Spread of smoke and fire gases via the ventilation system.

Gordonova, Polina

1997

Link to publication

Citation for published version (APA):

Gordonova, P. (1997). Spread of smoke and fire gases via the ventilation system. Division of Building Science,
Lund University.

General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors
and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the
legal requirements associated with these rights.

• Users may download and print one copy of any publication from the public portal for the purpose of private study
or research.
• You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal
Take down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove
access to the work immediately and investigate your claim.

LUNDUNI
VERSI
TY

PO Box117
22100L und
+4646-2220000
Lund University • Sweden
Lund Institute of Technology
Department of Building Science
Report TABK--97 I lOll

Spread of Smoke and Fire

Gases via the Ventilation
System
/

Polina Gordonova
Lund University
Lund University, with eight faculties and a number of research cen-
tres and specialized institutes, is the largest establishment for re-
search and higher education in Scandinavia. The main part of the
University is situated in the small city of Lund which has about
97 600 inhabitants. A number of departments for research and edu-
cation are, however, located in Malmo. Lund University was founded
in 1666 and has today a total staff of 5 850 employees and 38 200
students attending 50 degree programmes and 850 subject courses
offered by 170 departments.

Department of Building Science

The Department of Building Science is part of the School ofArchitec-
ture within the Faculty ofTechnology. The Department has two pro-
fessorial chairs, Building Science and Building Services. Research at
the Department is concentrated on energy management, climatic con-
trol and moisture problems. The main areas of research are:

• design and preformance of new low-energy buildings

• energy conservation in existing buildings
• utilization of solar heat
• climatic control
• climatic control in foreign climates
• moisture research
Spread of smoke and fire
gases via the ventilation
.system

Polina Gordonova

This report relates to Research Grant No 050-942 from the Swedish Fire Research Board to the
Department of Building Science, Lund University, Lund Institute of Technology, Lund, Sweden.
Spread of Smoke and Fire Gases via the Ventilation System

Keywords
Compartment fires, fire flow, fire pressure, fire simulation, smoke
spread, ventilation ducts.

© copyright Polina Gordonova and Department of Building Science,

Lund University, Lund Institute of Technology, Lund, Sweden, 1997
The English language corrected by L. J. Gruber BSc(Eng) MICE MIStructE

Printed by KFS AB, Lund 1998

Report TABK--97/1011
Spread of Smoke and Fire Gases via the Ventilation System
Lund University, Lund Institute of Technology, Department of Building Science

Lund Institute of Technology

Department of Building Science Telephone: int + 46 46 222 73 45
P.O. Box 118 Telefax: int + 46 46 222 47 19
SE-221 00 LUND e-mail: [email protected]
Sweden Homepage: http://www.bkl.lth.se

2
 Abstract

Abstract

Ventilation ducts provide possible routes for smoke spread be-

tween adjacent rooms in case of a compartment fire. However, the
probability of this and the conditions which lead to such a phe-
nomenon have not been fully examined yet. According to the rec-
ommendations in the building regulations/ codes of different coun-
tries, the only way smoke spread can be prevented is to install dif-
ferent kinds of dampers or smoke barriers and to shut down the
ventilation system in the event of fire, or to rely only on shutting
down the ventilation system. In order to improve the fire safety of
ventilation systems, an investigation has been made in this proj-
ect whether it is possible to make use of the ordinary existing
ventilation system to remove hot smoky gases from the compart-
ment, without spreading them to adjacent rooms. A method by
which the smoke spread via the ventilation system can be pre-
vented by running the fans is presented in this thesis.
Smoke spread is determined by the rate of heat release by the
fire, the airtightness of the building or room, and the layout of the
ventilation system. The heat released by the fire results partly in
a volumetric expansion which can be interpreted as a flow. The
project which focuses on four different types of rooms uses data
based on measured heat release rate-time curves and the chosen
characteristics of the room, in order to predict fire behaviour and
its consequences with the help of the computer simulation pro-
gram CFAST (HAZARD!). The airtightness (leakage of spaces),
and the durability of windows during the fire, have been investi-
gated.
The results obtained from simulations have been used to predict
the largest fire flow produced by the expansion, and with the help
of the computer program PFS which treats arbitrary flow sys-
tems, this has been compared to the "boundary" safe flow.

3
Spread of Smoke and Fire Gases via the Ventilation System

4
 Table of Contents

Table of Contents

Abstract 3

Contents 5

Preface and acknowledgements 9

Summary 11

Sammanfattning (in Swedish) 17

List of symbols and definitions 23

1 Introduction 27
1.1 Background 28
1.2 Objective and limitations 29
1.3 Comprised fields of knowledge 29

Section 1 Analysis of fire behaviour in single spaces 31

2. Fire behaviour in single spaces 31

2.1 Statistic estimation 31
2.2 Estimating of basic parameters in compartment fires 35
2.2.1 The stages of fire growth 35
2.2.2 t 2 fire growth 37
2.2.3 Measured heat release fires 38
2.2.4 Compartment fire model 41
2.3 Computer methods for predicting compartment fires 43
2.3.1 CFAST (HAZARD!) 43
2.3.2 DSLAYV 45

5
Spread of Smoke and Fire Gases via the Ventilation System

2.3.3 SMAFS 45
2.4 Airtighteness 46

3. Fire simulations 49
3.1 Input data used in simulations 49
3.1.1 Geometrical data 49
3.1.2 Fire scenario and fire position 51
3.1.3 Assumed airtighteness 51
3.2 Analysis of parameters evaluated by CFAST 52
3.2.1 Predicted variables 52
3.2.2 Peak pressures and peak temperatures
in temperatures in fire rooms 57
3.2.3 Mass and volume flow 59
3.3 Comparisons of predicted parameters with
experimental data 60
3.4 Analysis of parameters evaluated by DSLAYV
andSMAFS 61
3.5 Window breakage modelling 63
3.5.1 Study of the literature 64
3.5.2 Simple window breakage model 65
3.5.3 Window glass temperature model 66

4. Single peak fire flow models 71

4.1 Simple general model 73
4.1.1 Maximum fire flows predicted with DSLAYV
results 74
4.2 Analysis of flow prediction models 75
4.2.1 Flows, predicted by CFAST simulations 75
4.2.2 Model 1. Used linear segment fires as tn - fires 78
4.2.3 Model2 79
4.2.4 Model3 80
4.3 Flows predicted by computer simulations
and different models 82
4.4 Generalisation of predicted flows 85
4.4.1 Equations describing flows as a function of
volume 85

6
 Table of Contents
Section 2 Operation of ventilation systems
during fires 93

5. Study of the literature 93

5.1 Databases 94

6. Theoretical methods 97
6.1 Background 97
6.2 Flow model 98
6.3 Theoretical model for the analysis of ventilation
systems 99
6.3.1 Exhaust ventilation system 100
6.3.2 Exhaust and supply ventilation system without
air leakage 103
6.3.3 Exhaust and supply ventilation system with
air leakage 107

7. Design of ventilation systems 109

7.1 Characteristic ventilation systems 109
7.2 Fire compartments 110

8. Models and tools 113

8.1 PFS-program 113
8.2 Schematic model 114
8.2.1 Exhaust ventilation 117
8.2.2 Closed supply-exhaust ventilation
(without air leakage) 120
8.2.3 Open supply-exhaust ventilation
(with air leakage) 123
8.2.4 Open supply-exhaust ventilation
with a "zero" - duct 127
8.2.5 Some conclusions on different types
of flows and pressure estimation 131
8.3 Measures to prevent smoke spread via
ventilation ducts 132
8.3.1 Examination of possible measures to prevent
smoke spread 132
8.3.2 Analysis of risk of smoke spread in the
examined cases 133

7
Spread of Smoke and Fire Gases via the Ventilation System

9. "Real case" study 143

9.1 Discription of the main types of ventilation
systems 143
9.2 Designed air flows 145
9.3 Exhaust ventilation system 146
9.3.1 Exhaust ventilation system with
single outlet in each room 146
9.3.2 Exhaust ventilation system with
two terminals in each room 165
9.4 Supply-exhaust ventilation system
with a "zero-duct" 167
9.5 Supply-exhaust ventilation system 169

10. Evaluation of the toxicity of diluted smoke 173

10.1 Background 173
10.2 Evaluation of total volume of spread smoke 174
10.3 Evaluation of relative smoke spread
to a single room 174
10.4 Evaluation of CO - concentration in a single room 177
10.5 Dangerous levels of CO - concentration 177
10.6 Evaluation of CO- dose 178
10.7 Application 180

11. Evaluation of visibility of diluted smoke 183

11.1 Background 183
11.2 Evaluation of visibility in diluted smoke spread to a
single room 185

12. "Cold" case assessment 187

13. The "Alexander"-method 191

14. Discussions and conclusions 195

References 207

Appendix 211

8
 Preface and Acknowledgements

Preface and acknowledgements

Work on this thesis began in 1995 at the Department of Building

Science, Lund Institute of Technology, Lund University.

I would like to express my sincere thanks to my supervisor Lars

Jensen, the head of the Department Bertil Fredlund, Hakan
Hakansson, and to the staff of the Department of Building Sci-
ence.

In particular, I want to thank Sven Eric Magnusson, Goran

Holmstedt, Bjorn Karlsson, Hakan Franzich, Lars Goran
Bengtsson and the staff of the Department of Fire Safety Engi-
neering for their support, good ideas and help in supplying data
and information. Guest researcher Y an Zhenghua is especially ac-
knowledged for field model program fire simulation.

I would like to express my sincere thanks to Tor Aurell, the con-

sultant, who helped me in searching through certain databases.

I would also like to address my warmest thanks to the members of

the reference group: Staffan Bengtsson, Sven Palmkvist, Claes
Malmqvist och Hauker Ingasson for their support and long-term
interest.

This work was funded by the Swedish Fire Research Board, which
is gratefully acknowledged.

9
Spread of Smoke and Fire Gases via the Ventilation System

10
 Summary

Summary

The following study examines the problem of smoke spread via

ventilation ducts to adjacent room(s) in case of a compartment
fire. Although the problem is well known, the probability of this
and the conditions which lead to such a phenomenon have not
been fully examined yet. The main goal of this study is to examine
the possibility of operating the fans during different fires in four
types of buildings (compartment, hospitals, offices and hotels) and
to estimate the risk of smoke spread in different conditions.
Whether or not smoke spread will occur is determined by the
energy released by fire, the airtighteness of buildings and the lay-
out and properties of ventilation systems.
In this connection it is extremely significant to assess the rela-
tionship between typical fires in the above types of buildings, the
airtightness of the construction and the layout of the ventilation
system, and to give recommendations on good practice to diminish
or even eliminate the risk of smoke spread through ventilation
ducts to the adjacent rooms.
For the purpose of this study the following subjects were evalu-
ated:
most typical fire cases for chosen types of buildings,
internal and external airtightness of the chosen rooms,
window breakage,
peak fire flows,
fire compartment groups,
ventilation system design (principle features),
real ventilation system design,
measures for prevention of smoke spread via ducts,
toxicity of spread dilute smoke,
visibility of spread dilute smoke,
cold ventilation flows vs. real temperatures of ventilation flows.

11
Spread of Smoke and Fire Gases via the Ventilation System
According to statistical estimation of the principal dangers and
damage caused by fire, presented in different sources, the most
probable items in which fires initiate in the chosen rooms are: sofa
[6] and bed [7] in dwelling houses, bed [9] in hospitals, equipment
and furnishing [10] in offices and bed [6] in hotels. "Real" fire
growth curves, based on measured heat release rate vs. time
during full-scale burning tests of these items, are chosen to
simulate fires in spaces. The fires analysed are single-room fires,
limited to the item in which the fire started.
The objective of airtighteness evaluation is to summarise
knowledge about the real magnitude of leakage in relation to the
whole building envelope including interior building constructions,
for the studied types of enclosures, built at different times. For
the purpose of this study, air leakage in accordance with the
Swedish Building Regulations, BBR 94, applied to the whole
buildings envelope, can be accepted as a suitable and probable
value.
Different computer programs are used to simulate fire progress
in chosen rooms and the parameters obtained with these are com-
pared. CFAST (Consolidated Model of Fire Growth and Smoke
Transport), which is part of the computer program HAZARD I, is a
basic tool of fire simulations in the presented study. Two other
programs: DSLAYV and SMAFS (Smoke Movement and Flame
Spread) are also used for fire simulations in order to compare
predicted parameters. Both programs, CFAST and DSLAYV, use
a two-layer model of fire, while SMAFS is a field model program.
Comparison of parameters predicted by CFAST with those obtai-
ned with DSLAYV and SMAFS shows that the order of magnitude
of values is reasonable.
One of the most important issues is to estimate the characte-
ristic parameters/features that are produced by certain fires, and
to assess their influence on the pressure/flow ratio in the ventila-
tion system. Peak mass flows, peak temperatures and peak pres-
sures, predicted by computer simulations, are analysed and com-
pared to each other. Results are presented in the form of a table
and show peak values for temperature, exceeding 340°C in certain
cases, and peak values for pressure which vary widely between 30
and 2100 Pa.
In this connection, evaluation of both the time at which the
window breaks and the temperature in the fire room at which this

12
 Summary

occurs is of great importance. An analysis is made in order to

summarise present knowledge concerning the behaviour of gla-
zing in fire. In addition, a simple window breakage model is pre-
sented for the evaluation of the time of breakage and temperature
differences in the glass. Analysis shows that glass cracks due to
thermal stress if the temperature difference in the glass itself va-
ries between 58°C and 100°C. The corresponding temperature in
the fire room varies between 200°C and 400°C.
Volume fire flows, predicted by computer simulations for differ-
ent fires in studied rooms of different geometry, are analysed for
each type of room. A regression analysis of the largest predicted
fire flows is made with the statistical programs Minitab and EX-
CEL. Several theoretical models for the assessment of fire flows
as a function of heat released in assumed fires are also presented.
Comparison of flows predicted in such a way with those predicted
by CFAST leads to the generalisation of peak fire flows for each
type of room as a function of room volume.
Characteristic flow curves baced on the equations made with
the computer program EXCEL for the chosen buildings, of vol-
umes ranging from 30 m 3 to 240 m 3 , are presented. These volumes
are assumed to be those most usual for the described types of
buildings.
For the purpose of comparison, both the presented flows and
flows obtained with DSLAYV for l - fires are plotted in the same
diagram.
A search has been made through research publications in diffe-
rent databases in science and technology and via several search
program in Internet to ascertain the present state of knowledge
regarding ordinary mechanical ventilation systems during
compartment fires. Current publications on this subject in diffe-
rent scientific and technical periodicals were also followed, and an
attempt was also made to discuss the subject with researches via
Network's scientific news groups. The author was unsuccessful in
getting information on corresponding research with regard to the
operation of ventilation systems during a fire or, at least, in
discussing its possibilities.
For the purpose of evaluating smoke spread through ducts, a
computer program PFS, which treats arbitrary flow systems, was
used. The "boundary" safe flows assessed with PFS has been com-
pared with the predicted largest fire flows obtained in this study.

13
Spread of Smoke and Fire Gases via the Ventilation System
The boundary case for smoke spread via ventilation ducts can be
described by zero airflux in the adjacent room(s) and also zero
pressure in the connecting junction between the fire room and the
main duct. All flows and pressure exceeding the "boundary" va-
lues cause smoke spread.
A description of basic principles and theoretical model of ex-
haust, supply-exhaust (open and closed) mechanical ventilation
systems serving two rooms, which determine boundary conditions
of smoke spread through ventilation ducts, are presented in the
report.
The most typical ventilation systems used in the studied types
of buildings and fire compartments in these are discussed in the
report. The analysis focuses on
-exhaust ventilation system with one exhaust fan and a distri-
bution system consisting of one or two main ducts with branches
to each compartment,
-supply-exhaust ventilation system with supply and relief fans
and distribution ductwork to each compartment,
-supply-exhaust ventilation system with a "zero" -duct. This me-
ans that supply air is distributed via terminals directly to the
compartments, while relief (exhaust) air collects in corridors
("zero" -ducts) and is removed via relief dampers by the exhaust
ventilation system.
In all types of ventilation systems the plenum can be both verti-
cal and horizontal. These ventilation systems can include different
variations of system layouts, and the degrees of complexity of the-
se systems can be widely different.
The report presents an evaluation of boundary conditions for
smoke spread in the types of ventilation systems discussed above
with four branches, general measures for the prevention of smoke
spread via ventilation ducts, and a brief description of PFS-
elements and layout.
PFS provides the opportunity to describe resistance properties
of wyes and junctions in two different ways: as a quadratic
function and with the help of empirical equations. Both types are
tested. As this study is interested in "abnormal" flow behaviour,
i.e. changes in flow directions and the consequent decrease in flow
in certain parts of junctions to zero value, it is not obvious how
wyes would respond. Higher flows cause higher velocities in ducts,
and, as a result, higher Reynolds numbers. Along with increasing

14
 Summary
flow, turbulence increases pressure field described as quadratic
function. For these reasons, the resistance properties of wyes and
junctions described as a quadratic function is used in further
analysis.
Different solutions are analysed for the prevention of the smoke
spread to the adjacent room(s) via ventilation systems with four
branches. There are
shutting down the fan(s),
doubling the speed of rotation of the fan(s),
converting the supply fan into exhaust fan,
bypassing support plenum to the exhaust plenum
and a combination of these measures with ordinary fan function
and with each other.
Fire flow exceeding boundary fire flow, obtained in previous cal-
culations, is used as default flow. The studied cases do not cover
all possible measures for the prevention of smoke spread through
ventilation ducts, but they show method of evaluating the danger
due to this.
The "real case" study examines the risk of smoke spread
through the chosen ventilation systems in buildings of higher de-
gree of complexity. Ordinary design air flows are assessed in the
examined buildings. Different cases are studied in order to esti-
mate boundary fire flows and corresponding pressures in different
rooms. This problem is very numerically complex for such venti-
lation systems with numerous branches, because the flow in the
boundary room must be equal to zero, while other flows and pres-
sures are much higher then normal. Because of this, fire flows ob-
tained from fire simulations are used in smoke spread evaluation
in the studied ventilation systems. "Fires" are placed in different
rooms and the resulting smoke spread to the adjacent room(s), as
well as different measures of preventing this phenomenon, are
studied.
The problem of evaluation the toxicity of spread diluted smoke
is discussed in the report. This evaluation regarding the concent-
ration of CO is made step by step for an exhaust ventilation sys-
tem with a shut-down fan. An assessment of
the total volume of smoke spread,
relative smoke spread to a single room, including evaluation of
the proportion of
smoke reaching one of n rooms

15
Spread of Smoke and Fire Gases via the Ventilation System
the concentration of CO in a single room is presented in the
study. All rooms are assumed to be identical.
The concentration of CO, together with the duration of exposure
and the type of activity during which exposure takes place, de-
termines the CO-doze and toxic effects. Different recommenda-
tions about acceptable and dangerous GO-concentrations and also
CO-dozes are discussed. An evaluation of GO-concentrations in
ppm, corresponding to the IDLH (immediately dangerous to life
and health) CO level, limited to 1500 ppm during 30 min, which
reaches one of the n rooms is also presented.
Analogously with the evaluation of the toxicity of the diluted
spread smoke, evaluation of the visibility of this is discussed in
the report. The expressions presented give mean values of asses-
sed visibility in the space.
The whole study assumes that the examined flows, including fi-
re flows, have ambient parameters, "cold" flows. It is well known
and also as shown by compartment fire simulations that fire flow
temperatures can exceed 300°C - 340°C. This means that the as-
sessed smoke spread via ventilation ducts can differ from that in
reality. Evaluation of friction pressure drop in straight ducts
shows that it decreases to approximately one half as the tempera-
ture in K doubles. The pressure drop in wyes and junctions is mo-
re difficult to evaluate. At the same time, along with smoke dilu-
tion and heat loss downstream from the main duct, flow tempera-
tures decrease and "cold" flows become default ones.
One more method for the evaluation of possible smoke spread
via exhaust ducts is presented in the study. The "Alexander" met-
hod is a method for manual calculations, based on the" cut-off'
principle. The "cut-off' principle means, that a branch(es) at the
boundary conditions with zero pressure is cut off and the rest of
ventilation system is calculated once more regarding the new
pressure/flow conditions.
The problem examined in the present study appears complex,
and the assumptions, made with a certain lack of knowledge and
a paucity of empirical data, cannot diminish the uncertainty of the
results obtained. The results should not therefore be interpreted
as some sort of total solution to the elimination of smoke spread,
but should be considered as a guidance in such types of analysis.

16
 Sammanfattning

Sammanfattning

Projektet undersoker rokspridning via ventilationskanaler till an-

dra narliggande rum vid handelse av en brand. Trots att proble-
met ar valkant, ar sannolikheten av att fenomenet intraffar och
villkor som leder till detta inte utforskat an. Projektets syfte ar
att studera mojligheter for att utnyttja flaktar i drift vid handelse
av en brand i sma lokaler i bostader, sjukhus, kontor och hotell;
samt att uppskatta risken for rokspridning via ventilationssystem
i des sa byggnader.
Risk for rokspridning beror pa utvecklad varmeeffekt, byggnad-
ens lackage och ventilationssystemets egenskaper.
I detta sammanhang ar det viktigt att uppskatta ett samband
mellan de mest typiska branderna for ovan presenterade typer av
byggnader, byggnadens tathet och utformning av ventilationssys-
tem; samt att ge rekommendationer for minskning eller uteslut-
ning av risk for rokspridning via ventilationssystem.
For detta syfte ar foljande parametrar uppskattade:
de mest typiska branderna for de utvalda typerna av lokaler,
yttre och inre tathet hos brandceller,
fonstersprangning,
maximallt brand-flode,
brandcellsindelning,
forenklade ventilationssystem (principiella losningar ),
"verkliga" forgrenade ventilationssystem,
atgarder mot rokspridning via ventilationskanaler,
giftighet av utspadd rok,
siktbarhet av utspadd rok,
"kallt fall" kontra "hett fall".
Enligt statistisk data for brandskador av saval personer som
egendom, ar de mest sannolika brandforemalen i de analyserade
lokalerna: soffa och sang i bostader, sang i sjukhus, kontorsin-

17
Spread of Smoke and Fire Gases via the Ventilation System
redning i kontor och sang i hotell. ''Verkliga" brandkurvor
(varmeeffekt-tid samband) vilka erhallits vid fullstandig for-
branning av ovan presenterade foremal ligger till grund for rums-
brandssimuleringar. Brander antas vara begransade till ett rum
samt till det foremal, branden borjade i.
For att uppskatta yttre och inre tathet av analyserade lokaler
har en sammanstallning av kunskap betraffande hus byggda vid
olika tider genomforts. BBR 94:s krav pa hur stor en byggnads
otathet hogst far vara, tillampad pa hela omslutande skalet
inklusive inre konstruktioner, antas dimensionerande for det har
projektet.
For att uppskatta brandforlopp i en lokal har olika simulering-
sprogram anvants och resultatparametrar har jamforts. CFAST
(Consolidated Model of Fire Growth and Smoke Transport) som
ingar i dataprogrammet HAZARDl ar ett huvudsimuleringspro-
gram i det har projektet. Tva andra simuleringsprogram:
DSLAYV och SMAFS (Smoke Movement and Flame Spread) an-
vands ocksa for att kunna jamfora resultaten. Bade CFAST och
DSLAYV ar program baserade pa tvazonsbrandmodeller. SMAFS
ar ett faltmodelsprogram. Jamforelse mellan resultatparametrar
erhallna med CFAST med motsvarande parametrar erhallna med
DSLAYV och SMAFS visar varden inom samma storleksordning.
Maximalt brandmassflode, maximal temperaturer och maxi-
malt brandtryck erhallna med hjalp av simuleringarna har ana-
lyserats och jamforts med varandra. De maximala temperatur-
erna overskrider i vis sa fall 340°C och maximalt brandtryck veri-
erar avsevart mellan 30 och 2100 Pa.
I detta sammanhang verkar det viktigt att kunna uppskatta
tidpunkt och temperaturer i ett brandrum da fonster gar sander.
Sammanstallning av kunskap om brandpaverkan pa fonster ar
gjord. Dessutom presenteras en enkel modell for uppskattning av
tidpunkt och temperaturskillnad i sjalva fonsterglaset da det gar
sander. Analysen visar att glaset spricker p.g.a. termisk spanning
vid temperaturskillnader i glaset av 58°C - 100°C. Motsvarande
temperaturer i brandrum varierar mellan 200°C och 400°C .
Brandvolymflode erhallna med hjalp av simuleringar for olika
brander i de undersokta typer av byggnaderna med varierande lo-
kalgeometri analyseras i projektet for varje typ av lokaler var for
sig. Regressionsanalys av de maximala flodena ar gjord med hjalp
av statistiskt program Minitab och EXCEL. Dessutom presen-

18
 Sammanfattning
teras nagra teoretiska modeller skapade for denna analys. Mod-
eller syftar pa att uppskatta brandflode som funktion av brandef-
fekt, utvecklad i de studerade rummen. Generalisering av de pa
ovannamnda satt uppskattade brandflodena efter jamforelse av
des sa med brandflodena fran CFAST simuleringarna resulterar i
beskrivning av brandflode som funktion av rumsvolym. Ekva-
tioner erhallna med hjalp av dataprogrammet EXCEL visar bra
overensstammelse och kan anvandas for brandflodesuppskattning
i de studerade byggnadstyperna. Motsvarande brandflode -
rumsvolym kurvor for rumsvolym mellan 30 m 3 och 240 m 3 pre-
senteras i rapporten. Dessa rumsvolymer antas vara de vanligaste
i de studerade byggnaderna. Ett kombinerade brandflode -
rumsvolym diagram till innehallande kurvor ritade enligt bade
beskriven metod och for traditionella al brander presenteras i
rapporten.
Olika databaser i vetenskap och teknologi genomsoktes, veten-
skapliga artiklar har genomlasts samt olika sokprogram i Inter-
net har anvants for att skaffa information om kunskaper
berorande behandling av ventilationssystem vid handelse av en
brand. Dessutom har vetenskapliga och tekniska tidskrifter inom
aktuellt amnesomrade foljts. Ett forsok har ocksa gjorts for att
diskutera problemet med forskare via Network motsvarande
vetenskapliga nyhetsgrupper.
Risk for rokspridning via ventilationskanaler ar uppskattat
med hjalp av datorprogram for godtyckliga flodessystem PFS.
Gransbrandflode, uppskattat med PFS jamforas med maximalt
brandflode enligt brandsimuleringar. Gransfallet for rokspridning
via F -system innebar att franluftsflodet ar noll vid alla rum
raknat uppstroms fran dar brandrummet ansluter. Gransfallet for
rokspridning via TF-system innebar att tilluftsflodet till bran-
drum ar noll. Rokspridningen sker nar dessa tilluftsflode vander
och till alla rum raknat nerstroms fran dar brandrummet ans.:.
luter.
Grundlaggande principer och teoretiska modeller for upp-
skattning av gransbrandflode for tvarums F- och TF-
ventilationssystem (oppet och stangt) presenteras i rapporten.
De mest typiska ventilationssystemen i de undersokta
byggnaderna samt brandcellsindelning i dessa diskuteras.
Foljande typer av ventilationssystem analyseras:

19
Spread of Smoke and Fire Gases via the Ventilation System
-F-system med en fHikt och kanalsystem bestaende av en eller
tva huvudkanaler,
-TF -system med tva fHiktar och kanalsystem med grenar till
varje brandcell,
-TF-system med "noll" -kanal. Detta innebar att tilluft fors
genom tilluftsdon till varje brandcell, medan franluft samlas i
korridor ("noll"-kanal) och sedan fors bort darifran via franlufts-
don med hjalp av franluftsflakt.
Den har undersokningen ar begransad till principiell analys av
risk for rokspridning via ventilationssystem av ovan presenterade
typer med fyra grenar, och till motsvarande analys for roksprid-
ning via "verkliga" forgrenade ventilationssystem med samma
grenar.
Uppskattning av gransfall for rokspridning via ventilationssys-
tem i de studerade typerna av ventilationssystem med fyra
grenar, atgarder mot rokspridning och en kort beskrivning av da-
taprogrammet PFS presenteras i rapporten.
Det finns tva olika satt att beskriva tryckfallsfunktioner forT-
och X-stycken i PFS. Bada satten har provas. Projektet under-
soker en "onormal" flodesriktning iT- och X-stycken, d.v.s. en om-
vand flodesriktning i en av delarna och flodesminskning till noll-
flode i vissa delar av formstycke som foljd. Hur blir
tryckfallsforhallandet i formstycke i det har fallet ar inte kant.
Hogre flode resulterar i hogre hastighet i ventilationskanaler och
hogre Reynoldstal. Okande flodesturbulens resulterar i sin tur i
sakrare satt att beskriva tryck som kvadratisk funktion.
Beskrivning av T- och X-stycken i PFS som kvadratiska funk-
tioner antas dimensionerande i detta sammanhang.
Olika atgarder mot rokspridning via ventilationssystem ana-
lyseras i projektet:
avstangda flakt(ar ),
flakt(ar), som varvas upp till dubbla varvtalet,
tilluftsflakt konverterad till franluftsflakt,
bypass fran tilluft- till franluftkanal,
kombination av ovan presenterade atgarder tillsammans med
normal flakt och med varandra.
Brandflode som overskrider gransfallets £lode fran tidigare be-
rakningar anvands i denna analys. De undersokta fallen omfattar
ej alla mojliga atgarder mot rokspridning via ventilationssystem,

20
 Sammanfattning
men den presenterade metoden kan anvandas for uppskattning av
risk for sadan.
''Verkliga" fall analyserar risk for rokspridning via mer kom-
plicerade ventilationssystem typiska for de studerade byggnad-
styperna. Gransbrandflode och gransbrandtryck uppskattas for
olika ventilationssystem. Uppskattningen av gransfallet for rok-
spridning ar berakningstekniskt svart for forgrenade ventila-
tionssystem eftersom ett flode skall vara noll i en av grenar sam-
tidigt som andra flode och tryck ligger langt utover de normala
vardena. Brandflodena for respektive byggnadstyper uppskattade
och presenterade i form av diagram i detta projekt anvands for
undersokning av rokspridningsrisk via ventilationssystem.
"Brander" placeras i olika rum och rokspridning till de narlig-
gande rummen samt olika atgarder mot det fenomenet undersoks.
Uppskattning av giftighet av utspadd rok diskuteras i rappor-
ten. CO-nivan och CO-dosen kan uppskattas uppat med nagra
enkla berakningssteg for ett F -system med avstangd flakt. U pp-
skattningen av
total rokvolym som sprids,
andel av rok som sprids till ett enskilt rum, inklusive andel av
rok till ett utav n rum,
CO-niva i ett enskilt rum presenteras i rapporten. Alla rum an-
tas vara lika stora. CO-niva tillsammans med exponeringstid och
typ av aktivitet bestammer CO-dosen och forgiftningseffekter.
Olika rekommendationer angaende acceptabla och farliga CO-
nivaer och CO-doser diskuteras har. Uppskattning av CO-nivan i
ppm motsvarande IDLH (immediately dangerous to life and
health) begransat till 1500 ppm under 30 min vistelse, som nar ett
av n rum ocksa presenteras.
Uppskattning av medelsiktbarhet av rok ar analog med upp-
skattningen av CO-forgiftningsrisken. Presenterade ekvationer
ger medelvarde for siktbarhet.
Undersokningen ar gjord under forutsattningen att alla de ex-
aminerade flodena, inklusive brandflodena har normala tempera-
turer, d.v.s. "kalla" flodena. Samtidigt ar det valkant att tempera-
turer ar hogre an de normala temperaturerna och simuleringsre-
sultat i denna rapport visar att brandflodestemperaturer kan
overskrida 300°C - 340°C. Detta innebar att forvantad rok-
spridning via ventilationssystem enligt den presenterade ana-
lysen kan skilja sig fran den i verkligheten. Uppskattningen av

21
Spread of Smoke and Fire Gases via the Ventilation System
friktionstryckfall i raka ventilationskanaler visar att detta min-
skar c:a tva ganger samtidigt som lufttemperaturen i K okar tva
ganger. Tryckfall i motstycke ar svarare att uppskatta. Ett over-
slag visar att klarar man det kalla fallet, klarar man aven det
heta fallet. Sa smaningom tack vare rokutspadning och varmefor-
luster i ventilationskanaler, sjunker flodestemperaturerna och de
"kalla" flodena blir dimensionerande.
Ytterligare en metod for uppskattningen av eventuell rok-
spridning via F -system presenteras i rapporten. Den s.k.
"Alexandermetoden" ar en forenklad metod for handberakningar
baserad pa "borthuggningsprincip". "Borthuggningsprincipen" in-
nebar att den del av kanalsystem som kan utsattas for rok-
spridning huggas bort och resten av ventilationssystemet raknas
om med hansyn till de nya tryck/flode forhallandena.
Problemet som undersoks i denna rapport verkar komplicerad
och antaganden, gjorda med viss kunskapsbrist och brist av em-
piriska data, kan inte minska osakerheten av resultat. Resultaten
kan, alltsa, inte tolkas som en slags generelllosning for uteslut-
ning av rokspridning via ventilationssystem, utan skall betraktas
som en hjalp vid analyser inom detta omrade.

