen consumption rate needed to
release energy of QN assuming no oxygen in the fuel, can be obtained using: [30, 31]

QN
m& O2 = (3.2)
(∆H ) O

where (∆H ) o is the heat of combustion per unit mass of oxygen. For complex fuels, it
could be taken as 13.2 MJ/kg[32], while, for simple chemical formula fuels, it can be
obtained from the heat of combustion per unit mass of fuel.

The stoichiometric fuel to oxygen ratio, FOS, is:

m& PF (∆H ) O
FOS = = (3.3)
m& O2 (∆H ) F

The actual fuel to oxygen ratio in the fire plume is:

m& PF
FOP = (3.4)
m& eYO2

where m& e is the mass entrainment rate of the plume, and YO2 is the oxygen mass fraction.
Thus the equivalence ratio, φ, is:

15
 FOP (∆H ) F m& PF
φ= = (3.5)
FOS (∆H ) O m& eYO2

The actual heat release rate is considered to be related to φ as follows:

Q = f (φ )Q N (3.6)

In this model, f(φ) is obtained using the following simple relation used by CFAST: [3]

1
f (φ ) = (3.7)
max(φ ,1)

The fuel rich flammability is given by a limiting oxygen mass or volume fraction. In
order to make the calculation smooth near the fuel rich limit, following the CFAST
approach, a limiting coefficient CLOL is introduced: [3]

C LOL =
[ ( ) ]
Tanh 800 YO2 − YLOL − 4 + 1 (3.8)
2

where Tanh(x) is the hyperbolic tangent function of x, and YLOL is the limiting oxygen
mass fraction.[33] Thus Equation (3.5) should be adjusted as follows:

(∆H ) F m& PF
φ= (3.9)
(∆H ) O m& eYO2 C LOL

The actual fuel consumption rate can be obtained as follows:

Q
m& BF = (3.10)
(∆H ) F

And the oxygen consumption rate from air can also be obtained as follows:

Q
m& O2 = (3.11)
(∆H ) O

This model is similar to CFAST’s but re-expressed based on the concept of the
equivalence ratio φ.

16
3.2 Combustion Chemistry
In this model, the global combustion chemistry is written as follows:

mPF + mO2 → mtox1 + mtox 2 + mtox 3 + mTUCF + mCO2 + mCO + msoot + mH 2O (3.12)

where m PF is the mass of the pyrolyzed fuel, which is a user input,

mO2 is the mass consumption of oxygen, obtained from Equation (3.11)
mtox1 is the mass production of the first kind of toxic species,
mtox 2 is the mass production of the second kind of toxic species,
mtox 3 is the mass production of the third kind of toxic species,
mTUCF is the mass production of the total unburned combustible fuel,
mCO2 is the mass production of carbon dioxide,
mCO is the mass production of carbon monoxide,
msoot is the mass production of soot,
mH 2O is the water production.

The pyrolyzed fuel m PF is assumed to be composed of two parts: the combustible main
part mPFmain and the incombustible minimal part m PF min .

mPF = mPFmain + mPFmin (3.13)

mPFmain is the mass of the elements, carbon, hydrogen and oxygen of the pyrolyzed fuel:

mPFmain = mC + mH + mO (3.14)

m PF min is assumed to be composed of some harmful species, such as hydrogen chloride

and hydrogen cyanide, whose mass is much less than the total fuel mass and will not be
involved in the further combustion process. Thus m PF min from m PF in the left side of
Equation (3.12) will directly go into its right side. In this model, m PF min can be composed
of as many as three kinds of toxic species: mtox1 , mtox 2 and mtox 3 .

m PFmin = mtox1 + mtox 2 + mtox 3 (3.15)

The mass production of carbon dioxide, carbon monoxide and soot are combined to be
one new term mCRP , carbon-related products as follows:

17
mCRP = mCO2 + mCO + msoot (3.16)

where soot is assumed to be composed of carbon.

In terms of fuel consumption, m PF can be divided into two parts: burned pyrolyzate m BF ,
and unburned pyrolyzate mUF . Both m BF and mUF have the same composition as m PF .
And the main part of m UF will form the term m TUCF , while the main part of m BF will go
into the water and carbon-related products.

The approach in which that mTUCF is assumed to have the same element composition as
the main part of the pyrolyzate, will simplify the calculation of oxygen consumption,
although the unburned products of organic fuel are generally hydrocarbons without
oxygen content.

The above chemical equations can be integrated as the following single equation:

[ ( ) ( )]
mPF mBF mBFmain + mBFmin + mUF mUFmain + mUFmin + mO2 (3.17)
(
→ mPFmin mBFmin + mUFmin ) + m (m
TUCF UFmain ) + m (m
CRP CO2 )
+ mCO + msoot + mH 2O

The following part will present the calculation formulas of the combustion chemistry.

The composition of the pyrolyzed fuel is defined by the following five mass ratios, which
are input by the user of the model.

χ C = mC / m PF (3.18)

χ H = m H / m PF (3.19)

χ O = mO / m PF (3.20)

χ tox1 = mtox1 / m PF (3.21)

χ tox 2 = mtox 2 / m PF (3.22)

where χ C , χ H , χ O , χ tox1 and χ tox 2 are mass fractions of carbon, hydrogen, oxygen, the
first kind of toxic species and the second kind of toxic species in the pyrolyzed fuel,
respectively.

18
Based on the above five mass ratios, the toxic species production is then computed using:

m PF min = [1 − ( χ C + χ H + χ O )]m PF (3.23)

mtox1 = χ tox1mPF (3.24)

mtox 2 = χ tox 2 m PF (3.25)

mtox 3 = mPF min − (mtox1 + mtox 2 ) (3.26)

The mass ratios of the production of soot and carbon monoxide to the carbon related
products, γ CO and γ soot , are also defined by the user:

γ CO = mCO / mCRP (3.27)

γ soot = msoot / mCRP (3.28)

Then the production of the carbon-related products (CRP) of combustion can be obtained
from:

χ C mBF
mCRP = (3.29)
γ C / CRP

where m BF is obtained from Equation (3.10), which is assumed to have the same
composition as m PF , and γ C / CRP is the mass ratio of the carbon in the CRP products to
the total CRP products as follows:

12 12
γ C / CRP = γ soot +
γ CO + (1 − γ soot − γ CO )
28 44
(3.30)
3 12 8
= + γ CO + γ soot
11 77 11

Thus we can get the production of soot, carbon monoxide and carbon dioxide as follows:

msoot = γ soot mCRP (3.31)

mCO = γ CO mCRP (3.32)

mCO2 = (1 − γ soot − γ CO )mCRP (3.33)

19
Water production mH 2O and the total unburned combustible fuel mTUCF can be obtained as
follows:

m H 2O = 9 χ H m BF (3.34)

mTUCF = (χ C + χ H + χ O )(m PF − m BF ) (3.35)

3.3 Calculation of ( )O or ( )F
For a chemical reaction, there is a fixed relationship between the heat of combustion per
unit mass of burned fuel (∆H ) F and per unit mass of consumed oxygen (∆H ) o . If we
simply take (∆H ) F as a known value and (∆H ) o as 13.2 MJ/kg, then the oxygen
element in both sides of Equation (3.12) may not be balanced. So, for a fixed reaction
with a given (∆H ) F , (∆H ) o should be computed accordingly to ensure oxygen balance.

The relationship of (∆H ) F and (∆H ) O can be derived from the following combustion
equation without the unburned fuel production ( m PF = m BF ):

( )
m PF mPFmin + mPFmain + mO2 → m PFmin + mCRP + m H 2O (3.36)

By balancing the mass of oxygen of the above equation, the following relation can be
obtained:

χ O mPF + mO = mCRP (1 − γ C / CRP ) +

16
mH O (3.37)
2
18 2

Substituting Equations (3.29) and (3.34) into the above equation gives:

 χC 
mO2 =  (1 − γ C / CRP ) + 8χ H − χ O mPF (3.38)
 γ C / CRP 

Based on the definition of (∆H ) F and (∆H ) O , we have:

(∆Η )O mO 2
= (∆Η )F m PF (3.39)

Then we can get:

20
(∆Η )O = (∆Η )F (3.40)
χC
− χ C + 8χ H − χ O
γ C / CRP

In this model, (∆H ) F is the user input, and (∆H ) o will be calculated from Equation
(3.40). If (∆H ) F is input as zero, then (∆H ) o is taken as 13.2 MJ/kg, and (∆H ) F will be
calculated from (∆H) o by inversely using Equation (3.40).

In summary, by specifying mass pyrolysis rate m& PF , the various mass ratios χ C , χ H ,
χ O , χ tox1 , χ tox 2 , γ CO , γ soot , and the heat of combustion per unit mass of fuel for the
specified chemical reaction (∆H ) F , the heat release rate and species production and
consumption rates can be calculated using the above equations.

This formula can apply to pure hydrogen combustion by specifying χ H being one and
any other ratio being zero. This formulation can also apply to the case without CO2
production.

21
4. FLUID FLOW

4.1 Plume Entrainment

Fire-induced buoyant plume entrainment is a very important factor in modeling fire
growth and smoke spread in a building. A number of formulas can be found in the
literature.[34-38] A widely accepted consensus on entrainment models for large fires in
compartments does not yet exist. A review of existing models shows that they are based
primarily on data from smaller fires.[36, 37]

Some full-scale standard room fire experiments at NIST indicated that McCaffrey’s
model (used by CFAST) gives the best agreement with the measured entrainment rates,
although it does not account for the changing surrounding gas density.[37] In this model,
McCaffrey’s entrainment model has been used: [34]
CE 2
m& e  Ze 
= C E1  2 / 5  (4.1)
QC  QC 

where m& e is the plume entrainment rate,

Qc is the convection heat release rate,
CE1 and CE2 are constant coefficients as follows:

Ze
if 0≤ 2/5
< 0.08, C E1 = 0.011, C E 2 = 0.566 (4.2)
QC

Ze
if 0.08 ≤ 2/5
< 0.20, C E1 = 0.026, C E 2 = 0.909 (4.3)
QC

Ze
if 0.20 ≤ 2/5
, C E1 = 0.124, C E 2 = 1.895 (4.4)
QC

For atria or warehouses, Heskestad’s model is widely used.[37, 38] This correlation with
constant coefficients is also implemented in the model to allow users to select either of
the two models. Heskestad’s plume equations are as follows:

Z 0 = 0.166QC
2/5
(4.5)

22
if Z e ≤ Z 0 , m& e = 0.0056QC Z e / Z 0 (4.6)

Z e > Z 0 , m& e = 0.071QC Z e + 0.0018QC

1/ 3 5/3
if (4.7)

If fires burn along the walls or in corners, plume entrainment rate will be restricted. The
effects can be treated as follows: [39]

The mass entrainment rate for center fires can be expressed as a common function:

m& e = f (QC , Z e ) (4.8)

Then mass entrainment rate for wall and corner fires will be:

m& e = f (ωQC , Z e ) / ω (4.9)

where for corner fire, is 4, for wall fire, is 2, and for center fire, is 1.