22
 List of Symbols and Definitions

List of symbols and definitions

Symbols

Unless otherwise started, the following symbols are used

throughout this document:

A area, [m2 ]
b width, [m]
c specific heat of incompressible substance, [J/kg K]
c flow coefficient, [-]
d diameter, [m]
D optical density per meter, [m-1]
Ey elastic (Young's) modulus, [N/m2 ]
E energy, [J]
f coefficient of volumetric expansion, [(m3/s)/MWJ
h heat transfer coefficient, [W/m2 K]
l length, [m]
mb mass flow rate, [kg/s]
p pressure, [Pa]
q volume flow rate, [m3/s]
Q heat rate, [kW]
R resistance, [-]
s visibility range, [m]
t time, [s]
T temperature, [°C (K)]
V velocity, [m/s]
V volume, [m 3]
z height, [m]
Re Reynolds number, [-]
M{ mass loss of sample, [g]

23
Spread of Smoke and Fire Gases via the Ventilation System

Greek letters

a characteristic value for a certain fire progress,

[kW/s 2]
f3 thermal expansion, {K-1]
1J viscosity, [N s/m2 ]
.A thermal conductivity, [W/mK]
V kinematic viscosity, [m2/s]
p density, [kg/m3 ]
(j thermal stress, [N/m2 ]
L\ increment of

Definitions

For the purpose of this dissertation, [5] and [6] were used as the
main material to describe research in this domain. Nevertheless,
some other terms has been used in order to express more recent
concepts. The definitions of this dissertation are as follows:

Air leakage:
The uncontrolled flow of air through a component of the building,
or the building envelope itself, when a pressure difference is ap-
plied across the compartment.

Airtighteness:
A general descriptive term for the air permeability resistance of a
building.

Building envelope:
The inside of the exterior surfaces of the building.

Buoyant flow:
A gas flow which is caused directly or indirectly by gravity.

24
 List of Symbols and Definitions
Compartment fire:
Fire in enclosed spaces, which are commonly thought as rooms in
a building.

Constrained fire:
The availability of both fuel and oxygen are considered.

Fire flow:
Fire gas flow caused by temperature expansion of the heated gas
in the fire room in the case of fire.

Fire plume:
The generally turbulent buoyant flow, which includes any flames.

Forced (mechanical) ventilation:

Ventilation by means of fans.

Natural ventilation:
Ventilation using only natural driving forces such as differences
in air temperature and density, wind pressure.

Overflash:
The transition from a growing fire to a developed fire in which all
combustible items in the compartment are involved in fire.

Ventilation:
The process of supplying and removing air intentionally by natu-
ral or forced (mechanical) means to and from any space.

Unconstrained fire:
Fuel is burned without regard for oxygen availability.

25
Spread of Smoke and Fire Gases via the Ventilation System

26
 Introduction

1 Introduction

The problem of smoke spread via ventilation ducts in case of

fire has been examined in this work. Fire releases a large amount
of heat which causes the room air and smoke produced by com-
bustion to expand. The expansion produced by a room fire drives
some of the gas out of the room via all types of openings and
leaks. Apart from doors and windows, the most probable openings
through which the gas can flow out are ventilation terminals and
systems. A room can have all its doors and windows closed, but if
the ventilation ducts are open, the gas will leak through the venti-
lation system and the fire gases can spread to the adjacent rooms.
However, the probability of this and the conditions which lead
to such a phenomenon have not yet been fully examined. The
main purpose of this study was to examine the possibility of oper-
ating the fans during different fires in four types of buildings
(compartment, hospitals, offices and hotels) and to estimate the
risk of smoke spread in different conditions.

1.1 Background
According to the building regulations and codes in Sweden and in
other countries, until recently the only recommended way to pre-
vent the spread of smoke was to install different kinds of dampers
and smoke barriers and to shut down the ventilation system in
the event of fire, or to rely only on shutting down the ventilation
system designed accordingly to given pressure drop relations. The
disadvantage of this second method is that some fire gases will
always spread to adjacent rooms through the ventilation system.

27
Spread of Smoke and Fire Gases via the Ventilation System
The first attempt to analyse the problem was made in 1988 by
Lars Jensen. He presented the results of the analysis in [1]. The
main principles of this analysis will be discussed later in Section
2. Further studies in the field, made by Tobias Hielscher and
Bjorn Warelius, under the supervision of Lars Jensen, resulted in
dissertation [2] in 1992. The next work which expanded the prin-
cipal ideas of the above reports was published in 1993 [3]. In 1994
the Swedish Board of Housing, Building and Planning, Boverket,
published a report on this issue [19]. A handbook which summa-
rised up-to-date experiences on the subject was published in 1996,
[16]. The results and theoretical bases of all these works will be
discussed in Section 2.

1.2 Objective and limitations

After certain theoretical studies and practical tests in order to
solve the problems associated with the use of fans during a fire, in
order to reduce and in certain cases to eliminate the risk of fire
gas spread through ventilation ducts, the main principles and
methods are now known. Application of these basic principles
largely depends on the characteristic parameters, estimated for
different fires in different types of buildings. The main building
types chosen for this work are

- dwelling houses
-hospitals
-offices
-hotels.
The main objective is to assess the relationship between typical
fires in the above types of buildings, the airtightness of the con-
struction and the layout of the ventilation system, and to give rec-
ommendations on good practice to diminish or even eliminate the
risk of smoke spread through ventilation ducts to the adjacent
rooms. In this connection, one of the most important issues to es-
timate are the characteristic parameters/features that are pro-
duced by certain fires, and to assess their influence on the pres-
sure/flow ratio in the ventilation system.
The aim of this study is to clarify the following:

28
 Introduction

design fire case

design fire spread requirement in relation to flashover
design internal and external airtightness of the chosen rooms
durability of the ventilation system in case of fire
method for assessing what measures are adequate against fire
gas spread
proposal for Swedish Building Codes concerning fire gas spread.

This report to some extend can also be used as a textbook in

teaching in this field.

1.3 Comprised fields of knowledge

The study is interdisciplinary in nature and comprises different
fields of engineering knowledge: fire protection engineering,
building science, and building services. It can be illustrated with
the following schematic figure:

Studied field

Fire _________ -------~~-~~ding

protection science

Building services

29
Spread of Smoke and Fire Gases via the Ventilation System

The whole study is divided into two sections, one of which exam-
ines fire behaviour and estimates characteristic parameters in the
chosen types of enclosures, and the other one examines the ability
of the ventilation system to spread smoke and fire gases via ducts.

30
 Section 1. Analysis of Fire Behaviour in Single Spaces
Fire Behaviour in Single Spaces

Section 1 Analysis of fire

behaviour in single
spaces

2 Fire behaviour in single

spaces

2.1 Statistical estimation

In order to evaluate the principal dangers and damage caused by
fire in the chosen types of enclosures, and to assess the most
probable items in which fire starts, the following analysis was
made.
This Subsection discusses injuries to persons and damage to
property caused by fire. It deals both with the cause and the
magnitude of this damage. There are large gaps in the informa-
tion regarding the probability of the outbreak of fire in relation to
e.g. the area of the space and the number of people. Thus, it ap~
pears to be impossible to analyse this risk, but some data for its
assessment are given in this section.
The number of deaths in fire in Sweden varies from 100 to 150
persons per year. The majority of all fatalities occur in dwelling
houses (76%). The most usual cause of such fatal fires is obviously
smoking.
The danger is also associated with remaining in the burning
space for a certain time. Table 2.1 sets out the risk assessment ac-

31
Spread of Smoke and Fire Gases via the Ventilation System
carding to English statistical sources for death and property dam-
age for each hour of remaining in the burning space in different
types of buildings.

Table 2.1 Risk of death and property damage in case of fire in differ-
ent types of premises [4].

Type of building Damage per person Deaths per person

8 8
10 exposed hours 10 exposed hours
Dwelling house 0.9 0.9
Hospital 0.55 0.29
Office 0.55 0.05
Hotel 20 3.6

It is seen from Table 2.1 that the most dangerous fires, regarding
both injuries and the risk of death while remaining in the burning
space for a certain time, are those in hotels, but the danger in
other types of buildings cannot be ignored. At the same time it is
much more difficult to evaluate the number of injures than deaths
and to define the type of injury. SRV 1993 reports about 530 in-
jures in 1992.
To evaluate the most probable fire behaviour in the examined
cases, it is important to assess the objects which can cause igni-
tion.
Table 2.2 provides information about the objects which cause fa-
tal fires in dwelling houses.

32
 Section 1. Analysis of Fire Behaviour in Single Spaces
Fire Behaviour in Single Spaces

Table 2.2 Objects which cause fatal fires in dwelling houses [4].

O~ecl %
Bedclothes 36
Sofa, armchair 14
Clothes (old people, handicapped) 11
Flammable liquid (suicide) 6
Food 3
Building construction 6
(electrical fault, defective fireplace etc.)
Other 4
Cannot be identified 20

As seen from Table 2.2, the most probable objects which cause fa-
tal fires in dwelling houses are bedclothes and sofas. Table 2.3
presents data regarding the probability of fire hazard annually in
different types of buildings.

Table 2.3 Probability of fire hazard in different types of buildings [4].

Type of building Probability function of Probability of fire

one fire per year per year at A =
2
1500m
Hospital 0.0007 A0 · 75 0.17
Office 0.000059 A 0 ·9 0.042
Hotel 0.00008 Al.O 0.12

It is seen from Table 2.3 that the most probable fires are those in
hospitals and the least probable are fires in offices. In order to
predict the fire behaviour in the building, it is significant to assess
the flashover phenomenon. Table 2.4 gives some data on limita-

33
Spread of Smoke and Fire Gases via the Ventilation System
tions of fire hazard to the object in which the fire started and to
the room in which it started.

Table 2.4 Fire behaviour related to the spread of fire from the object
in the room of fire origin and spread of the compartment
fire to another room [4].

Type of Total Limited to Spread to Spread out-

building numbe the object in other objects side the
r of which the in the room room of fire
fires fire started, of fire ori- origin,
% gin, %
%
Dwelling 52024 46 46 8
house
Hospital 2243 73 26 1
Hotel 5244 36 53 11
Other 20502 25 63 11

As can be seen from Table 2.4, the majority of all fires remain in
the room of fire origin. At the same time, there is the risk of fire
spread to the other objects in the same room.
All the data on the statistical evaluation of fire behaviour in
chosen rooms is taken from [4].
To sum up the statistical data in this part regarding the danger
of fire in the analysed buildings, the most probable fire behaviour
is limited to the room of fire origin, and the most probable objects
where the fire starts are bedclothes and sofas in dwelling houses.
These data are very important in predicting fire behaviour in
buildings, which will be discussed in this section.

34
 Section 1. Analysis of Fire Behaviour in Single Spaces
Fire Behaviour in Single Spaces

2.2 Estimation of basic parameters in

compartment fires
The ability to predict different parameters such as pressure, heat
release rate, temperature, gas flow through openings, in com-
partment fires is of great significance for the prediction of fire gas
spread. In this section, the fundamental principles underlying the
prediction of compartment fire parameters will be summarised
along with a number of solution methods.

2.2.1 The stages of fire growth

Compartment fires are often discussed in terms of growth stages.
Figure 2.1 shows a theoretical curve of temperature vs. time along
with the growth stages.

I
I
I
w POST-F LASHOVER
tf)
FLASHOVER
re I
w I
re I
::>
1-
<! I
re
w
I
FULL V-DEVELOPED FIRE I
0.
~
I
'W I
1- I
I
I DECAY
I
I

TIME--+

Figure 2.1 General description of theoretical compartment fire (in ab-

sence of fire control) [7].

The growth stages can be defined as follows:

Ignition
Growth
Flashover

35
Spread of Smoke and Fire Gases via the Ventilation System
Fully developed fire
Decay.

Ignition stage: The period during which the fire begins.

Growth stage: The stage during which the fire grows.
At the beginning of this stage the growth of fire is not influenced
by the compartment. The fire can be described in terms of its rate
of energy and combustion product generation. If sufficient fuel
and oxygen are available (fuel-controlled fire), the fire will con-
tinue to grow, causing the temperature and rate of heat release to
rise, until it changes into a ventilation-controlled fire.
Flashover: The stage characterised by the total in-
volvement of all combustible items. The flashover stage can be de-
scribed in terms of the temperature at which the radiation from
the hot gases in the compartment will ignite all the combustible
items in the room. Gas temperatures of 500-600°C are common.
This gas temperature is equal to a radiation intensity of 12.5 - 20
2
kW/m floor.
Fully developed fire: The stage during which the heat release
rate is the greatest. During this stage the fire is said to be strictly
ventilation-controlled. If there are some openings in the compart-
ment, the unburned fuel will leave the compartment with the gas
flow, mix with oxygen and burn outside the compartment.
Decay stage: The stage during which the heat release
rate declines and the fuel is consumed. The fire changes from
ventilation to fuel controlled.
While many fires will not conform to this idealisation, it pro-
vides a useful framework for the discussion of compartment fires.
All fires include the ignition stage; beyond that they may fail to
grow, or they may be affected by different activities before passing
through all the stages listed above. It can be extremely important
to differentiate between well ventilated fires (fuel-controlled) and
ventilation-controlled fires. All fires consume oxygen. A period
when fire growth depends only on the amount of fuel, i.e. when
there is sufficient air and oxygen, is called well-ventilated or fuel-
controlled fire. The stage of fire when its growth depends on the
amount of air (oxygen), i.e. when there is insufficient air, is called
ventilation-controlled fire stage. These processes can play a deci-
sive role in fire growth stages, particularly in compartment fires.

36
 Section 1. Analysis of Fire Behaviour in Single Spaces
Fire Behaviour in Single Spaces
In small compartments of the analysed types, real fires can be dif-
ferent from the above case.

2.2.2 f fire growth

The curves in Figure 2.2 give a schematised picture of fire growth
during complete burning of combustibles (samples) in terms of
temperature as a function of time. It will typically be found that
there are limited experimental data available regarding the
burning rate of a specific burning item. Nevertheless, there are
different ways in which fire progress can be described. One of
these, the simplified method, is based on the assumption that the
growth phase of many fires can be characterised as a function of
the square of time. This idealised fire energy release rate is de-
scribed by
(kW) (2.1)

where Q is in kW and tin seconds. a is in kW/s 2 and has a charac-

teristic value for a certain fire progress. Its value for four types of
fires as well as time to reach 1MW heat release is presented in
Table 2.5.

Table 2.5 Fire growth parameter and time to reach 1 MW heat re-
lease for different types of idealised fires.

Type of fires a, kW/s 2 Time to reach 1 MW, s

Slow 0.00293 584
Medium 0.01172 292
Fast 0.0469 146
Ultrafast 0.1876 73

Note that each a value is four times as high as the previous one
and time to reach 1 MW heat release is two times as high as the
previous one.

37
Spread of Smoke and Fire Gases via the Ventilation System

100 200 300 400 500 600 700 800 900

time after ignition (sek)

Figure 2.2 Four types of fire progress [30].

This method of evaluating fire growth is used especially for indus-

trial buildings and warehouses. Owing to its simplicity, sufficient
agreement with test results and lack of available data, this
method can be applied to some other types of fires. However, this
description of fire progress can be questioned when fire behaviour
is more precisely defined.

2.2.3 Measured heat release fires

In order to evaluate fire behaviour more precise, another type of
fire growth curve is widely used. These type of data are based on
the results of full-scale tests in which different items were burnt
in the laboratory and heat release rate as a function of time was
measured.
According to the results in Subsection 2.1, the following items
were assumed to initiate fires:
- in dwelling houses sofa [6] and bed [7]
- in hospitals bed [9]
- in offices equipment and furnishing [10]

38
 Section 1. Analysis of Fire Behaviour in Sing le Spaces
Fire Behaviour in Single Spaces
- in hotels bed [6].
Figures 2.3- 2.7 present fire growth curves for these items.

RHR(kW)

4000

1500

1000

/''--.....'-
500
J ....________________________ .
1800 2400

time (s)

Figure 2.3 Heat release rate/time for burning sofa [6].

RHR(kvV)

400

200

time (s)
Figure 2.4 Heat release rate I time for burning bed [7].

39
Spread of Smoke and Fire Gases via the Ventilation System

Mass Burning Rate (kg/s) RHR(kW)

400

300

200

100

120 240 360 480 600 720

time (SI

Figure 2.5 Heat release rate/time for burning hospital bed [9].

RHR

-without sprinklers
---------------------- 'vvith sprinklers

1000

500

tune (s)

Figure 2. 6 Heat release rate I time for burning equipment and furniture
in office [10].

40
 Section 1. Analysis of Fire Behaviour in Single Spaces
Fire Behaviour in Sing le Spaces

RHR(kW)

2000

1500

1000

500

600 1200 1800 2400

time (s)

Figure 2. 7 Heat release rate I time for burning bed [6].

The ability to predict different parameters in compartment fires is

of great significance for this project and for fire protection engi-
neering.

2.2.4 Compartment fire model

In order to calculate or predict parameters generated by a com-
partment fire, a description or model of the fire phenomenon must
be created. Such a model is also an idealisation of real hazards.
Different models divide the room into different numbers of control
volumes depending on the desired level of accuracy. The most
common fire model is the zone model. It generally uses two vol-
umes to describe the room- an upper layer and a lower layer.
An arbitrary fire, which starts at some point below the ceiling,
releases energy and products of combustion. These are time de-
pendent variables. The hot products of combustion form a plume
which rises to the ceiling due to its buoyancy. While rising, the
plume draws in ambient air from within the compartment. This

41
Spread of Smoke and Fire Gases via the Ventilation System
decreases the temperature of the plume and, at the same time, in-
creases the volume flow rate. After reaching the ceiling, the plume
spreads out and forms a hot gas layer which descends as gases
continue to flow into it. Hot gases collect at the ceiling and fill the
room from the top. As the hot gas layer descends and reaches
openings in the compartment, hot gas flows out and the ambient
"cool" air flows in through the openings. This description of com-
partment fire is referred to as a two-layer or zone model. This
two-layer model has evolved from observation of such layering in
real-scale fire experiments. While these experiments show some
variation in conditions within the layer, these are small compared
with the differences between the layers. These two-layer models
can thus provide a fairly realistic simulation under most condi-
tions. This model assumes that the predicted parameters such as
temperature, pressure, smoke and gas concentration, and layer
height above the floor, are uniform at any time within this volume
which is referred to as the control volume.
There are different mathematical models based on different as-
sumptions, which describe the two-layer fire model with the help
of differential equations. The detailed analysis of such mathe-
matical models is beyond this study, but it can be interesting to
emphasise that two of them, constructed by Zukoski and McCa-
frey respectively, are used in different computer programs, dis-
cussed further.
Other types of models include field models. These models divide
the room into thousands or even hundreds of thousands of grid
points. Such models can predict the variation in conditions within
layers. They are used when highly detailed calculations are es-
sential.
The main compartment fire phenomenon is illustrated sche-
matically in Figure 2.7.

42
 Section 1. Analysis of Fire Behaviour in Single Spaces
Fire Behaviour in Single Spaces

r layer
Ceil jet

Plume
Lower layer

Figure 2.8 Two-layer model.

2.3 Computer methods for predicting

compartment fires
Very complicated processes of fire development which are de-
scribed with the help of a large number of mathematical expres-
sions necessitate the use of a computer program. There are differ-
ent computer programs which simulate hazard phenomena. These
simulations provide estimates of different elements of hazard de-
velopment, such as fire growth, smoke generation, temperature
rise, species production, layer descent. Some programs are con-
fined to the room of fire origin, others can estimate processes in
multiple rooms. This study is limited to single-room compartment
fires. The programs described below were chosen to simulate haz-
ard development for the types of fires analysed in this study.

2.3.1 CFAST (HAZARDl)

CFAST (Consolidated Model of Fire Growth and Smoke Trans-
port), which is part of the computer program HAZARDl [15],
predicts the conditions within a structure, resulting from a user-
specified fire. The program is developed and written at the Center
for Fire Research, the National Institute of Standards and Tech-

43
Spread of Smoke and Fire Gases via the Ventilation System
nology, USA. The program uses the two-layer model described in
Subsection 2.2.4 in order to calculate the distribution of evolving
smoke, fire gases and temperature throughout a building during a
fire. In CFAST, each room is divided into two layers, all parame-
ters in which are assumed to be exactly the same at every point.
Since these layers represent the upper and lower parts of the
room, conditions within the room vary only vertically, from floor
to ceiling, and not horizontally. CFAST is based on solving a set of
equations that predict state variables (pressure, temperature,
heat release, species, etc.) based on the enthalpy and mass flux
over small periods of time. These equations are derived from the
conservation equations for energy, mass and momentum, and the
ideal gas law. The fire model involves an interdisciplinary consid-
eration of physics, chemistry, fluid mechanics and heat transfer.
The model is particularly well suited for predicting parameters
and for studies of changes, both subtle and large, within a single
room. The necessary approximations required by operational
practicality result in the introduction of uncertainties in the re-
sults. The fire plume model description follows McCafrey's plume
model.
Limitations:
- an important limitation of CFAST is the absence of a fire
growth model. The program utilises a user specified fire,
-the two zone model involves a mixture of established theory,
approximations and empirical correlation,
-burning is constrained by the available oxygen. An oxygen co
bustion chemistry scheme is employed only in constrained
(ventilation-controlled) fires. Predictions of concentrations of
chemical products are based on user-specified hydrocarbon r
tios,
-some coefficients, used in the equations, are empirically de
termined values.
The only mechanisms provided in zone models to move enthalpy
and mass into the upper layer of a room are two types of plumes:
those formed by the burning item in the room of fire origin, and
those formed by the jet of upper layer gases escaping through an
opening.

44
 Section 1. Analysis of Fire Behaviour in Single Spaces
Fire Behaviour in Sing le Spaces

2.3.2 DSLAYV
DSLAYV is a computer program which predicts the conditions
within a single space, resulting from a user-specified fire. The
program uses the two-layer model described in Subsection 2.2.4 in
order to calculate the smoke filling process, temperature through-
out a space during a fire, pressure rise and other parameters. A
steep thermal gradient separates the two layers. For a given fire
size, specified in terms of heat release rate, the conservation
equations of mass and heat are solved for the layers, yielding the
smoke layer height, temperature and the radiant heat conditions
in the room of fire origin as a function of time. The fire plume
model description follows Zukoski's plume model. Only incombus-
tible linings are allowed for in the model. The surfaces enclosing
the lower layer are heated, but the air layer is considered to be at
ambient temperature. DSLAYV is an interactive program. A dia-
logue between the user and the program is governed by com-
mands adapted to the concepts of the user. The program is devel-
oped and written by Bengt Hagglund at the National Swedish De-
fence Research Institute, and is presented in [18].

2.3.3 SMAFS
SMAFS (Smoke Movement and Flame Spread) is a field model
program which is designed to simulate the fluid flow and com-
bustion processes in a general sense, especially those involved in
fire. The program is used at the Department of Fire Safety Engi-
neering, LTH, and simulations are made by guest researcher Yan
Zhenghua, which is gratefully acknowledged.

The simulation is based on the simultaneous solution of a large

number of differential equations. Consideration is mainly given to
the following physical and chemical processes:
1. turbulent fluid flow,
2. radiation,
3. combustion,
4. heat transfer inside wall boundaries, and
5. pyrolysis of solid combustible materials.

45
Spread of Smoke and Fire Gases via the Ventilation System
The purpose of developing this package is to have a theoretical
tool to predict turbulent flow and turbulent combustion which are
of great interest to industry and safety design. It includes the fol-
lowing models: turbulence model, combustion model, radiation
model, solid material pyrolysis model, heat transfer wall function,
etc. The limitation is determined by the accuracy and validity of
the incorporated models.
With the support of its pre-processor and post-processor, the
package is quite easy to use. The computer power requirement is
mainly dependent on the complexity of the problem under study.

In this study CFAST was chosen as the basic tool of fire simu-
lation owing to the fact that it is a modern program under con-
stant improving. DSLAYV was used for test fire simulations in
order to control and compare results. Both CFAST and DSLAYV
are zone-model based programs. As the zone-model based pro-
grams predict the average gas parameters in the various regions,
but do not provide information on local conditions which may be
obtained from field model studies, another program, SMAFS, was
used to estimate the danger of flashover effects, and for evaluat-
ing the temperature field near the ventilation terminal. These
simulations and their results will be discussed further.

2.4 Airtighteness
The expansion of gas caused by heat released during the fire in-
creases the pressure in the compartment and drives some gas out
of the room through leakage paths. All openings through which
gas flows out are called vents. If a window is open in the room of
fire origin little flow occurs through the remainder of the building.
When doors and windows are closed and there are no ventilation
openings, flow will move towards cracks and leaks wherever they
may be in the room. All these flows are initially nonbuoyant. As
the fire grows, hot gas flows buoyantly out of the place of origin,
while cold gas flows in below (Figure 2.8). To evaluate gas flux
(mass flow) out of the burning room with the help of the simula-
tion programs, it is essential to know the sizes of such vents.

46
 Section 1. Analysis of Fire Behaviour in Single Spaces
Fire Behaviour in Single Spaces
Air leakage occurs as a consequence of building design, work-
manship and materials used. The amount of air leakage depends
on the magnitude of the driving forces in combination with the
size and shape of the leakage paths. In the normal case, these
driving forces are pressure differences across the structure, tem-
perature differences between the inside and outside of the struc-
ture and, if there is one, the ventilation system.
For new houses, the standard requirements specified in Swed-
ish Building Regulations, BBR 94, must be followed. According to
BBR 94, air infiltration should not exceed 0.8 lls,m 2 (3 m/h) for
dwellings and 1.6 l/s,m 2 (6 m/h) for other types of buildings at 50
Pa pressure difference. This requirement relates to the area of the
building envelope. The building envelope area is defined as the
area exposed to outside air.
The objective of this evaluation is to summarise knowledge
about the real magnitude of leakage, in relation to the whole
building envelope including interior building constructions, for the
types of enclosures, built at different times, which are described
in Subsection 1.2.
Air leakage, normally divided into external and internal leak-
age, is of interest for this study in terms of average values for the
whole building envelope of rooms or apartments. Tracer gas
measurements and pressurisation measurement, performed in
dwelling houses built in 1988 (Per Levin, [9]) have shown the fol-
lowing results.
Measured internal air leakage between apartments has been
found to account for 12- 33%, and in some cases up to 50%, of the
total air leakage at 50 Pa. Thus, small internal leaks have been
found to exist.
Measured air leakage for the whole apartment has shown good
agreement with the Swedish Building Regulations requirements
for the external building envelope even for a corner apartment.
This study includes several experimentally built energy effi-
cient apartment buildings of extremely good airtightness. The
techniques used for building construction and building services
systems are representative of those used for new Swedish multi-
family buildings.
In different investigations, made both in Sweden and abroad,
the problem of air leakage in single family houses was analysed

47
Spread of Smoke and Fire Gases via the Ventilation System
and air infiltration under different conditions was measured
(Technical Note AIVC [27], Ake Blomsterberg [10]). All these
studies relate to comparatively newly built dwellings of good air-
tightness, and show very good agreement with Code require-
ments.
The analysis described above are not sufficient for this study
which comprises different types of buildings. In particular, there
are no data for older buildings. At the same time, the information
from international scientific documents published abroad cannot
be used as straightforward data, due to Swedish construction
traditions and airtightness requirements which are different from
those in other countries.
As a reasonable assumption in this respect, the air leakage ac-
cording to the Swedish Building Regulations, BBR 94, applied to
the whole buildings envelope, can be accepted as a suitable and
probable value.

48
 Section 1. Analysis of Fire Behaviour in Single Spaces
Fire Simulations

3 Fire simulations

In order to predict fire behaviour in the selected enclosures, a se-

ries of simulations was made with the help of different computer
programs. The purpose of these simulations was to find reason-
able peak values for different parameters, such as temperature of
the hot gas layer, layer descent, pressure rise due to volume ex-
pansion and, what is most important, peak flow out of the fire
room. This mass/volume flow affects the behaviour of the ventila-
tion system, which can result in spread of smoke and hot gases to
adjacent rooms via ventilation duct network.

3.1 Input data used in simulations

3.1.1 Geometrical data

As described in Chapter 2, the CFAST (HAZARD!) packet was
chosen to simulate hazards in four types of buildings: dwellings,
hospitals, offices and hotels. The geometrical data is presented in
Table 3.1. All walls in all types of buildings have the same data,
which is gypsum.

49
Spread of Smoke and Fire Gases via the Ventilation System

Table 3.1 Geometrical data for the examined cases.

Type of building Area, m 2 Height, m Volume, m 3

12 2.4 28.8
20 2.4 48
56 2.4 134.4
Dwelling house * 62.2 1) 2.4 149.3
75 2.4 180
85 2.4 204
14.42 ) 2.5 36
14.42) 3.75 54
Hospital 202) 2.5 50
20 2) 3.75 75
30 3.75 112.5
10.5 3 31.5
Office 15 3 45
25 3 75
40 3 120
16 2.4 38.4
Hotel 24 2.4 57.6
24 3.5 84
64 3.5 224
* - Geometrical data for apartments correspond to average statistical size
for students' lodgings, one-, two- and three-room apartments of standard
height.
1
) Average size of apartment per person in Sweden according to [11].
2
) Geometrical data for hospital rooms according to [12].

The main purpose of this data was to choose the most typical sizes
of the assumed types of buildings.

50
 Section 1. Analysis of Fire Behaviour in Single Spaces
Fire Simulations

3.1.2 Fire scenario and fire position

The bed and sofa fires are assumed to be initiated in dwelling
houses, the hospital bed fire in hospitals, the furnishing and
equipment fire in offices and the bed fire in hotels. Flaming com-
bustion is assumed. The fire growth model in the rooms discussed
above conforms to the description of initial fires, given in Chapter
2 for "real fires". This was considered to give a more realistic pic-
ture of fire hazards in the examined types of buildings.
The position of the object in the room plays an important role in
estimating fire behaviour as it determines fire plume wall and
corner effects. Owing to the fact that few experimental data are
available and the errors of such simulations are therefore difficult
to estimate, different cases were tested in this study in order to
assess the process. The cases tested were fire growth of the same
fire, in the same room, placed in the centre of the room, in the
corner of the room, and near one of the walls. The fire location af-
fects plume expressions, which results in different values of pre-
dicted parameters.

3.1.3 Assumed airtightness

The room is closed except for leakage paths at the sill of the closed
doorway (normal case). Other leakage paths, at the top of the
doorway and vertically along the sides of the doorway, were also
tested.
The size of vents is determined by requirements in the Swedish
Building Regulations as discussed in Chapter 2. If a pressure
drop, ~p, exists across a leakage path of a certain area, A, with a
fluid density, p, the flow through the vent has volume flow

q=CA~p
/UP (3.1)

In this formula the flow (or constriction) coefficient can be set at C

= 0.68 for an orifice. The leakage paths can thus be estimated to
be equal to 1.25 cm 2/m 2 of building envelope for dwellings and 2.5
cm 2/m 2 of building envelope for other types of buildings.

51
Spread of Smoke and Fire Gases via the Ventilation System
It is assumed that there is no ventilation system and, therefore,
no influence on the fire progress.

3.2 Analysis of parameters evaluated by

CFAST

3.2.1 Predicted variables

In this study the following set of variables, predicted with the
simulation program CFAST, to be shown:
-upper layer gas temperature,
- room pressure,
-layer interface position,
- heat release rate,
-upper layer gas species concentration (oxygen),
- vent flow in and out of the room.
Although there are certainly other variables of interest, these
will provide an indication of the characteristic fire behaviour in
the fire room.
Table 3.1 shows predicted variables in the form of diagrams and
a table for a fire in a 4x4x2.4 m hotel room.
This example was chosen to show the way the predicted results
from simulations are evaluated. All in all, 29 fires were simulated
according to the principles described in Table 3.1 and Subsection
3.1.2. Fire behaviour for the examined cases was simulated in
rooms of different area and height, and with different fire place-
ment. The results of these simulations in the form of diagrams are
given in Appendix A. All the simulated fires have shown that only
the ignition and growth stages occurred.