As the entrainment formula is an empirical relation from experiments, its application is

limited to the range of the experimental data. Mowrer and Williamson have reported that
for heat release rates ranges of 50 to 1000 kW, the mass entraiment rates for center, wall,
and corner fires are in the range of (0.64-1.06), (0.48-0.85), and (0.35-0.67) kg/s,
respectively.[39] To use equations 4.8 and 4.9 outside these ranges, it is necessary to limit
the maximum entrainment rate based on energy balance:

QC
m& e ≤ − m& PF (4.10)
C P (TI − TL )

where TI is assumed to be the interface temperature. Equation (4.10) implies that the
averaged plume temperature at the interface should be greater than TI. In this model, TI is
obtained based on Cooper’s N percent rule. [14, 40]

4.2 Door/Window Vent Flow

Mass flow through a vertical-vent is driven by the pressure difference between the two
sides of the vent, and it can be calculated by integrating Bernoulli’s equation along the
vertical direction of the vent. However, the flow through a vent may not be
unidirectional, i.e., there may be some gas flowing in and some flowing out of the room.
In these cases, the integral limit is divided into several parts, each part having the same
flow direction. If the vent is rectangular, then for any sub-divided part, we can obtain the
following formula from Bernoulli’s equation:[2, 22]

23
  Pt + Pt Pb + Pb 
C vent AV 2 ρ  
2
m& V = (4.11)
3  Pt + Pb 
 

where m& V is the mass flow rate through this part of the vent,
Pt and Pb is the pressure difference at the top and bottom position of this part,
ρ is the gas density of source side,
AV is the area of this part of vent,
Cvent is the coefficient of vent flow, which is taken as 0.7 in this model.

In order to divide a vent into several uni-directional parts, first we determine the positions
of the neutral planes and flow directions, which can be estimated by calculating the
pressure difference between the two sides of the vent.

Room 1 Opening
Room 2
U2 (upper layer)
L2 (lower layer)
U1 (upper layer)
L1 (lower layer) Zso
Zi2

Zi1 Z
Zsi
Floor of room 2

Zf2
Floor of room 1
Zf1

Reference height

Figure 4.1 Basic vertical-vent (door/window) configuration

In Figure 4.1, Zi1 and Zi2 are the interface heights of the two compartments connected by
a vent. Zso and Zsi are the soffit height and sill height of the vent with respect to the floor
height of No.1 compartment, respectively. For each compartment, the pressure at the
floor level is considered as the reference pressure. At any elevation Z, the pressures in the
two compartments P1(Z) and P2(Z) can be obtained as follows:

24
P1( Z ) = P1( Z f 1 ) − g {min(Z − Z f 1 , Zi1 ) ρ L1 + max[(Z − Z f 1 − Zi1 ), 0]ρU 1 } (4.12)

P 2( Z ) = P 2( Z f 2 ) − g {min(Z − Z f 2 , Zi2 ) ρ L 2 + max[(Z − Z f 2 − Zi2 ), 0]ρU 2 } (4.13)

∆P( Z ) = P1( Z ) − P 2( Z ) (4.14)

where ∆P(Z) is the pressure difference between the two sides of the vent at height Z,
Zf1 and Zf2 are the floor elevations of the two compartments.

Z, Zf1 and Zf2 are defined with respect to the same reference height. The positions of the
neutral planes are defined as the points where ∆P(Z)=0. The number of neutral planes in
an opening can range from zero to three (see Figure 4.2).

When hot smoke (U1) flows out of the vent, it may entrain air from the cool lower layer
(L2) of the neighboring room and transport it into the upper layer. This process may be
modeled by considering the vent flow as a plume induced by buoyancy. The virtual
plume location can be estimated from inversely applying McCaffrey’s correlation. [3]

∆ P (z )
F lo w
D ir e c tio n

P1 P2

N o N e u tr a l P la n 1 N e u tr a l P la n 1 N e u tr a l P la n

2 N e u tr a l P la n 2 N e u tr a l P la n 3 N e u tr a l P la n

Figure 4.2 Possible neutral planes in an opening

25
QC = C P (TU 1 − TL 2 )m& U 1− L 2 (4.15)

1 / 0.566 1 / 0.566
 m& U 1− L 2   m& 
ZV = Q 2/5
C
  , if 0.0 <  U 1− L 2  ≤ 0.08 (4.16)
 0.011QC   0.011QC 

1 / 0.909 1 / 0.909
 m& U 1− L 2   m& 
ZV = Q 2/5
C
  , if 0.08 <  U 1− L 2  ≤ 0.20 (4.17)
 0.026QC   0.026QC 

1 / 1.895 1 / 1.895
 m& U 1− L 2   m& 
ZV = Q 2/5
C
  , if 0.20 <  U 1− L 2  (4.18)
 0.124QC   0.124QC 

Z e = Z12 + Z V (4.19)

where QC is the equivalent convection heat release rate,

m& U 1− L 2 is the mass flow rate from layer U1 to layer L2,
ZV is the location of the virtual point plume,
Z12 is the real entrainment distance of the smoke flow given as follows:

Z 12 =
[
( Zi 2 + Z f 2 ) − max ( Zi1 + Z f 1 ), ( Zi si + Z f 1 ) ]
2

+
[
max 0, ( Zi 2 + Z f 2 ) − ( Zso + Z f 1 ) ] (4.20)

Then from Ze and QC, plume entrainment rate from L2 to U2 can be calculated using
Equations (4.1) to (4.4).

For Heskestad’s plume model, similar method can be used. The relevant equations are as
follows:

Z 0 = 0.166QC
2/5
(4.21)

Z 0 m& U 1− L 2 m& U 1− L 2
ZV = , if ≤1 (4.22)
0.0056QC 0.0056QC

3/5
 m& − 0.0018QC  m& U 1− L 2
Z V =  U 1− L 2 1/ 3  , if >1 (4.23)
 0.071QC  0.0056QC

26
Similarly, Ze and QC, plume entrainment rate from L2 to U2 can be calculated using
Equations (4.5) to (4.7).

Additionally, when cool gas (L2) enters the neighboring zone (U1), it will behave like an
inverse plume, and will entrain upper layer gas into the lower layer under some
conditions. In this model, the above equations are also used to calculate the inverse plume
entrainment rate, and an equation similar to Equation (4.10) is used to limit the maximum
plume entrainment rate.

4.3 Ceiling Vent Flow

Vertical gas flow through a horizontal-vent involves two flow-driving mechanisms,
buoyancy and pressure; while horizontal gas flow through a vertical-vent, such as a door,
is considered to be driven only by pressure difference. It is not appropriate to directly use
Bernoulli’s equation for ceiling vent flow because in addition to pressure difference,
buoyant force has to be considered, which may lead to bi-directional flow. Cooper gives a
calculation model for unstable flow through shallow, horizontal, circular vents under
high-Grashof-number conditions, which is the case encountered in a building fire.[41-43]
This has been included in this model as follows:

VT PT, ρT
L

VB PB, ρB
D

Figure 4.3 Basic horizontal-vent (ceiling vent) configuration

Let:

PH = max( PT , PB ), PL = min( PT , PB ) (4.24)

∆P = PH − PL ≥ 0 (4.25)

∆ρ = ρ T − ρ B > 0 (4.26)

27
ρ = ( ρ B + ρT ) / 2 (4.27)

ε = ∆ρ / ρ , if PT ≥ PB
 (4.28)
 ε = −∆ρ / ρ , if PT < PB

T = (TT + TB ) 2 (4.29)

ν = 0.04128 × 10 −7 T 2.5 (T + 110.4) (4.30)

2 gD 3 ε
Gr = (4.31)
ν2

where is the kinematic viscosity, and Gr is the Grashof number. If Gr is greater than
2 × 107 then Cooper’s formulae can be used to calculate ceiling vent flow as follows.

The following step is to judge whether it is under the condition of flooding or not; i.e.,
whether a bi-directional flow exists.

Π FL = 0.2427(1 + ε / 2) exp(1.1072ε ) (4.32)

∆PFL = (4 g∆ρD)Π FL (4.33)

FrH , FL = 0.1754 exp(0.5536ε ) (4.34)

V H , FL = (2 gD ε )1 / 2 AV FrH , FL (4.35)

If ∆P ≥ ∆PFL , then unidirectional flow is expected, and the relevant calculation formulae
are as follows:

{
V H = [1.19252 + 11.3569(∆p / ∆p FL − 1)]
1/ 2
}
− 0.092025 VH , FL (4.36)

VL = 0 (4.37)

V N = VH (4.38)

where V is the volume flow rate through the vent,

subscripts H, L and N refer to flow from the high pressure side to the low pressure
side, from the low to the high side and the net flow rate.

28
If ∆P < ∆PFL , then bi-directional exchange flow will exist, and the relevant calculation
formulas are as follows:

V EX , MAX = 0.055(4 / π ) AV ( gD ε )1 / 2 (4.39)

9.4 − [1 + 87.36(1 − ∆P / ∆PFL )]

1/ 2

VN = V H , FL (4.40)
8.4

[
V L = 0.6465(1 − ∆P / ∆PFL ) 2 − 1.6465(1 − ∆P / ∆PFL ) VEX , MAX] 2
(4.41)

VH = VL + V N (4.42)

In this model, if the condition using the above Cooper’s model is not satisfied, then the
uni-directional Bernoulli’s equation is directly used as follows:

2∆P
V H = C CV AV (4.43)
ρH

where CCV is the coefficient of ceiling vent flow, which is taken as 0.61 in this model.

4.4 Mechanical Ventilation

Mechanical ventilation is considered using a specification model. Figure 4.4 shows a
schematic of a mechanical ventilation opening. Through the opening, smoke can be
extracted out of the room to ambient, or ambient air can be supplied into the room. Two
parameters can be specified in this model. One is the Zext, the vertical extension of the
opening away from the ceiling surface, another is the mechanical volume or mass
ventilation rate.

In the case shown in Figure 4.4, initially the smoke interface is above the opening
elevation, and the exhausted gas is lower layer air only. If the plume entrainment rate at
the elevation of the exhaust opening is greater than the exhaust rate, then the interface
will be formed under the opening, and after that time, the exhausted gas will be smoke
only. If the plume entrainment rate at the elevation of the exhaust opening is less than the
exhaust rate, and the smoke exhaust system is assumed to be ideally effective, then the
interface will be formed at the opening elevation, and the exhausted gas is assumed to be
composed of two parts, smoke and lower layer gas. For this case, the following
formulation has been used in the model to identify how much gas is extracted from each
zone.

For mass flow rate specification:

29
m& MU = min(m& Fan , m& max ) (4.44)

m& ML = m& Fan − m& MU (4.45)

For Volume flow rate specification:

m&
V&MU = min(V&Fan , max ) (4.46)
ρU

V&ML = V&Fan − V&MU (4.47)

where m & max is the maximum smoke exhaust rate from the upper layer. m& MU and m& ML are
mass flow rate exhausted from upper layer and lower layer, respectively. m& Fan is the
specified mass exhaust rate of the exhaust fan. V& and V& are volume flow rate
MU ML

exhausted from upper layer and lower layer, respectively. V&Fan is the specified volume
exhaust rate of the exhaust fan.

The above method of limiting the maximum smoke exhaust rate is also helpful in
ensuring the numerical stability and efficiency, when the exhausted gas is composed of
two parts, smoke and the lower layer gas. Sometimes m& max is difficult to calculate. For
the situation shown in Figure 4.4, where a ceiling vent is not provided and the soffit of
the door is lower than the opening, m& max is the plume entrainment rate at the elevation of
the opening.

Zext

Hot Layer

Cool Layer

Fuel

Figure 4.4 Schematic of the mechanical ventilation opening

30
4.5 Stair Shaft Flow
When dealing with multi-storey buildings, it is important to compute the mass or volume
flow rate through a stair shaft, because a stair shaft is the main path of smoke spread from
floor to floor. Achakji and Tamura conducted research work on pressure drop
characteristics of typical stairshafts in high-rise buildings [44]. It was found that pressure
drop through a stairshaft can be represented by a pressure loss across an equivalent
horizontal orifice located between floors of a frictionless stair shaft.

Based on Achakji and Tamura’s experimental data, a correlation for calculating the area
of the effective orifice was obtained and used in CONTAM, a model used to analyze
contaminant migration in a building [45]. This correlation is as follows:

For open treads:

Aei = 0.089 hi AS (1.0 − 0.14 d i ) (4.48)

For closed treads:

Aei = 0.083 hi AS (1.0 − 0.24 d i ) (4.49)

As shown in Figure 4.5, the parameters in the above two equations are as follows: Ae i is
the effective orifice area on the ith floor, hi is the height of the shaft between the ith floor
and (i-1)th floor, d i is the density of people on the stairs between the ith floor and (i-1)th
floor, and AS is the cross-sectional area of the stair shaft.

Aei

ith Floor

hi

AS

(i-1)th Floor

Aei-1

Figure 4.5 Schematic of an effective ceiling vent for a stair shaft

31
Equations (4.48) and (4.49) were used in previous version of FIERAsmoke. However, it
was found that there is a problem in the two equations with the relationship between Ae i
and hi . Following equations (4.48) and (4.49), the higher hi is, the larger the area Ae i is.
This is contradictory to the real physical laws of fluid flow. As hi represents the length
of friction in the stairshaft, it is expected that increasing hi will increase friction and
decrease the effective area Ae i . After carefully re-investigating the original paper [44]
referenced by CONTAM, it was found that the following equation is given:

Aei 1
= 1
(4.50)
AS
 h  2
C CV  k i 
 De 

where C CV is the coefficient of discharge as in section 4.3, taken as 0.6 in reference [44]; k
is the friction pressure loss coefficient; De is equivalent diameter, which can be
calculated by the stairshaft cross-sectional area A S and the perimeter of the shaft Γ as
follows:

4 AS
De = (4.51)
Γ

Equation (4.50), however does not include the effect of the density of people d i . To
account for this, the approach used in Equations (4.48) and (4.49) is followed, resulting in
the following equation that is used in the model:

Aei (1.0 − β d i )
= 1
(4.52)
AS
 h 2
C CV  k i 
 De 

where, β is taken as 0.14 for open tread case, 0.24 for closed tread case. According to
reference,[44] k is taken as 29 for open tread case and 32 for closed tread case. In the
model, after the equivalent area Ae i is obtained from Equation (4.52), Cooper’s [43]
ceiling vent flow model is used to calculate the volumetric flow rate through the ceiling
vent, as mentioned in section 4.3.

32
4.6 Species Concentration
Suppose at time t in a well-stirred gas layer, there are NS kinds of species, the total mass
is m, and the mass fraction of species K is YK, K=1, 2, …, NS.

Then, at the next time step t+∆t, the concentration of the Kth species will be
approximated as follows:

(YK ) t + ∆t =
[ ] , K=1, 2, …, NS.
mYK + ∆t ∑ m& e , K + ∑ m& V , K
m + ∆t [∑ m& + ∑ m& ]
(4.53)
e V

where ∑ m& e and ∑ m& V are the total net mass flow rate entering the layer (negative for
flowing out) by plume entrainment and vent flow. ∑ m& e, K and ∑ m&
V ,K are the total net
mass flow rate of species K entering the layer by plume and vent flow.

33
5. HEAT TRANSFER

5.1 Conduction Heat Transfer

To calculate conduction heat transfer through the compartment boundaries, a one-
dimensional conduction model is used. The governing equation is as follows:

∂T K ∂ 2T
= , 0 ≤ x ≤ L, t ≥ 0 (5.1)
∂t ρC ∂x 2

Due to the complexity of the building geometry, it is assumed that heat is transferred to
the outside and not to the gas of the neighboring rooms as in CFAST. Then the following
initial and boundary conditions can be given:

0 ≤ x ≤ L, t = 0, T = T ( x, 0) (5.2)

∂T
x = 0, −k = hIC (Tg − T (0, t )) + q IR (5.3)
∂x

∂T
x = L, −k = −hOC (Ta − T ( L, t )) − qOR (5.4)
∂x

where hIC and hOC are heat transfer coefficients of the inside and outside surface of the
wall, and q IR and qOR are the net radiation heat fluxes from gas received by inside and
outside surfaces of the wall. The T(x, 0) is the temperature profile at the initial time, from
which we can predict the temperature profile at any future time by coupling conduction
with convection and radiation.

5.2 Convection Heat Transfer in Non-Fire Rooms

Convection heat transfer in a non-fire room may be considered to be a natural convection
system. The following equations for turbulent convection heat transfer are included in
this model:[3, 46, 47]

Qconv = qconv Aw (5.5)

34
where Qconv and qconv are the convection heat loss and flux to a surface of a wall, ceiling
or floor, and Aw is the heat transfer area,

q conv = h(Tg − Tw ) (5.6)

where Tg is the temperature of gas, and Tw is the temperature of the surface of a wall,
ceiling or floor. h is the convection heat transfer coefficient given as follows:

K
h= Nu (5.7)
l

K is the gas conductivity, l is the characteristic length, and Nu is the Nusselt number. l, K
and Nu numbers can be obtained as below:

l = Aw (5.8)

K = 2.72 × 10 −4 (T )
0.8
(5.9)

Nu = C conv (Gr Pr )
1/ 3
(5.10)

where Cconv is a constant. For a wall Cconv=0.13, for a heated floor or a cooled ceiling
Cconv=0.021, for a cooled floor or a heated ceiling Cconv=0.21. T is the averaged
temperature as follows:

Tg + Tw
T =
2

In Equation (5.10), Pr is the Prandtl number of air, which is taken as a constant 0.7. Gr is
the Grashof number:

gl 3 ∆T
Gr = (5.11)
Tν 2

where

∆T = Tg − Tw (5.12)

And is the kinematic viscosity calculated using Equation (4.30).

35
5.3 Ceiling Jet-Induced Heat Transfer
In the fire room, ceiling jet-induced convection heat transfer should be separately
evaluated due to its specific magnitude. Heat loss due to a ceiling jet can be calculated
using the following formulae: [48-50]

Qceil = ∫ q ceil dA (5.13)

A

q ceil = h(Tad − Tceil ) (5.14)

where Qceil is gas heat loss through the ceiling surface,

qceil is the convection heat transfer flux,
h is the heat transfer coefficient,
Tad and Tceil are characteristic gas and ceiling surface temperatures.

 −1 / 2 −2 / 3 r r
8.82 Re H Pr [1 − (5 − 0.284 Re H ) H ], 0 ≤ H < 0.2
0.2


h 
~ = −1.2
r
− 0.0771
(5.15)
h 
− 0.3 −2 / 3  r  H r
0.283 Re H Pr   , 0.2 ≤
 H r
+ 0.279
H
 H
~
where h is the characteristic heat transfer coefficient,
r is the radius from the stagnant point,
H is the vertical distance between the fire source and the ceiling.

 r r
10.22 − 14.9 H , 0 ≤ H < 0.2

Tad − TU 
2/3
= (5.16)
TU QH* 
  r  r
8.39 f  H , 0.2 ≤ H
  

0.8 1.6
 r   r 
1 − 1.1  + 0.808 
 r  H H
f = (5.17)
H
0.8 1.6 2.4
 r   r   r 
1 − 1.1  + 2.2  + 0.69 
H H H

36
~
h = ρ U C P gH QH*
1/ 3
(5.18)

1/ 3
gH HQH*
Re H = (5.19)
νU

ν U = 0.04128 × 10 −7 × TU5 / 2 /(TU + 110.4) (5.20)

Q′
QH* = (5.21)
ρU C P TU gH H 2

 ΩΓ
QC 1 + Ω , Z fire < Z i < Z ceil

Q′ =  (5.22)
Q , Z ≥ Z , Z = Z
 C fire i i ceil

Ω and Γ are given as follows, which reflect the effects of a smoke layer on the ceiling jet.