52
 Section 1. Analysis of Fire Behaviour in Single Spaces
Fire Simulations

Table 3.1 CFAST results in the form of diagrams and table for fire
behaviour in a 4x4x2.4 m hotel room (beginning).

Time (s) Temperature Pressure, Interface, m Heat release,

Pa kW
oc'
0 22.0 -0.004 2.40 0.00
10 37.4 1.17 2.29 7.66
20 43.9 3.67 2.10 14.0
30 47.1 6.34 1.85 19.0
40 51.4 8.00 1.63 22.7
50 56.0 8.26 1.45 25.0
60 63.0 19.6 1.30 37.0
70 73.2 35.3 1.15 50.5
80 85.6 54.5 0.986 65.5
90 100.0 75.7 0.821 82.0
100 116.0 96.9 0.656 100.0
110 136.0 125.0 0.509 123.0
120 160.0 136.0 0.392 147.0
130 184.0 135.0 0.273 172.0
140 208.0 125.0 0.157 198.0
150 230.0 90.3 0.06 225.0
160 252.0 76.8 0.024 258.0
170 276.0 67.1 0.01 292.0
180. 246.0 -685 0.160 24.5
190 201.0 -181 0.355 17.5

53
Spread of Smoke and Fire Gases via the Ventilation System

Table 3.1 CFAST results in the form of diagrams and table for fire
behaviour in a 4x4x2.4 m hotel room (end).

Time (s) Species, 1U,

ffivent 1-2, 1L,
ffivent 1-2, ffivent2-1,2U, 2-1
02% 1-2 kg/s 1-2 kg/s kg/s
0 20.7 0.00 0.000 0.00
10 20.5 0.00 0.021 0.00
20 20.4 0.00 0.037 0.00
30 20.4 0.00 0.049 0.00
40 20.3 0.00 0.055 0.00
50 20.2 0.00 0.056 0.00
60 20.0 0.00 0.086 0.00
70 19.7 0.00 0.115 0.00
80 19.4 0.00 0.143 0.00
90 19.0 0.00 0.169 0.00
100 18.6 0.00 0.191 0.00
110 18.0 0.00 0.217 0.00
120 17.1 0.00 0.226 0.00
130 16.1 0.00 0.225 0.00
140 15.0 0.00 0.216 0.00
150 13.7 0.057 0.110 0.00
160 12.3 0.097 0.04 0.00
170 10.7 0.105 0.016 0.00
180 9.83 0.00 0.00 0.510
190 9.80 0.00 0.00 0.262

54
 Section 1. Analysis of Fire Behaviour in Single Spaces
Fire Simulations
300.0 200.0

0
~

[
--1 u
0.0 '--------'---__L.----'-----'-----' -800.0
0.0 500.0 0.0 500.0

_,1
n= 1 Time (s) n= 1 Time (s)
3000

I r
·a j

0.0
~
0.0 500.0
n= 1 Time (s)

22.0
0.2

f-
-- 1 u 0 --1 u '1- 2

0
..._____
rn
..._____

~ 6
..._____

c
(j)
(.)

(j)
.E
0..

6.0
0.0
0.0 500.0
n= 1 Time (s) 0.0
n= 1 Time (s)
500.0

0.3
-- 1 L 1- 2
1.0
2 L 2- 1

i i

~
~
.E .E

0.0 0.0
0.0 500.0 0.0 500.0
n= 1 Time (s) n= 2 Time (s)

1- fire room, 2- adjacent space, n- room number

U- upper layer, L -lower layer, mvent 1-2- mass flow from the fire room
to the adjacent space

55
Spread of Smoke and Fire Gases via the Ventilation System
As can be seen from Table 3.1, the fire in this specific room can be
analysed in the above chosen terms. Peak temperature of 276°C is
reached 170 s after ignition. Peak heat release rate, 292 kW, is
reached at the same time. The lower limit of oxygen, which is as-
sumed at 12% in CFAST, determines the point below which
burning will not take place. As can be seen, the rate of burning
decreases as the oxygen level drops. Pressure reaches its peak of
136 Pa after 120 s, and causes fire gas and smoke to move
through the leakage paths with a mass flow which reaches its
peak value at the same time, 120 s. As smoke travels away from
the fire, its temperature drops due to heat transfer and dilution.
The upper layer reaches the floor after 150 s, and once the buoy-
ant outward flow exceeds gas expansion due to the fire, the pres-
sure at the floor falls below atmospheric. Outside air thus flows in
at the bottom 180 s after ignition.
As can be seen, in a ventilation controlled compartment fire
which is extinguished due to lack of oxygen, there is a limit to the
heat developed, the temperature rise and the volume of smoke
produced.

3.2.2 Peak pressure, peak heat release rate and peak

temperatures in fire rooms
As mentioned before, the peak temperature and peak pressure in
the fire room can be predicted with the help of simulations. Some
of these peak values are presented in Table 3.2.

56
 Section 1. Analysis of Fire Behaviour in Single Spaces
Fire Simulations

Table 3.2 Peak pressure, peak heat release rate and peak tempera-
ture, simulated with CFAST for the examined cases.

No Type of building V ' m3 Heat re- Temp, p,Pa

lease rate, oc
kW
1 48.00 300 281 575
2 Dwelling house 134.40 1035 322 2057
3 (sofa) 149.28 1035 344 1856
4 180.00 1298 344 1758
5 204.00 1569 345 1619
6 28.80 160 240 172
7 Dwelling house 134.40 422 208 86.1
8 (bed) 149.28 442 206 77.1
9 180.00 500 202 74.3
10 204.00 500 197 68.2
11 36.00 225 244 76.7
12 50.00 225 257 96.5
13 Hospital 54.00 287 260 125
14 75.00 287 220 108
15 112.50 287 185 121
16 31.50 187 331 122
17 Office 45.00 224 318 40
18 75.00 275 290 28.3
19 120.00 350 155 48.5
20 38.40 309 262 187
21 Hotel 57.60 528 286 300
22 84.00 410 315 410
23 224.00 1167 293 548

57
Spread of Smoke and Fire Gases via the Ventilation System
In the table, the peak values for temperature, heat release rate
and pressure are the maximum values obtained in these simula-
tions. These parameters may reach their peaks at different times.
As can be seen in Table 3.2, the temperature in the upper layer
of the majority of examined rooms varies between 150 and 300°C,
which is acceptable. In several cases, in dwellings, offices, and ho-
tels, it exceeds 300°C. Pressure in the majority of the rooms is
relatively low and in dwellings (bed), hospitals and offices it does
not exceed 135 Pa. In dwellings where the sofa is burning, the
pressure is considerably higher- up to 2057 Pa.
As this value seems high enough to be discussed, some assess-
ments have been made in order to predict some more "real" pres-
sures in these rooms. Simulations were made of the same fires as
before in the above rooms, on the assumption that ventilation
terminals were not blocked in addition to leakage paths. The re-
sults have shown considerably lower pressures in dwellings with
sofa-burning fires. The highest value does not exceed 647 Pa, as
against 2057 Pain the model assumed above.
The peak heat release rate shows considerable variation be-
tween 160 and 1569 kW.

3.2.3 Mass and volume flow

Mass flow is the dominant source term for predicting the probabil-
ity of smoke spread to the adjacent rooms via ventilation ducts
and other paths. It fluctuates very rapidly and is highly sensitive
to changes in the environment. Flow through vents comes in two
varieties. There can be a vertical flow through horizontal vents.
This occurs if there are vents in the floor or ceiling of the com-
partment. The other flow is a horizontal flow through vertical
vents such as ventilation terminals and cracks in the walls. These
horizontal flows are of interest in this study as all leakage paths
are assumed to be concentrated around the doors.
Prediction of vent flow that is governed by the pressure difference
across a vent is based in CFAST on the equation

(3.2)

where b is width and z height of the vent.

58
 Section 1. Analysis of Fire Behaviour in Single Spaces
Fire Simulations
CFAST defines the limits of integration with neutral planes, i.e.
the height at which flow reversal occurs, and physical boundaries
such as sills. As hot gases from the source compartment leave that
compartment and flow into an adjacent space, a door jet can exist
which is analogous to a normal plume.
Ventilation systems operate with volume flows. The term "flow"
or "fire flow" will be used to mean volume flow. Combination of
two parameters - mass flow and its temperature at a certain time
-determines the magnitude of volume flow. Peak values of flows,
predicted by source simulations, are used later in assessing the
probability of smoke spread in ventilation systems.

3.3 Comparisons of predicted by CFAST

parameters with experimental data
Comparison of model predictions with experimental measure-
ments serves the purpose of determining the accuracy in predict-
ing variables in relation to hazard. As the present study is based
only on computer simulations, it seems reasonable to compare re-
sults from these with some reference values obtained during ex-
periments. Some experimental data relating to this area are pre-
sented in [15]. Simulations and experimental tests were made for
a fire under conditions different from those in the present study.
Certain results for a single-room fire case are limited to a special
type of fire- with a burning wall involved. Such fires are not typi-
cal of the studied hazards, but considering the general lack of ex-
perimental data on the subject they cannot be ignored. A short
summary of such a comparison, presented in [15], is described in
this Subsection.
Predicted temperatures for the single-room tests show obvious
similarities with the measured values. Peak values occur at simi-
lar times, with comparable rise and fall for most comparisons. In
general, upper layer temperature and interface position predicted
by the model are somewhat higher than experimental measure-
ments, with the differences ranging from -110 to 180°C for the
temperature and -0.1 to -0.3 m for the interface position. These
peak temperatures, varying between 590 and 970°C, occur at ap-

59
Spread of Smoke and Fire Gases via the Ventilation System
proximately 500 s after the start of the fire. Before this time,
measured and predicted temperatures are always close. For all
the tests, the time to peak values predicted by the model is on av-
erage within 25 s of experimentally measured values.
Predicted values of the single-room test concentrations of 0 2 are
lower than those measured experimentally (on average 5% low).
For the pressure predicted in the single room with certain leak-
age, agreement can be considered reasonable only for small leak-
age areas. Otherwise, both temperatures and pressures change by
more than a factor of two. For the type of cases examined in the
present study, with relatively small leakage areas, agreement can
be considered reasonable.
For the mass flow through vents in a single-room test, the
simulated values were somewhat underpredicted by the model.
However, owing to the scarcity of experimental data and the
complexity of comparing the numerous variables in a complex fire
model, true validation of the model cannot be made (true valida-
tion of a model would involve proper statistical analysis of a vari-
ety of compared variables). Nevertheless, in view of the lack of ex-
perimental data and the results of the comparison discussed
above, it may be assumed that fire parameters predicted by
CFAST agree quite well with measurements.

3.4 Analysis of parameters evaluated by

CFAST, DSLAYV and SMAFS
Hazard parameters predicted by the simulation program CFAST
are based on the analysis of a two-zone model and McCaffrey' s
plume equations. As described before, this model assumes homo-
geneous layers with no variation in conditions within the layer. In
other words, the parameters characterising both layers are aver-
age values of variables. Another type of program based on a field
model provides an opportunity to predict variations in conditions
within the layer. SMAFS (simulation of fires in enclosures) is such
a program. For the purpose of evaluating the results obtained
with CFAST, a special case was studied.
Bengt Hagglund of the Swedish Defence Research Establish-
ment FOA has simulated different l-fires in compartments of cer-

60
 Section 1. Analysis of Fire Behaviour in Single Spaces
Fire Simulations
tain size with the help of DSLAYV, and his results are presented
in [16]. One of the cases from this study was chosen as a reference
case and the similar fire scenario in a similar room under similar
conditions was simulated with the help of SMAFS and CFAST. A
fire scenario for medium fast fire in a 6 x 5 x 2.5 m room was
tested.
The purpose of this comparison was to see how well the magni-
tude of the mass flow predicted by CFAST agrees with those pre-
dicted by the two other programs.
Some parameters predicted by these three programs are shown
in Table 3.3.

Table 3.3 Comparison of temperatures and mass flow, predicted with

the three simulation programs DSLAYV, CFAST and
SMAFS for a medium fast fire in a compartment of75 m 3 .

DSLAYV CFAST SMAFS

Temperature, oc I time, s 200 I 180 313 I 200 377 I 300
Mass flow, kgls I time, s 0.250 I 180 0.296 I 200 0.36 I 300
Volume flow, 0.333 I 180 0.488 I 200 0.663 I 300
m 31s I time, s

Mass flows predicted by DSLAYV, CFAST and SMAFS are peak

values, obtained at a certain time in the studied cases. As seen in
Table 3.3, both mass and volume flows predicted by SMAFS
simulations for the medium fast fire in the compartment of 75m3
are higher than those predicted by CFAST, which is in turn
higher than those predicted by DSLAYV. Flows obtained with
SMAFS cannot be considered as determining values because they
are very approximate and are the results of only one simulation,
which is not enough for any conclusions to be drawn. Unlike
CFAST and DSLAYV, SMAFS assumes no limit to oxygen. This
means that combustion proceeds even at low values of oxygen in
the room, in other words if there is available oxygen in the room.
This explains the "continuation" in fire growth compared with
then, predicted by DSLAYV and CFAST.

61
Spread of Smoke and Fire Gases via the Ventilation System
SMAFS is a field model program which divides the room into
thousands or even hundreds of thousands of grid points, as de-
scribed in Chapter 3. Such models can predict the variation in
conditions within layers. In this case it was used to determine pa-
rameters under the ceiling near the possible ventilation termi-
nals. Results from SMAFS simulations cannot be used directly in
smoke spread assessment, but they can show the trend in such
evaluations, such as:
- the pressure rise in the burning room can be approximately at
least double that predicted by CFAST
- temperatures near ventilation terminals can attain higher
values than those predicted by the two-zone model; it is however
difficult to draw any conclusions regarding flashover effects in
ducts due to lack of data on the subject
- mass and volume flows can exceed values predicted by two-
zone models.
One of the important conclusions in this connection is that it is
reasonable for safety evaluation to prefer higher parameters of
the kind, if possible, to lower ones.

3.5 Window breakage modelling

As seen from Table 3.1, a compartment fire can go out by the de-
pletion of available oxygen. The CFAST fire model assumes that
the envelope of the room of fire origin is perfectly airtight except
for leakage paths. Window breakage sets limits for the "closed"
fire scenario.
If a window in the building envelope were to be broken during
the fire, the whole fire scenario would change and other fire
phases and hazard parameters would be observed. The fire model
discussed above and the whole fire behaviour would not therefore
be valid any longer. Before breaking, the window acts as a wall,
and after breaking, as a vent for the escape of toxic fire gases and
as an inlet for outdoor air. As the breaking of window glass due to
thermal stress and pressure in compartment fires is a very com-
monly observed phenomenon, it is necessary to predict the time at
which the window breaks and the ambient conditions which make
this possible. CFAST can not predict breakage of window glass

62
 Section 1. Analysis of Fire Behaviour in Single Spaces
Fire Simulations
during a fire and it assumes that the building envelope is of solid
construction.
The results of simulations should therefore be analysed further
concerning glazing behaviour. There are a lot of parameters which
can play an important role in the assessment of window breakage:

-temperature in the room,

- pressure in the room,
- outdoor temperature,
-window geometry and size,
- number of panes of glass,
-quality of glass,
- construction of the window,
- rapidity of fire,
- direct/indirect contact with flames,
-vertical temperature gradient in the room,
- window bars,
- glass absorption,
-etc.

3.5.1 Study of the literature

The initiation and propagation of cracks in glass and the breaking
of glass due to the heat developed by a fire is a very complicated
problem, and its analysis is beyond this study. In order to evalu-
ate window behaviour and to estimate or assume some character-
istics, the following study of the literature relating to this was
made. This study is based on different sources, but mainly on the
summary made by Joachim Bergstrom in his report [13], which
carries out a survey of the literature, including 61 reports, com-
puter programs and articles in order to summarise present
knowledge concerning the behaviour of glazing in fire. Certain ex-
perimental data have been published, and the first computer pro-
gram "BREAK-I" has been written by A. Joshi and P.J.Pagni [14].
Several conclusions and general statements can be made.

- There is a lack of data on window behaviour during real fires,

theoretical models are based on constant temperature in the
burning room, experimental data are insufficient.

63
Spread of Smoke and Fire Gases via the Ventilation System
-No unequivocal conclusions, with a satisfactory level of uncer-
tainty, can be drawn regarding window breakage for different
window constructions.
- There are no data on window behaviour during a fire as a
function of two parameters, temperature rise and pressure.
- The simulation program "BREAK-1" examines single-pane
window behaviour, which is an unusual construction for real win-
dows in northern climates.

But
- All studies have shown that glass cracks due to thermal stress
if the temperature difference in the glass itself varies between
58°C and 100°C.
- Window glazing breaks at 200-400°C in the burning room.

3.5.2 Simple window breakage model

A theoretical analysis can be made regarding the risk of window
breakage in the case of a compartment fire.
The one dimensional thermal stress, ~in glass due to a tem-
perature difference iJT between a large and a small area of glass
can be written as
(3.3)

where f3 is thermal expansion (K -1 ), EY is Young's modulus/elastic

modulus (N/m2 ) and r is the proportion of glass heated.
The window glass breaks when the thermal stress is larger then
the tensile strength, ab.
Only the temperature difference between different parts of
glass matters. The geometry is normally the central part of a win-
dow vs. frame/edges, but can also be inside surface vs. outside sur-
face.
Some typical values for window glass are
f3 8.5 - 9.5 10 -s (K -1),
Ey 7.0- 7.25 10 10 (N/m2 ),
ab 2.0 - 13.8 10 7 (N/m2 ).
The corresponding temperature difference which causes glass
breakage ranges from 31 K to 213 K, which agrees with experi-
mental data presented above.

64
 Section 1. Analysis of Fire Behaviour in Single Spaces
Fire Simulations
A model is needed to predict the window glass breakage tern-
perature. Such a model has been developed and it is presented in
Subsection 3.5.3.

3.5.3 Window glass temperature model

Physical model

Ts

Tft)

pAc

Assumptions and simplifications:

- total heat transfer for both radiation and convection between fire
and window glass is given by a constant hi (W/m2 K)
- total heat transfer for both radiation and convection between
outside air and window glass is given by a constant ho (W/m2 K)
- infinite heat conduction in window glass between inside and
outside surfaces
- heat conduction from window glass to the frame is described
through area rcrd and assumed distance r I 2
-the window glass temperature is connected to a single heat ca-
pacity

65
Spread of Smoke and Fire Gases via the Ventilation System
C=1C r 2 d pc J/K
Derived model parameters are
Qi =1C r2 hi W/K

W/K

Qc =1C r 2 d A I (r I 2) =21C r d A W/K

Heat balance equation for a circular disc of glass is

where T/t) is glass temperature at time t, T jt) is fire temperature

at point t and Ts is a start and surrounding temperature.
Assuming, that Ts=O and, hence, Tg(O)=O, and simplifying the
model with parameters T, time constant, and g, the gain,

we have

(3.5)

Heat conduction, Q<Y is much smaller than heat convection, Qi and

Qo' and can be neglected. Thus
1Cr 2 dpc dpc
2 =
1Cr (hi +ha) (hi +ha)
The time constant determines in a simple way how fast the state
of the models changes. In our case it determines the change in
glass temperature. For the case with a temperature step it seems
that the glass temperature would achieve a steady state after the
time interval of one time constant.
Notice that steady state temperature is not a fire temperature
T, but a reduced value gTr. Notice also that the reduced tempera-
ture parameter g with Qc = 0 becomes g = hJ (hi+h).

66
 Section 1. Analysis of Fire Behaviour in Single Spaces
Fire Simulations
In the case of a linearly increasing fire temperature the product
gkT tells how much the glass temperature can be delayed after
the reduced fire temperature.
For the window glass of d=0.004 m, p=2800 kg/m3, c=BOO J!kg,K,
2 2
A=O.Sl W/m,K and assuming that hi=16 W/m K and h =6 W/m K, 0

gives T=410 s=7 min. Fire dynamics is fast.

Fire temperature cases

k, k=T,d constant
Tft)= kt, k=T,d/td linear oc (3.6)
{
kt2, k=T,d!t/ quadratic

where T,d is a specific fire temperature at the specific time t d'

This can be illustrated with Figure 3.1.

Tft),
(°C)
Tfd Constant

Linear

t (min)

Figure 3.1 Schematic evaluation of temperature for different fire pro-

gresses.

Solving (3.4) along with fire temperature functions (3.5) for natu-
ral start condition gives the following expressions

Tft)=(l-e -tJT)gT,d constant

67
Spread of Smoke and Fire Gases via the Ventilation System

T jt)=kt-kT(l-e -tiT) linear

T jtJ=kt2-2kTt +2k~(l-e -tiT) quadratic

Solving the above presented equations for the window glass with
p=2800 kg/m 3, c=800 J/kg K, A=0.81 W/m K and assuming that
temperature rise Tr=300°C in the fire room is achieved at different
times after ignition, t, the values presented in Table 3.4 are ob-
tained for different values of the thickness, d.

Table 3.4 Temperature differences in studied window glass of differ-

ent thickness at different time points.

Tlr(t), K
td, Constant Linear Quadratic
mini d,mm
3 4 5 6 3 4 5 6 3 4 5 6
2 71 56 46 39 38 29 24 20 25 20 16 14
3 97 78 65 56 53 42 34 29 37 29 24 20
4 118 97 82 71 67 53 44 38 48 37 31 26
5 136 114 97 85 79 64 53 46 57 45 37 32

Table 3.4 shows that as the thickness of glass increases, the tem-
perature differences in the glass itself decrease. Standard window
glass thickness are 3, 4 and 5 mm.
The above presented temperatures can be interpreted as tem-
perature differences, as the start temperature for this case is as-
sumed to be equal to zero. The highest values of temperatures for
the same thickness of glass at the same time interval are observed
for constant fires, and the lowest for quadratic fires.
The most realistic fires are quadratic fires, i.e. Q:::: t 2 •
The temperature rise is less then 60 K for all cases with quadratic
temperature progress. This simple analysis shows that the win-
dows will not crack during first phase of the fire growth.
The radiation absorption length for glass is about 1 mm. This
means that most of the radiation is absorbed close to the entering

68
 Section 1. Analysis of Fire Behaviour in Single Spaces
Fire Simulations
surface. The radiation intensity reduces with a factor of e -z for
each mm. This increases the normal temperature difference be-
tween the inside and outside surfaces in case of fire (heavy radia-
tion).
Owing to the lack of knowledge and experimental data, it ap-
pears difficult to make a proper analysis of window breakage
conditions, but it may be assumed for the purpose of this study
that a temperature interval of 200-300°C in the burning room can
be considered a "dangerous" temperature interval. As the goal of
this research is to provide the most reliable parameters (volume
flows) which determine the progress of fire in a space, one more
assumption can be made regarding the peak temperature that oc-
curs in the fire. Although the "dangerous temperatures" for win-
dow breakage should not exceed 300°C, even higher temperatures
can be accepted in the upper layer in the room of fire origin in or-
der to predict the peak volume flow out of the room.

69
Spread of Smoke and Fire Gases via the Ventilation System

70
 Section 1. Analysis of Fire Behaviour in Single Spaces
Single Peak Fire Flow Models

4 Single peak fire flow

models

The main purpose of this chapter is to test a simple method and

algorithm for the prediction of peak fire flows which are
illustrated by Figure 4.1. This method has been developed for the
present study.
The natural question is whether peak fire flow can be calculated
with explicit expressions if the type of fire or the heat release
curve and the room volume are known?
The answer is yes if the following assumptions are made:

Each MJ heat release consumes a (m 3 ) of room air

Each MW heat release produces f(m 3/s) fire flow to the room

The share of room air volume used by the fire is given by the
factor r. This parameter can also be interpreted as the oxygen
reduction factor.
The fire flow depends directly on the rate of heat release. The
heat released is accumulated as released energy. The released
energy corresponds to the given amount of consumed room air.
The volume of room air limits the fire at a certain time. Thus the
peak fire flow is determined indirectly by the heat release at this
time.
The problem can be solved if the rate of heat release is a simple
function that can be integrated. One example is a tn-function. It is
also found that a heat release function described by linear
segments can be treated. A minor drawback is that the calculation
has to be made segment by segment until a solution is found.

71
Spread of Smoke and Fire Gases via the Ventilation System

Q(t), Fire progress (here

(W) described as a linear
function of time) produces
certain heat at time t.
Q(t) = f(t)

E(t) Energy released by the fire

(J) during the period oft
seconds.
E(t) = tj Q(t) dt
0

Energy can be expressed

in terms of the volume of
room air consumed and also
in terms of the air
0.1Vroom utilisation factor r.
V(t) = a E(t) I r

t (s)

The fire flow is proportional

to the heat release rate at
qwith vent timet.
q without vent. q/t) = fQ(t)

t (s)

Figure 4.1 Logical sequence of problem solution.

72
 Section 1. Analysis of Fire Behaviour in Single Spaces
Single Peak Fire Flow Models
The above schematic process description corresponds to a linear
fire progress. A more detailed and general analysis is presented in
this chapter.

4.1 Simple general model

There are different ways in which fire progress can be described.
A general description can be made on the assumption that the
growth phase of a fire can be characterised as a function of time to
the n-th power. These idealised fire energy release rates are
described as follows:
Q(t)=atn w (4.1)

where Q is in Wand t is in seconds, a is in W/sn and is specific for

a certain fire progress. The energy released by fire during the
period oft seconds is defined by
t

E(t) = f Q(t)dt J (4.2)

0

Substituting equation 4.1 in 4.2 leads to

E (t) =a tn+l I (n + 1) J (4.3)

Energy can be expressed in terms of the volume of air consumed

in the room

aE(t)=V(t) (4.4)

where a is a coefficient which expresses the volume of air that is

consumed by fire up to timet.
All oxygen in the room cannot be consumed. This air/oxygen
utilisation can be described by the factor r, normally around 0.5.
Equation (4.4) is modified to
aE(t)=r V (4.5)

Substituting of E(t) by (4.3) in (4.5) gives

a a tn+l /(n+l)=r V (4.6)

73
Spread of Smoke and Fire Gases via the Ventilation System
Solving equation (4.6) fort, we have
t=((n + 1) rV I a)) 11 n+l s (4.7)

Energy released by the fire in a certain room during time t

produces flow due to gas expansion

q(t)=fQ(t) (4.8)

where f ((m 3 Is) I MW) is a coefficient of volumetric expansion due

to energy input.
Substitution of(4.1) and (4.7) into (4.8) gives
q(t) =fa [(n + l)r V I a ar'n+l m 31s (4.9)

q(t) = f[(n + l)r I ar'<n+l) av<n+l) vnl(n+l) m31s (4.10)

Therefore, if

n=O q~ voal m 31s (4.11)

n=1 q~
Vl/2 a112 m 31s (4.12)

n=2 q~
V 213 al/3 m 31s (4.13)

n=3 q~
v314 al/4 m 31s (4.14)

One remark on the above numbers is that the peak fire flow
increases less than in proportion to the room volume. This means
that a fire in a small room creates a relatively large peak fire flow.

4.1.1 Maximum fire flows predicted with DSLAYV

results
In [16], Bengt Hagglund presented the results of twenty four
2
simulations of t -fires in closed rooms of different sizes. The fire
flows generated by these fires were analysed. The MINITAB
program [17] was used for this purpose. The resulting equation
showed good agreement with predicted flows:
q = 0.28 V o.s3 a 0.425 (4.15)

74
 Section 1. Analysis of Fire Behaviour in Single Spaces
Single Peak Fire Flow Models
The coefficient of determination, R2 , is equal to 99.3%, and R2
adjusted for degrees of freedom, R 2(adj), is equal to 99.2%.
Equation (4.15) can be rewritten with certain approximation, as
q~ 0.3${i (4.16)

A more precise conclusion is that the flow q ( 4.16) predicted by

DSLAYV corresponds to the flow q described by (4.12) and (4.13)
when the value ofn for atn fires is between 1 and 2.
Flows predicted by the model (4.15) are slightly larger then
those predicted by DSLAYV. The difference between them can
however be considered acceptable.

4.2 Analysis of fire flow prediction models

The flows predicted by CFAST and described in Chapter 2 are
generated by real fires in buildings of different types and sizes. In
order to generalise these predicted flows and to describe them as a
function of some variables well known to the user, the following
analysis is made.

4.2.1 Flows, predicted by CFAST simulations

For the purpose of the study, the peak flows, q (m 3/s), predicted by
CFAST simulations were analysed.

75
Spread of Smoke and Fire Gases via the Ventilation System

Table 4.2 Results of CFAST simulations (beginning).

No Type of V, t, 02, mb, T, T, p, q,

buildings ms s % kg/s oc K kg/m m 3/s
3

1 48.0* 120 10.5 0.250 281 554 0.6368 0.39

2 48.0 * 120 10.9 0.235 263 536 0.6582 0.36
3 Dwelling 48.0 * 120 10.5 0.250 280 553 0.6378 0.39
4 house 134.4 150 10.8 1.170 322 595 0.5929 1.97
5 (sofa) 149.3 150 11.3 1.210 312 585 0.6031 2.01
6 180.0 160 10.4 1.350 327 600 0.5880 2.30
7 204.0 170 9.3 1.460 345 618 0.5709 2.56
8 28.8 200 16.8 0.079 153 426 0.8282 0.10
9 Dwelling 134.4 400 13.0 0.216 184 457 0.7720 0.28
10 house 149.3 440 11.3 0.221 199 472 0.7475 0.30
11 (bed) 180.0 450 12.4 0.257 185 458 0.7703 0.33
12 204.0 410 15.3 0.289 151 424 0.8321 0.35
13 36.0 260 10.9 0.127 244 517 0.6824 0.19
14 50.0 260 13.4 0.230 207 480 0.7350 0.31
15 Hospital 54.0 270 13.8 0.189 213 486 0.7260 0.26
16 75.0 270 15.7 0.247 175 448 0.7875 0.31
17 112.5 270 17.4 0.421 138 411 0.8584 0.49
18 31.5 * 100 6.3 0.118 331 604 0.5841 0.20
19 31.5 * 60 15.9 0.163 209 482 0.7320 0.22
20 31.5 * 70 10.1 0.118 331 604 0.5841 0.20
21 45.0 * 130 5.3 0.135 318 591 0.5970 0.23
22 Office 45.0 * 80 16.9 0.223 174 447 0.7893 0.28
23 75.0 * 170 5.4 0.159 290 563 0.6266 0.25
24 75.0 * 100 16.6 0.261 158 431 0.8186 0.32
25 120.0 140 16.6 0.273 155 428 0.8243 0.33

76
 Section 1. Analysis of Fire Behaviour in Single Spaces
Single Peak Fire Flow Models

Table 4.2 Results ofCFASTsimulations (end).

No Type of V, t, 02, mb, T, T, p, q,

buildings m3 s % kg/s oc K kg/m 3
m 3/s
26 38.40 250 10.5 0.301 262 555 0.6357 0.31
27 Hotel 57.60 270 10.7 0.331 286 559 0.6311 0.52
28 84.00 280 12.7 0.483 269 542 0.6509 0.74
29 224.0 190 16.2 1.109 300 573 0.6157 1.45
* - Hazards with different placement of burning items.

Table 4.2 shows that in certain cases peak flows are obtained
when the temperature of the upper hot gas layer is higher then
300°C. Although such temperatures can be dangerous with regard
to window breakage, they are considered acceptable (see
Subsection 3.5).
Note that the values of oxygen in the room at the time
corresponding to maximum flows vary quite a lot even in the same
room, depending on fire placement.
It is seen from Table 4.2 that peak fire flow occurs shortly after
ignition. This period of time varies between 1 and 5 min. One
exception is a fire in a dwelling with a bed as the source of
ignition, where this period can be as long as 7 min 30 sec.
All fire flows presented in Table 4.2 are obtained for closed
spaces with assumed leakage paths at the sill of the closed
doorways. Some other leakage paths were also tested- at the top
of the door and vertically along the sides of the door. Mass flows
through leakage at the sill were larger then those at the other
locations (5-20%) which depends on higher density of air at the sill
of the door. At the same time temperature, pressure, heat release
rate and interface position were insignificantly affected. The
location of fire in the room affects fire progress as burning can be
constrained by the available oxygen. Wall and corner effects
describes with an "image method" in CFAST model, which uses
different coefficients- 1/2 open fire for a wall fire and%, open fire
fore a corner fire.