TU 2/3
1− + CT Qeq*
TL
Ω= ,CT = 9.115 (5.23)
(TU − TL ) / TL

1.04599Ω + 0.360391Ω 2
Γ= for σ > 0
1. + 1.37748Ω + 0.360391Ω 2
(5.24)
=0 for − 1 < σ ≤ 0

3/ 2
 0.21QC 
Q = 
*
eq
 (5.25)
C T m&
 P L e

In this model, the ceiling surface is converted into an equivalent circular surface with the
same area, and the plume impingement point is assumed to be at the center of the surface.
The ceiling surface temperature is assumed to be spatially uniform.

37
5.4 Convection Heat Transfer in Fire Rooms
For the lower wall and floor, the convection heat transfer model is the same in the fire
room as in the non-fire room. For the ceiling surface, the model introduced in the above
section is applied to calculate ceiling jet convection heat transfer. For the upper wall, the
non-fire room convection model may underestimate the heat transfer due to the wall jet.
Thus, in this model, a simple treatment is introduced as follows.[6]

Let us denote the equivalent radius of the ceiling as RC, and the smoke thickness as ZS.
If RC≥ZS, then:

q = (qconv + Qceil / Aceil ) / 2 (5.26)

If RC<ZS, then:

q = ( RC × Qceil / Aceil + Z S × qconv ) /( RC + Z S ) (5.27)

where qconv is the heat transfer flux as described in Equation (5.6),

Qceil and Aceil are the heat loss to the ceiling surface and the ceiling surface area.

5.5 Radiation Heat Transfer

5.5.1 Derivation of Equations

Radiation heat transfer is a very important mechanism in compartment fires, especially in
the fire room. As shown in Figure 5.1, a two-surface model is applied in the fire room,
where the ceiling and upper wall surface is considered to be one surface (upper surface),
and the floor and lower wall surface is considered to be another surface (lower surface).
In order to simplify the calculation process, in the fire room, the flame is assumed to be a
sphere with its center located at the position of half flame height above the fuel bed, and
the radiation flux of the sphere to any direction is assumed to be uniform. Additionally,
the sphere is assumed to be the third differential emitting blackbody surface interacting
with the upper surface, lower surface and upper layer gas.[6, 22, 51]

The smoke layer is considered to be an absorptive medium, and the lower layer is
considered to be transparent. The following radiation heat transfer equation for each
surface element k is used:[52]

 δ kj 1− ε j 
[ (δ kj − Fkjτ kj )σT j4 − Fkjα kjσTg4 ]
N N

∑ 

j =1  ε
− Fkj
ε
τ  q
kj  j = ∑ (5.28)
j j  j =1

38
 Upper Surface (1)
Smoke layer (g)

Lower Layer

LowerSurface (2)

Figure 5.1 Schematic of the radiation geometry

where qj is the net radiation heat transfer rate of surface j,

δkj is Kronecker function (1 for k = j and 0 for k ≠ j),
εj is emissivity of surface j,
Fkj is the view factor from surface k to surface j,
τ kj , α kj are geometrical mean transmittance and absorptance from surface k to j,
N is the number of surfaces in the compartment (in this model N = 2)

Thus, for each surface, we can get the following relations:

1 1 − ε1  1− ε2
k = 1,  − F11 τ 11 q1 − F12 τ 12 q 2 =
 ε1 ε1  ε2 (5.29)
(1 − F11τ 11 )σT14 − F12τ 12σT24 − ( F11α 11 + F12α 12 + F1F α 1F )σTg4 − F1F τ 1F σTF4

1 − ε1  1 1− ε2 
k = 2, − F21 τ 21 q1 +  − F22 τ 22 q 2 =
ε1  ε2 ε2  (5.30)
− F21τ 21σT14 + (1 − F22τ 22 )σT24 − ( F21α 21 + F22α 22 + F2 F α 2 F )σTg4 − F2 F τ 2 F σTF4

In the above, the last term of the right side of each equation accounts for the assumed
flame radiation. For convenience, the value of the radiation fraction instead of flame
temperature is used, and the following relation is applied:

F1F τ 1F σTF4 = QF FF 1τ 1F / A1 (5.31)

F2 F τ 2 F σTF4 = QF FF 2τ 2 F / A2 (5.32)

39
where QF is the radiation power of the flame, which is a fraction of the total heat release
rate of the fire.

Using the relations of transmittance and absorptance,

τ ij = τ ji (5.33)

α ij = α ji (5.34)

α ij + τ ij = 1 (5.35)

and assuming

α 11 = α 12 = α F 1 = ε g (5.36)

αF2 = 0 (5.37)

the following formulae for calculating radiation heat transfer can be derived:

 W1ε 1
 q1 = Z
 Wε
(5.38)
q 2 = 2 2
 Z

Z = F21 F12 (1 − ε g ) 2 (1 − ε 1 )(1 − ε 2 )

(5.39)
− [1 − F11 (1 − ε g )(1 − ε 1 )][1 − F22 (1 − ε 2 )]

W1 = {[1 − (1 − ε g ) F11 ][1 − (1 − ε 2 ) F22 ] − [(1 − ε 2 )(1 − ε g ) 2 F12 F21 ]}σT14

− (1 − ε g ) F12 ε 2σT24
− {1 + (1 − ε 2 )[(1 − ε g ) F12 F21 − F22 ]}ε g σTg4 (5.40)
QF
− (1 − ε g )[ F21 − ε 2 ( FF 2 − F22 )]
A1

40
W2 = {[1 − (1 − ε 1 )(1 − ε g ) F11 ](1 − F22 ) − (1 − ε 1 )(1 − ε g ) 2 F12 F21 ]}σT24
− (1 − ε g ) F21ε 1σT14
− {[1 − (1 − ε 1 )(1 − ε g ) F11 ] + (1 − ε 1 )(1 − ε g )}F21ε g Tg4 (5.41)

−
QF
{FF 2 + (1 − ε 1 )(1 − ε g )[ FF1 F12 (1 − ε g ) − FF 2 F11 ]}
A2

Qg = −( A1q1 + A2 q2 ) + QF (5.42)

where T and A are the temperature and surface area,

subscripts 1, 2 refer to upper surface and lower surface, respectively,
g and F refer to smoke layer and flame, respectively,
Qg is the net radiation heat gain rate of smoke layer.

5.5.2 Calculation of View Factors

It is simple to get the values of F11, F12, F21 and F22 as follows:

F12 = Aint / A1 (5.43)

F21 = Aint / A2 (5.44)

F11 = 1 − F12 (5.45)

F22 = 1 − F21 (5.46)

As mentioned above, the flame is assumed to be a black sphere radiating uniformly.

Shown in Figure 5.2, point F is assumed to be the center of the sphere from which heat is
radiated uniformly. The view factor from pointer F to area A (OBCD) is obtained:

cosθ
ξ =∫ dA (5.47)
A 4π ( FK )
2

Integrate the above equation, we get:

1  1 
ξ= arcsin   (5.48)
4π  ( x 2 + 1)( y 2 + 1) 

OF OF
where x = , y=
OB OD

41
From Equation (5.48), we can get the view factor from F to any finite rectangular plane.

D C
F
θ
K

O B

Figure 5.2 A schematic of flame radiation to a rectangular surface

5.5.3 Gas Emissivity

Gas emissivity for carbon dioxide, water vapour and soot can be obtained as follows: [52,
53]

ε = ε soot + ε g − ε soot ε g (5.49)

ε g = ε CO + ε H O − ε D
2 2
(5.50)

1
ε soot = 1 −
(1 + 350 C soot Le Tg )4
(5.51)

where Csoot is the volume fraction of soot in the smoke layer,

Le is the equivalent mean beam length of the smoke layer,
ε D is emittance reduction resulting from spectral overlap.

ε CO = 6.17( PCO Le)1 / 3 / Tg

2 2
(5.52)

ε H O = 0.674[1 − exp(−1.32 β )]
2
(5.53)

β = PH O Le (300 / Tg )[ Pg + (5 300 / Tg + 0.5) PH O ]

2 2
(5.54)

where Pg is the gas pressure in atmospheres.

PCO2 and PH 2O are the partial pressure of CO2 and H2O.

ε D = F (Tg )G (ξ ){log[101.3( PCO + PH O ) Le]}2.76

2 2
(5.55)

42
F (Tg ) = −1.0204 × 10 −6 Tg + 0.002449Tg − 0.23469
2
(5.56)

G (ξ ) = ξ /(10.7 + 101ξ ) − ξ 10.4 / 111.7 (5.57)

ξ = PH O /( PH O + PCO )
2 2 2
(5.58)

43
6. NUMERICAL METHOD

6.1 Introduction
As described in Chapter 2, each compartment has four independent governing variables
or equations. Basically, the governing equations can be solved numerically using two
approaches. In the first approach, the four variables of each room are independently
solved, and then all the variables are iterated room by room. This method is referred to as
the Series method. In the second approach, all the variables are solved simultaneously for
all the compartments. This method is referred to as the Parallel method.

The Parallel method was applied in the previous version of the model. It was working
well for the cases with a few compartments. With a large number of compartments,
however, this method was not converging well. Thus the Series method has been
implemented in this version of model. This method has the advantage that it is easier to
identify the sources of convergence problems, easier for ODE solver to converge, and
easier to implement specific calculation methods for specific compartments. Obviously,
the Series method has the disadvantage that the transfer of information from one
compartment to its neighboring ones is slower. Possibly some kind of combination of
Series method and Parallel method is a better way in solving a large system of ODEs,
which will be investigated further.

For the Series method, the calculation procedure is compartment-based, i.e., each time,
only one compartment is calculated with the conditions of the other compartments being
kept fixed. If we have known the solution at time t, then we select the step size ∆t, and let
the numerical integration proceed to t + ∆t for each compartment using the ODE solver.
Once sweeping through all the compartments, loop back to repeat the process until a kind
of specified precision criterion is satisfied or a specified number of iterations are finished.
There are two layers of iteration. The outer layer step size is ∆t, which is selected by the
model. The inner layer iteration step size is the integration step size used by the ODE
solver. The inner layer iteration step size should certainly be less than or equal to the
outer layer step size.