77
Spread of Smoke and Fire Gases via the Ventilation System
Owing to the complications in the program in determining fire
location according to given fire position co-ordinates, different
cases for the same rooms were tested. For the purpose of
generalisation, fire locations which give cases highest fire flow
values were chosen.
A regression analysis of the flows presented in Table 4.2 for the
studied cases was made with the statistical program Minitab [17].
Owing to the paucity of data and generalisation of the whole
range of predicted flows, the results are poor since there are
different fire models which determine different fire flows.
The results of the analysis for each fire model are set out below.

Table 4.3. Results of regression analysis with Minitab.

Type of Regression equation R2% R2(adj),

building '
%
Dwelling house q = 0.0018 V1. 39 98.7 98.4
(sofa)
Dwelling house q = 0.01 y0.68 99.8 99.7
(bed)
Hospital q = 0.013 V0 •76 88.0 84.0
0 34
Office q = 0.066 V • 76.2 72.2
0 85
Hotel q = 0.016 V • 97.7 96.5

Table 4.3 shows good agreement (R2 and R2(adj)) between flows
predicted by CFAST and the regression line in the case of fires in
dwellings and fires in hotels. The paucity of data could influence
the results of this analysis.

4.2.2 Modell. Used linear segment fires as tn-fires

The actual fires used for fire simulation and shown earlier in
Figures 2.3 - 2.6 can be described as curves consisting of linear
segments. Such functions cannot be easily used to calculate the
peak fire flow, as Subsection 4.1. The fire curves used have been

78
 Section 1. Analysis of Fire Behaviour in Single Spaces
Single Peak Fire Flow Models
approximated as tn-fires using two points on the fire curve. The
exponent of interest n is calculated as

The selected points and calculated exponent n are shown in

Table 4.4.

Table 4.4 Values ofn for atn model.

Type of building tl, s Ql, t2' s Q2' n

kW kW
Dwelling house (sofa) 60/ 100/ 120/ 300/ 1.584/
110* 260* 150* 1035* 4.454*
Dwelling house (bed) 150 49.9 270 160 1.982
Hospital 150 75 270 262 2.128
Office 50 105 100 187 0.833
Hotel 200 100 300 900 5.419
*-Compartment fires in "larger" flats.

It will be seen from Table 4.4 that the values of n exhibit

considerable variation between 0.833 and 5.419.
Flows predicted with the help of this method are presented in
Table 4.5.

4.2.3 Model 2
The heat release rate given as a linear segment function can for
the n-th segment be stated as
Q(t) = Qn-l + kn-1 f w (4.17)

where t is the relative time in the interval (0, tn-tn_1 ) and the slope
kn_ 1 is given by
W/s (4.18)

79
Spread of Smoke and Fire Gases via the Ventilation System
It is possible to calculate when the room air is consumed to the
extent of the factor r. This gives the relation similar to (4. 7) and a
second order equation in t.
(4.19)

Where En_ 1 is energy released up to time tn_ 1 and is simply

integrated from Q(t). If the relative time t is larger than the time
interval, the calculation of (4.19) has to be repeated for the next
time and segment interval until the solved relative time is within
the actual time interval. All tested fires have been described with
only three segment for the increasing part of the fire.

-·-·····"·'····'·················i----

Figure 4.2 Fire growth curve assumed for Model 2.

Flows predicted with the help of this method are presented in

Table 4.5.

4.2.4 Model 3
The two previous models assume that no volume is driven out of
the room during timet.
This subsection presents a model which predicts flows according
to the principle outlined in Subsection 4.2.1. It describes the
studied hazards in terms of atn -fires on the assumption that a
certain volume is driven from the burning room during timet.

80
 Section 1. Analysis of Fire Behaviour in Single Spaces
Single Peak Fire Flow Models
We have
rV I (1 + E(t) I CT0 ) =a E( t), and

r V =a E ( t) + a E 2 ( t) I CTa ,
By solving the last equation for E(t) and substituting it in the
equation for Q and q (algorithm 1, Subsection 4.2.2), the
corresponding flow can be calculated.
Flows predicted with the help of this method depend on the
value of C. The major problem in this method is to evaluate C.
Due to the complexity of the problem, method 3 was not chosen
as a working method.

81
Spread of Smoke and Fire Gases via the Ventilation System

4.3 Flows predicted by computer simulations

and different models
The table below summarises the results and conclusions of this
chapter.

Table 4.5 Flows predicted by CFAST, model 1 and model 2

(beginning).

No Type of V, q q 1' q 2' f, q2

building ms CFAST m /s 3 3 3
m /s (m /s)/ (edited),
MW m 3/s
'
m 3/s
1 48.0 * 0.39 0.42 0.40 1.30 0.40
2 48.0 * 0.36 0.42 0.40 1.20 0.40
3 Dwelling 48.0 * 0.39 0.42 0.40 1.30 0.40
4 house 134.4 1.97 2.09 1.66 1.80 1.93
5 (sofa) 149.3 2.01 2.27 1.77 1.87 2.08
6 180.0 2.30 2.54 1.97 1.70 2.36
7 204.0 2.56 2.70 2.12 1.60 2.56
8 28.8 0.10 0.16 0.14 1.00 0.14
9 Dwelling 134.4 0.28 0.44 0.42 0.80 0.42
10 house 149.3 0.30 0.48 0.44 0.70 0.44
11 (bed) 180.0 0.33 0.54 0.49 0.75 0.49
12 204.0 0.35 0.59 0.52 1.00 0.52
13 36.0 0.19 0.22 0.22 0.83 0.22
14 50.0 0.31 0.28 0.28 1.70 0.30
15 Hospital 54.0 0.26 0.30 0.29 1.00 0.29
16 75.0 0.31 0.37 0.36 1.20 0.37
17 112.5 0.49 0.48 0.47 1.90 0.56

82
 Section 1. Analysis of Fire Behaviour in Single Spaces
Single Peak Fire Flow Models

Table 4.5 Flows predicted by CFAST, modell and model2 (end).

No Type of V, q qp q 2' f, q2
building ma CFAST, m /s 3 3 3
m /s (m /s)/ (edited),
m 3/s MW m 3/s
18 31.5 * 0.20 0.23 0.23 1.10 0.23
19 31.5 * 0.22 0.23 0.23 1.80 0.23
20 31.5 * 0.20 0.23 0.23 1.30 0.23
21 45.0 * 0.23 0.27 0.26 1.00 0.26
22 Office 45.0 * 0.28 0.27 0.26 1.80 0.34
23 75.0 * 0.25 0.34 0.30 1.00 0.30
24 75.0 * 0.32 0.34 0.30 1.73 0.41
25 120.0 0.33 0.42 0.42 1.43 0.42
26 38.4 0.31 0.46 0.41 1.50 0.41
27 Hotel 57.6 0.52 0.65 0.56 1.30 0.56
28 84.0 0.74 0.90 0.73 1.70 0.84
29 224.0 1.45 2.06 1.29 1.61 1.58
*-Hazards with different placement of burning items.

q1 flows predicted by the model 1

q2 flows predicted by the model 2
q 2 (edited) flows predicted by the model2 (edited).
Table 4.5 shows flows predicted by CFAST and by both models.
Flows predicted by model 1 are overestimated over almost the
whole range of values, with the exception of cases 14, 17, 22 (in
italics). Overestimation varies between 9% and 61%.
The method described in Section 4.1 for predicting flows due to
an at;n fire in a certain room, which in the case of this study is
based on a real fire scenario, does not yield realistic values and
cannot be used. This can be illustrated with the following
(exaggerated) figure.

83
Spread of Smoke and Fire Gases via the Ventilation System

Q(W)

Real fire

Error field

t (s)

Figure 4. 8. Prediction of energy released by fire during the period oft

seconds.

As seen from Figure 4.8, the error in estimating energy released

by fire during the period t (s) can differ considerably in an
arbitrary "real fire" case from the t2 - fire growth case. Note that
this figure shows the growth stage of fire.
Flows predicted by model 2 show better agreement with those
predicted by CFAST, but there are more underestimated cases for
larger rooms (in italics). These underestimated flows appear to be
directly related to the value of f assumed in the model (1.0
(m 3/s)/MW). When the value off exceeds 1.6, the results show
lower flows. Portion of heat released by a fire which is absorbed by
the room air increases in larger rooms. When in such cases it is
assumed that f = 1.6 (m 3/s)/MW, the flows presented in the
column "q model2 (edited), m 3/s" are obtained.

84
 Section 1. Analysis of Fire Behaviour in Single Spaces
Single Peak Fire Flow Models

4.4 Generalisation of predicted flows

Flows predicted by the above simulation program and different
models were presented in Table 4.5. A regression analysis made
by Minitab for flows predicted by CFAST showed insufficient
agreement for fires in hospitals and offices (see Table 4.3). A very
brief comparison of flows predicted by CFAST with experimental
data showed underestimated predicted flows. At the same time,
prediction of flows by the field model program SMAFS showed
that flows can actually exceed those predicted by CFAST. It seems
reasonable to prefer higher flows to lower ones. In this connection,
flows predicted by the edited model 2 can be considered to be
designed flows.

4.4.1 Equations describing flows as a function of

volume
In order to describe predicted flows, the computer program
EXCEL was used. The resulting equations and their statistical
analysis are presented in Figures 4.9- 4.12.

85
Spread of Smoke and Fire Gases via the Ventilation System

Dwelling houses (sofa)

2.5
y = 1.4922Ln(x) - 5.380
2
R =1
2

1.5

0.5

0
0 50 100 150 200

Dwelling houses (bed)

3
q (m /s)
3.-------------------------------------------~

2.5

2 - y =0.1902Ln(x)- 0.5021-
R2=0.9946 I

1.5
y =0.0022x + 0.0915
2
R =0.9821
L__----~---------

0.5

o+--------+--------,_--------r--------+------~

0 50 100 150 200 250

V(m)

Figure 4.9 Predicted flows (rhombus) and trend lines for dwelling
houses.

86
 Section 1. Analysis of Fire Behaviour in Single Spaces
Single Peak Fire Flow Models

Hospitals
3
o (m /s)

0.6

0.5

0.4

0.3

0.2

0.1

0
0 50 100 150 200

Figure 4.10 Predicted flows (rhombus) and trend line for hospitals.

Offices
3
q (m /s)

0.6

0.5

0.4

0.3

0.2

0.1

0
0 50 100 150 200 250

V(ml

Figure 4.11 Predicted flows (rhombus) and trend lines for offices.

87
Spread of Smoke and Fire Gases via the Ventilation System

Hotels
1.8 , - - - - - - - - - - - - - - - - - - - - - - - - - - ,

1.6
0 7677
y = 0.0256x '
2
1.4 R = 0.9898

1.2

0.8

0.6 y = 0.0061x + 0.2283

2
R = 0.9838
0.4

0.2

0+----~----+----r-----+---~

0 50 100 150 200

Figure 4.12 Predicted flows (rhombus) and trend lines for hotels.

It is seen from the Figures 4.9 - 4.12 that linear equations

describe predicted flow values in all the studied cases with
sufficient agreement. In certain cases better agreement was
obtained for the same variables with other types of equation. The
equations which describe the phenomena with the best accuracy
are presented in Table 4.6.

Table 4.6 Equation describing flows in the buildings.

Type of building Equation

Dwelling house (sofa) q = 1.4922 ln (V) - 5.3806 1.0000
Dwelling house (bed) q = 1.1902 ln (V) - 0.502 0.9946
Hospital q = 0.0043 V+ 0.0627 0.9831
Office q = 0.2214ln (V)- 0.5291 0.9873
Hotel q = 0.0256 V0" 7677 0.9898

Flows obtained in dwellings with a burning sofa are essentially

larger then those obtained in dwellings with a burning bed. As it

88
 Section 1. Analysis of Fire Behaviour in Single Spaces
Single Peak Fire Flow Models
is the most dangerous case that is of interest for flow assessment,
only the case with a sofa as the burning item will be presented
below for fires in dwellings.
Flow obtained for a 100m3 hospital room according to Table 4. 6
agrees with that presented in [34].
As the equations presented in Table 4.6 show sufficient
agreement with predicted flows, they can be applied to
corresponding types of buildings of different sizes, i.e. they can be
used for generalisation of the results obtained. Characteristic flow
curves for the chosen buildings are presented in Figure 4.13.
Space volume varies from 30 m 3 to 240 m 3 . These volumes are
assumed to be those most usual for the described types of
buildings. Flows in spaces smaller then 30 m 3 can be assumed to
be equal to those for rooms of 30 m 3 .

Flows

-+-Dwelling houses

~Hospitals

.......,._Offices

··· •·· ... ·hotel

0 50 100 150 200 250

V(m)

Figure 4.13 Flows in the chosen types of buildings.

Hotel rooms are usually furnished with a bed which was

assumed to be the item that initiates fire. This statement can be
true for relatively small rooms. In larger hotel rooms and suites,
which can also be furnished with a sofa, another curve for
dwellings can be used in order to evaluate flow in the fire room.

89
Spread of Smoke and Fire Gases via the Ventilation System
Curves of (4.18) for different al - fires, simulated with the
computer program DSLAYV, are plotted in the figure below.

2
Flows (<X{: -fires)

3
q (m/s)

2.5

2
-::K- q 1,medium
1.5 ........_ q2, fast
--+- q3 ultrafas

0.5

0
0 50 100 150 200 250
3
V(m)

2
Figure 4.14 Flows for different at - fires according to the equation
(4.18).

Curves in the above figures are all plotted in the figure below for
purposes of comparison.

90
 Section 1. Analysis of Fire Behaviour in Single Spaces
Single Peak Fire Flow Models

Flows (all cases)

-w- Dwelling houses

-+--Hospitals

~Offices

hotel

-*-- q 1,medium
---q2, fast

.......f-- q3, ultrafast

0 50 100 150 200 250

V(ml

Figure 4.15 Flows according to the equations in Table 4.3 and equation
(4.18).

Figure 4.15 shows that flows obtained for offices agree very well
with those predicted for the medium fast al - fires. Flows
obtained for hospitals correspond to values between those
predicted for medium and fast al -fires over the whole range of
3
volumes. Flows obtained for hotels up to approximately 100 m
are insignificant lower than those predicted for fast al - fires,
3
while flows in hotels larger than 100 m correspond to values
2
between those predicted for fast and ultrafast at - fires.
3
Flows obtained for dwellings up to approximately 100 m are
2
lower than those predicted for ultrafast at - fires. In contrast,
3
flows in dwellings larger than 100 m have higher values than
2
those predicted for ultrafast at - fires.

Analyses presented in this Section 1 consider fire progress in a

single space limited to the object in which it started. According to
the Table 2.4, between 23% and 53% of all fires spread to other
objects in the room of fire origin. This case was not examined in
the present study. It could be interesting to examine such cases.
However, the results of the study show that it is relatively
difficult even in the case of the studied fires to assess whether

91
Spread of Smoke and Fire Gases via the Ventilation System
windows in the building envelope will be broken, as the
temperatures which develop in the hot gas layer during the
examined fires of one - four minutes' length can exceed the
temperature level which is dangerous for glass breakage. This
means that fire flow generation process which is of interest for
this particular study can be limited so far to the above presented
cases.

92
 Section 2. Operation of Ventilation Systems during Fires
Study of Literature

Section 2. Operation of
ventilation systems
during fires

5 Study of the literature

In order to ascertain the current state of knowledge regarding

smoke spread through a ventilation duct network, the following
study was made.
According to the building regulations and codes in different
countries the only recommended way to prevent the spread of
smoke is to install different kinds of dampers and smoke barriers
and to shut down the ventilation system in the event of fire, or to
rely only on shutting down the ventilation system, designed ac-
cording to given pressure drop relations. Other solutions were also
allowed in Sweden until 1967. The operation of fans during fires
was not excluded by the Building Code SBN 1967. Subsequent
regulations recommended that fans should be turned off in case of
fire. In the present Swedish Building Regulations [8], 1994, it is
recommended that under certain conditions the existing ventila-
tion system, which removes hot smoky gases from the compart-
ment without spreading them to adjacent rooms, should be run. A
handbook [16], 1996, further improves the safety design of venti-
lation systems and proposes some calculation methods in order to
reduce and even eliminate the risk of smoke spread via ventila-
tion ducts to adjacent rooms.

93
Spread of Smoke and Fire Gases via the Ventilation System

The state of knowledge in this regard and research publications

have been searched through different databases in science and
technology and via several search programs in Internet. In addi-
tion, current publications on this matter in different scientific and
technical periodicals were followed, and an attempt was also
made to discuss the subject with researchers via scientific news
groups on the Net. The last one failed, as no researchers have re-
sponded.

5.1 Databases
All in all seven major databases were searched through twice with
a one year interval. These databases are as follows:
-IBSEDEX, which covers all aspects of building serv-
ices world-wide, published since 1960. It includes information on
technical books, reports, conference papers, journal articles and
standards,
-ICONDA, which covers the world-wide literature on
all fields of building construction and design, energy conservation,
civil and structural engineering, architecture and town planning,
published since 1976. Sources of information are technical books,
conference papers and journal articles,
-COMPENDEX, which is a comprehensive interdis-
ciplinary engineering database, covering different areas of engi-
neering, civil, energy, mechanical, environmental, etc. Sources of
information are technical reports, conference papers and proceed-
ings, published since 1970,
-BRIX-FLAIR, which covers technical information
published since 1960,
-TRADE and INDUSTRY Database, which covers
technical information for industrial needs published since 1976,
-CORDIS, which includes nine different databases,
provides information on all aspects of EU R&D activities includ-
ing current calls for proposals and tenders, events, publications
and activities in preparation. Publications contain abstracts and
bibliographic information on publications, reports and scientific
papers arising from EU research activities as well as other scien-

94
 Section 2. Operation of Ventilation Systems during Fires
Study of Literature
tific and technical documents in the fields of the environment, en-
ergy and information sciences,
- AIRBASE, which is the Bibliographic Database of
the International Energy Agency's Air Infiltration and Ventilation
Centre. It contains abstracts of articles and publications relating
to energy efficient ventilation and includes topics on ventilation
strategies, design and retrofit methods, calculation techniques,
standards and regulations, measurement methods, indoor air
quality and energy implications. Entries are based on articles and
reports published in journals, internal publications and research
reports.
The information search showed that there are few articles on
this issue. Some of these are comparatively old, from the 60s and
early 70s, which makes them nearly unobtainable. The author
was unsuccessful in getting information on corresponding re-
search with regard to the operation of ventilation systems during
a fire or, at least, in discussing its possibilities.

95
Spread of Smoke and Fire Gases via the Ventilation System

96
 Section 2. Operation of Ventilation Systems during Fires
Theoretical Methods

6 Theoretical methods

In this chapter the main principles of treating ordinary ventila-

tion systems during compartment fires will be discussed. On the
basis of up-to-date research results on the subject, as well as
knowledge about smoke spread via ventilation ducts, the following
analysis has been performed.

6.1 Background
The main principles of running ordinary ventilation systems dur-
ing compartment fires were formulated in [1], [2], and [3]. The
principal points in these reports are as follows ..
Whether or not smoke spread will occur is determined by the
energy released by the fire, the airtighteness of buildings and the
layout of the ventilation system. There are three main types of
ventilation systems: exhaust, supply and exhaust, and natural
ventilation. If a natural ventilation system serves each separate
room with a separate duct, smoke spread will never occur unless
the duct itself leaks. In both exhaust and supply-exhaust me-
chanical ventilation systems, on the contrary, spread of smoke via
ducts is always possible as a certain number of rooms are con-
nected by those ducts.
Assuming that air in the burning room is heated from Tr toTe,
at every point in the room during a fuel controlled fire, the air
volume V will increase by
(~!If -l)V (6.1)

This means that the volume of smoke is of the same magnitude as

the room volume and is limited by it.

97
Spread of Smoke and Fire Gases via the Ventilation System

A very rough estimate of ftre gas flow can be made as follows:

(6.2)

If the temperature, measured in K, doubles over a period of one

or two minutes, the mean flow according to (6.2) will be V/60 and
V/120 respectively. These flows are much larger than ordinary
ventilation flows in the room. For example, in a room of 30 m3
with 10 lis (36 m3/h) air flow, the estimated fire gas flow is 500 lis
(1800 m3/h).
The conditions when smoke from the burning room starts
spreading via ducts to the adjacent rooms can be defined as limit
or boundary case conditions. There is only one boundary case,
whether it is based on the pressure in the fire room or the fire
flow. This boundary case can be determined by a certain pressure
or flow in the burning room, at which smoke will start to spread
from the fire room to the adjacent room(s) via the ventilation duct.
The boundary case for smoke spread via ventilation ducts can be
described by zero air flux in the adjacent room(s). While pressure
rise in the fire room depends on different parameters, such as
airtightness and the number of openings, flow is a function of the
energy released over a certain period of time, as discussed in
Chapter 3, and was chosen as a characteristic value in this study.
Expressions and models presented in this Subsection are the
summary of algorithms and conclusions discussed in [1], [2] and
[3].

6.2 Flow model

A pressure difference, created by a fan, causes airflow. Airflow
in the ventilation system is governed by the total pressure pt ,
which is a sum of the dynamic and static pressure. The total pres-
sure diminishes downstream because of friction in the ductwork
and pressure losses in fittings.
In order to describe an arbitrary ventilation system the follow-
ing variables are used:

pi pressure drop in a component i

98
 Section 2. Operation of Ventilation Systems during Fires
Theoretical Methods
qi flow in a component i
Ri resistance (turbulent) of a component i

For turbulent flow

(6.3)

For two resistances connected in series for turbulent flow

Rs=Rl+R2 (6.4)

and for parallel coupling

R-:·5 =R-:·5 +R-;_0·5 (6.5)

For a flow junction, the sum of all inlet and outlet flows is equal to
zero, which gives
0= L junction qi (6.6)

For an arbitrary loop duct, the sum of all pressure looses in the
loop is always equal to zero, which gives
O=Lzoop Pi (6.7)

6.3 Theoretical model for the analysis of

ventilation systems
A very short description of basic principles and a theoretical
model of ventilation systems, determining boundary conditions of
smoke spread through ventilation ducts, is presented in this sub-
section. In the following analysis certain common assumptions
were made:
the ventilation system serves two rooms
all pressure drop are turbulent
the pressure, produced by the fan is equal to 1 and do not de-
pend on the flow fluctuations
total flow is normalised to 1
flow in room A is equal to q
flow in room B is equal to 1-q

99
Spread of Smoke and Fire Gases via the Ventilation System

room A is a "burning room".

6.3.1 Exhaust ventilation system

The layout of an exhaust ventilation system for two rooms and its
model are presented in Figure 6.1.

p=O

q 1- q

ROOM A ROOM B
p=-b

p= -a

p='-1

p=O

Figure 6.1 Layout of exhaust ventilation system for two rooms and its
model [3].
In the following analysis further assumptions were made:

total pressure drop is normalised to 1

pressure before the plenum (collecting duct) is equal to -a
pressure in both rooms is equal to -b

All the resistance R. can be described as

1

(6.8)

100
 Section 2. Operation of Ventilation Systems during Fires
Theoretical Methods
R 2 = (a-b) I (1- q) 2 (6.9)

(6.10)

(6.11)

(6.12)

In case of fire, smoke spread will occur if pressure in the junc-

tion c changes from -a to 0 (zero). This results in increased flow q
in room A and reduced flow in room B from (1-q) to 0 (zero). If flow
in the main duct is equal to qk, inlet flow through the facade is
equal to q f', and fire pressure in the room a is equal to p b' the fol-
lowing expressions are valid:
for the main duct
1=R 5 q~ (6.13)

for the branch to the room A

Pb=R1q~ (6.14)

for the facade to the room A

Pb =R3q; (6.15)

This gives
q~ =1 I (1- a) (6.16)

q; =(a-b)/ ((1-a)b) (6.17)

The total needed fire flow

qb =qk + qf (6.18)

and the corresponding pressure rise is

(6.19)

Some important conclusions can be drawn:

fire flow does not depend on the ventilation flow

101
Spread of Smoke and Fire Gases via the Ventilation System

fire flow is larger, than the total ventilation flow (>1)

large pressure drop in the main duct requires large fire flow
and high pressure in the burning room. This statement shows
that the ordinary way to prevent smoke spread via the ventilation
duct is to make the pressure drop in the branch much larger than
that in the main duct.
qr> qk if a/ b > 2. This statement can be interpreted as a ra-
tio of the two flows
qf/ qk, if

q1 I q k =(a I b- 1) o.s
If the pressure drop in the main duct is high, the danger of smoke
spread is significant. A relatively little fire flow can be enough to
cause spread of smoke and fire gases to the other room.

102
 Section 2. Operation of Ventilation Systems during Fires
Theoretical Methods

6.3.2 Exhaust and supply ventilation system without

air leakage
The layout of an exhaust and supply ventilation system for two
rooms without air losses and its model is presented in Figure 6.2.

p=O
p=a

p=b

ROOM A ROOM B

p=O

P= -c

P= -d

p=O

Figure 6.2 Layout of an exhaust and supply ventilation system for two
rooms without air leakage and its model [3].

In the following analysis the following assumptions were made in

addition to the general ones presented in Subsection 6.3:
pressure in the rooms is equal to 0
pressure in the supply plenum (collecting duct) is equal to -a
pressure before the supply plenum (collecting duct) is equal to
-b
pressure before the exhaust plenum (collecting duct) is equal to
-c
pressure in the exhaust plenum (collecting duct) is equal to -d
pressure in both rooms is equal to 0

103
Spread of Smoke and Fire Gases via the Ventilation System

Parameters Rican be calculated from the following expressions

2
R1 =q I b (6.20)

R2 = (1- q) 2 I b (6.21)

2
R5 =q I c (6.22)

R6 = (1 - q) 2 I c (6.23)

R7 =11 (a-b) (6.24)

R8 =1 I (d- c) (6.25)

In case of fire with fire flow q b in room A, smoke spread will occur
if the inlet flow q 1 is equal to 0 (zero). The boundary condition for
smoke spread from the burning room A, assuming that the pres-
sure drop in terminals does not change, gives
(6.26)

Using the subscripts i for supply air, o for exhaust air and n for
ordinary operation, the last equation can be rewritten as
qb = (Ri I Ro + 1)o.s qn (6.27)

IfR. =R
l 0

(6.28)

Pressure in the burning room, which causes spread of smoke, is

(6.29)

Some important conclusions can be drawn :

fire flow does not depend on the pressure drop in registers

fire flow is approximately ~2 times higher than ordinary
ventilation flow
pressure in the burning room is always higher than the
pressure drop in terminals
pressure in the burning room is lower than pressure
produced by the fan.

104
 Section 2. Operation of Ventilation Systems during Fires
Theoretical Methods
In this case two fans can be interpreted as only one fan, because
both the duct network and the building are closed systems.
Switched off the exhaust fan affects the evaluated fire flow and
pressure for boundary case as d=O in Figure 6.2.
Switched off the supply fan will cause spread of smoke which
does not depend on the pressure in the fire room, because an out-
door air terminal is closed. The exhaust fan will have no effect in
this case if the inlet terminal is tight.
The proportion of smoke, k, which spreads to the other room is
k=f I (1+ f) (6.30)

where
(6.31)

If both fans are switched off, this would also result in smoke
spread, which is similar to the previous case.
The smoke goes the normal way through R 5 and the non desired
"wrong" way through R 1, R 2, the other way, and R6'
If equal flows goes both ways, q 1=q 2 =q I 2, then
R 1=R 5=R2 =R 6 , and
f= 11-../3,
k = 1 1( -../3 + 1) = 0.366

Conclusion: a lot of smoke goes the wrong way.

If only one out of ten ordinary flows goes non desired way (from
a small room to a big one), then
R 1=R5 and R 2=R 6 ,
R1 q; =R q; 9
2
2
and R 5 q; =R q; 9
6
2

R2 =R 6 = R1 I 81 = R5 I 81
According to (6.30), (6.31)
f=(RIR (1+1181+1181)) 0 ·5 ~1 andkz0.5

Conclusion: 50% smoke goes the non desired way to the rest of the
building or rooms.

105
Spread of Smoke and Fire Gases via the Ventilation System

If nine out of ten ordinary flows go the non desired way (from a
large room to a small one), then
R 1=R5 andR2 =R 6 ,
R 1 qn2 92
=R 2 qn2 an d R5 qn2 92
=R 6 qn2
R2 =R 6 = R1 I 81 = R5 I 81
According to (6.30), (6.31)
f =(RI 81R (1 I 81 + 1 + 1)
05
' ;::: 0.79 and k ;:::0.07

Conclusion: 7% smoke goes the non desired way to the rest of the
building or rooms.

To sum up the above estimate of the risk of spread of smoke to the

adjacent room(s) in the case with the supply fan or both fans
switched off, it is seen that
the maximum proportion of smoke goes the non desired way
from a small burning room to a large one, or from one of several
rooms to other rooms,
a lot of smoke goes the non desired way from one burning room
to another equal one,
a small amount of smoke goes the non desired way from a large
burning room to a small one.
a little share goes non desired way from a big burning room to a
small one.

106
 Section 2. Operation of Ventilation Systems during Fires
Theoretical Methods

6.3.3 Exhaust and supply ventilation system with air

leakage
The layout of an exhaust and supply ventilation system for two
rooms with air leakage and its model are presented in Figure 6.3.

Figure 6.3 Layout of exhaust and supply ventilation system for two
rooms with air leakage, and its model.

Air leakage is assumed to be equal to ordinary ventilation flow q n

at the normalised pressure difference pi.
This gives
(6.32)

(6.33)

The open system boundary flow can be approximately calculated as

the boundary fire flow of the closed systems with the leakage flow ad-
ded. The leakage flow is easily calculated from the fire pressure and
the airtightness of the rooms.

107
Spread of Smoke and Fire Gases via the Ventilation System

(6.34)

(6.35)

Pressure in the fire room for the boundary case is approximately

Pb =b, (6.36)

and the corresponding flow is

qb= 2 o.5 qn + (Pb I Pi )o.5 qn. (6.37)

Some important conclusions can be drawn:

-fire flow varies between .V2 and .V2+.V10 power air

flow, depending on the pressure drop in terminals (b
changes between 0 and 1) and on the normalised
pressure difference pi = 0.1
-fire flow is larger than in the case without air
leakage
-pressure in the burning room is always higher than
pressure drop in terminals
-pressure in the burning room is lower than the
pressure produced by the fan
-pressure in the burning room here is smaller than in
the case without air losses.

The above model was tested and confirmed experimentally. Experi-

mental data for the three-room case and its comparison with the vari-
ables predicted by the model are presented in [2]. The agreement can
be considered reasonable.

108
 Section 2. Operation of Ventilation Systems during Fires
Design of Ventilation Systems

7 Design of ventilation
systems

7.1 Characteristic ventilation systems

In very general terms, evaluation of the risk of smoke spread via venti-
lation ducts in an arbitrary building can be shown in Table 7 .1.

Table 7.1 Evaluating of the risk of smoke spread in buildings via the
ventilation systems duct network.

Ventilation system Running Not running

Buildings envelope, whole little great
Buildings envelope, dam- none little
aged

As can be seen, running the ventilation system during a fire can

reduce the risk of smoke spread via ventilation ducts.
Mechanical ventilation systems can be generally divided into
three major system groups:
-exhaust ventilation system
-supply-exhaust "closed" ventilation system (without
air losses)
-supply-exhaust "opened" ventilation system (with air
losses).

109
Spread of Smoke and Fire Gases via the Ventilation System
As this project is focused on certain types of buildings, the most
typical ventilation systems used in them will be discussed later
on.
One of the most typical ventilation systems in residential
buildings is an exhaust ventilation system with one exhaust fan
and a distribution system consisting of one or two main ducts with
branches to each flat. Another type of ventilation system, supply-
exhaust ventilation system, with supply and exhaust fans and dis-
tribution ductwork to each flat is also used in dwellings.
In both types of ventilation systems plenums can be both vertical
and horizontal.
Hospitals are usually ventilated by supply-exhaust ventilation
systems with branches connected to the dampers in each hospital
section.
Offices can be ventilated in different ways with the help of or-
dinary supply-exhaust ventilation systems with dampers in each
of the departments or rooms. "Zero"-ducts can be also used in such
buildings. This means that supply air is distributed via terminals
direct to office rooms, while exhaust air collects in corridors
("zero"-ducts) and is removed via exhaust terminals by the ex-
haust ventilation system.
Hotels are usually ventilated by supply-exhaust ventilation sys-
tems with branches connected to the dampers in each hotel room
or each hotel suite.

These ventilation systems can include different variations of

system layout, and the degree of complexity of these systems can
be widely different. The problems of air distribution in spaces and
the types of terminals, and their position in the rooms, are not
discussed here. Problems concerning the ventilation plant, its
construction and components, including air leakage in the heat
exchanger, are beyond this study.

7.2 Fire compartments

In order to define the term "adjacent room" in evaluating the risk
of smoke spread via ventilation ducts in the building, the question
on fire compartmentation seems to be of interest. The layout of a
ventilation system greatly depends on the building's inherent

110
 Section 2. Operation of Ventilation Systems during Fires
Design of Ventilation Systems
compartmentation, for the purpose of the present study - on fire
compartmentation.
According to Swedish Building Codes [8] and Quality assurance
requirements [20] a fire compartment is defined as a room or a
group of rooms, in which activity has no direct connection with the
activities in the rest of the building.
-In dwellings it embraces each self-contained flat.
-In hospitals it comprises each ward.
-In offices it comprises all rooms in each office.
-In hotels it comprises each hotel room or hotel suite.
Since the layout of the ventilation system generally conforms to
the layout of compartments, the risk of the spread of smoke to
adjacent fire compartments via ducts connecting different fire
compartments will be discussed with reference to fire compart-
mentation.
The problem of smoke spread within a fire compartment can
also be analysed in certain cases. In such cases the problem will
be confined to the evaluation of the risk of smoke spread to adja-
cent rooms, instead of to adjacent fire compartments.
In this study the risk of smoke spread via ducts to adjacent fire
compartments will be evaluated.

111
Spread of Smoke and Fire Gases via the Ventilation System

112
 Section 2. Operation of Ventilation Systems during Fires
Models and Tools

8 Models and tools

The following analysis of the schematic models, evaluation of the

risk of smoke spread and estimation of smoke spread in "real"
ventilation systems are based on the main flows produced by fires
in the examined types of buildings, which were set out in Section
1, along with the features presented in Chapters 5 and 7 of Sec-
tion 2.

8.1 PFS program

For the purpose of evaluating of smoke spread through ducts, a
computer program PFS was used. This program, developed and
written by Lars Jensen, is based on the semi graphical circuit
drawings and treats arbitrary flow systems of any structure and
layout, any media, and any problem: design or investigation. The
program manual for PFS includes three major reports and four
reports with applications [21]-[27]. The following assumptions
were made concerning ventilation systems analysed in the present
study:
-all ventilation systems are mechanical
-all pressure losses are governed by expressions for the "cold"
flow (under normal conditions)
-all ordinary ventilation flows correspond to fire compartment
grouping
-all room flows are equal.

113
Spread of Smoke and Fire Gases via the Ventilation System

8.2 Schematic model

The main principles of describing the ventilation system, and its
analysis, will be presented in this part. A simpler study of differ-
ent ventilation systems, serving four rooms, and the expected be-
haviour of these systems under certain conditions, will also be
shown. The main purpose of this study is to show how smoke
spread via ventilation ducts can be predicted for a certain venti-
lation system with the help of the computer program PFS. The
schematic layout of the analysed ventilation systems is shown in
Figure 8.1.

114
 Section 2. Operation of Ventilation Systems during Fires
Models and Tools

Exhaust ventilation system

Supply-exhaust ventilation system (closed)

Supply-exhaust ventilation system (open)

Figure 8.1 Schematic layout of four-room ventilation system.

115
Spread of Smoke and Fire Gases via the Ventilation System
Some of the usually occurring PFS elements and symbols are as
follows:
"+" define flow direction downwards and to the right as
positive
"" define flow direction upwards and to the left as
negative
h?O printout pressure
h?200 describes a fire with a flow of 200 lis, downwards or
to the right, and prints out pressure
h?-200 describes flow of 200 lis, upwards or to the left, and
prints out pressure
h,100:q describes pressure of 100 Pa and prints out flow
h?:q describes boundary case conditions and prints out
pressure and flow in a room
q,O describes boundary case's zero flow
s,value describes some initial flow for calculation, which
facilitates convergence
FF:hq< describes a fan with flow direction upwards and to
the left, and prints out pressure and flow produced by
this fan
FF:hq> describes a fan with a flow direction downwards and
to the right, and prints out pressure and flow
produced by this fan
t,10,50:q describes an unit with quadratic pressure loss of 10
Pa at flow of 50 1/s and prints out flow
d,200,6 describes 6 m long ventilation duct of 200 mm
diameter
UD notation for outdoor air intake
TD notation for transfer air device.

Calculations were made for the fan with the following character-
istics.

Pressure, Pa 100 100 0

Air flow, 1/s 0 200 BOO

There are four rooms with 50 lis air flow in each. Air leakage of 50
1/s at 50 Pa pressure difference is assumed in this case.
PFS provides the opportunity to describe resistance properties of
wyes and junctions in two different ways: as a quadratic function

116
 Section 2. Operation of Ventilation Systems during Fires
Models and Tools

and with the help of empirical equations. This is ruled by the con-
trol operator con=1 and con=2 respectively. Both types were
tested.

8.2.1 Exhaust ventilation

The layout of the exhaust ventilation system with the total air
flow of 200 1/s at 100 Pa pressure is presented in Figure 8.2. Five
cases were studied: a "normal" case and four boundary cases with
fire in each of the rooms in turn. Results are presented in Table
8.1. and Figure 8.3 (for con=1).

Table 8.1 Flows and pressure for boundary cases in rooms of fire ori-
gin for the examined exhaust ventilation system.

Case Flow in the fire Pressure in the

room, 1/s fire room, Pa
con=1 con=2 con=1 con=2
Normal 0 0 -10 -10
Boundary case for room 1 949 455 1551 392
Boundary case for room 2 483 559 444 597
Boundary case for room 3 810 957 1342 1772
Boundary case for room 4 641 1350 921 4053

As Table 8.1 shows, flows and pressures for boundary cases differ
significantly from one fire room to another. This generally de-
pends on the resistance properties of wyes and the resulting pres.,.
sure drop in these. As calculated in Subsection 6.1.3, the expected
flows and pressures for boundary cases are considerably higher
than in the normal (ordinary) case. Boundary flows and pressure
drops predicted with con=1 are higher then those for con=2, ex-
cept for fire case in room 1. According to Table 8.2 flow values
show sufficient agreement with theoretically predicted qb. The
deviation is due to the resistance properties of wyes and the re-
sulting pressure drop in these.

117
Spread of Smoke and Fire Gases via the Ventilation System
begin
flow 1/s
pressure Pa
control bend=1 con=1 duct=1 i tmax=99 table=12 dxs=25
set UD=t,10,50:q 160=d,160 200=d,200
fan FF 100:0 100:200 0:800
160, 6 160, 6 200,6 200,6 FF:hq
s,200
160,3 160, 3 160,3 160,3
h?-50 h?-50 h?-50 h?-50
h?O

160,6 160,6 200,6 200,6 FF:hq

s,200

h?:q

160, 6 160,6 200,6 200,6 FF:hq

s,200

h?O

160,6 160,6 200,6 200,6 FF:hq

s,250

h?O

160, 6 160, 6 200,6 200,6 FF:hq

s,200

h?O

end

Figure 8.2 PFS-layout of exhaust ventilation system for four rooms.

118
 Section 2. Operation of Ventilation Systems during Fires
Models and Tools
begin
flow 1/s
pressure Pa
control bend=1 con=1 duct=1 itmax=99 table=12 dxs=25
set UD=t,10,50:q 160=d,160 200=d,200
fan FF 100:0 100:200 0:800
160, 6 160,6 200,6 200,6 FF:hq
s,200 h:100.00
160, 3 160, 3 160,3 160,3 q:200.00
h?-50
h=85.062
h?O
h=-10.00

160,6 160,6 200,6 200,6 FF:hq

s,200 h:82.733
q:404.77

h=1551. 0
q:948.64

160,6 160,6 200,6 200,6 FF:hq

s,200 h:97.652
q:245.84

h?O
h= 0.000

160, 6 160,6 200,6 200,6 FF:hq

s,250 h:95.164
q:282.25

h?O
h= 0.000

160,6 160,6 200,6 200,6 FF:hq

s,200 h:101.30
q: 161.21

h?O
h= 0.000

end block problems elements 162

Figure 8.3 PFS-result of exhaust ventilation system for four rooms.

119
Spread of Smoke and Fire Gases via the Ventilation System

8.2.2 Closed supply-exhaust ventilation (without air

leakage)
The layout of the closed supply-exhaust ventilation (without air
leakage) system with the total air flow of 200 lis at 100 Pa pres-
sure is presented in Figure 8.4. Five cases were studied: a
"normal" case and four boundary cases with fire in each room in
turn. The results are presented in Table 8.2 and Figure 8.5 (for
resistance ofwyes as a quadratic function, con=1).

Table 8.2 Flows and pressures for boundary cases in rooms of fire
origin for the examined closed supply-exhaust ventilation
system.

Case Limit fire flow, Limit fire pres-

lis sure, Pa
con=1 con=2 con=1 con=2
Normal 0 0 0 0
Boundary case for room 1 76 82 81 83
Boundary case for room 2 72.5 68 80 79
Boundary case for room 3 71 68 81 82
Boundary case for room 4 66 61 71 81

As Table 8.2 shows, boundary values for both flows and pressures
are considerably lower than those in the case with exhaust venti-
lation, and the difference between con=l and con=2 cases is in-
significant. As calculated in Subsection 6.1.4, the expected flow for
the boundary case is qb =~2 qn, which in the present case is 71
lis. According to Table 8.2, the difference between flows and pres-
sures for the boundary cases is insignificant from one fire room to
another, and flow values show sufficient agreement with the theo-
retically predicted qb. The deviation is due to the resistance prop-
erties ofwyes and the resulting pressure drop in these.

120
 Section 2. Operation of Ventilation Systems during Fires
Models and Tools
begin
flow l/s
pressure Pa
control bend=1 con=1 duct=1 itmax=20
set 160=d,160 200=d,200
fan FF 100:0 100:200 0:800
160, 6 160, 6 200,6 200,6 FF:hq<

160,3 160,3 160,3 160,3

q,O t,50.2,50:q t,63.3,50:q t,63.8,50:q
h?:q

F:hq>

160, 6 160,6 200,6 200,6 FF:hq<

s,-200

h?O

t,38.1,50:q t,SS.O,SO:q t,69.5,50:q t,95.1,50:q

160,3 160,3 160,3 160, 3

F:hq>

160,6 160,6 200,6 200,6 FF:hq<

o, 3

h?O

t,38.1,50:q t,SS.O,SO:q t,69.5,50:q t,95.1,50:q

160,3 160,3 160,3 160, 3
s,200
F:hq>

160, 6 160,6 200,6 200,6 FF:hq<

160,3 160,3 160,3 160,3

t,55.6,50:q t,50.2,50:q t,63.3,50:q q, 0
h?O

s, 50
t,38.1,50:q t,SS.O,SO:q t,69.5,50:q t,95.1,50:q
160,3 160,3 160,3 160,3

F:hq>

end
Figure 8.4 PFS-layout of closed supply-exhaust ventilation system for
four rooms.

121
Spread of Smoke and Fire Gases via the Ventilation System

F:hq>
h:99.201
q:217.62
end block 1 problems 5 elements 257
Figure 8.5 PFS-result of closed supply-exhaust ventilation system for
four rooms.

122
 Section 2. Operation of Ventilation Systems during Fires
Models and Tools

8.2.3 Open supply-exhaust ventilation (with air

leakage)
The layout of the open supply-exhaust ventilation (with air leak-
age) system with the total air flow of 200 l/s at 100 Pa pressure is
presented in Figure 8.6. Five cases were studied: a "normal" case
and four boundary cases with fire in each room in turn. The re-
sults are presented in Figure 8.6 and Figure 8. 7 (for resistance of
wyes as a quadratic function, con=1).

Table 8.3 Flows and pressures for boundary cases in rooms of fire
origin for the examined open supply-exhaust ventilation
system.

Case Limit fire flow, Limit fire pres-

1/s sure, Pa
con=1 con=2 con=1 con=2
Normal 0 0 0 0
Boundary case for room 1 140 144 78 77
Boundary case for room 2 134 130 76 76
Boundary case for room 3 134 132 79 80
Boundary case for room 4 128 128 77 90

As Table 8.3 shows, the boundary values for both flows and pres-
sures are considerably lower than those in the case with exhaust
ventilation, but higher than those in the case with a closed sup-
ply-exhaust ventilation system, and the difference between con=1
and con=2 cases is insignificant. As calculated in Subsection 6.1.4,
the expected flow for the boundary case follows (6.37) and the fire
flow varies between --.12 and (--.12+--.110) of ordinary air flow. Fire flow
in the present case should vary between 71l/s and 229l/s. Accord-
ing to Table 8.2, the difference between flows and pressures for
boundary cases is insignificant from one fire room to another and
flow values show sufficient agreement with theoretically predicted

123
Spread of Smoke and Fire Gases via the Ventilation System
qb. The deviation is due to the resistance properties of wyes and
the resulting pressure drop in these.
Boundary fire flows, leakage flows and duct flows for resistance
of wyes as a quadratic function, con= 1, are shown in more detail
in Table 8.4.

Table 8.4 Fire flows, leakage flows and duct flows for boundary cases
in rooms of fire origin for the examined open supply-
exhaust ventilation system.

Case Total flow Flow in the Leakage

in the fire duct to the flow out
room, fire room, from the
lis lis fire room,
lis
Boundary case for room 1 139.5 77.1 63.4
Boundary case for room 2 134.4 72.8 61.6
Boundary case for room 3 133.4 70.5 62.9
Boundary case for room 4 128.2 66.1 62.1

Table 8.4 shows, that flows in the duct in this case agree with
those presented in Table 8.2 for the closed supply-exhaust venti-
lation system. Leakage flow can be evaluated with the help of
(6.35).

124
 Section 2. Operation of Ventilation Systems during Fires
Models and Tools
begin
flow 1/s
pressure Pa
control bend=l con=1 duct=1 itmax=20
set leak=t,50,50:q 160=d,160 200=d,200
fan FF 100:0 100:200 0:800
160,6 160, 6 200,6 200,6 FF:hq<
s,200
160,3 160,3 160,3 160, 3
q,O t,50.2,50:q t,63.3,50:q t, 63. 8, 50:q
h?:q

leak

t,38.1,50:q t,55.0,50:q t,69.5,50:q t,95.1,50:q

160,3 160,3 160,3 160, 3

F:hq>

160,6 160,6 200,6 200,6 FF:hq<

s,200
160,3 160,3 160,3 160,3
t,55.6,50:q q, 0 t,63.3,50:q t,63.8,50:q
h?O

leak

t,38.1,50:q t,55.0,50:q t,69.5,50:q t,95.1,50:q

160,3 160,3 160,3 160, 3

F:hq>

160, 6 160, 6 200,6 200,6 FF:hq<

s,200
160,3 160,3 160,3 160, 3
t,55.6,50:q t,50.2,50:q q,O t,63.8,50:q
h?O

leak

t,38.1,50:q t,55.0,50:q t,69.5,50:q t,95.1,50:q

160,3 160,3 160,3 160,3
s,200
F:hq>

160, 6 160, 6 200,6 200,6 FF:hq<

s,200
160,3 160,3 160,3 160, 3
t,55.6,50:q q, 0 t,63.3,50:q q, 0
h?O

leak

t,38.1,50:q t,55.0,50:q t,69.5,50:q t,95.1,50:q

160,3 160,3 160,3 160,3

F:hq>

end
Figure 8.6 PFS-layout of open supply-exhaust ventilation system for
four rooms.

125
Spread of Smoke and Fire Gases via the Ventilation System
begin
flow 1/s
pressure Pa
control bend=1 con=1 duct=1 itmax=20
set 1eak=t,50,50:q 160=d,160 200=d,200
fan FF 100:0 100:200 0:800

t,38.1,50:q t,55.0,50:q t,69.5,50:q t,95.1,50:q

48.608 48.614 48.629 66.099
160,3 160,3 160,3 160,3
s,200
F:hq>
h:99.472
q:211.95
end block problems elements 280

Figure 8.7 PFS-result of open supply-exhaust ventilation system for

four rooms.

126
 Section 2. Operation of Ventilation Systems during Fires
Models and Tools

8.2.4 Open supply-exhaust ventilation with a "zero"

-duct
The layout of the open supply-exhaust ventilation system with a
"zero" duct, with the total air flow of 200 1/s at 100 Pa pressure, is
presented in Figure 8.8. Five cases were studied: a "normal" case
and four boundary cases with fire in each room in turn. The re-
sults are presented in Table 8.5 and Figure 8.9 (for resistance of
wyes as a quadratic function, con=1).

Table 8.5 Flows and pressures for boundary cases in rooms of fire
origin for the examined open supply-exhaust ventilation
system with a "zero" duct.

Case Limit fire flow, Limit fire pres-

1/s sure, Pa
con=1 con=2 con=1 con=2
Normal 0 0 0 0
Boundary case for room 1 177 175 79 77
Boundary case for room 2 176 175 77 77
Boundary case for room 3 179 180 81 81
Boundary case for room 4 179 180 79 81

As Table 8.5 shows, boundary values for flows are somehow

higher than those in the case of ordinary supply-exhaust ventila-
tion system. According to Table 8.5, the difference between flows
and pressures for boundary cases is insignificant from one fire
room to another. The deviation is due to the resistance properties
of wyes and the resulting pressure drop in these!.
This case can be analysed in a more detailed way with the help
of equation 6.27. Ro is a subscript for exhaust air up to the corri-
dor. Figure 8.8 shows that Rt z 60 Pa and Ro z 15 Pa. Hence,
qbk =(Ri / Ro + 1)0.5 q n =50.5 q n'

127
Spread of Smoke and Fire Gases via the Ventilation System
and in this case qbk:=::115 1/s. For leakage paths in the case of fire
yields

=50( 80)0.5
q bf 50

and in this case qbr:=::65 1/s. Total expected boundary fire flow is 180
1/s. This agrees very good with the presented results (Table 8.5).
Note, that in the open supply - exhaust ventilation with a
"zero" -duct smoke spreads via exhaust ducts and collects in the
corridor.

128
 Section 2. Operation of Ventilation Systems during Fires
Models and Tools

begin
flow 1/s
pressure Pa
control bend=1 con=1 duct=1 itmax=99 dxs=20
set 160=d,160 200=d,200 TD=t,15,50:q leak=t,50,50:q
fan FF 100:0 100:200 0:800

160, 6 160,6 200,6 200,6 FF:hq<

160,3 160,3 160, 3 160,3

q, 0 t,50.2,50:q t,63.3,50:q t,63.8,50:q
leak

h?:q

t,70.4,200 FF:hq>
00, 6 s,200

160,6 160, 6 200,6 200,6 FF:hq<

160,3 160,3 160,3 160, 3

t,55.6,50:q q, 0 t,63.3,50:q t,63.8,50:q
leak

h?O

t,70.4,200 FF:hq>
00,6 s,200

160,6 160,6 200,6 200,6 FF:hq<

160,3 160,3 160,3 160,3

t,55.6,50:q t,50.2,50:q q,O t,63.8,50:q
leak

h?O

t,70.4,200 FF:hq>
00,6 s, 200

160, 6 160,6 200,6 200,6 FF:hq<

160,3 160,3 160,3 160, 3

t,55.6,50:q t,50.2,50:q t,63.3,50:q q, 0
leak

h?O

FF:hq>
00,6 s,200

end

Figure 8.8 PFS-layout of open supply-exhaust ventilation system for

four rooms with a "zero-duct".

129
130
 Section 2. Operation of Ventilation Systems during Fires
Models and Tools

8.2.5 Some conclusions on different types of flows

and pressure estimation
The analysis of boundary flows and pressures obtained in the
types of ventilation systems examined above, with two different
ways of defining the resistance of wyes (as a quadratic function,
con=1, and with the help of empirical equations, con=2), shows
that
-all results agree quit well with predicted theoretical values,
- for the exhaust ventilation system, all values obtained for
con=1 are somewhat higher than those obtained for con=2, except
for the case with fire in room 1 where the opposite result is ob-
tained,
- in all cases for the supply-exhaust ventilation systems', the
values obtained for con=1 differ insignificantly and without a
clear trend from those obtained for con=2.
It seems reasonable to use both control operators in calculating
the simulated cases for supply-exhaust ventilation systems of all
types, while the choice of operator for exhaust ventilation systems
is not so simple. As a matter of fact, it is more reliable to use
lower boundary flow values than to use higher ones. Regarded in
that light, con=2 should be chosen as a default operator. One of
the reasons for not using con=2 in evaluating the boundary fire
flow is that all pressure drops in wyes are based on empirical
measurements for normal flow directions. As this study is inter-
ested in "abnormal" flow behaviour, i.e. changes in flow directions
and the consequent reduction in flow in certain parts of junctions
to zero, it is not obvious how wyes would respond. Higher flows
cause higher velocities in ducts, and, as a result, higher Reynolds
numbers. Along with increasing flow turbulence, the pressure
field described as a quadratic function increases.
For these reasons, control variable con= 1 will be used in further
analysis.

131
Spread of Smoke and Fire Gases via the Ventilation System

8.3 Measures to prevent smoke spread via

ventilation ducts
It will have been seen from Subsection 8.2 that fire flows and cor-
responding pressures in fire rooms, caused by smoke spread via
ventilation ducts, can be calculated with the help of the PFS pro-
gram for different ventilation systems. Flows exceeding boundary
flows will always cause smoke spread. The open supply-exhaust
ventilation system with air leakage, presented in Figure 8.5 for
the case of fire in room 1 which produces 200 Vs fire flow, as well
as some measures to prevent smoke spread through the system,
will be presented in this part. The flow of 200 Vs which is larger
than the boundary flow (Table 8.3) was chosen arbitrarily.

8.3.1 Examination of possible measures to prevent

smoke spread
The PFS layout and PFS results for flow directions in the open
supply-exhaust ventilation system serving four rooms, in the case
of fire in room 1 (fire flow 200 Vs), are presented in Figures 8. 7 -
8.12. Seven possible solutions are studied. Table 8.4 presents data
regarding the examined measures.

132
 Section 2. Operation of Ventilation Systems during Fires
Models and Tools

Table 8.4 Data regarding the examined measures to prevent smoke

spread

Case Supply fan Exhaust fan

1 normal normal
2 rotational speed doubled rotational speed doubled
3 closed closed
4 closed normal
5 closed with bypass to the normal
mmaustfan
6 closed with bypass to the rotational speed doubled
exhaust fan
7 converted into exhaust fan normal

Case 1 is an ordinary case. Case 3 corresponds to natural venting.

Cases 2 and 6 use the same fan(s), with the rotational speed dou-
bled which is a very hard measure and can be applied in e.g.
ventilation systems with reduced normal capacity.

8.3.2 Analysis of risk of smoke spread in the

examined cases
Case 1.
As expected, smoke spreads to the adjacent room 2 (34.55 1/s),
where it mixes with supply air and is removed from the room via
the exhaust system. As a general observation, it must be empha-
sised that the whole pressure and flow configuration in both venti-
lation systems changes due to the fire in the room 1 (some pres-
surisation of all rooms, increased air losses through leakage
passes, decreased air flow, produced by the supply fan (141.3 1/s
instead of 200 1/s in normal case), and increased supply air flow to
each room).

133
Spread of Smoke and Fire Gases via the Ventilation System
Case 2.
In case of fire, usual fans can be used in an unusual way. This
case examines fans with the rotational speed doubled (FF,2). The
fans produce increased flows and pressures, and there is no smoke
spread at all via ducts.
Case 3.
This case studies the behaviour of the systems with both fans
shut down (natural ventilation). It shows total spread of smoke to
all rooms through all branches.
Case 4.
This case examines the behaviour of the whole system, with the
supply fan shut down and the exhaust fan in operation. The calcu-
lation shows that smoke spreads to all three rooms via supply
ducts.
Case 5.
This case studies the behaviour of the systems with the supply
fan shut down and the supply plenum connected to the exhaust
plenum and to the exhaust fan. The result is no smoke spread at
all and insignificant underpressure, 40-52 Pa, in three "clear"
rooms.
Case 6.
This case studies the behaviour of the systems with the supply fan
shut down and the supply plenum connected to the exhaust ple-
num, combined with doubling the rotational speed of the exhaust
fan. The result is no smoke spread at all and significant under-
pressure, up to 228 Pa, in three "clear" rooms. The last observa-
tion can be decisive in choosing this measure since the underpres-
sure, converted into door opening force, should never exceed 130
N.
Case 7.
This case examines the theoretical possibility of converting the
supply fan into an exhaust fan. The calculation shows that there
is no spread of smoke via the ventilation system, but underpres-
sure of up to 72.5 Pa can occur in rooms. This case can be com-
bined with increasing n times the rotational speed of one or both
fans, which will result in lower pressure in the rooms.
A very rough estimate of smoke spread can be made with the
help of flow direction and PFS layout, which are shown in Figures
8.9 and 8.12.

134
 Section 2. Operation of Ventilation Systems during Fires
Models and Tools

The studied cases do not cover all possible measures to prevent

smoke spread via ventilation ducts, nor do they examine the be-
haviour of the systems with fire dampers.
The main principles set out in this part will be used in evaluat-
ing "real" ventilation systems.

135
Spread of Smoke and Fire Gases via the Ventilation System
begin
flow 1/s
pressure Pa
control bend=1 con=1 duct=1 itmax=20
set leak=t,SO,SO:q 160=d,160 200=d,200
fan FF 1 0 0 : 0 1 0 0 : 2 0 0 0 : 8 0 0

160,6 160,6 200,6 200,6 FF:hq<

"Case 1"
s,-200
160,3 160,3 16013 160,3
t,55.6,50:q t,50.2,50:q t,63.3,50:q t,63.8,50:q
h?200

leak

t,38.1,50:q t,55.0,50:q t,69.5,50:q t,95.1,50:q

160,3 160, 3 160, 3 160,3
s,200
F:hq>

160,6 160, 6 200,6 200,6 FF, 2 :hq<

"Case 2" s,-200
160,3 160, 3 160,3 160, 3
t,55.6,50:q t,50.2,50:q t,63.3,50:q t,63.8,50:q
h?200

leak

t,38.1,50:q t,55.0,50:q t,69.5,50:q t,95.1,50:q

160,3 160,3 160,3 160,3

F, 2 :hq>

160,6 160,6 200,6 200,6 h?O

"Case 3" s,20
160,3 160,3 160, 3 160, 3
t,55.6,50:q t,50.2,50:q t,63.3,50:q t,63.8,50:q
h?200

leak

t,38.1,50:q t,55.0,50:q t,69.5,50:q t,95.1,50:q

160,3 160, 3 160, 3 160, 3
s,20 h?O
00, 6

160,6 160,6 200,6 200,6 h?O

"Case 4"
160, 3 160, 3 160,3 160,3
t,55.6,50:q t,50.2,50:q t,63.3,50:q t,63.8,50:q
h?200

leak

t,38.1,50:q t,55.0,50:q t,69.5,50:q t,95.1,50:q

160,3 160,3 160, 3 160,3
s,200
F:hq>

Figure 8.7 PFS-layout of open supply-exhaust ventilation system for

four rooms with 200 lIs flow in the fire room and four cases
1-4.

136
 Section 2. Operation ofVentilation Systems during Fires
Models and Tools

Figure 8.8 PFS-result for open supply-exhaust ventilation system for

four rooms with 200 lIs flow in the fire room and cases 1-4.

137
Spread of Smoke and Fire Gases via the Ventilation System
begin
flow 1/s
pressure Pa
control bend=l con=l duct=l itmax=20
set
fan ~~aki8 6 ~g.s~b3: 2 ~go=g;~gg 200=d,200

160,6 160,6 200,6 200,6 FF:hq<

~+++++++++++++v-------------v-------------v-------;~=2CiCi--
"case 1"
-160,3 +160,3 +160,3 +160,3
+ + +
-t,55.6,50:q +t,50.2,50:q +t,63.3,50:q +t,63.8,50:q
h?200
-
r- h?O
+
;-+
h?O
+
h?O;-+
+++++++++++-] 00000000000+] 00000000000+] 00000000000+]
+
;-+
lea!: I leak I
lea!: I
leak
-----------+ -----------+ -----------+ -----------+
I
L+ L+ L+ L+
+ + + +
+t,38.1,50:q +t,55.0,50:q +t,69.5,50:q +t,95.1,50:q
+ + + +
+160,3 +160,3 +160,3 +160,3
+ + + + s,200
lk+++++++++++++£+++++++++++++.!1+++++++++++++.!1+++++++++++++++
160,6 160,6 200,6 200,6 FF:hq>
160,6 160,6 200,6 200,6 FF,2:hq<
"case 2" r------------v-------------y-------------y-------;~=2oo--

+160,3 +160,3 +160,3 +160,3

+ + + +
+t,55.6,50:q +t,50.2,50:q +t,63.3,50:q +t,63.8,50:q
+ + + +
h?200 1 +] h?O
++++++++++++ ;-+]
h?O r+J 00000000000+
00000000000+ 00000000000+ h?O r+J
leak I
-----------+
leak I
-----------+
leak I
-----------+
leak
-----------+
I
L+ L+ L+ L+
+ + + +
+t,38.1,50:q +t,55.0,50:q +t,69.5,50:q +t,95.1,50:q
+ + + +
+160,3 +160,3 +160,3 +160,3
+ + + + s,200
lk+++++++++++++£+++++++++++++.!1+++++++++++++.!1+++++++++++++++
160,6 160,6 200,6 200,6 FF,2:hq>

160,6 160,6 200,6 200,6 h?O

w++ ++++++++++ +if+++++++++++++if+++++++++++++ifO o o o o o o o o o o o o o o
"case 3" - + + + s,20
-160,3 +160,3 +160,3 +160,3
+ + +
-t,55.6,50:q +t,50.2,50:q +t,63.3,50:q +t,63.8,50:q
- + + +
1
h?200 -J 00000000000+
+++++++++++- h?O r+J 00000000000+
h?O r+J 00000000000+
h?O r+J
leak I
-----------+
lea!: I
------------
leak I
------------
leak
------------
I
L+ L- L- L_
+
+t,38.1,50:q -t,55.0,50:q -t,69.5,50:q -t,95.1,50:q
+ - -
+160,3 -160,3 -160,3 -160,3
+ - - - s,20 h?O
lk+++++++++++++£+++++++++++++.!1+++++++++++++&000000000000000
160,6 160,6 200,6 200,6
160,6 160,6 200,6 200,6 h?O
w+++++ ++++++++if+++++++++++++if+++++++++++++ifO o o o o o o o o o o o
"case 4" - + + +
-160,3 +160,3 +160,3 +160,3
+ + +
-t,55.6,50:q +t,50.2,50:q +t,63.3,50:q +t,63.8,50:q
- + + +
1
h?200 -J 00000000000+
+++++++++++- h?O r+J 00000000000+
h?O r+J 00000000000+
h?O r+J
lea!: I
-----------+
lea!: I
++++++++++++
leak I
++++++++++++
leak
++++++++++++
I
L+ L+ L+ L+
+ + + +
+t,38.1,50:q +t,55.0,50:q +t,69.5,50:q +t,95.1,50:q
+ + + +
+160,3 +160,3 +160,3 +160,3
+ + + + s,200
lk+++++++++++++&+++++++++++++&+++++++++++++&+++++++++++++++
160,6 160,6 200,6 200,6 FF:hq>

Figure 8.9 PFS-layout and flow direction for open supply-exhaust ven-
tilation system for four rooms with 200 lIs flow in the fire
room and four cases 1-4. "+" shows flow direction down-
wards and to the right, while"-" shows flow direction up-
wards and to the left. "0" corresponds to no flow.

138
 Section 2. Operation of Ventilation Systems during Fires
Models and Tools

begin
flow 1/s
pressure Pa
control bend=1 con=1 duct=1 itmax=20
set leak=t,SO,SO:q 160=d,160 200=d,200
fan FF 1 0 0 : 0 10 0 : 2 0 0 0 : 8 0 0

160,6 160,6 200,6 200,6

"Case 5" s, -200
160,3 160, 3 160,3 160, 3
t,55.6,50:q t,S0.2,50:q t,63.3,50:q t,63.8,50:q
h?200 200,6

leak

t, 38 .1, 50: q t,SS.O,SO:q t,69.5,50:q t,95.1,50:q

160,3 160,3 160,3 160, 3
s,200 200,1
F:hq>

160, 6 160,6 200,6 200,6

"Case 6" s,-200
160,3 160,3 160,3 160,3
t,55.6,50:q t,50.2,50:q t,63.3,50:q t,63.8,50:q
h?200 200,6

leak

t,38.1,50:q t,SS.O,SO:q t,69.5,50:q t,95.1,50:q

160,3 160,3 160,3 160,3
s,200 200,1
F,2:hq>

160, 6 160,6 200,6 200,6 FF:hq>

"Case 7" s,200
160,3 160, 3 160,3 160, 3
t,55.6,50:q t,50.2,50:q t,63.3,50:q t,63.8,50:q
h?200

leak

t,38.1,50:q t,55.0,50:q t,69.5,50:q t,95.1,50:q

160,3 160,3 160,3 160, 3

F:hq>

end

Figure 8.10 PFS-layout of open supply-exhaust ventilation system for

four rooms with 200 lIs flow in the fire room and cases 5, 6
and 7.