The solution procedure is shown in Figure 6.1. For each compartment, if the
compartment has smoke or fire, then the ODEs solver is used to integrate. the ODEs from
t to (t+∆t); and if the compartment has no temperature rise, then the non-linear algebraic
equation solver is used. After sweeping through all the compartments the process is
repeated until convergence is achieved. Also as shown in Figure 6.1, for each step of the
ODE solver, the mass and energy source terms of the governing equation set are
computed. As shown in Figure 6.1 insert A, there are two subgroups under the “Mass”

44
and “Energy” source terms. The left subgroup contributes to both of the source terms,
mass and energy change rates. And the right subgroup represents heat transfer process,
which does not contribute to the source term of mass change rate but energy change rate.
Also, for each step of the non-linear algebraic equation solver, the flow rate of each vent
connected to the compartment being solved is also computed using the related equations
given in Section 4.

Initialization

Select step size ∆t

Solve for the 1st room

Algebraic
Room has smoke
No equation
or fire?
solver

A Yes
Vent Steffenson's
ODE solver flow Method

Source LSODA
terms solver

Finished all
Next room No
the rooms?

Yes

Next loop No Converged

Yes

Last time
Next step No
step?

Yes

End

Source
terms
A
Heat transfer process
1. Conduction
Mass Energy 2. Ceiling Jet
3. Convection
4. Radiation
Mass and energy change rates
1. Plume
2. Forced ventilation
3. Natural ventilation
4. Combustion

Figure 6.1 Schematic Chart of Numerical Calculation

45
6.2 Numerical Properties of the ODE System
There are many ways in selecting the four independent governing equations for each
compartment. The equation set form implemented in the current model is composed of
the following four variables, pressure, upper layer enthalpy, upper layer mass and lower
layer mass. As analyzed by Forney for the equation of mass, this formulation does not
have the vanishing denominator problem of the density or temperature equation. [29]
However, quantities derived from enthalpy and mass such as density are only valid when
a layer volume has a significant number of accurate digits.

One significant property of the differential equations set, that describes the zone fire
modeling, is the presence of multiple time scales. For examples, combustion chemical
reaction has the fastest time scale in the order of milliseconds. Parameters related to fluid
flow have a scale about 0.1 second. Conduction has a time scale of several seconds or
even longer. In mathematics, this property is known as stiffness, which means the
equation set has a very large ratio of the maximum eigenvalue to its minimum eigenvalue
of the related Jacobian matrix.

Stiffness property will affect the choice of the numerical method used to solve the
governing equations.[54] The use of Runge-Kutta or Adams methods in solving stiff
problems would dictate prohibitive small step sizes to control local errors and avoid
instabilities of the computations. Other special numerical solutions are sought to solve
this class of stiff ODE.[29, 54] Backward differentiation formulas (BDFs) are generally
used to effectively solve the stiff problems.[55, 56]

In fact, to maintain numerical stability, if non-stiff methods such as those of Runge-Kutta

or Adams are applied to solve the equation set, the choice of the step size will be
dominated by stability, not accuracy. This property requires the application of special
numerical solution methods.[29, 54]

There are a number of ODE solvers available to solve the stiff ODE system.[56] In this
model, an ODE solver called LSODA, developed by Petzold and Hindmarsh, is used to
integrate the equation set numerically.[57] LSODA is a descendant of so-called DIFSUB, a
pioneering stiff ODE solver, developed by Gear using BDF method, but LSODA is more
powerful. Due to its implicitness, an iteration process composed of prediction and
correction should be applied. A Newton iterative method is used to speed up the
converging process. In this algorithm, the order and time-step size are automatically
chosen by estimating the local truncation error.

Both the model and its interface are coded using MS Visual Basic Version 5. Thus the
ODE solver LSODA was converted to VB from FORTRAN.

46
6.2.1 Stiff Equations
As mentioned above, the governing equation set of a fire zone is a set of first-order, stiff
ordinary differential equations for initial value problems. The following paragraphs
intend to shed some light on the nature of the stiff problems.

A general form of an n first order ODE equation set can be given as follows:

dY
= F (t , Y ) (6.1)
dt

with the initial condition:

Y (t 0 ) = Y0 (6.2)

where Y is a column vector:

Y = ( y1 , yn )
T
y 2 , L, (6.3)

F is the right-side function of the equation set and it is also a column vector:

F = ( f 1, fn )
T
f 2 , L, (6.4)

The derivative of F with respect to Y is called Jacobian Matrix as follows:

 ∂f1 ∂f1 ∂f1 

 ∂y L
∂y 2 ∂y n 
 1 
∂f ∂f 2 ∂f 2 
∂F  2 L
=  ∂y ∂y 2 ∂y n  (6.5)
∂Y  1
L L L L
 ∂f ∂f n ∂f n 
 n L 
 ∂y1 ∂y 2 ∂y n 

This matrix has n eigenvalues, denoted as λi , i = 1, 2, L, n. In terms of physics, they are

related to the time scale of the concerned parameters. In terms of mathematics, it is
related to the numerical stability and the choice of step size. Whether an equation set is
stiff can be determined by evaluating the eigenvalues of the Jacobian matrix. If
Re(λi ) < 0, and max Re(λi ) >> min Re(λi ) then it is a stiff equation set.

6.2.2 BDF Method

Let us denote:

47
  y1 y2 L yn
 hy ′ hy ′2 L 
hy n′
 1 
Z = L L L 
L (6.6)
 h q (q) q
h (q) h q (q) 
 y1 y2 L yn 
 q! q! q! 

where Z is a (q+1) by n matrix to be solved using the BDF method. q is the order of the
method, and h is the step size. Gear proposed gave the following procedure to integrate
stiff ODE system: [55]

Z K( 0+)1 = PZ K (6.7)

[ ]T
[
G ( Z K( m+)1 ) = hF (t K +1 , ( Z K( m+)1 ) 0 ) − ( Z K( m+)1 )1 ]
T
(6.8)

T
 ∂F 
−1

Z ( m +1)
K +1 =Z (m)
K +1 − L  - l1 I + l 0 h
∂
 G Z K +1
( m)
( )
 (6.9)
 Y  

Z K +1 = Z K( m+)1 (6.10)

where the superscripts and subscripts of Z denote the iteration number and grid number,
respectively. Z 0 and Z 1 is the 0th and 1st rows of Z. L is a constant column vector,
[l 0 , l1 , L, l q ]T , whose value is dependent on the order of the method. I is the unit
 j
matrix. P is the Pascal triangle matrix whose (i, j) element is  , q ≥ j ≥ i ≥ 0 .
i

In the above procedure, the Newton iteration method has been used, which will speed up
the iteration convergence. In the above procedure the Jacobian matrix has to be inverted.
In LSODA, this is achieved by using matrix triangular factorization method.

6.2.3 Error Estimation and Automatic Control of Step Size and Order
Let us denote:

T
 h q y1q h q y 2q h q y nq 
a =  , , L, 
 (6.11)
 q! q! q! 

48
In Gear’s method, the local truncation error of each step is estimated as:

∇a i
ε = C q +1 q! (6.12)
ωi Max

where Cq+1 is a constant coefficient depending on the order q, ∇a i is the backward

difference of the ith component of the column vector a. i is the ith component of the
column vector , whose component is a weight of the corresponding component of
vector a. • Max is the Max-norm. is taken as the maximum value of Y seen so far.
Thus if the above error is less than a specified error criterion, then the step is accepted.
Otherwise it is rejected.

Based on the estimate of local truncation error, the step size to use for the next step or to
repeat the rejected step is estimated to be h, can be estimated as:
1 /( q +1)
 
 
1  ε 1 
α=  
, for order q (6.13)
1.2 C q +1 q! ∇a i
 
 ωi Max 


1 /( q + 2 )
 
 
1  ε 1 
α= , for order q + 1 (6.14)
1.4  C q + 2 q! ∇ ai
2 
 
 ωi 
Max 

1/ q
 
 
1  ε 1 
α= , for order q − 1 (6.15)
1.3  C q q! ai 
 
 ω i Max 

Then, the order corresponding to the largest is chosen, and the step changed properly.

6.3 Solution of the Constant Temperature Compartments

Considering the fact that, for compartments without smoke (smoke has not propagated
here), generally we do not have to solve a system of ODE equations, we just have to
solve a non-linear algebraic equation of pressure based on mass conservation. In this

49
code, a mechanism is implemented to judge whether it is necessary to solve the
differential equations based on the estimation of the temperature differences between the
solved compartment and its connected neighboring compartments. If one of the
temperature differences is considerably large, then the ODE solver is called. Otherwise, a
non-linear algebraic equation of pressure based on the constant mass assumption is
solved as follows:

NV

∑ (m&
K =1
K ( P0 , PK ) ) = 0 (6.16)

where m & K is the mass flow rate through vent K, which is connected to the solved
compartment. m & K is a function of the pressure difference between the two sides of the
vent, in this model, Bernoulli’s equation is still used to calculate the mass flow rate as
described earlier in this report. NV is the number of vents connected to the solved
compartment. P0 is the pressure of the solved compartment, and PK is the pressure of the
compartment connected to the solved compartment through vent K. This equation is
numerically solved using Steffenson Acceleration method.[58]

6.4 Solution of the Physical Sub-models

From the equation set, it can be found that, besides the four independent governing
variables, there are other four parameters to be solved for each compartment. They are
the four source terms, upper and lower layer mass variation rates, upper and lower layer
energy variation rates. These four parameters can be obtained from the calculation of the
relevant fire sub-models, including fire growth model, smoke flow model and heat
transfer model, as given in Sections 3, 4 and 5. In the following of this section, the
solution methods for the conduction and ceiling jet-induced heat transfer will be
presented.

6.4.1 Conduction
Discretization: The current model implements one dimensional heat conduction model.
A uniform grid space is applied to discretize the partial equation into linear algebraic
equations. The method used to obtain the discrete equations is presented in the following
section.

Let us define the Biot and Fourie numbers as follows:

hIC ∆x
Bi I = (6.17)
K

50
 hOC ∆x
BiO = (6.18)
K

K∆t
Fo = (6.19)
ρC∆x 2

where Bi I and Bi O are Biot numbers of inside and outside surface of the wall,
x and t are spatial and time step sizes,
, C and K are density, specific heat and conductivity of the wall material.

The conduction equation can be discretized as follows:

For internal grid points:

− FoTnj−+11 + (1 + 2 Fo)Tnj +1 − FoTnj++11 = Tnj (6.20)

For inside surface grid point:

q 
(1 + 2 Fo + 2 FoBi I )T1 j +1 − 2 FoT2j +1 = 2 FoBi I Tg + T1 j + 2 FoBi I  IR 
 hIC 
(6.21)

For outside surface grid point:

q 
− 2 FoTNj−+11 + (1 + 2 Fo + 2 FoBio )TNj +1 = 2 FoBiO Ta + TNj + 2 FoBiO  OR  (6.22)
 hOC 

In the above three equations, T refers to temperature, its superscripts and subscripts refer
to grid coordinates of time and space.

TDMA Method: The discretized equations can be written in the following general form:

− ai Ti −1 + bi Ti − ci Ti +1 = d i , i = 1, 2, L , N (6.23)

T0 = 0
(6.24)
TN +1 = 0

ai, bi, ci and di can be obtained from the coefficients of Equations (6.20)-(6.22). This is a
linear system of N variables, whose coefficient matrix is a triangle diagonal matrix. So
TDMA algorithm is used as follows:

51
For i=1, we have:

b1T1 − c1T2 = d1 (6.25)

Let us denote:

c1
β1 = (6.26)
b1

d1
υ1 = (6.27)
b1

Then we have:

T1 = β 1T2 + υ1 (6.28)

For i=2, we have:

− a 2T1 + b2T2 − C 2T3 = d 2 (6.29)

Substitute Equation (6.28) into (6.29) and let us denote:

c2
β2 = (6.30)
b2 − a 2 β 1

d 2 + a 2υ1
υ2 = (6.31)
b2 − a 2 β 1

Then we obtain:

T2 = β 2T3 + υ 2 (6.32)

Similarly, we have the following formulas:

ck
β 0 = 0, β k = , k = 1, 2, L , N (6.33)
bk − a k β k −1

d k + a kυ k −1
υ 0 = 0, υ k = , k = 1, 2, L , N (6.34)
bk − a k β k −1

52
TN +1 = 0, T j = β j T j +1 + υ j , j = N , N − 1, L, 1 (6.35)

6.4.2 Ceiling Jet Heat Transfer.

As mentioned in Section 5, for the ceiling jet-induced convection heat transfer, the heat
flux has to be integrated over the ceiling surface, and the ceiling surface is assumed to be
circular in this model. This is a problem of one-dimensional definite integral.

Rc
Qceil = ∫ f (r )dr
r =0
(6.36)

f (r ) = 2πrq ceil (r ) (6.37)

where q ceil (r ) is convection heat flux induced by ceiling jet, which can be estimated
using Equations (5.14)-(5.25).

In this implementation, the trapezoidal formula is applied as follows:

N
 f (ri ) + f (ri −1 ) 
Qceil = ∑ (ri − ri −1 )  (6.38)
i =1  2 

ri is the radius of grid point i. i=0 refers to the center of the circular surface, and i=N
refers to the edge of the circular surface.

As shown in Equations (5.15) and (5.16), the heat flux formula for impingement area and
the outside area are different, so the integration area should be divided into two parts
accordingly.

53
7. VERIFICATION EXAMPLES

In this section, three examples of model predictions are compared to three sets of
experimental data.

7.1 Standard Single Compartment

Dembsey, Pagni, and Williamson reported results of twenty fire experiments suitable for
model comparison. [59] The experiments were done in a single compartment with
dimensions of 2.5 m × 3.7 m in plan and 2.5 m in height. This is similar in size, geometry
and construction to the standard fire test compartment. The compartment had a single
doorway, 0.76 m wide × 2.0 m high, centered on one of the shorter sides. A 0.61 m ×1.22
m porous surface burner was placed into the compartment with its porous surface being
0.61 m above the floor. The propane-fired burner supplied heat at a steady rate between
330 and 980 kW for the duration of each experiment. Furthermore, three sets of
experimental data with heat release rates (HRR) at 330 kW, 630 kW and 980 kW were
compared to two comprehensive compartment fire models, CFAST and FIRST. In this
study, the FIERAsmoke model predictions are compared to the data of reference [59]. The
input and output data of the model are shown in Tables 7.1 and 7.2, respectively.

In Table 7.2, for each of the three experiments, there are four rows to describe the
experimental data, CFAST’s prediction, FIRST’s prediction and the prediction of the
current model, respectively. Fire duration of each experiment is provided, which is taken
as the fire simulation time. The fire end time is the point at which the model predictions
are compared to the experimental data. The CPU time used by each of this model for
each case run on a Pentium 166 PC is also presented in the table.

Analysis of the Results

It is shown that this model gives favorable results for the parameters compared,
especially for upper layer temperature, interface height, neutral plane height and vent
flow rates.

However, the model overestimates the floor incident radiation heat flux and the lower
wall surface temperature. This may be due to the two-surface radiation heat transfer
model. This will also be affected by the value of the radiation fraction.

Furthermore, the lower layer gas temperature is obviously underestimated, especially

when the HRR is raised. As already described, in this model, two factors affect the lower
layer temperature. One is the convection heat transfer from the lower wall and floor, the
other is the inverse cold plume entrainment when outside cold air enters the lower layer
through the upper layer. However, in real fires, there are other mechanisms affecting the

54
lower layer temperatures, such as mass mixing between upper and lower layer and
radiation absorption of the contaminated lower layer, which has not been accounted for in
this model.

Table 7.1 Zone Model Input Data for Single Compartment Example

Fuel Properties
Fuel Propane (C3H8) χH 0.18
Heat of Combustion (MJ/kg) 44 χC 0.82
Radiation Fraction[59, 60] 0.27 γ t 0
Nominal HRR (kW) 330, 630, 980 γ 0
Ambient Conditions
Pressure (Pa) 101300 Temperature (°C) 20
Thermal Properties of Compartment Boundaries
Ceiling Wall Floor
Thickness (m) 0.066 0.054 0.044
Density (kgm-3) 449 449 770
Conductivity (Wm-1K-1) 0.10 0.10 0.14
Specific heat (Jkg-1K-1) 1090 1090 900
Surface Emissivity 0.9 0.9 0.9
Dimensions
Depth (m) Width (m) Height (m)
Compartment 2.5 3.7 2.5
Vent N/A 0.76 2.0
Burner’s surface location 1.25 1.85 0.61

55
 Table 7.2 Comparison of Model Prediction with Reference [59]
TU, TL: gas temperatures of upper and lower layers, TWU, TWL: interior surface temperatures of upper and
lower walls, TFLR: interior surface temperature of floor, ZI, ZN: interface & neutral plane heights, ∆PF: floor
pressure difference to ambient, m & AV : average vent mass flow rate, q IF : floor radiant incident heat flux.

Parameters TU TL TWU TWL TFLR ZI ZN ∆PF m& AV q IF

Units (°C) (°C) (°C) (°C) (°C) (m) (m) (Pa) (kg/s) (kW/m2)
HRR=330 (kW), Duration Time=30 (min) , Run Time=17 (s)
EXPTL. 370 84 315 187 180 1.12 1.04 -1.97 0.87 3.8
CFAST 457 48 426 372 334 1.31 1.08 -1.11 0.64 16
FIRST 278 26 245 N/A N/A 1.15 N/A N/A 0.67 N/A
FIERAsmoke 408 80 359 337 249 1.29 0.98 -1.88 0.73 11
HRR=630 (kW), Duration Time = 35 (min) , Run Time=24 (s)
EXPTL. 610 179 566 342 377 1.04 0.94 -3.88 1.03 13
CFAST 730 88 711 672 655 1.18 0.96 -2.13 0.78 57
FIRST 476 36 376 N/A N/A 1.10 N/A N/A 0.71 N/A
FIERAsmoke 585 104 585 545 391 1.06 0.97 -2.44 0.83 31
HRR=980 (kW), Duration Time=20 (min), Run Time=31 (s)
EXPTL. 796 236 728 471 551 0.99 0.92 -4.36 1.01 33
CFAST 901 118 885 852 841 0.96 0.95 -2.82 0.89 107
FIRST 642 61 483 N/A N/A 1.09 N/A N/A 0.74 N/A
FIERAsmoke 785 122 785 754 531 0.89 0.93 -2.87 0.89 71

7.2 One-Compartment with Mechanical Ventilation

A series of experiments were conducted at NRC to investigate the effectiveness of smoke
management system in an atrium. [61] Three sets of data are presented in this report. [62]

A schematic diagram of the experimental facility used in this study is shown in Figure
7.1. The facility is a large compartment with dimensions of 9 m × 6 m × 5.5 m height.
The interior wall surface of the compartment was insulated using 25mm thick rock fiber
insulation.

56
 Smoke Exhaust

Opening
Smoke Layer Extension

Air Inlet Air Inlet

Fuel

Figure 7.1 Schematic of the Test Facility

A fan was used to supply fresh air into the compartment through openings in the floor
around the walls. These openings had a width of 0.1m, and a total length of 22.8 m. Inlet
air was supplied to the four sides of the room through a duct system in the under-floor
space. Thirty-two exhaust inlets with a diameter of 150 mm were located in the ceiling of
the compartment. These inlets were used to extract the hot gases from the compartment
during the tests. All exhaust ducts were connected to a central plenum. A 0.6 m diameter
duct was used between the plenum and an exhaust fan.

A square propane sand burner was used as the fire source. The burner was capable of
simulating fires ranging from 15 kW to 1,000 kW. The heat release rate of the fire was
determined using two methods. One was based on the volume flow rate of propane
supplied to the burner. The other was based on the oxygen depletion method.

Twelve CO2 inlets and fifteen thermocouples (TCs) were located at one of the room
quarter points at various heights. The TCs, together with the CO2 measurements, were
used to determine the depth of the hot layer in the compartment. The volume flow rate,
temperature, CO2 and oxygen concentrations were measured in the main exhaust duct.
These measurements were used to determine the heat release rate of the fire, as well as to
calculate the exhaust rate of the ventilation system. A pitot tube and TC, located at the
center of the duct, were used to determine the volumetric flow rate in the duct.

57
The basic input data used in the model are shown in Table 7.3. It includes fuel properties,
thermal properties of the compartment boundaries, and the compartment dimensions. In
total forty tests were conducted at NRC, from which three sets of test data are presented
in this report for analysis and comparison with model predictions.