139
Spread of Smoke and Fire Gases via the Ventilation System
begin
flow 1/s
pressure Pa
control bend=1 con=1 duct=1 itmax=20
set leak=t,50,50:q 160=d,160 200=d,200
fan FF 100:0 100:200 0:800

end block problems elements 401 warnings 8

Figure 8.11 PFS-result for open supply-exhaust ventilation system for

four rooms with 200 lIs flow in the fire room and cases 5, 6
and7.

140
 Section 2. Operation of Ventilation Systems during Fires
Models and Tools
begin
flow 1/s
pressure Pa
control bend=1 con=1 duct=1 itmax=20
set leak=t,50,50:q 160=d,160 200=d,200
fan FF 100:0 100:200 0:800

160,6 160,6 200,6 200,6

rr+++++++++++++v+++++++++++++v+++++++++++++v+++++++++++++il
"case 5" - - - - s,-200+
-160,3 -160,3 -160,3 -160,3 +
+
-t,55.6,50:q -t,50.2,50:q -t,63.3,50:q -t,63.8,50:q +
+
1 -J h?O +200,6
h?200
+++++++++++- r - J 00000000000-
00000000000- h?O r - J 00000000000-
h?O r-J +
leak I
-----------+
leak
++++++++++++
I leak
++++++++++++
I leak
++++++++++++
I +
+
+

L+ L+ L+ L+ +
+ + + + +
+t,38.1,50:q +t,55.0,50:q +t,69.5,50:q +t,95.1,50:q +
+ + + + +
+160,3 +160,3 +160,3 +160,3 +
+ + + + s,200 +200,1
lb+++++++++++++£+++++++++++++£+++++++++++++£+++++++++++++£+++
160,6 160,6 200,6 200,6 FF:hq>

160,6 160,6 200,6 200,6

rr+++++++++++++v+++++++++++++v+++++++++++++v+++++++++++++il
"case 6 - - - s,-200+ -
-160,3 -160,3 -160,3 + -160,3
+
-t,55.6,50:q -t,50.2,50:q -t,63.3,50:q -t,63.8,50:q +
+
+200,6
+~l;~~+++++=J ogzgooooooo=J og68ooooooo=J og68ooooo60=] +
leak I leak I
leak leak
++++++++++++ ++++++++++++ ++++++++++++ ++++++++++++
I +
+
+
I
L+ L+ 4 4
+ + + +
+t,38.1,50:q +t,55.0,50:q +t,69.5,50:q +t,95.1,50:q
+ + + +
+160,3 +160,3 +160,3 +160,3 +
+ + + + s,200 +200,1
lb+++++++++++++£+++++++++++++L4++++++++++++£+++++++++++++£+++
160,6 160,6 200,6 200,6 FF, 2 :hq>

160,6 160,6 200,6 200,6 FF:hq>

rr+++++++++++++v+++++++++++++v+++++++++++++v+++++++++++++++
"case 7" - - - - s,200
-160,3 -160,3 -160,3 -160,3
-t,55.6,50:q -t,50.2,50:q -t,63.3,50:q -t,63.8,50:q
- - - -
h?200 1 -J h?O r - J 00000000000-
h?O r - J 00000000000-
h?O r-J
+++++++++++- 00000000000-
leak I
-----------+
leak
++++++++++++
I leak
++++++++++++
I leak
++++++++++++
I
L+ L+ L+ L+
+ + + +
+t,38.1,50:q +t,55.0,50:q +t,69.5,50:q +t,95.1,50:q
+ + + +
+160,3 +160,3 +160,3 +160,3
+ + + + s,200
lb+++++++++++++£+++++++++++++£+++++++++++++£+++++++++++++++
160,6 160,6 200,6 200,6 FF:hq>

end

Figure 8.12 PFS-layout and flow direction for open supply-exhaust ven-
tilation system for four rooms with 200 lIs flow in the fire
room and four cases 5, 6 and 7. "+" shows flow direction
downwards and to the right, while"-" shows flow direction
upwards and to the left. "0" corresponds to no flow.

141
Spread of Smoke and Fire Gases via the Ventilation System

142
 Section 2. Operation of Ventilation Systems during Fires
"Real Case" Study

9 "Real case" study

The study presented in Chapter 8 refers to a simple ventilation

system serving four rooms. Such systems are not usual in reality.
The types of buildings examined in this study are usually
equipped with much more advanced systems, the degree of com-
plexity of which can be different. In this chapter, smoke spread
will be evaluated in the most typical ventilation systems installed
in the examined types of buildings.

9.1 Description of the main types of

ventilation systems
In spite of the fact that a great variety of different ventilation sys-
tems are used in the studied types of buildings, the most widely
used systems can be described.
The mechanical ventilation systems usual in dwellings (blocks
of flats) can be of both exhaust and supply-exhaust type.
The duct network of the exhaust ventilation system can have one
vertical plenum, exhausting air via branches and terminals from
each flat, or two vertical plenums, exhausting air from different
rooms in the same flat separately (e.g. one from the kitchen, and
one from toilet and bathroom). An exhaust fan in this case ex-
hausts air at a rate which sets up a slight negative pressure in the
ventilated space to ensure that the required quantity of outside
air is introduced. The outside air enters the space through outdoor
louvers and terminals.

143
Spread of Smoke and Fire Gases via the Ventilation System
Exhaust branches from each space can also pass vertically
through the building and connect to the plenum just before the
fan.
These types of exhaust ventilation systems can be combined with
supply ventilation system with a similar duct layout.
The design of ventilation systems in hospitals is much more
complicated than in other types of buildings because of the variety
of services they provide. Mechanical ventilation systems in hospi-
tals are usually of the supply-exhaust type. Each ward is sup-
plied with air through one or several terminals, and air is ex-
hausted via terminals in lavatories and wash rooms. In some
wards, operating theatres, delivery rooms, labour rooms and
nurseries, both supply and exhaust systems can deliver and ex-
haust air directly from the room served.
Accommodation in hotels and motels is usually in the form of
single rooms with toilet and bathroom adjacent to a corridor,
flanked on both sides by other guest rooms. Mechanical ventila-
tion systems in hotels are usually of the supply-exhaust type.
Each hotel room or each suite is supplied with air through one or
more terminals, and air is exhausted via terminals in lavatories
and bathrooms.
The variety of functions and range of design criteria applicable
to office buildings have allowed the use of almost every available
ventilation system. While multi-storey structures are discussed
here, the principles are similar for all sizes of office buildings. Me-
chanical ventilation systems in offices are usually of the supply-
exhaust type described before. Another type of ventilation system
which can also be used in offices is the "zero duct" ventilation sys-
tem in which the corridor, the "zero duct", collects vitiated air
through transfer air devices from all rooms in the office and trans-
fers it to the exhaust plenum in the ceiling through terminals.
Some of the above systems will be analysed further in Chapter
10 in connection with the danger of CO poisoning when smoke
spreads to adjacent spaces.

144
 Section 2. Operation of Ventilation Systems during Fires
"Real Case" Study

9.2 Designed air flows

The quantities of ventilation air in most types of buildings are
specified in Building Codes. [8] requires minimum outdoor air of
0.35 l/s,m2 in all types of rooms. In rooms for sleeping the mini-
mum outdoor air required is 4 l/s and person. In a room or part of
a room with more then casual occupation the minimum outdoor
air recommended is 7 l/s and person. There are requirements for
minimum exhaust air flow also:

10 l/s air flow from a kitchen, high sped extraction rate min
75% of extraction capacity
15 l/s air flow from a pantry
10 l/s air flow from a bathroom with openable window (if
area exceeds 5m2 , required air flow increases by
1 l/s, m 2 )
15 l/s air flow from a bathroom without window (if area
exceeds 5m2 , required air flow increases by
1 l/s,m2 )
10 l/s air flow from a toilet

According to these air flow requirements, the ordinary air flow in

an average flat and hotel room can be assumed to be 30 l/s (15 l/s
for kitchenette or kitchen cubicle and 15 l/s for bathroom without
openable window).
Rooms and occupancy in office buildings can vary significantly.
Occupancy in e.g. accounts or other sections where clerical work is
done can be approximately one person per 15m2 of floor area. In
waiting rooms and conference rooms occupancy can be as high as
one person per 1.5 m 2 • According to air flow recommendations (7
l/s and person) and assuming average occupancy of 6-7 persons,
air flow in offices is 50 l/s.
In Section 1, flows obtained in all the examined types of build-
ings are shown in Figure 4.13. Values from Figure 4.13 are used
for a simplified estimation of fire flows for the volumes set out in
Table 9.1.

145
Spread of Smoke and Fire Gases via the Ventilation System

Table 9.1 Fire flows obtained in different types of buildings of certain

volumes.

Room volume, Fire flows in different types of building, m 3/s

Dwelling Hospital Office Hotel
ma houses
50 0.48 0.3 0.30 0.55
100 1.5 0.5 0.45 0.85
150 2.0 0.7 0.60 1.20*
200 2.5 0.9 0.65 1.50*
* - Flows according to curve for hotels.

9.3 Exhaust ventilation system

The analysis of exhaust ventilation systems described in Subsec-
tion 8.2.1 comprises only schematic models. In this section more
complicated ventilation systems will be analysed.

9.3.1 Exhaust ventilation system with single outlet in

each room
The layout of the examined exhaust ventilation systems is pre-
sented in Figure 9 .1.

146
 Section 2. Operation of Ventilation Systems during Fires
"Real Case" Study

plenum 1

"" plenum2

Figure 9.1 Layout of exhaust ventilation system for four floors and ten
rooms on each floor.

The layout of the exhaust ventilation system with the total air
flow of 1200 1/s at 107 Pa pressure, which exhausts via two verti-
cal plenums from four floors, and further via branches from ten
rooms on each floor, is presented in Figure 9.2. The exhaust fan
has the following characteristics:

Pressure, Pa 100 125 75

Air flow, lis 200 750 1500

Results for a "normal" case are presented in Figure 9.3.

147
Spread of Smoke and Fire Gases via the Ventilation System
begin
flow 1/s
pressure Pa
control bend=1 con=1 duct=1 itmax=99 table=3
set UD=t,10,30
fan FF 100:200 125:750 75:1500

d,160,4 d,160,4 d,200,4 d,200,4 d,250,4

s,150
d,160,3 d,160,3 d,160,3 d,160,3 d,160,3
t,42.1,30 t,47.8,30 t,51.9,30 t,60.2,30 t,62.6,30

UD:hq UD:hq UD:hq UD:hq UD:hq

d,160,4 d,160,4 d,200,4 d,200,4 d,250,4

s,150

s,150

s,150

s,150

s,150

Figure 9.2 PFS-layout of exhaust ventilation system according to figure

9.1 (beginning).

148
 Section 2. Operation of Ventilation Systems during Fires
"Real Case" Study
d,160,4 '160, 4 d,200,4 d, 200,4 d, 250,4
s,150
d,160,3 d,160,3 d,160,3 d,160,3 d,160,3
t,50.7,30 t,56.5,30 t,60.5,30 t,68.9,30 t,71.3,30

UD:hq UD:hq UD:hq UD:hq UD:hq

d,160,4 d,160,4 d,200,4 d,200,4 d,250,4

s, 150

d,400,4:hq d,400,4:hq

FF:hq<
s,-1200 ,630,8 d,400,8

end

Figure 9.2 PFS-layout of exhaust ventilation system according to figure

9.1 (end).

149
Spread of Smoke and Fire Gases via the Ventilation System
begin
flow 1/s
pressure Pa
control bend=l con=l duct=l itmax=99 table=3
set UD=t,10,30
fan FF 100:200 125:750 75:1500

,160,4 d,160,4 d,200,4 d,200,4 d,250,4

d,160,3 d,160,3 d,160,3 d,160,3

t,42.1,30 t,47.8,30 t,51.9,30 t,60.2,30

UD:hq UD:hq UD:hq UD:hq

h: 9. 994 h:10.006 h: 9.993 h:10.003
q:-29.99 q:-30.01 q:-29.99 q:-30.00
d,160,4 d,160,4 d,200,4 d,200,4
s,150

UD:hq
h:10.003
q:-30.00 q:-30.00
d,200,4 d,200,4 d,250,4
s,150

UD:hq
h: 9.997
q:-30.00 q:-30.00
d,200,4 d,200,4 d,250,4
s,150

UD:hq
h: 9. 998
q:-30.00 q:-30.00
d,200,4 d,200,4 d,250,4
s,150

s,150

Figure 9.3 PFS-result of exhaust ventilation system according to figure

9.1 (beginning).

150
 Section 2. Operation of Ventilation Systems during Fires
"Real Case" Study
d,160,4 d,160,4 ,200,4 d,200,4 d, 250,4
s,150
d,160,3 d,160,3 d,160,3 d,160,3 d,160,3
t,50.7,30 t,56.5,30 t,60.5,30 t,68.9,30 t, 71. 3, 30

UD:hq UD:hq UD:hq UD:hq UD:hq

h:10.002 h: 9.997 h:10.000 h: 9. 996 h: 9.995
q:-30.00 q:-30.00 q:-30.00 q:-29.99 q:-29.99
d,160,4 d,160,4 d,200,4 d,200,4 d, 250,4
s, 150

d, 400,4 :hq d,400,4:hq

h:-2.531 h:-2.531
q: 600.00 q:600.00

FF:hq<
h:- 06.600 d,630,8 ,400,8
q; -1200

end block 1 problems elements 316

Figure 9.3 PFS-result of exhaust ventilation system according to figure

9.1 (end).

151
Spread of Smoke and Fire Gases via the Ventilation System
Different cases were studied in order to figure out boundary fire
air flows and corresponding pressures in different rooms. This
problem is very numerically complex for such ventilation systems
with several branches, because the flow in the boundary room
must be equal to zero, while other flows and pressures are much
higher then normal. There is an impossible case in exhaust venti-
lation systems, where zero air flow can never be achieved owing to
the entraiment effect in the side branch of a converging wye. This
problem was studied in [27].
The above presented statements agree sufficiently well with the
results from test simulations. In some simulations the program
did not manage to get any result at all and informed about differ-
ent types of faults. Because of this it seems more reasonable to
test the behaviour of the ventilation system regarding smoke
spread to the adjacent rooms by using flows predicted in Section 1
and Table 9.1.
According to Building Code requirements for minimum outdoor
air which are 0.35 l/s,m2 , and taking into account the assumed 30
1/s air flow in and out of each room along with usual height of
dwellings in Sweden, 2.4 m, the volume of a ventilated room is
approximately 200 m 3 • According to Table 9.1, the flow produced
by fire in e.g. hotel rooms of this volume is 1500 1/s.
"Fires" were placed in different rooms and the behaviour of
ventilation systems was studied with the PFS-program. According
to the results obtained, a flow of 1500 1/s from the majority of
burning rooms will definitely cause spread of smoke to all adja-
cent rooms on the same floor which are served by the same branch
as the burning room. One of the PFS-results with fire flow of 1500
1/s in the first room in branch 5 is set out in Figures 9.4.

152
Section 2. Operation of Ventilation Systems during Fires
"Real Case" Study

153
Spread of Smoke and Fire Gases via the Ventilation System
begin
flow 1/s
pressure Pa
control bend=1 con=1 duct=1 i tmax=99 table=3 dxs=25
set UD=t,10,30
fan FF 100:200 125:750 75:1500

d,160,4 d,160,4 d,200,4 d,200,4 d,250,4

s,150
d,160,3 d,160,3 d,160,3 d,160,3 d,160,3
t,42.1,30 t,47.8,30 t,51.9,30 t,60.2,30 t,62.6,30

UD:hq UD:hq UD:hq UD:hq UD:hq

h: 8.979 h: 8.995 h: 8.988 h: 8.998 h: 9. 002
q:-28.43 q:-28.45 q:-28.44 q:-28.46 q:-28.46
d,160,4 d,160,4 d,200,4 d,200,4 d,250,4
s,150

d,250,4
s,150

d,250,4
s,150

d,250,4
s,350

UD:hq
h:-12889
q:19.143 q:1077.0
d,160,4 d,160,4 d,200,4 d,200,4 d,250,4
s,150

UD:hq
h: 8.233
q:-27.22 q:-27.22 q:-27.23 q:-27.23 q:-27.24
Figure 9.4 PFS-result of exhaust ventilation system according to figure
9.1 with a fiow of 1500 lIs in the first room in branch 5
(beginning).

154
 Section 2. Operation of Ventilation Systems during Fires
"Real Case" Study
d,l60,4 d,l60,4 d,200,4

d,l60,3 d,l60,3 d,l60,3

t,50.7,30 t,56.5,30 t,60.5,30

UD:hq UD:hq UD:hq UD:hq

h: 8.130 h: 8. 129 h: 8.137 h: 8.137
q:-27.05 q:-27.05 q:-27.06 q:-27.06
d,l60,4 d,l60,4 d,200,4 d,200,4 d, 250,4
s,l50

d,400,4:hq d,400,4:hq
h:-2.297 h:-3.845
q:569.11 q: 753.34

, 630, 8 d,400,8

end block 1 problems 1 elements 317 warnings 11

Figure 9.4 PFS-result of exhaust ventilation system according to figure

9.1 with a flow of 1500 l/ sin the first room in branch 5
(end).

155
Spread of Smoke and Fire Gases via the Ventilation System
begin
flow 1/s
pressure Pa
control bend=1 con=1 duct=1 itmax=99 table=3 dxs=25
set UD=t,10,30
fan FF 100:200 125:750 75:1500

d,160,4 d,160,4 d,200,4 d,200,4 d,250,4

!!'+++++++++++!+++++++++++!+++++++++++!+++++++++++!+++++++++~~t~~++t+

-d,160,3 -d,160,3 -d,160,3 -d,160,3 -d,160,3

-t,42.1,30 -t,47.8,30 -t,51.9,30 -t,60.2,30 -t,62.6,30

[~] [~] [~] [~] [~]

-UD:hq -UD:hq -UD:hq -UD:hq -UD:hq

d,160,4 d,160,4 d,200,4 d,200,4 d,250,4

rr+++++++++++;r+++++++++++;r+++++++++++;r+++++++++++;r++++++++++++++++++++
- - - - - s,150 2
-d,160,3 -d,160,3 -d,160,3 -d,160,3 -d,160,3
-t,46.8,30 -t,52.5,30 -t,56.6,30 -t,64.9,30 -t,67.3,30

[~] [~] [~] [~] [~]

-UD:hq -UD:hq -UD:hq -UD:hq -UD:hq

d,160,4 d,160,4 d,200,4 d,200,4 d,250,4

rr+++++++++++;r+++++++++++;r+++++++++++;r+++++++++++;r++++++++++++++++++++
- - - - - s,150 3
-d,160,3 -d,160,3 -d,160,3 -d,160,3 -d,160,3
-t,50.1,30 -t,55.9,30 -t,59.9,30 -t,68.3,30 -t,70.7,30

[~] [~] [~] [~] [~]

-UD:hq -UD:hq -UD:hq -UD:hq -UD:hq

d,160,4 d,160,4 d,200,4 d,200,4 d,250,4

rr+++++++++++;r+++++++++++;r+++++++++++;r+++++++++++;r++++++++++++++++++++
- - - - - s,150 4
-d,160,3 -d,160,3 -d,160,3 -d,160,3 -d,160,3
-t,60.2,30 -t,66.0,30 -t,70.0,30 -t,78.4,30 -t,80.8,30

-UD:hq -UD:hq -UD:hq -UD:hq -UD:hq

d,160,4 d,160,4 d,200,4 d,200,4 d,250,4

r----------;_r----------;_r----------lt-----------!+++++++++~~;~t++++~+

+d,160,3 +d,160,3 +d,160,3 +d,160,3 -d,160,3

+ + + +
+t,42.6,30 +t,48.4,30 +t,52.4,30 +t,60.8,30 -t,63.2,30
+ + + +

c:J +
CJ+
CJ+
c:J +~:!~~~E:J
+ +
+UD:hq +UD:hq +UD:hq +UD:hq +UD:hq
+ + + + +

d,160,4 d,160,4 d,200,4 d,200,4 d,250,4

rr+++++++++++;r+++++++++++;r+++++++++++;r+++++++++++;r++++++++++++++++++++
- - - - - s,150 6
-d,160,3 -d,160,3 -d,160,3 -d,160,3 -d,160,3
-t,47.3,30 -t,53.1,30 -t,57.1,30 -t,65.5,30 -t,67.9,30

[~] [~] [~] [~] [~]

-UD:hq -UD:hq -UD:hq -UD:hq -UD:hq

Figure 9.5 PFS-flow direction for exhaust ventilation system according

to figure 9.1 with a flow of 1500 lIs in the first room in
branch 5 (beginning).

156
 Section 2. Operation of Ventilation Systerns during Fires
"Real Case" Study
d,160,4 d,160,4 d,200,4 d,200,4 d,250,4
~r+++++++++++if+++++++++++if+++++++++++if+++++++++++if++++++++++++++++++++
- - - - - s,150 7
-d,160,3 -d,160,3 -d,160,3 -d,160,3 -d,160,3
-t,50.7,30 -t,56.5,30 -t,60.5,30 -t,68.9,30 -t,71.3,30

[~] [~] [~] [~]

-UD:hq -UD:hq -UD:hq -UD:hq -UD:hq

d,160,4 d,160,4 d,200,4 d,200,4 d,250,4

1f+++++++++++if+++++++++++if+++++++++++if+++++++++++if++++++++++++++++++++
- - - - - s,150 8
-d,160,3 -d,l60,3 -d,160,3 -d,160,3 -d,160,3
-t,60.8,30 -t,66.6,30 -t,70.6,30 -t,79.0,30 -t,81.4,30

[~] [~] [~] [~] [~]

-UD:hq -UD:hq -UD:hq -UD:hq -UD:hq

d,250,1 d,250,1
r-----1- r-----:s-
+d,250,4:hq +d,250,4:hq
+ +
+ d,250,1 + d,250,1
!------2- !------6-
+d,315,4:hq +d,315,4:hq
+ +
+ d,250,1 + d,250,1
!------3- !------7-
+d,400,4:hq +d,400,4:hq
+ +
+ +
+ d,250,1 + d,250,1
!------:;- ~------8-
+d,400,4:hq +d,400,4:hq
+ +
+ +
+ +
+ +
_::~~~~---------1--------------------~
s,-1300 d,630,8 d,400,8

end

Figure 9.5 PFS-flow direction for exhaust ventilation system according

to figure 9.1 with a flow of 1500 lIs in the first room in
branch 5 (end).

157
Spread of Smoke and Fire Gases via the Ventilation System
Figures 9.4 and 9.5 show that in the studied case all flows in all
adjacent rooms are positive (supply). It means that a flow of 1500
lis exceeds the boundary flow and causes spread of smoke to all
adjacent rooms on the same floor.
The layout of the ventilation system plays an important role in
this kind of assessment, since the resistance properties of wyes
and bends affect the resulting pressure drop and thus the flow
distribution in these (e.g. Table 8.1 -Table 8.3). Normal ventila-
tion flow of 30 lis can ventilate even smaller spaces. The assumed
fire flow of 1500 lis seems high regarding smoke spread through
the ventilation system. In this case it can be interesting to assess
flow in the burning room that can be as close as possible to
boundary flow. According to Table 8.1 the lowest expected bound-
ary flow in a four-room duct system is the flow from the last but
one room (room 2). Calculations show that a flow of 550 lis from
each last but one room in the present system layout, connected to
each floor branch (rooms No 2, Figure 9.1), will cause smoke to
spread to the adjacent room 1 on the same floor which is served by
the same branch. At the same time there are some differences in
"boundary" flows between different branches, with the lowest flow
of 450 lis in branch 5 situated furthest away from the fan. This
calculation is presented in Figure 9.6. Some other fire cases were
studied for the present ventilation system. Flows of 1100 lis in the
nearest room of each branch cause smoke to spread to all other
four rooms on the same floor or a significant reduction, down to
zero, in flows in other rooms, which is characteristic of flows close
to boundary values. One of the calculations is presented in Figure
9.7.

158
Section 2. Operation of Ventilation Systems during Fires
"Real Case" Study

159
Spread of Smoke and Fire Gases via the Ventilation System
begin
flow 1/s
pressure Pa
control bend=1 con=l duct=1 itmax=99 table=3 dxs=25
set UD=t,10,30
fan FF 100:200 125:750 75:1500
d,160,4 d,160,4 d,200,4 d,200,4 d,250,4
s, 150
d,160,3 d,160,3 d,160,3 d,160,3 d,160,3
t,42.1,30 t,47.8,30 t,51.9,30 t,60.2,30 t,62.6,30

UD:hq UD:hq UD:hq UD:hq UD:hq

h: 9.648 h: 9. 661 h: 9. 650 h: 9.660 h: 9. 661
q:-29.47 q:-29.49 q:-29.47 q:-29.49 q:-29.49
d,160,4 d,160,4 d,200,4 d,200,4 d, 250,4
s, 150

UD:hq
h: 9. 662
q:-29.49 q:-29.49
d,200,4 d,200,4 d,250,4
s,l50

d,160,4 d,160,4 d,200,4 d,200,4 d,250,4

s, 150

d,250,4
s, 350

d,250,4
s,150

Figure 9.6 PFS-result of exhaust ventilation system according to figure

9.1 with a flow of 450 lIs in the last but one room in branch
5 (beginning).

160
 Section 2. Operation of Ventilation Systems during Fires
"Real Case" Study
d,160,4 d,160,4 d, 200, 4 d,200,4 d,250,4
s, 150
d,160,3 d,160,3 d,160,3 d, 160,3 d, 160,3
t,50.7,30 t,56.5,30 t,60.5,30 t,68.9,30 t,71.3,30

UD:hq UD:hq UD:hq UD:hq UD:hq

h: 9. 344 h: 9. 341 h: 9.346 h: 9.343 h: 9.345
q:-29.00 q:-28.99 q:-29.00 q:-29.00 q:-29.00
d,160,4 d,160,4 d,200,4 d,200,4 d,250,4
s, 150

d,400,4:hq d,400,4:hq
h:-2.451 h:-2.976
q:589.64 q: 655.25

,400,8

end block 1 problems 1 elements 317 warnings 11

Figure 9.6 PFS-result of exhaust ventilation system according to figure

9.1 with a flow of 450 lIs in the last but one room in branch
5 (end).

161
Spread of Smoke and Fire Gases via the Ventilation System
begin
flow 1/s
pressure Pa
control bend=1 con=1 duct=1 itmax=99 table=3 dxs=25
set UD=t,10,30
fan FF 100:200 125:750 75:1500

d,160,4 d,160,4 d,200,4 d,200,4 d,250,4

s, 150
d,160,3 d,160,3 d,160,3 d,160,3 d,160,3
t,42.1,30 t,47.8,30 t,51.9,30 t,60.2,30 t,62.6,30

UD:hq UD:hq UD:hq UD:hq UD:hq

h: 9.330 h: 9. 343 h: 9.334 h: 9.344 h: 9. 346
q:-28.98 q:-29.00 q:-28.98 q:-29.00 q:-29.00
d,160,4 d,160,4 d,200,4 d, 200, 4 d,250,4
s, 150

UD:hq
h: 9. 348
q:-29.00 q:-29.01
d,160,4 d,160,4 d,200,4 d,200,4 d,250,4
s, 150

s,150

d,250,4
s,350

g~:~~924
q:789.39
d,160,4 d,160,4 d,200,4 d,200,4 d,250,4
s,150

UD:hq
h: 8. 786
q:-28.12 q:-28.12 q:-28.13 q:-28.12 q:-28.13
Figure 9. 7 PFS-result of exhaust ventilation system according to figure
9.1 with a flow of 1100 lIs in the nearest room in branch 5
(beginning).

162
 Section 2. Operation of Ventilation Systems during Fires
"Real Case" Study
d,160,4 d,160,4 d,200,4 d,200,4 d,250,4
s,150
d,160,3 d,160,3 d,160,3 d,160,3 d,160,3
t,50.7,30 t,56.5,30 t,60.5,30 t,68.9,30 t,71.3,30

UD:hq UD:hq W:~ W:~ W:~

h: 8.756 h: 8.753 h: 8.760 h: 8.759 h: 8.763
q:-28.07 q:-28.07 q:-28.08 q:-28.08 q:-28.08
d,160,4 d,160,4 d,200,4 d,200,4 d,250,4
s,150

d,400,4:hq d,400,4:hq
h:-2.378 h:-3.389
q:579.92 q:703.30

,630,8 d,400,8

end block 1 problems 1 elements 317 warnings 11

Figure 9. 7 PFS-result of exhaust ventilation system according to figure

9.1 with a flow of 1100 lIs in the nearest room in branch 5
(end).

Figure 9.6 shows, that flow in one adjacent room has a positive
value, i.e. air flows out of the room. The pressure drop in the other
parts of the ventilation system and flows in the other branches
and terminals are also affected by the fire flow in the tested room.
Figure 9.7 shows that flows in all four rooms adjacent to the
burning room have small positive values, i.e. air flows out of the
room. In this case fire flow is very close to the boundary flow. The
pressure drops in the other parts of the ventilation system and
flows in the other branches and terminals are also affected by the
fire flow in the tested room.
Calculations show that there are some differences in fire flow,
causing smoke to spread to the adjacent room(s), in different

163
Spread of Smoke and Fire Gases via the Ventilation System
branches (serving different floors) in similar rooms. Flows causing
smoke spread via the same branch can differ significant from case
to case. Note, that fire flows from different rooms can cause smoke
to spread to different numbers of rooms served by the same
branch.
Some measures to prevent smoke spread were discussed in
Chapter 8. Two of these were tested for all the above fire flows,
with the rotational speed of the exhaust fan doubled and with the
fan switched off. The results are presented in Table 9.1.

Table 9.1 Data regarding the examined measures for prevention of

smoke spread via exhaust ventilation system

Exhaust fan Smoke spread

To room(s), To other rooms
same branch
Normal Yes No
Doubled rotational speed No No
Switched off Yes Yes (to the whole
building)

The case when the rotational speed of the fan is doubled is a very
hard measure and can be applied e.g. in ventilation systems with
reduced normal capacity. The case when the fan is switched off
corresponds to natural ventilation.

164
 Section 2. Operation of Ventilation Systems during Fires
"Real Case" Study

9.3.2 Exhaust ventilation system with two terminals

in each room
In certain types of buildings another type of exhaust ventilation
system can be used. Such systems exhaust air from each of the
compartments via two separate terminals, connected to two sepa-
rate plenums. For example, one terminal is placed in the bath-
room and the other one in the kitchen. The layout of the examined
exhaust ventilation system is presented in Figure 9.8.

to plenum 1

"" ""
"" ""
"" ""
"" ~~--~--~--~~

to plenum2

Figure 9.8 Layout of exhaust ventilation system with two plenums for
four floors and five rooms on each floor.

Two different cases were studied. In the first case air is ex-
hausted at the rate of 15 lis from each of two terminals in the
room. In the second case air is exhausted at different rates, 10 lis
and 20 lis, from the terminals. The total exhaust air flow from the
room is 30 lis for both cases. A fan with the same characteristics
is installed in all studied cases.
The exhaust fan has the following similar characteristics as the
fan in Subsection 9.3.1.
The exhaust ventilation system drives out the total air flow of
600 lis at 123 Pa pressure, which exhausts via two vertical ple-

165
Spread of Smoke and Fire Gases via the Ventilation System
nums from four floors, and further via branches from five rooms
on each floor.
Different "fire" cases were studied in order to test the behaviour
of ventilation system regarding the risk of smoke spread to the
adjacent rooms. "Fires" were placed in different rooms and the
behaviour of the ventilation systems was studied for two major
cases with fires in the last but one room and the nearest room
with the help of PFS-program. Flows predicted in Section 1 and
Table 9.1 were used for the purpose of this study.
For the first ventilation system with equal air flows
(symmetrical case), air flows of 1100 1/s from the last but one room
and 2350 1/s from the nearest room cause smoke to spread to adja-
cent room (s).
For the second ventilation system with unequal air flows
(asymmetrical case), air flows of 700 1/s from the last but one room
and 1450 1/s from the nearest room cause smoke to spread to adja-
cent room (s).
This example shows that even small changes in the layout of
the ventilation system can affect the behaviour of the whole sys-
tem regarding the risk of smoke spread.