Table 7.3 Basic Input Data of the Test Facility

Fuel Properties (Propane)

Mass fraction of hydrogen in the fuel 0.18 Heat of Combustion (MJ/kg) 44
Mass fraction of carbon in the fuel 0.82 Radiation Fraction 0.27
Thermal Properties of Dimensions (m)
Compartment Boundaries
Thickness (m) 0.025 Length Width Height
Density (kgm-3) 130 Compartment 9 6 5.5
Conductivity (Wm-1K-1) 0.08 Floor Vent 22.8 0.1 N/A
Specific heat (Jkg-1K-1) 900 Burner’s location 4.5 3.0 0.2
Surface Emissivity 0.9

Experimental data and predicted results for the three tests are shown in Tables 7.4 to 7.6.
Some points should be noted. Both the data of heat release rate and exhaust rate are time-
dependent and piece-wise linear. The given heat release rate is not based on oxygen
depletion method but fuel (propane) flow rate (complete combustion is assumed).
Exhaust rate is fitted from the data measured in the main duct. As shown in Tables 7.4 to
7.6, the first three parts of the data, i.e., basic settings, heat release rates and exhaust
rates, are taken as input data of the zone model together with the data of Table 7.3.

The experimental results, shown in these tables, are the interface heights, hot layer
temperatures, and hot layer CO2 concentrations. The interface height is obtained both
from the temperature and CO2 concentration profiles based on their maximum gradient.
The smoke layer temperature and CO2 concentration are then calculated by averaging the
measured data within the smoke layer.

The last two parts of the Tables 7.4 to 7.6 are the predicted results using both
McCaffrey’s and Heskestad’s plume models.

Analysis of the Results

It is shown that, the results of the zone model compare also well with the experimental
data.

58
For small to medium heat release rates, the interface heights determined using the
temperature profiles have no significant difference from those using the concentration
profiles. For the larger HRR, the CO2 concentration profiles give considerable higher
interface heights. Longer duct extension into the room results in higher hot layer
temperature rise, and generally, higher hot layer CO2 concentration and lower interface
height.

For the larger heat release rates, the predicted concentration is considerably higher than
the experimental data if the temperature profile is used to determine the interface height.
But if the CO2 concentration profile is used, then much better agreement is obtained.
Temperature was found to be the parameter with the best agreement to the experimental
data.

Table 7.4 Results of Test 1 with Mechanical Ventilation

Zi --- Interface Height, TU --- Smoke Layer Temperature, C CO2 --- Smoke Layer CO2 Concentration

Basic Settings Exhaust Rate

Initial Temperature (ºC) 24 Time 0 3 31 36 45
(min.)
Extension of Ports (m) 0 Rate (kg/s) 4.79 4.79 4.39 3.88 3.88
Heat Release Rate (150-250-600)
Time 0 2 15 16 30 31 45
(min.)
HRR (kW) 0 150 150 250 250 600 600
Experimental Results Experimental Results
(Maximum Temperature Gradient) (Maximum Concentration Gradient)
Time (min.) 15 30 42 Time (min.) 15 30 42
Zi (m) 3.38 3.38 2.62 Zi (m) 3.88 3.38 2.88
TU (ºC) 46 70 143 TU (ºC) 48 70 145
3 3
C CO (m /m %) 0.15
2
0.25 0.60 C CO 2
0.16 0.25 0.64
(m3/m3%)
Simulation Results Simulation Results
(McCaffrey’s Plume) (Heskestad’s Plume)
Time (min.) 15 30 42 Time (min.) 15 30 42
Zi (m) 3.90 3.58 3.03 Zi (m) 4.87 4.22 3.05
TU (ºC) 50 70 149 TU (ºC) 50 70 149
3 3
C CO (m /m %) 0.15 0.26 0.69 C CO 0.15 0.25 0.69
2 2

(m3/m3%)

59
 Table 7.5 Results of Test 2 with Mechanical Ventilation
Zi --- Interface Height, TU --- Smoke Layer Temperature, C CO2 --- Smoke Layer CO2 Concentration
Basic Settings Exhaust Rate
Initial Temperature (ºC) 26 Time (min.) 0 40 45 52.5
Extension of Ports (m) 1 Rate (kg/s) 4.79 4.22 3.76 3.76
Heat Release Rate (25-150-250-600)
Time 0 2.5 15 16 27.5 28.5 40 41 52.5
HRR (kW) 0 25 25 150 150 250 250 600 600
Experimental Results Experimental Results
Time (min.) 15 27.5 40 52.5 Time (min.) 15 27.5 40 52.5
Zi (m) 4.25 3.88 3.38 2.25 Zi (m) 3.62 3.88 3.12 2.88
TU (ºC) 33 53 76 150 TU (ºC) 32 53 72 155
3 3
C CO2 (m /m %) 0.07 0.18 0.27 0.55 C CO2 0.08 0.18 0.27 0.66
Simulation Results Simulation Results
Time (min.) 15 27.5 40 52.5 Time (min.) 15 27.5 40 52.5
Zi (m) 4.30 3.82 3.51 2.98 Zi (m) 4.30 4.30 4.10 2.99
TU (ºC) 31 53 73 154 TU (ºC) 35 57 74 155
3 3
C CO (m /m %) 0.03
2
0.15 0.27 0.71 C CO2
0.06 0.18 0.27 0.71

Table 7.6 Results of Test 3 with Mechanical Ventilation

Zi --- Interface Height, TU --- Smoke Layer Temperature, C CO2 --- Smoke Layer CO2 Concentration
Basic Settings Exhaust Rate
Initial Temperature (ºC) 7 Time (min.) 0 48 54 60
Extension of Ports (m) 2 Rate (kg/s) 5.13 4.56 3.99 3.99
Heat Release Rate (25-150-250-600)
Time 0 5 20 21 32.5 33.5 45 46 60
HRR (kW) 0 25 25 150 150 250 250 600 600
Experimental Results Experimental Results
Time (min.) 20 32.5 45 60 Time (min.) 20 32.5 45 60
Zi (m) 3.62 3.38 3.12 2.12 Zi (m) 3.12 2.88 3.38 2.88
TU (ºC) 17 37 62 131 TU (ºC) 16 35 64 137
3 3
C CO2 (m /m %) 0.10 0.2 0.30 0.55 C CO2 0.10 0.20 0.32 0.66
Simulation Results Simulation Results
Time (min.) 20 32.5 45 60 Time (min.) 20 32.5 45 60
Zi (m) 3.30 3.30 3.30 3.07 Zi (m) 3.30 3.30 3.30 3.12
TU (ºC) 15 42 59 129 TU (ºC) 20 52 71 129
3 3
C CO (m /m %) 0.05 0.20 0.30 0.67 C CO2 0.09 0.28 0.38 0.67
2

60
7.3 Two Compartments
Cooper, Harkleroad, Quitiere, and Rinkinen reported a series of multi-compartment fire
experiments.[40] In this study, a set of that experimental data is compared to model
predictions. The test space involved two compartments, a burn room and a corridor. The
heat release rate was a linearly growing fire with the relation: Q=C t, C=30 kW/min.
There was a 2.0 m high and 1.07 m wide door between the burn room and the corridor. In
addition, a 0.15 m × 0.94 m hole to the ambient space was provided in a wall surface of
the corridor. The input data for the model are shown in Table 7.7. Figures 7.2 to 7.4
present the comparisons for averaged gas temperature rise, interface height and pressure
difference between the two compartments near the ceiling height. The run time on a
Pentium 166 PC for this case is 140 seconds, the outer layer step size is 2 seconds, and
the number of iterations is 1.

Analysis of the Results

Figures 7.2 to 7.4 show that the model also gives favorable results, when compared to the
experimental data.
It is shown that the predicted gas temperature is higher than the experimental data. This is
consistent with the 330 kW case of Table 7.2, where the predicted Tu is also higher than
the experimental data.
Table 7.7 Zone Model Input Data for Two-Compartment Example

Fuel Properties
Fuel Methane (CH4) χ 0.25
Heat of Combustion (MJ/kg) 50.0 χ 0.75
Radiation Fraction[63] 0.24 γ 0
Nominal HRR (kW) 30t (t: min.) γ 0
Ambient Conditions
Pressure (Pa) 101300 Temperature (°C) 25
Thermal Properties of Compartment Surfaces
Ceiling/Wall Floor
Thickness (m) 0.013 1.0
Density (kgm-3) 790 1950
Conductivity (Wm-1K-1) 0.2 0.8
Specific heat (Jkg-1K-1) 900 860
Emissivity 0.9 0.9
Dimensions
Depth (m) Width (m) Height (m)
Burn room 3.3 4.3 2.36
Corridor 2.4 11.1 2.36
Vent from burn room to corridor N/A 1.07 2.0
Vent from corridor to ambient N/A 0.94 0.15
Burner’s surface location 1.65 2.15 0.24

61
 Model Exptl.

1.8
1.6
1.4
Pressure (Pa)

1.2
1
0.8
0.6
0.4
0.2
0
0 25 50 75 100 125 150 175 200 225 250 275 300
Time (Seconds)

Figure 7.2 Predicted and Experimental Results for Pressure Difference Between
the Burn Room and Corridor Near the Ceiling

Model-BR Exptl-BR Model-CR Exptl-CR

120
Temperature (Deg. C)

100

80

60

40

20

0
0 25 50 75 100 125 150 175 200 225 250 275 300
Time (Seconds)

Figure 7.3 Predicted and Experimental Results for Averaged Gas Temperature of
Burn Room (BR) and Corridor (CR)

62
 Model-BR Exptl-BR Model-CR Exptl-CR
2.50
Interface Height (m)

2.00

1.50

1.00

0.50

0.00
0 25 50 75 100 125 150 175 200 225 250 275 300 325 350
Time (Seconds)

Figure 7.4 Predicted and Experimental Results for the Interface Heights of the Burn
Room (BR) and Corridor (CR)