Some measures to prevent smoke spread were discussed in

Chapter 8. Two of these, the case with exhaust fan with rotational
speed doubled, and the case with the fan switched off, were tested
for all the above fire flows. The results are presented in Table 9.2.

Table 9.2 Data for the examined measures to prevent smoke spread
via double-plenum exhaust ventilation system

Exhaust fan Smoke spread

To room(s), To other rooms
same branch
Normal Yes No
Doubled rotational speed No No
Shut down Yes Yes (to the whole
building)

166
 Section 2. Operation of Ventilation Systems during Fires
"Real Case" Study
The case with a fan whose rotational speed is doubled is a very
hard measure and can be applied. in e.g. ventilation systems with
reduced normal capacity. The case with the fan shut down corre-
sponds to natural ventilation.

9.4 Supply-exhaust ventilation system with a

"zero-duct"
Such types of ventilation systems is often used in different office
buildings, as described in Subsection 9.1. In these ventilation sys-
tems corridors acts as collectors or "zero-ducts". Exhaust air from
all rooms in the office is removed through transfer air devices,
collects in corridors, and is in turn transported to the exhaust
plenum through corridor terminal. The layout of the examined
ventilation system is presented in Figure 9.9.

to supply plenum

to exhaust plenum

Figure 9.9 Layout of supply-exhaust ventilation system with "zero-

duct" for four floors and four rooms on each floor.

Both supply and exhaust fans have similar characteristics as the

fan in Subsection 9.3.1.
Different cases were studied in order to figure out boundary fire
air flows and corresponding pressures in different rooms. "Fires"

167
Spread of Smoke and Fire Gases via the Ventilation System
were placed in different rooms and the behaviour of the ventila-
tion system was studied with the PFS-program. Though such
problems are usually very numerically complex for the majority of
ventilation systems with several branches, calculations succeeded
in this case. Boundary fire flow for the majority of "fire" cases was
140 lis. Smoke spreads to the adjacent rooms via supply ducts,
while higher values of gas flows spread directly to the corridor
("zero-duct") through transfer air devices.

Some measures for the prevention of smoke spread were dis-

cussed in Chapter 8. The results of smoke spread evaluations for
fire flow of 450 lis are presented in Table 9.3.

Table 9.3 Data regarding the examined measures for the prevention
of smoke spread via supply-exhaust ventilation system with
"zero-duct"

Supply Exhaust Smoke spread

fan fan To room(s), To other
same branch rooms
Normal Normal Yes No
Doubled rota- Doubled rota- No No
tional speed tional speed
Closed with by- Normal No No
pass to the ex-
haust fan
Closed Normal Yes Yes
Closed with by- Closed with Yes Yes
pass to the outside bypass to the
outside
Converts into ex- Normal Yes No
haust fan

The case when the rotational speed of the fan is doubled is a very
hard measure and can be applied e.g. in ventilation systems with
reduced normal capacity. The case when the fan is switched off
corresponds to natural ventilation.

168
 Section 2. Operation of Ventilation Systems during Fires
"Real Case" Study
The studied cases do not cover all possible measures to prevent
smoke spread via ventilation ducts, nor do they examine system
behaviour with fire dampers.

In all the above cases, part of the smoke produced by a fire

spreads directly to the corridor through transfer air devices,
where it mixes with ambient air. Evaluation of the degree of dilu-
tion can be of great importance in risk assessment. This problem
will be discussed in Section 10.

9.5 Supply-exhaust ventilation system

These types of ventilation systems are the most common and are
used in all types of examined buildings, as described in Subsection
9.1. Each fire compartment is supplied with air through one or
more terminals, and air is exhausted via terminals in the same
space. The layout of the examined ventilation system is presented
in Figure 9.10.

to supply plenum

to exhaust plenum

Figure 9.10 Layout of supply-exhaust ventilation system for four floors

and four rooms on each floor.

169
Spread of Smoke and Fire Gases via the Ventilation System
Both supply and exhaust fans have similar characteristics as
the fan in Subsection 9.3.1.
Different cases were studied in order to calculate boundary fire
air flows and corresponding pressures in different rooms. "Fires"
were placed in different rooms and the behaviour of the ventila-
tion system was studied with the PFS program. Though such
problems are usually very numerically complex for the majority of
ventilation systems with several branches, calculations succeeded
in this case. Boundary fire flow for the majority of "fire" cases was
130 lis. All fire flows which exceed the boundary flow will cause
smoke to spread to the adjacent room(s) via supply ducts. Accord-
ing to Table 9.1, fires in all types of buildings with a volume
larger then 30 m3 will cause spread of smoke through supply
ducts.

Some measures to prevent smoke spread were discussed in

Chapter 8. Some of these were tested. The results of smoke spread
evaluations for fire flow of 450 lis are presented in Table 9.4.

Table 9.4 Data regarding the examined measures for the prevention
of smoke spread via supply-exhaust ventilation system

Supply fan Exhaust fan Smoke spread

To room(s), To other
same branch rooms
Normal Normal Yes No
Doubled rotational Doubled rota- Yes No
speed tional speed
Closed with bypass Normal Yes No
to the exhaust fan
Closed with bypass Doubled rota- No No
to the exhaust fan tional speed
Closed Normal Yes Yes
Closed with bypass Closed with Yes No
to the outside bypass to the
outside
Converted into ex- Normal No No
haust fan

170
 Section 2. Operation of Ventilation Systems during Fires
"Real Case" Study
The case when the rotational speed of the fan is doubled is a
very hard measure and can be applied e.g. in ventilation systems
with reduced normal capacity. Note, that doble speed results in
eight times higher moter effect. The case when the fan is switched
off corresponds to natural ventilation.
The studied cases do not cover all possible measures to prevent
smoke spread via ventilation ducts, nor do they examine system
behaviour with fire dampers.

In all the above cases, part of the smoke produced by a fire

spreads directly to the corridor through transfer air devices,
where it mixes with ambient air. Evaluation of the degree of dilu-
tion can be of great importance in risk assessment. This problem
will be discussed in Chapter 10.
All the above studied ventilation systems comprise equal spaces
with similar air flows. In reality, the air flows in different rooms
served by a ventilation system can differ from each other. In such
cases the whole behaviour of the system in case of fire in a room
can be more complicated and unpredictable than those discussed
above. A detailed analysis of such systems is needed to assess the
risk of smoke spread via the ventilation system.

171
Spread of Smoke and Fire Gases via the Ventilation System

172
 Section 2. Operation of Ventilation Systems during Fires
Evaluation of the Toxicity of Diluted Smoke

10 Evaluation of the toxicity of

diluted smoke

10.1 Background
In certain cases, where predictions show that comparatively low
volumes of smoke will spread to adjacent rooms, another situation
can be discussed. It can be reasonable to look at the degree of di-
lution of smoke instead of totally forbidding the spread of smoke
to adjacent rooms. In this connection, the toxicity of the diluted
smoke may have to be evaluated. It is well known that roughly
two-thirds of all deaths from enclosure fires can be attributed to
the presence of carbon monoxide (CO) which is known to be the
dominant toxicant in fire deaths. In the literature, some astonish-
ing facts concerning fatal fires can be found. One of the most ex-
treme cases is the fatal town house fire in Sharon, USA, 1987. Al-
though the fire was essentially localised in the first floor kitchen,
deaths occurred on the second floor. One of the victims had an ex-
traordinarily high level of carboxyhaemoglobin, suggesting expo-
sure to very high concentrations of CO. This fact and the later as-
sessment of the highest CO concentration produced by enclosure
fires are discussed in [28] and [29]. In these papers very special
cases of fire in an enclosure with ceiling and upper walls lined
with plywood were studied. The most dramatic effect was the
rapid growth and peak level of CO observed. Dry CO concentra-
tions reached 7% at the front sampling position and 14% at the
rear 200 s after ignition. The enclosure was fuelled by a natural
gas flow corresponding to 400 kW. The above reported levels are
extremely high and represent an unusual CO/C02 ratio. The tem-
perature observed was higher than 500°C. The fire was under-
ventilated (ventilation-controlled fire).

173
Spread of Smoke and Fire Gases via the Ventilation System

10.2 Evaluation of total volume of spread

smoke
The concentration of CO spreading to the adjacent room(s) was
evaluated step by step for an exhaust ventilation system.
Assuming that the volume of the room of fire origin is V (m3 )
and expansion is characterised by TJTr- 1, (6.1), the volume of
the hot spread smoke is
Vr = (Te I Tf - l)V
and after cooling down to a normal temperature it is equal to

(10.1)

10.3 Evaluation of relative smoke spread to a

single room
The share of this smoke volume which reaches another room can
be assessed for a simplified ventilation system with the fan shut
down. Several rooms are connected to a plenum at the same point.
The ductwork can be part of both the exhaust ventilation system
and the supply ventilation system.
Assuming, that
n number of rooms of equal size/volume, connected to
the same point before the plenum
a branch resistance
b facade resistance
c total resistance, = 1
qs flow in the plenum
qr flow in other n -1 rooms
qk fire flow in the branch
qr fire flow through the facade
qb total fire flow

The following simple expressions can be drawn up

174
 Section 2. Operation of Ventilation Systems during Fires
Evaluation of the Toxicity of Diluted Smoke

(10.3)

The ratio between q r and q b expresses the proportion of the

smoke volume which spreads from the fire room into the other
room. This ratio, r, represents the proportion of smoke volume in
the following assessment. A simple way to calculate this ratio is to
assign some flow value to the other rooms as q r and then calculate
backwards. First of all the pressure in the common junction, pg, is
calculated as
(10.4)

The flow qs in the plenum is calculated with the help of pressure

Pg
(10.5)

Fire flow in the branch is calculated according to (10.2) and the

fire pressure pb is then calculated
(10.6)

The next step is to calculate facade flow with the help of fire pres-
sure and the following expression
(10.7)

The proportion of smoke volume r

r=qr I qb =qr I (q 1 + qs + (n-l)qr) (10.8)

The proportion of smoke volume r as a function of parameters a, b

andn
1
r 05
(10.9)
05 2
( ( n - 1 + (a + b) + ((a +bb + a ( (a + b) o.s + n - 1) ) ·

The proportion of smoke volume r can be calculated for several

combinations of parameters.

175
Spread of Smoke and Fire Gases via the Ventilation System
Table 10.1 shows that the proportion of smoke volume, r, reaching
one of the rooms decreases with increasing number of rooms, n,
and decreasing airtighteness of the facade, b.

Table 10.1 Proportion r of smoke volume which spreads to the one of

the n rooms according to (10.9)

r
a b n=2 n=5 n=lO n=20
5 1.0 0.212 0.047 0.027 0.014
5 0.5 0.163 0.037 0.021 0.011
5 0.2 0.111 0.026 0.014 0.008
5 0.1 0.082 0.020 0.011 0.006

The above expressions can also be used to evaluate the propor-

tion of smoke volume reaching another room in the case when
both fans in a supply-exhaust system are shutdown. Perfect sym-
metry is assumed and the above expressions each describe half of
the ventilation system. The only parameters which must be
changed is b. The parameter b in this case describes only one half
of the total facade. This means that the previously defined b is
replaced by 4b because one half of the total flow gives the same
pressure drop as before.
The proportion r of smoke volume can be calculated for the case
with double spread for several combinations of parameters.
Table 10.2 shows that the proportion r of smoke volume reach-
ing one of the rooms decreases with increasing number of rooms,
n, and decreasing airtighteness of the facade, b. In general, more
smoke spreads to the adjacent room in this case that that in Table
10.1.

176
 Section 2. Operation of Ventilation Systems during Fires
Evaluation of the Toxicity of Diluted Smoke

Table 10.2 Proportion of smoke volume r according to (10.9), which

spreads to one of then rooms (double spread)

r
a b/4 n=2 n=5 n=lO n=20
5 1.0 0.344 0.200 0.118 0.064
5 0.5 0.273 0.153 0.088 0.047
5 0.2 0.195 0.106 0.060 0.032
5 0.1 0.149 0.080 0.045 0.024

10.4 Evaluation of CO- concentration in a

single room
Assuming that
cb eo-concentration in the fire room and in spreading
smoke (average value)
cr CO- concentration in other rooms

c,
Te
=(- Te
- 1) I (1 + (T - 1)) r cb -Vb = (1 - TJ Vb
- ) r cb - (10.10)
T1 1 V, Te V,

or for room of the same volume, Vb=Vr

c, = ( T~ - 1) I (1 + (T~ - 1)) r c =(1 - T)

b
~
rc b (10.11)
f f e

The expansion (Te IT{ 1) can vary from 1 to 2, which corresponds

to temperature change from 20°C to 313°C and 606°C, respec-
tively. Note that all rooms are assumed to be equal. It is also pos-
sible, of course, to calculate the case with rooms of different size.

177
Spread of Smoke and Fire Gases via the Ventilation System

10.5 Dangerous levels of CO- concentration

Different recommendations can be drawn up concerning accept-
able and dangerous CO concentrations in diluted smoke. Different
concentrations of CO cause different human reactions.

Table 10.3 Dangerous levels if GO-concentrations

CO con- Type of Other corn- CO- Time to Ref

cent- fire pounds, dose, Incapacita
ration, % ppm* tion,
ppm m1n min
02 C0 2
Death 10000 1 30
Dangerous 3000 30 30
!Unconscious 1500-2000 60 30
ne ss
!Unconscious 1000-8000 27000 7
ne ss
1500 smolde 15-21 hours 7
ring
10000 flaming 10-21 0-10 a few 7
0-30000 fully deve HCN <1 7
loped 0-1500 if possible
pp m
Acceptable 2000 >15 <5 4

Information about the toxic effects of CO in respiration air varies

a lot both in terms of concentration and dose. Thus it seems diffi-
cult to draw unequivocal conclusion concerning the risk due to
GO-concentration.

178
 Section 2. Operation of Ventilation Systems during Fires
Evaluation of the Toxicity of Diluted Smoke

10.6 Evaluation of CO- dose

The CO concentration calculated according to (10.10) may be
found not to be dangerous in itself. It is the CO dose or the prod-
uct of CO concentration, duration of exposure and the type of ac-
tivity during which exposure takes place which determines the
toxic effects. According to [7] it is not concentration in the smoke
that directly determines how someone will be affected, but the
concentration that has accumulated in the blood in the form of
carboxyhaemoglobin. Direct measurements are not feasible and
indirect measurements of exposure involve a certain degree of un-
certainty. A high level of activity and, as a result, intensive respi-
ration is more dangerous than slow level of activity, e.g. a sleeping
person.
The fatal dose of CO is around 27000 ppm*min at concentra-
tions ofmax. 8000 ppm [7].
There is another criterion, IDLH (immediately dangerous to life
and health), limited to 1500 ppm during 30 min, which gives a to-
tal of 45000 ppm *min. The concentrations presented in Table
10.3, converted into CO dose, show a significant spread.
If the time for one air change, T, is known, the CO dose can be
easily calculated
(10.12)

It is assumed. that the eO-concentration is achieved during a

few minutes, after which fire "chokes", or pressure relieves in the
fire room. The eO-concentration decreases exponentially accord-
ingto
(10.13)

Integrating of c(t) from 0 to infinity gives expression (10.12) for

the CO-dose. The time of one air change can be simply calculated
for case of mechanical ventilation, but it is more difficult for the
case of only infiltration and exfiltration.
This case is suitable for exhaust ventilation systems with a dis-
tribution chamber and shut down fan, and even for supply venti-
lation systems with shut down fan. A distribution chamber in a
supply ventilation system can provide protection against a dan-

179
Spread of Smoke and Fire Gases via the Ventilation System
gerous CO dose, but at the expense of the risk of several rooms
being subjected to equal smoke spread.

10.7 Application
CO concentrations of smoke reaching a single room can be calcu-
lated according to (10.10) and Table 10.1 for rooms of different de-
grees of airtighteness and different CO concentrations in the
smoke itself. Assuming expansion (T)Trl) equal to 1, which cor-
responds to temperature change from 20°C to 313°C, we have
er= 0.5rcb

An average value of parameter cb as can be assessed at 1%. Corre-

sponding values converted into ppm are set out in Table 10.4.
Note that all rooms are assumed to be equal.

Table 10.4 CO concentration in ppm which spreads to one of n rooms

at an average value cb = 1%

CO,ppm
b n=2 n=5 n=lO n=20
1.0 1060 235 135 70
0.5 815 185 105 55
0.2 555 130 70 40
0.1 410 100 55 30

All values are acceptable according to IDLH since CO concentra-

tion level does not exceed 1500 ppm (shaded area).
Values of mean CO concentration cb other than 1% result in
higher values. Corresponding CO concentrations reaching a single
room, for 14% CO in the smoke from the fire room, are presented
in Table 10.5.

180
 Section 2. Operation of Ventilation Systems during Fires
Evaluation of the Toxicity of Diluted Smoke

Table 10.5 CO concentration in ppm which spreads to one of n rooms

at an average value cb = 14%

CO,ppm
b n=2 n=5 n=lO n=20
1.0 14840 3290 1890 980
0.5 11410 2590 1470 770
0.2 7770 1820 980 560
0.1 5740 1400 770 420

All values which do not exceed CO concentration level of 1500

ppm are acceptable according to IDLH (shaded area).

181
Spread of Smoke and Fire Gases via the Ventilation System

182
 Section 2. Operation of Ventilation Systems during Fires
Evaluation of Visibility of Diluted Smoke

11 Evaluation of visibility of
diluted smoke

11.1 Background
As discussed in Subsection 10, it may be reasonable in certain
cases where the predicted spread of smoke to adjacent rooms has
comparatively low values to look at the degree of dilution of the
smoke instead of totally forbidding the spread of smoke to adja-
cent rooms. In this connection, evaluation of risk due not only to
the toxicity of diluted smoke but also to visibility may be neces-
sary. Visibility through smoke can be of great importance for the
survival of an individual in a fire. Visibility depends on many dif-
ferent factors, such as wavelength of the light, whether the sign is
emits or reflects light, illumination of the room(s), the visual acu-
ity of the individual, the absorption coefficient of the smoke, etc.
However, there are methods for smoke visibility estimation. One
of them, for safe level, is presented in Swedish Buildings Codes
[8]; the height of the smoke layer above the floor must not be less
than 1.6 m+ 10% of room height. This smoke visibility can be re-
garded as safe if it is combined with other limiting parameters,
such as level of heat radiation and air temperature. Another
method is presented in [5]; visibility scan be described as

!~
!5__ for light- emitting sign
(11.1)
s= for light- reflecting sign

183
Spread of Smoke and Fire Gases via the Ventilation System
where k is the smoke extinction coefficient and is a product of an
extinction coefficient per unit mass, km, and mass concentration of
the smoke, m.
k=k m m (11.2)

For example, Seader and Einhorn obtained km values of 7.6 m 2/g

for smoke produced during flaming combustion of wood and plas-
tics and a value of 4.4 m 2/g for smoke produced during pyrolysis of
these materials. There are a lot of uncertainty factors in such
smoke visibility assessment. The data is based e.g. on the subjects
viewing the smoke through glass (the irritant effect was elimi-
nated). The value km is based on a limited number of small-scaled
experiments with a polychromatic light source, etc.
Another method for estimating the visibility uses the empirical
mass optical density data. Optical density per meter, D, describes
as

D=-D_m_Mf_ (11.3)
VC
where Dm is the empirical mass optical density, m 2/g, .1M is the
mass loss of the sample, g, and Vc is the volume of the fire room,
m 3 • This method is regarded to be more reliable and more direct.
According to [32], [33] lowest visibility value in familiar spaces
is 3-5 m, but never less then 2 m, and in unfamiliar spaces it is
15-20 m.

184
 Section 2. Operation of Ventilation Systems during Fires
Evaluation of Visibility of Diluted Smoke

11.2 Evaluation of visibility in diluted smoke

spread to a single room
The evaluation of visibility in smoke in another room as a result
of smoke spread via ventilation ducts is analogous to evaluation of
GO-concentration in smoke, presented in Subsection 10.
Assuming that the smoke in the fire room and also in the other
room which it reaches is well mixed, that some limit of weakness
of the comfort conditions is common for all rooms and that

sb is visibility in the fire room

sr is visibility of diluted smoke reaching one of the rooms
rh is smoke concentration in the fire room
rr is concentration of diluted smoke reaching one of the rooms,
we have
(11.4)

(11.5)

Visibility will in actual fact be greater because smoke concen-

tration will be higher near the ceiling and will gradually decrease
downwards.

185
Spread of Smoke and Fire Gases via the Ventilation System

186
 Section 2. Operation of Ventilation Systems during Fires
"Cold" Case Assessment

12 ''Cold'' case assessment

The whole study assumes that all flows have ambient parameters,
"cold flows". In fact, according to results obtained in Section 1,
flow temperatures can exceed 300 °C. Although all evaluations of
smoke spread via ventilation ducts take the temperature factor
into consideration in dealing with volume flows, the problem of
calculating pressure drop in ventilation systems can be discussed
further. According to [31], Fanning friction factor f for fully devel-
oped tube flow is inversely proportional to the Reynolds number
based on mean velocity and tube diameter. At the same time, the
Darcy - W eisbach friction factor fD is four times as large as f.
fv=4f (12.1)

Different degree of flow turbulence show different empirical rela-

tions between fD and ReD', where ReD=UD I v. Kinematic viscosity
is V=1J I p, m 2/s, where 1J is viscosity.
25
fv :=4 · 0.079Re-:· (12.2)

Pressure drop per unit length, l, is

M fv pU2
=-·-- Palm (12.3)
1 d 2
Substituting (12.2) into (12.3) gives
M 17
o.25 P o.15u 1.75
-~-= 0.158 dl.25

or

(12.5)

187
Spread of Smoke and Fire Gases via the Ventilation System
or, as a function of flow
M _ o.25 p o.75 ql.75
17
-~-= 0.158 d4.75 (12.6)

4
2.5 ·103 < q/dv < 2.5 ·10
and
M V0.25p 1.75

-~-= 0.158 d4./: (12.7)

4 6
2.5 ·10 < q/dv < 1.3 ·10
An empirical relation that holds at high Reynolds numbers is
2
fv =:4 X 0.046Re-;}· (12.8)

Substituting (12.8) into (12.3) gives

M 11 0.2 P o.8ul.8
-=
l
0.092
d"
12 (12.9)

or
M vo.2p Ul.8
-=
l
0.092
d"
12 (12.10)

or, as a function of flow

M o.2 p o.8 1.8
17
-=
l
0.092
d .
48 q (12.11)

4 6
2.5 ·10 < q/dv < 1.3 ·10
and
M
-=
l
0.092
v0.2p
4;
d .
1.8
(12.12)

4 6
2.5 ·10 < q/dv < 1.3 ·10
Expressions (12.11) and (12.12) describe fully developed turbu-
lent flow with high Reynolds numbers, which is characteristic for
high temperature flows.
For fully developed turbulent flows through ducts with cross
sections other than round, provided ReD should be replaced by the
Reynolds number based on hydraulic diameter, ReDh'

188
 Section 2. Operation of Ventilation Systems during Fires
"Cold" Case Assessment
The relationship between temperature, density p and v0 ·2 p-term
from (12.12) and (12.10) is illustrated in Figure 12.1

p 1.4,...------------------------rQ.14 V0.2.p

1.2 0.12

0.1

0.8 0.08

0.6 - - - - - - - - 0.06

0.4 - - - - - - - 0.04

0.2 - - - - - - - - - - - - - - - - - - 0.02

o~---+---+----~-----~--~~---4o

0 50 100 150 200 250 300

Temperature (C)
02
Figure 12.1 Density p and the v p-term vs. temperature for fully devel-
·

oped turbulent flows.

Figure 12.1 shows that density p and the v0·2 p-term decrease by
approximately a factor of two as temperature in K doubles. This
means that the pressure drop per meter length of duct (friction
pressure drop in straight ducts) also decreases by a factor two.
This seems to indicate that all ventilation system calculations
for the "cold" case are overestimated. At the same time, as smoke
is diluted downstreams in the main duct, flow temperatures de-
crease and "cold" flows are the default values.
The pressure drop in wyes and junctions is more difficult to
evaluate, as data is not available.

189
Spread of Smoke and Fire Gases via the Ventilation System

190
 Section 2. Operation of Ventilation Systems during Fires
"Alexander" Method

13 The ''Alexander'' -method

Another method for evaluating the ability of a ventilation sys-

tem to ventilate a burning room without smoke spread to adjacent
room(s) via the ventilation ducts is developed by Lars Jensen and
is presented in [16]. It is a method for manual calculations and is
named "Alexander" method after Alexander the Great and myth
of the Gordian knot. The method is based on the "cut-off' princi-
ple. A branch(es) with zero pressure (boundary conditions), serv-
ing the burning room, is cut off and the rest of the ventilation sys-
tem is calculated once more regarding the new pressure/flow pro-
portions. This calculation results in new total pressure loss and
flow values. On the bases of these new parameters and the fan
curve, new total pressure and flow for the abridged ventilation
system are assessed. The fire flow for such a system can be calcu-
lated.
The algorithm of this method is shown for an exhaust ventila-
tion system with six equal branches. The schematic layout of the
examined ventilation system is presented in Figure 13.1.

b d f

Figure 13.1 Layout of exhaust ventilation system.

where
qn is a normal ventilation flow

191
Spread of Smoke and Fire Gases via the Ventilation System
a-f roonas
G,H junctions.
In the ordinary case the present ventilation systena is regulated
and the pressure loss in branch a -G is equal to pressure loss in
branch b-G. In the case of fire, pressure in the burning roona, e.g.
in roona a, rises and affects pressure in branch a-G. The boundary
or linait condition, when snaoke frona roona a starts to spread to the
adjacent roona b, is reached when pressure in junction G beconaes
equal to zero. The described naethod uses this phenonaena and
replaces the layout presented above with an abridged ventilation
systena layout, shown in Figure 13.2

.G H

I
c d e f
qn qn qn qn

Figure 13.2 Layout of abridged exhaust ventilation system.

Part of the naain duct a-G and branch G-b is cut offinG and is
replaced by sonae fictitious "roona" with atnaospheric pressure and
flow of 2qn. If the pressure loss in branch G-H is pP the "new" flow
frona roona c is
q=qn (pl /pc-H)O.S

where PcHn is pressure loss in G-H deternained for the nornaal case.
This principle is used to calculate "new" flows in each branch
for the abridged ventilation systena. The resulting total flow qa
and the resulting total pressure loss Pa are plotted in Figure 13.3.

192
 Section 2. Operation of Ventilation Systems during Fires
"Alexander" Method

qs qa Flow, 1/s

Figure 13.3 Pressure-flow diagram.

Pressure Ps and flow qs are new fan values for this abridged venti-
lation system. The corresponding boundary fire flow in branch G-
H (instead of assumed flow of 2q n) can be calculated
q b = 2q n q s I q a
where
q b is boundary fire flow.
The corresponding boundary fire pressure can be calculated
2
Pb = Pa-G (qb I qn )
where
Pa-G is pressure loss In branch a-G at the normal ventilation
flow, qn.

193
Spread of Smoke and Fire Gases via the Ventilation System

194
 Discussions and Conclusions

14 Discussions and
conclus ions

Introduction
The theoretical studies along with computer simulations have
provided knowledge regarding smoke production of different typi-
cal fires in certain types of buildings, as well as knowledge regard-
ing spread of smoke to adjacent room(s) via ventilation ducts.
Application of this basic knowledge largely depends on the charac-
teristic parameters estimated for different types of buildings. The
main building types chosen for this work are dwelling houses,
hospitals, offices and hotels. The study is interdisciplinary and
comprises different fields of knowledge: fire protection engineer-
ing, building science and building services.
Whether or not smoke spread will occur is determined by factors
such as the typical fire scenario in the above types of buildings,
the airtightness of the construction, window breakage as a result
of heat stress and the layout of the ventilation system.

Fire scenario and rooms geometry

Summing up statistical data presented in different sources re-
garding the danger of fire in the analysed buildings, it was as-
sumed that fire is limited to the room of fire origin and to the ob-
ject in which the fire started. The most probable objects where the
fire starts are bedclothes and sofas in the dwelling houses.
In order to evaluate fire behaviour in the analysed types of
rooms, a fire growth curve is used which is based on the results of

195
Spread of Smoke and Fire Gases via the Ventilation System
full-scale tests in which different items were burnt in the labora-
tory and heat release rate as a function of time was measured.
The following items were assumed to initiate fires: sofa and bed
in dwelling houses, bed in hospitals, equipment and furnishing in
offices and bed in hotels. Data provided by these fire growth
curves is used as basic fire data in evaluating fire progress in the
examined rooms with the help of computer programs.
The geometry of the studied types of rooms is chosen in diapa-
son of 30 -224m3 for the standard room areas and room heights
for each type of buildings. Doors and windows are closed and un-
damaged, and the only vents are leakage paths.

Computer programs for predicting compart-

ment fires
Fire simulations are made with the help of the computer program
CFAST (Consolidated Model of Fire Growth and Smoke Trans-
port), which is part of the computer program HAZARDL It is
based on the two-layer fire model. A comparison of parameters ob-
tained by CFAST for single-room fires, with measured values,
shows that pressures, 02-concentrations and mass flows are
slightly underpredicted, while there is good agreement regarding
temperatures.
In order to compare the magnitude of parameters obtained with
CFAST with those predicted with the two-layer fire model com-
puter program DSLAYV and with the field model program
SMAFS, a special case of a compartment medium fast at2 -fire in a
single room of 6x5x2.5 m is simulated. Both mass and volume
flows predicted by SMAFS are 32% higher than those predicted by
CFAST, which is in turn 35% higher than those predicted by
DSLAYV. Flows obtained with SMAFS cannot be considered as
determining values because they are very approximate.
Owing to the scarcity of experimental data and simulation re-
sults, and the complexity of comparing the numerous variables in
a complex fire model, true validation of the CFAST model is not
possible in this study (true validation of a model would involve
proper statistical analysis of a variety of compared variables). A
special study of the issue is required. Different locations of burn-
ing items are also tested.

196
 Discussions and Conclusions

Airtightness
Validation of the magnitude of internal and external leakage
paths in the studied buildings is an important issue for evaluation
of both fire progress in the room and predicted parameters.
The study of relevant literature has shown that the air leakage
according to the Swedish Building Regulations BBR 94 (1.25
cm2/m2 for dwellings and 2.5 cm2/m2 for other types of studied
buildings), applied to the whole building envelope, can be accepted
as a suitable and probable value.
The analysed data are not sufficient for less robust validation
as they cover mostly dwellings, while this study comprises differ-
ent types of buildings. In particular, there are no data for older
buildings. At the same time, due to Swedish construction tradi-
tions and airtightness requirements which are different from
those in other countries, the information from international sci-
entific documents published abroad cannot be used as straight-
forward data.

Window breakage
As the CFAST fire model assumes that the envelope of the fire
room is perfectly airtight during the entire fire progress, except
for the leakage paths, the evaluation of the time and ambient
conditions at which window breakage occurs seems important.
A study of the literature, together with the simple window
breakage model presented in this work, shows that the tempera-
ture difference of 58-100°C in the glass itself can be assumed to be
dangerous for glass cracking. The most probable temperatures in
the fire room which cause this phenomenon vary between 200 and
400°C.
For this study, the temperature of 350°C in the fire room is ac-
cepted as the limiting temperature value. The uncertainty of such
assumptions is aggravated by lack of data on window breakage as
a result of the simultaneous effect of both high pressure and high
temperature in the enclosure.
Further research on the issue appears necessary.