63
8. REFERENCES

[1] Jones, W.W. & Forney, G.P., Improvement in predicting smoke movement in
compartment structures. Fire Safety Journal 21(1993)269.
[2] Jones, W.W., A multi-compartment model for the spread of fire, smoke and toxic
gases. Fire Safety Journal 9(1985)55.
[3] Peacock, R.D., Forney, G.P., Reneke, P., Portier, R. & Jones, W.W., CFAST, the
Consolidated Model of the Fire Growth and Smoke Transport. NIST Technical
Note 1299 (1993).
[4] Jones, W.W. & Peacock, R.D., Refinement and experimental verification of a
model for fire growth and smoke transport. Second International Symposium on
Fire Safety Science, 1988.
[5] Quintiere, J.G., Compartment fire modeling. The SFPE Handbook of Fire
Protection Engineering, Quincy, Mass.: National Fire Protection Association,
1995, p. 3-125.
[6] Fu, Z. & Fan, W., A zone-type model for a building fire and its sensitivity analysis.
Fire and Materials, 20(1996) 215.
[7] Friedman, R., An international survey of computer models for fire and smoke.
Journal of Fire Protection Engineering, 4(1992)81.
[8] Jones, W.W., The Evolution of Hazard, the fire hazard assessment methodology.
Fire Technology, 2nd quarter, 1997, p. 167.
[9] Walton, W.D., Zone computer fire models for enclosures. The SFPE Handbook of
Fire Protection Engineering, Quincy, Mass.: National Fire Protection Association,
1995, p. 3-148.
[10] Kawagoe, K., Fire behavior in rooms. Building Research Institute, Ministry of
Construction (Japan), Report No. 27, Tokyo, 1958.
[11] Cox, G., Compartment fire modeling, Chapter 6 of “ Combustion Fundamentals of
Fire”, edited by Cox, G., Academic Press, 1995.
[12] Janssens, M., Room fire models, Chapter 6 of “Heat Release in Fires”, edited by
Babrauskas, V. and Grayson, S.J., Elsevier Applied Science, 1992.
[13] Thomas, P.H., Hinkley, P.L., Theobald, C.R. and Simms, D.L., Investigations into
the flow of hot gases in roof venting. Fire Research Technical Paper No.7, HMSO,
London, 1963.
[14] Peacock, R.D., Davis, S. & Babrauskas, V., Data for room fire model
comparisons. Journal of Research of the NIST, 96(1991)411.
[15] Fowkes, N.D., “A mechanistic model of the 1973 and 1974 bedroom test fires” in
Croce, P.A., ed., “A study of room fire development: The second full-scale
bedroom fire test of the home fire project (July 24, 1974)”, Vol. II, FMRC Tech.
Rept. No. 21011.4, pp. 8-50, 1975.

64
[16] Quintiere, J.G., “Growth of fire in building compartments” in Robertson, A.F., ed.,
Fire Standards and Safety, ASTM STP614, American Soc. For Testing and
Materials, pp. 131-167, 1977.
[17] Pape, R. & Waterman, T., Modification to the FIRES pre-flashover room fire
computer model, IITRI Project J6400, IIT Res. Inst., Chicago, 1977.
[18] Mitler, H.E., The physical basis for the Harvard computer fire code. Home Fire
Project Report 34, Harvard University, 1978.
[19] Barnett, J.R., The WPI/Fire room fire computer model. Proceedings of the 1st
Asian Conference on Fire Science and Technology, Hefei, China 1992, P. 313.
[20] Tanaka, T., A model of multi-room fire spread. IR 83-2718, National Bureau of
Standards, Washington, D. C., 1983.
[21] Mitler, H.E. & Rockett, J.A., Users’ guide to FIRST, a comprehensive single-
room fire model. NBSIR 87-3595, NBS, Gaithersburg, MD, 1987.
[22] Jones, W.W. & Peacock, R.D., Technical Reference Guide for FAST version 1.8.
NIST Technical Note 1262, NIST, Gaithersburg, MD, 1989.
[23] Cooper, L.Y. & Forney, G.P., The consolidated compartment fire model (CCFM)
computer code application CCFM.VENTS, Parts 1, 2, 3 and 4. NIST,
Gaithersburg, MD, 1990.
[24] Yung, D., Hadjisophocleous, G.V., Proulx, G. & Kyle, B.R., Cost effective fire
safety upgrade options for a Canadian government office building. Proceedings of
the International Conference on Performance Based Codes and Fire Safety Design
Methods, Ottawa, Canada, 1996.
[25] Hadjisophocleous, G.V. & Yung, D., A model for calculating the probabilities of
smoke hazard from fires in multi-story buildings. Journal of Fire Protection
Engineering, 4(1992)67.
[26] Takeda, H. & Yung, D., Simplified fire growth models for risk-cost assessment in
apartment buildings. Journal of Fire Protection Engineering, 4(1992)53.
[27] Hadjisophocleous, G.V. and Torvi, D.A., Development of a Fire Risk Assessment
Computer Model for Light Industrial Buildings. Canadian Society of Mechanical
Engineering (CSME) Forum: Volume 3, 1998, Ryerson Polytechnic University,
Toronto, ON, pp. 305-310
[28] Wylen, G.J.V. & Sonntag, R.E., Fundamentals of Classical Thermodynamics, 2nd
edition, John Wiley & Sons, 1978.
[29] Forney, G.P., Analyzing and exploiting numerical characteristics of zone fire
models. Fire Science and Technology, 14(1994)49.
[30] Huggett, C., Estimation of the rate of heat release by means of oxygen
consumption. Journal of Fire and Flammability, 12(1980).
[31] Parker, W.J., Calculation of the heat release rate by oxygen consumption for
various applications. Journal of Fire Sciences, 2(1984)380.
[32] Standard Test Method for Heat and Visible Smoke Release for Materials and
Products Using Oxygen Consumption Calorimeter, ASTM E1354-90, ASTM,
Philadelphia, PA 1990.

65
[33] Morehart, J.H., Zukowski, E.E. & Kubota, T., Characteristics of large diffusion
flames in a vitiated atmosphere. 3rd International Symposium on Fire Safety
Science, Edinburgh, 1991.
[34] McCaffrey, B.J., Momentum implications for buoyant diffusion flames.
Combustion and Flame, 52(1983) 149.
[35] Cetegen, B.M., Zukoski, E.E. & Kubota, T., Entrainment in the near and far field
of fire plumes. Combustion Science & Technology, 39(1984)305.
[36] Zukoski, E.E., Mass flux in fire plumes. Proceedings of the Fourth International
Symposium on Fire Safety Science, International Association for Fire Safety
Science, 1994.
[37] Dembsey, N.A., Pagni, P.J. & Williamson, R.B., Compartment near-field
entrainment measurements. Fire Safety Journal, 24(1995) 383.
[38] Heskestad, G., Fire plumes. The SFPE Handbook of Fire Protection Engineering,
Quincy, Mass.: National Fire Protection Association, 1995, p2-9.
[39] Mowrer, F. W. & Williamson R. B., Estimating room temperatures from fires
along walls and in corners. Fire Technology, 23(1987)133.
[40] Cooper, L.Y., Harkleroad, M., Quintiere, J.G. & Rinkinen, W.J., An experimental
study of upper hot layer stratification in full-scale multi-room fire scenarios.
ASME Journal of Heat Transfer, 104(1982)741.
[41] Cooper, L.Y., Combined buoyancy and pressure-driven flow through a shallow,
horizontal, circular vent. Journal of Heat Transfer, 117(1995)659.
[42] Cooper, L.Y., VENTCF2: an algorithm and associated FORTRAN 77 subroutine
for calculating flow through a horizontal ceiling/floor vent in a zone-type
compartment fire model. Fire Safety Journal, 28(1997)253.
[43] Cooper, L.Y., Calculating combined buoyancy- and pressure-driven flow through
a horizontal, circular vent - Application to a problem of steady burning in a
ceiling-vented enclosure. Fire Safety Journal, 27(1996)23.
[44] Achakji, G.Y. and Tamura, G.T., Pressure drop characteristics of typical
stairshafts in high-rise buildings. ASHRAE Transactions, 100 (1994) 1223.
[45] Walton, G. N., CONTAM96 User Manual. NISTIR 6056, Gaithersburg, MD,
1997.
[46] Kreith, F., Principles of Heat Transfer, 3rd edition, Intext Educational Publishers,
New York, 1976.
[47] Holman, J.P., Heat Transfer, McGraw-Hill, 1986, p. 323
[48] Cooper, L.Y., Heat transfer from a buoyant plume to an unconfined ceiling.
ASME Journal of Heat Transfer, 104(1982)446.
[49] Cooper, L.Y., The buoyant plume-driven adiabatic ceiling temperature revisited.
ASME Journal of Heat Transfer, 108(1986)822.
[50] Motevalli, Y. & Ricciuti, C., Characterization of the confined ceiling jet in the
presence of an upper layer in transient and steady-state conditions. NIST-GCR-92-
613, 1992.
[51] Fu, Z. & Fan, W., Calculation of the radiation heat transfer in a zone model for
building fires. Journal of China University of Science and Technology,
26(1996)375. (in Chinese)

66
 [52] Siegel, R. & Howell, J., Thermal Radiation Heat Transfer, 2nd edition, Hemisphere
Publishing Corporation, 1981.
[53] Dietenberger, M.A., Technical Reference and User’s Guide for FAST/FFM
Version 3, NIST-GCR-91-589.
[54] Rehm, R.G. & Forney, G.P., A note on the pressure equations used in zone fire
modeling. NISTIR-4906, 1992.
[55] Gear, C.W., Numerical Initial Value Problems in Ordinary Differential Equations,
Prentice Hall, Englewood Cliffs, NJ, 1971.
[56] Shampine, L.F., Numerical Solution of Ordinary Differential Equations, Chapman
& Hall, New York, 1994.
[57] Petzold, L., Automatic selection of methods for solving stiff and nonstiff systems
of ordinary differential equations. SIAM J. SCI. STAT. COMPUT., 4(1983)136.
[58] Mathews, J.H., Numerical Methods for Mathematics, Science, and Engineering,
Prentice Hall, Englewood Cliffs, New Jersey, 1992.
[59] Dembsey, N.A., Pagni, P.J. & Williamson, R.B., Compartment fire experiments:
comparison with models. Fire Safety Journal, 25(1995)187.
[60] Tewarson, A., Generation of heat and chemical compounds in fires. In the SFPE
Handbook of Fire Protection Engineering, 1st ed., Ch. 1-13. National Fire
Protection Association, Quincy, MA, USA, 1988.
[61] G. Lougheed & G. Hadjisophocleous, Investigation of Atrium Smoke Exhaust
Effectiveness. ASHRAE Annual Meeting, Boston, 1997.
[62] Hadjisophocleous, G.V. & Fu, Z., “Modeling smoke conditions in large
compartments equipped with mechanical exhaust using a two-zone model”.
International Symposium on Engineering Performance-Based Fire Codes, Hong
Kong, September 8, 1998.
[63] McCaffrey, B., Measurements of the radiative power output of some buoyant
diffusion flames, Western States Section, Combustion Institute, WSS/CI 81-15,
Pullman, Wash., 1981.

67

View publication stats