197
Spread of Smoke and Fire Gases via the Ventilation System

Peak fire flows

The assumption in predicting peak fire volume flows by CFAST
simulations is that leakage paths are located at the sills of door-
ways. Mass flows with leakage paths at the sill of the doorway
were somewhat larger then those with leakage paths at the other
locations (5-20%) (due to the differnce in air density). At the same
time temperature, pressure, rate of heat release and interface po-
sition were insignificantly affected by the location of leakage
paths. For the purpose of generalisation, the fire locations chosen
were those which give the highest fire flow values in these cases.
A regression analysis of these flows made with the statistical
program Minitab shows good agreement with the regression line
in the case of fires in dwellings and fires in hotels, while the
agreement with fire flows in hospitals and offices, predicted in
such a way, was not good enough.
A simple model which approximates fire curves as a linear seg-
ment function and is based on evaluation of fire flow, which de-
pends directly on the heat release effect, accumulated into re-
leased energy, which in turn corresponds to the given volume of
consumed room air, gives peak fire flow values in good agreement
with the regression line over the whole range of the studied types
of buildings. Although there is an uncertainty in such an assess-
ment, it seems reasonable to use reported fire flows as default
peak flow values in the described rooms, as they are evaluated
with a margin for overestimation. Peak fire flows are at least ten
times higher than ordinary ventilation flows.
Characteristic flow curves for the chosen buildings, of volumes
ranging from 30m3 to 240m3 , are presented. These volumes are
assumed to be those most usual for the described types of build-
ings.
For the purpose of comparison, both the presented flows and
flows obtained with DSLAYV for .f- fires are plotted in the same
diagram. It shows that flows obtained for offices agree very well
with those predicted for the medium fast .f - fires. Flows obtained
for hospitals have values between those predicted for medium and
fast .f - fires over the whole range of volumes. Flows obtained for
3
hotels up to approximately 100m are insignificantly lower than

198
 Discussions and Conclusions
those predicted for fast l - fires, while flows in hotels larger than
3
100m have values between those predicted for fast and ultrafast
2 3
t - fires. Flows obtained for dwellings up to approximately 100m
are lower than those predicted for ultrafast l - fires. In contrast,
3
flows in dwellings larger than 100 m have higher values than
2
those predicted for ultrafast t - fires.

Other parameters predicted with fire simula-

tions
All compartment fire simulations show that only the ignition and
growth stages of a fire occur.
Values of oxygen in the rooms studied at the time when the cor-
responding peak fire flows were obtained vary quite a lot even in
the same room depending on fire location.
Peak fire flow occurs shortly after ignition. The time after igni-
tion varies between 1 and 5 min. Exceptions are fires in dwellings
with a bed as ignition item, where this period can be as long as 7
min 30 sec.
The peak temperature in the upper layer of the majority of the
examined rooms does not exceed 340°C.
The pressure in the majority of rooms is relatively low, except
for dwellings with a sofa as the burning item where the pressure
is considerably higher, up to 2057 Pa. "Real" pressures, obtained
in rooms with terminals, do not exceed 647 Pa.

Ventilation system design

In natural ventilation systems which serve each separate room
with a separate duct, smoke spread will never occur except due to
leakage of the duct itself.
The risk of smoke spread via mechanical ventilation systems
depends mainly on the type of ventilation system.
General theoretical methods show that the highest risk of
smoke spread to the adjacent room(s) via ducts can be expected in
open supply-exhaust ventilation systems, while the lowest risk
can be expected in exhaust ventilation systems with the pressure
drop in branches larger than in the main duct.

199
Spread of Smoke and Fire Gases via the Ventilation System
In mechanical ventilation systems designed according to given
pressure drop relations, in which prevention of smoke spread re-
lies only on shutting down the ventilation system, some smoke
spread to the adjacent room(s) will always occur.
The proportion of smoke which reaches adjacent room(s) in me-
chanical systems in which the supply fan or both fans are
switched off depends on the number of rooms and their sizes. The
maximum amount of smoke spreads to the adjacent room(s) from
the burning room in cases where there are two equal rooms (with
fire in one of them), a lot of smoke spreads from a small burning
room to the rest of the building, and little smoke spreads from a
large burning room (or several burning rooms) to the small room.
The magnitude of fire flow is not discussed here. The term
"adjacent room(s)" in this study can refer to any fire compartment
on the same floor or on another floor nearest to the burning room,
served by the same ventilation system.

Boundary conditions
Exhaust and supply-exhaust (open and closed) systems and sup-
ply-exhaust systems with "zero"-duct ventilation are studied with
the help of the semi-graphical computer program PFS which
treats arbitrary flow systems of any structure and layout, any
media, and any problem, design and investigation.
Boundary conditions, e.g. boundary fire flows when smoke from
the burning room starts spreading via ducts to the adjacent room
(zero air flow in the adjacent room(s)), are in good agreement with
values expected according to the general models.
As this study is interested in "abnormal" flow behaviour, i.e.
changes in flow directions and consequent decrease in flow in cer-
tain parts of junctions to zero value, it is not obvious how wyes
would respond.
PFS provides the opportunity to describe the resistance proper-
ties of wyes and junctions in two different ways: with a quadratic
function and with the help of empirical equations. Both types are
tested, and the resulting boundary fire flows are different. Pres-
sure loss as a quadratic function, which takes account of high
Reynolds numbers in duct flows in case of fire, is chosen as the de-

200
 Discussions and Conclusions
fault. More detailed research into the behaviour of wyes under
such conditions seems important.

"Real-case"
The types of buildings examined in this study are equipped, as
usual, with ventilation systems, the degree of complexity of which
can be different. Evaluation of boundary fire flows in these venti-
lation systems is very numerically complex. Assessment of the
risk of smoke spread via ducts seems more manageable when fire
flows typical for the studied buildings presented in this work are
used.

Measures to prevent the spread of smoke

Different solutions for the prevention of smoke spread to the adja-
cent room(s) via the ventilation system are analysed; these are
shutting down the fan(s), doubling the speed of rotation of the
fan(s), converting the supply fan into exhaust fan, bypassing the
support plenum to the exhaust plenum, and a combination of
these measures with ordinary fan function and with each other.
These measures do not cover the entire range of measures for
the prevention of smoke spread via ducts, nor do they examine the
behaviour of systems with fire dampers. There is no total solution
of the problem. Different measures have shown different results
for different ventilation system layouts. Each must be tested
separately for the actual type of ventilation system and layout.
Doubling the speed of rotation of ordinary fan(s) is a very hard
measure and can be applied e.g. in ventilation systems with re-
duced normal capacity. Shutting down the fan(s) corresponds to
natural ventilation, and spread of smoke occurs in all the exam-
ined systems.

Toxicity and visibility of the diluted smoke

In certain cases, where predicted spread of smoke to adjacent
rooms shows comparatively low values, another situation can be
discussed.

201
Spread of Smoke and Fire Gases via the Ventilation System
It can be reasonable to look at the degree of dilution of the
smoke instead of totally forbidding the spread of smoke to adja-
cent rooms. In this connection, the risk of toxicity and visibility of
diluted smoke may have to be evaluated. Evaluation of the con-
centration of CO in the smoke spread is then important. In cer-
tain underventilated fires, with the ceiling and upper walls lined
with plywood, dry CO concentrations are reported to reach 14%.
These reported levels are extremely high and represent an un-
usual CO/C0 2 ratio. Generally, CO concentrations increase dra-
matically in underventilated fires.
Evaluation of the concentration of CO which spreads to the ad-
jacent room(s), made step by step for exhaust and supply-exhaust
ventilation systems, comprises evaluation of the total volume of
smoke spread, evaluation of the proportion of this smoke volume
which reaches another room, evaluation of the concentration of
CO in a single room, and evaluation of the CO - dose.
The CO-dose is a product of GO-concentration, the duration of
exposure, and the type of activity during which exposure takes
place.
There are different criteria which determine the fatal dose of
CO. Information about the toxic effects of CO in respiration air
varies a lot both in terms of both concentration and dose. Thus it
seems difficult to draw unequivocal conclusions about the risk
rate due to GO-concentrations. IDLH (immediately dangerous to
life and health), limited to 1500 ppm during 30 min, is used as the
default value in this study.
Visibility through smoke can be of great importance to an indi-
vidual attempting to survive a fire. The evaluation of the visibility
of smoke in another room, as a result of smoke spread via venti-
lation ducts, is analogous to evaluation of the concentration of CO
in the spread smoke. In actual fact, visibility will be higher than
assessed, because the concentration of smoke will be higher near
the ceiling and will gradually decrease downwards.

"Cold" case assessment

The whole study assumes that all flows have ambient parameters,
"cold flows". In fact, according to results obtained in fire simula-
tions, flow temperatures can exceed 340°C. Although all evalua-

202
 Discussions and Conclusions
tions of smoke spread via ventilation ducts take into consideration
the temperature factor in dealing with volume flows, the pressure
drop per meter length of duct (friction pressure drop in straight
ducts) at such temperatures seems to decrease to one half of ordi-
nary values. This statement seems to indicate that all ventilation
system calculations made for the "cold" ventilation systems are
over estimated. The pressure drop in wyes and junctions is more
difficult to evaluate due to the paucity of data.
Further research in this field is needed. At the same time, along
with smoke dilution downstream from the main duct, flow tem-
peratures decreases and "cold" flows become the default ones.

Manual methods
One more method for evaluating of possibility og smoke spread via
ducts is presented in the study. The "Alexander" method is a
method for manual calculations, based on the" cut-off' principle.
The "cut-off' principle means that a branch(es) at the boundary
conditions with zero pressure is cut off and the rest of the venti-
lation system is calculated once more with respect to the new
pressure/flow conditions. This calculation results in new total
pressure loss and flow values. On the bases of these new parame-
ters and the fan curve, the new total pressure and flow are as-
sessed for the abridgedventilation system. The fire flow for such a
system can be calculated. The method is named after Alexander
the Great and myth of the Gordian knot.

Application
To evaluate smoke spread via ventilation systems in case of fire,
the following procedures can be applied.
For ventilation systems of all types, the risk of smoke spread
via ducts can be evaluated step by step with the help of computer
programs as follows:
"Boundary" fire flow:
step 1 the ventilation system studied is described with the
help of a computerprogram,
step 2 the "boundary" fire flow in the room studied or in
one of the rooms connected with the branch located

203
Spread of Smoke and Fire Gases via the Ventilation System
furthest away from the fan are evaluated. It means
that the supply flow to the room studied in the case
of supply and exhaust ventilation system is replaced
by the zero flow. In the case of exhaust ventilation
system, the flow(s) in the exhaust branch calculated
upwards from the "fire" room is replaced by the zero
flow,
step 3 the "boundary" flow(s) is compared with a fire flow(s)
(Figure 4.13) for the room of corresponding volume in
the corresponding type of building. If the "boundary"
flow is higher than the fire flow from the Figure 4.13,
smoke does not spread via the ventilation system. If
the "boundary" flow is lower than the fire flow from
the Figure 4.13, smoke spread occurs via the ventila-
tion system.

"Real" fire flow:

step 1 the ventilation system studied is described with the
help of a computer program,
step 2 the fire flow in the room of certain volume studied in
the type of building studied is evaluated with the
help of Figure 4.13,
step 3 the ventilation system studied is tested for risk of
smoke spread with the help of those fire flows
replacing normal ventilation flows in the
room(s) studied.

"Alexander" method

This method for manual calculations, described for the exhaust

ventilation systems in Section 13, can be applied for smoke spread
evaluations via ventilation systems with lower degree of complex-
ity.
Note that smoke spread via ventilation systems should be
evaluated separately for each ventilation system of actual layout
and properties.
Different measures preventing smoke spread via ventilation
ducts can then be tested separately for the ventilation system
studied in conjunction with Subsection 8.3.

204
 Discussions and Conclusions
In certain cases, when toxicity of assessed smoke spread via ex-
haust ventilation system and supply and exhaust ventilation sys-
tem are of interest, it can be rapidly evaluated using Table 10.4
and Table 10.5 respectively.

Future research
This work along with other reports [1], [2], [3] and [27] is a pio-
neering research attempts in regard to the evaluation of smoke
spread via ventilation ducts in ventilation systems in operation.
Owing to the great uncertainty of the assumptions made, the
paicity of experimental data and lack of knowledge regarding cer-
tain fire processes, it is necessary to investigate the problem fur-
ther.
The fire scenario in the buildings studied, with more than one
burning item involved, is of interest as such fires can produce fire
flows different from those predicted in this work.
The issue of the reliability of the parameters predicted by
CFAST, along with true validation of the model, should be ana-
lysed in a proper way.
The problem of flashover in smoke in ventilation ducts or ter-
minals, i.e. ignition of unburned fuel in fire gases, must be studied
properly, probably with the help of field model computer pro-
grams.
Window breakage phenomena as a result of both high tempera-
tures and the influence high pressures must be studied.
The "cold" case as against the real high temperature smoke flow
in ventilation systems must be examined in greater detail, since
no examination has been made as yet of pressure drop in junc-
tions and wyes in cases with "abnormal" flow behaviour, i.e.
changes in flow directions and consequent decrease in flow in cer-
tain parts of junctions to zero value.
The problem of smoke spread via ventilation ducts in other
types of buildings, e.g. in industrial buildings and stores, seems
important. The possibility of using ordinary ventilation systems
along with other safety measures (smoke venting, sprinklers, etc.)
in the event of fire in such buildings can improve safety, and the
economic consequences of it are far from negligible.

205
Spread of Smoke and Fire Gases via the Ventilation System
The problem of eliminating or reducing smoke spread via ducts
to the adjacent room(s) is connected with evaluation of the toxic-
ity and visibility of the diluted smoke. This matter needs further
research.

206
 References

References

1. L. Jensen, "Enkel systemanalys av rokspridning i

ventilationssystem, Rapport BKL 1988:15" (1988) (in
swedish).

2. T. Hielscher, B. Varelius, "Brandspridning via

Ventilationssystem - Smokespread through Ventilation
Systems", Lund (1992) (in swedish).

3. L. Jensen, "Spridning av rok och brandgaser i

ventilationssystem", Rapport TABK--93/3011, Lund
(1993) (in swedish).

4. "Brandskydd, Boverkets Byggregler, Teori och Praktik",

Brandskyddslaget, LTH-Brandteknik, Stockholm
(1994) (in swedish).

5. "The SFPE Handbook of Fire Protection Engineering", Society of

Fire Protection Engineers, National Fire Protection
Association, First Edition (1988).

6. "The SFPA Handbook of Fire Protection Engineering, 2nd

edition", ISBN: 0-87765-354-2, USA (1995).

7. S. Sardqvist, "Initial fires, RHR, Smoke Production and CO

Generation from Single Items and Room Fire Tests",
Lund (1993).

8. BBR 94, Boverkets Byggregler, Foreskrifter och allmant rad,

BFS 1993:57, Boverket (1993) (in swedish).

207
Spread of Smoke and Fire Gases via the Ventilation System
9. P. Levin, "Building Technology and Air Flow Control in
Housing", Stockholm (1991).

10. A. Blomsterberg, "Ventilation and Air Tightness in Low-rise

Residential Buildings", Stockholm (1990).

11 Arsstatistik bok 1993 (in swedish).

12. B. Hagglund, "A Numerical Simulation of Fires in Hospital

Rooms", National Defence Research Institute, FOA
Rapport C 20684-2,.4 (1987).

13 J.Bergstrom, "Brandpaverkan pa Glaskonstruktioner,

Litteraturstudie samt forslag till forskning",
BRANDFORSK, Stockholm (1995) (in swedish).

14. Aruna A. Joshi, Patrick J. Pagni, "Users' Guide to BREAK1,

the Berkeley Algorithm for Breaking Window Glass in a
Compartment Fire", University of California, Berkeley
(1991).

15. "An Update Guide for HAZARD 1 Version 1.2", Building and
Fire Research Laboratory Gaithersburg, U.S.Department
of Commerce, (1994).

16. B. Backvik, S.Bengtson, T. Fagergren, 0. Granberg, L.Jensen,

"En Handbok om Brandskyddsteknik for
Ventilationssystem", Stockholm (1996) (in swedish).

17. Minitab Inc.: Minitab Handbook, Second Edition. Boston, PWS

Publishes (1988).

18. B. Hagglund, "Simulating Fires in Natural and Forced

Ventilated Enclosures", National Defence Research
Institute, FOA Rapport C 20637-2.4 (1986).

19. "Skydd mot brandgasspridning via ventilationssystem med

flaktar i drift", Boverket, (1994) (in swedish).

208
 References
20. "Kvalitetsansvarig tekniska foreskrifter, Niva K", Bygginfo,
(1995) (in swedish).

21. L.Jensen, "Beskrivningssprak for PFS", Rapport TABK--

95/7020, Institutionen for byggnadskonstruktionslara,
LTH, Lund (1995) (in swedish).

22. L.Jensen, "Arbetsbank for PFS", Rapport TABK--96/7033,

Institutionen for byggnadskonstruktionslara, LTH, Lund
(1996) (in swedish).

23. L.Jensen, "Berakningsuttryck for PFS", Rapport TABK--

95/7022, Institutionen for byggnadskonstruktionslara,
LTH, Lund (1995) (in swedish).

24. L.Jensen, "Tillampningsexempel med PFS", Rapport TABK--

95/7023, Institutionen for byggnadskonstruktionslara,
LTH, Lund (1995) (in swedish).

25. L.Jensen, "Brandtekniska berakningar med PFS", Rapport

TABK--95/7024, Institutionen for
byggnadskonstruktionslara, LTH, Lund (1995) (in
swedish).

26. L.Jensen, "Brandtekniska ovningsexempel med PFS", Rapport

TABK--95/7025, Institutionen for
byggnadskonstruktionslara, LTH, Lund (1995) (in
swedish).

27. L.Jensen, "Undersokning av rokspridning via

ventilationssystem med PFS", Rapport TABK--96/7035,
Institutionen for byggnadskonstruktionslara,
LTH, Lund (1996) (in swedish).

28. William M. Pitts, Eric L. Jonsson and Nelson P. Bryner,

"Carbon monooxide formation in fires by high-
temperature anaerobic wood pyrolysis",
Twenty-Fifth Symposium (international) on
Combustion/The Combustion Institute, 1994/pp.1455-
1462.

209
Spread of Smoke and Fire Gases via the Ventilation System

29. William M. Pitts, "Reactivity of product gases generated in

idealised enclosure fire environments", Twenty-Fifth
Symposium (international) on Combustiontrhe
Combustion Institute, 1992/pp.1737-1746.

30. J. Ondrus, "Brandforlopp", Lund (1989) (in swedish).

31. A. Bejan, "Heat transfer", John Wiley & Sons, Inc. (1993).

32. T. Jin, ''Visibility through fire smoke, Part 5", Fire Research
Institute of Japan, Tokyo (1976).

33. T. Jin, T. Yamada, "Irritating effects of fire smoke on

visibility", Fire Science and Technology, Vol. 5 No 1,
pp. 79-90, (1985).

34. O.Wallin, "Rokkontroll i sjukhus- rokevakuering vid brand i

vardrum", Installationsteknik, KTH, Stockholm,
(1987) (in swedish).

210
 Appendix

Appendix

The following figures correspond to Table 4.5, Subsection 4.3.

211
Spread of Smoke and Fire Gases via the Ventilation System

300.0 600.0
--1 u

-50.0 -100.0
0.0 500.0 0.0 500.0
n= 1 Time (s) n= 1 Time (s)
3.0 .300.0
--1 u --1 u

1~ I
t>, ·a
5!

0.0
500.0 0.0 500.0
n= 1 Time (s) n= 1 Time (s)

25.0
0.2
-- 1 u 02 --1 1-2

'--._ (;)
'--._
E
Q_
Q_
6
'--._

c
(!) >
2 .E
(!)
Q_

0.0 0.0
0.0 500.0 0.0 500.0
n= 1 Time (s) n= 1 Time (s)
0 ..3
0.0
--1 L 1- 2 2 'L 2- 1 1

I
V; (;)
'--._
'--._
0'
Cl'
6 6

•E
.
0.0 0.0
0.0 500.0 0.0 500.0
n= 1 Time (s) n= 2 Time (s)

Dwelling house (sofa) 1

212
 Appendix

300.0 ,------,.---,.---,.---_-_,_-----,;-;-;-u--,1ooo.o
1

0
o_

l::'
Q_

-50.0
0.0 500.0 0.0 500.0
n= 1 Time (s) Time (s)
3.0 .-----------,.-----------,----;-----,--, ---,-c-:ol
--1U
300.0
--1 u

f
·I

0.0
0.0
0.0 500.0
n= 1 Time (s) 0.0 500.0
n= 1 T'1me (s)

22.0 0.2

1-
-.-1U02 1 ti 1- 2 I

1-
0

""'- m
""'-
Ql
E 6
0.

I
0.

""'-
c
Q)

2
Q)
0.

6.0 0.0
0.0 500.0 0.0 500.0
n= 1 Time (s) n= 1 Time (s)
0.3 1.0
--1 L ,_ 2 2 L 2- 1

(j) m
""'-
ry ""'-
Ql
6 6

.E

0.0
0.0
n= 1 Time (s)
500.0
0.0
0.0
n= 2
~ Time (s)
500.0

Dwelling house (sofa) 2

213
Spread of Smoke and Fire Gases via the Ventilation System

300.0 600.0
--1 u

-50.0 -100.0
0.0 500.0 0.0 500.0
n= 1 T1me (s) n= 1 Time (s)
3.0 ~-~--~----.--~--- 300.0
--1 u -'- 1 u

~
i
Q1
~
([)
>,
·a
_Q

0.0
500.0
n= 1 Time (s) 0.0 500.0
Time (s)

25.0 0.2
-- 1 u 02 -- 1 1-2
1-
u
"'- en
~
E CJ'
0.
0.
CS

"'-
c
~
~

0.0
0.0
0.0 500.0
Time (s) 0.0 500.0
n= 1 Time (s)
0 ..3 0.0
--1 L 1- 2 2 l 2- 1

"'
~
({) en
~
Q1 CJ'
6 CS

.E

0.0 ,___ _L___

0.0
____cL___ ___[_ __L_...::~-..1

500.0
0.0 L'===:=:r::====:r::::==:::::::r====:::r:===-
n= 1 Time (s) 0.0 500.0
n= 2 Time (s)

Dwelling house (sofa) 3

214
 Appendix

400.0 2500.0 , - - - , - - - - - - , - - - - - - , - - - - . . , . . - - - - ,
--1 u --1 u

0.0 -500.0
0.0 500.0 0.0 500.0
n= 1 Time (s) n= 1 Time (s)
1200.0
3.0
--1 u --1 u

~
i
_c
.'2'
1'
(\)
·a
>-
_Q

0.0 0.0
0.0 500.0 0.0 500.0
n= 1 Time (s) Time (s)

25.0 1.0
--
~ 21
I
--1 I)
I

"l
1-
u
-....._ m
-....._
E
Q_
-1 ~
Q)

Q_ I N

"-.. i -'-
c
1
I
(\)

2
(\)
Q_

-1.0 I
500.0
n= 1 Time (s) 0.0 500.0
n= 1 Time (s)
2.0 0.0
--1 L 1- 2 2 L 2- 1'

m
"-..
00
"-..
Q)

6 ~

.E .E

0.0 0.0
0.0 500.0 0.0 500.0
n= 1 Time (s) n= 2 Time (s)

Dwelling house (sofa) 4

215
Spread of Smoke and Fire Gases via the Ventilation System

400.0 2000.0 ,------.----.----,----.-_-_-..:::....,.,·

--1

i a.

0.0 -500.0
0.0 500.0 0.0 500.0
n= 1 Time (s) Time (s)

3.0 1200.0
--1 -- 1 u

~.2"'
1! I
iG ·o
>,
.Q

0.0 0.0
0.0 500.0 0.0 500.0
n= 1 Time (s) n= 1 Time (s)

25.0 1.0
-- 1 u 02 --1 u 1-2
1-
u
........ if)
........
0>
E
Q. 6
a.
........
~
c

2 .E
~

0.0 -1.0
0.0 500.0 0.0 500.0
n= 1 Time (s) n= 1 T"1me (s)
2.0
0.0
--1 L 1- 2 2 L 2- 1

if) if)
........ ........
0>
0>
6 6

.E

0.0 0.0
0.0 500.0 0.0 500.0
n= 1 Time (s) n= 2 Time (s)

Dwelling house (sofa) 5

216
 Appendix

400.0 2000.0
--1 --1

l a.

0.0 -500.0
0.0 500.0 0.0 500.0
n= 1 Time (s) n= 1 Time (s)

3.0 1400.0
--1

~.'2'
<1.)
s:.
I
<1.) ·a
:>-
.!2

0.0 0.0
0.0 500.0 0.0 500.0
n= 1 Time (s) n= 1 T'1me (s)

25.0

f-
I
u 0, 1.0
1 u 1- 2

"-. lf)
-......_
E
Q_
a.
6
"-.

c
<1.)
0
Q;
a.

0.0
-1.0
0.0 500.0
n= 1 Time (s) 0.0 500.0
n= 1 T'1me (s)
2.0 0.1
1L 1- 2 2 L 2- 1

--:;; --:;;
-......_ "-.
0'
6 .:s

~
.E

0.0
500.0 0.0 500.0
Time (s) n= 2 Time (s)

Dwelling house (sofa) 6

217
Spread of Smoke and Fire Gases via the Ventilation System

400.0 ,----,-----,-----,-----_-_,_---::-u:-:--1 2ooo.o ,----,-----.---.----:_-_,_--:-

1 -:-:u"
1

0.0 -500.0
0.0 500.0 0.0 500.0
n= 1 Time (s) Time (s)
3.0 2000.0
-'-1 u -- 1 u

~ .Q' I
1:
'a
a:>
>,
_Q

0.0
0.0
0.0 500.0
0.0 500.0 n= 1 Time (s)
n= 1 Time (s)

25.0 1.0
--1U02 -1u 1- 2
r-
0

'--... -;;;-
'--...
ry
E
Q_
Q_
6
'--...

c
Ql
2
Ql
Q_

-1.0
500.0
n= 1 Time (s) 0.0 500.0
n= 1 Time (s)
2.0 0.1
--1 L 1- 2 2 L 2- 1

(;)
'6; z ry
6 6

.E

0.0 0.0
0.0 500.0 0.0 500.0
n= 1 Time (s) n= 2 Time (s)

Dwelling house (sofa) 7

218
 Appendix

250.0 500.0

--1 u
0.0 c_l- - - - ' - - - - - ' - - - - - ' - - - - - ' - - - - - - - '
0.0 500.0
n= 1 T1me (s) 0.0 500.0
n= 1 Time (s)

3.o .---.-----.-----.-----.::;::==-----=-,-:-:lu 200.0

·a

0.0
0.0
0.0 500.0
n= 1 Time (s) 0.0 500.0
n= 1 Time (s)

22.0 0.1
-' 1 U021 I 1()l 'F2 I

1-
u
""- Ul
""-
ry
E
Q. 6
Q_

""-
c >

~ .E
~

6.0 00
0.0 500.0 0.0 500.0
n= 1 Time (s) n= 1 Time (s)

0.1 03
--1L 1- 2 2 L 2- 1

Ul
"--,
0'
<ry
6 6

.E .E

0.0
0.0
,\
0.0 500.0
o.o n= 1 Time (s)
500.0 n= 2 Time (s)

Dwelling house (bed) 8

219
Spread of Smoke and Fire Gases via the Ventilation System

250.0 200.0
-1

1
Q
0

~
!
0.0 L _ __ ___L_ _ _ L _ _ _. . J __ _. . L __ ___;
--1 u ~
0.0 500.0 0.0 500.0
n= 1 Time (s) n= 1 Time (s)

500.0
--1

1
·a

0.0
500.0 0.0 500.0
n= 1 Time (s) n= 1 Time (s)

22.0 1.0
--1 u 'I- 2
f-
u
~ Ul
~
C)1
E
Q_ ~
Q_

c
Q)
~
.E
2
Q)
Q_
-- 1 u 02
8.0 -1.0

0.0 500.0 0.0 500.0

n= 1 Time (s) n= 1 T'1me (s)
0 ..3 1.0
-- 2 L 2- 1

Ul Ul
~ ~
C)1 C)1
~ ~

>
.E .E

- 2
0.0 0.0
0.0 500.0 0.0 500.0
n= Time (s) n= 2 Time (s)

Dwelling house (bed) 9

220
 Appendix

250.0 200.0
--1'U

0
s
I
2::-
-
~
Q_

--1 u
OD
0.0 500.0 0.0 500.0
n= 1 Time (s) n= 1 Time (s)
3.0 500.0
--1

~ .<:2'
2 1
(jj ·o
>,
_Q

0.0
0.0 500.0 500.0
n= 1 Time (s) n= 1 Time (s)

22.0 1.0
1 'u

~:
f-
0

""-. (/)
""-.
V'
E
Q_ 6
Q_

......_

c
<lJ !
(.)
.E
v
Q_
--1U02
8.0 -1.0
0.0 500.0 0.0 500.0
n= 1 Time (s) n= 1 Time (s)
0.3 1.0
--2 L 2- 1

(/)
~ry
""-.
V'
6 6

>
.E .E

0.0 0.0
o.o n= 1 Time (s)
500.0 0.0
n= 2 Time (s)
500.0

Dwelling house (bed) 10

221
Spread of Smoke and Fire Gases via the Ventilation System

250.0 200.0
--1

--1 u
0.0 '--------'-----'----'-------'------'
0.0 500.0 0.0 500.0
n= 1 Time (s) n= 1

3.o c - - - r - - 1~-'1 ul 500.0

'? J
~
1'
;I

1...
·o
<lJ
A
.2

0.0
0.0 '---------'-----'----'-----....J_-------1 o.o 500.0
n= 1 Time (s)
0.0 500.0
n= 1 Time (s)

22.0 lO
-- 1 u 1-2
f-
u
(;)
~
Ql
6

c >
<lJ
(.) .E
Q;
0.

g_o ' - - - - - - - L - - - - - ' - - - - - ' - - - _ _ l _ - - - - 1 -1.0

0.0 500.0 0.0 500.0
n= 1 Time (s) n= 1 Time (s)
1.0
-- 2 L 2- 1

(;)
~
Ql
6

.E .E

o_o ""-------'-----'------'-------'------'-' 0.0

0.0 500.0 0.0 500.0
n= 1 Time (s) n= 2 Time (s)

Dwelling house (bed) 11

222
 Appendix

200.0 100-D

--1 u
0.0 -500.0 t _ __ ____j__ _ _--l-_ __ j_ _ __ J

r--
0.0 800.0 0.0 800.0
n= 1 Time (s) n= 1 Time (s)
3D
500.0 1 u,

~
.2'
~
~ 'o
>,
_Q

0.0 0.0
0.0 800.0 0.0 800.0
n= 1 Time (s) n= 1 T'1me (s)

22.0 1.0
-- 1 u 1- 2

Ul
"--..
E
0.
0.
J£ N

c
([)

~ .E
([)
0.

8.0 L__----'-------'-------'-------___j -10

0.0 800.0 0.0 800.0
n= 1 Time (s) n= 1 Time (s)
1.0
-- 2 L 2- 1

Ul
"--..
CJl
6

.E
.E
--1L 1-2
0.0
0.0
~ _ ____[__ _ __L__-\;;;;;=~=;;;;--__J
800.0
0.0 1---------.J' "---
c___ ____j__ _ _--L-_ _ _c _ __ ____j

n= 1 Time (s)
0.0 800.0
n= 2 : 1me (s)

Dwelling house (bed) 12

223
Spread of Smoke and Fire Gases via the Ventilation System

300.0
__ , u 200.0

0
D..

I ~
0.

--1 u
0.0 L _ _ . . . . J __ ___t._ _ j __ _L...._ __j__ _

-300.0 '-------'-----'----'-------'-----'--~
0.0 600.0
n= 1 Time (s) 0.0 600.0
n= 1 Time (s)

0.0 L _ _ . . . . J __ __j__ _ j __ _ j __ _ _ j __ _

0.0 600.0
n= 1 Time (s) 0.0 600.0
n= 1 Time (s)

22.0
- ' - 1 1.! 1- 21
1-
u 021

"-
E
z
o>
0. 6
0.

c
(l)

2 .E
(l)
0.

8.0 L _ _ . . . . J __ ___t._ _ j __ _ j __ ____l_ _

0.0 L..__ _L__ __L_~======___j

0.0 600.0 o.o 600.0

n= 1 Time (s) n= 1 Time (s)
1.0
L 2- 1

.E .E

0.0 t===:::c:==::i._ _j-===:c::==::::c=-_j 0.0 I...---'-----1..--I--.JI-..__ __.._ __J

0.0 600.0 0.0 600.0

n= 1 Time (s) n= 2 Time (s)

Hospital13

224
 Appendix

250.0 200.0

--1 u
0.0 ' - - - - - ' - - - - ' - - - - - - - ' - - - - ' - - - - - - ' - - _ _ _ _ j -600.0 L...__ _ j __ _. L __ _ j __ __L__ ___L_ _

0.0 600.0 0.0 600.0

n= 1 Time (s) n= 1 Ttme (s)
3.o .-----.---.-------,---.--=::::.::.---:,-:-:-u 250.0
-'-1 u1

'o

0.0 L __ _ L_ _. L __ _ j __ __L__ _..L_ ____j 0.0 L __ _ L_ _L___ _J__ _ j _ __ _..L_ __j

0.0 600.0 0.0 600.0

n= 1 Time (s) Time (s)

22.0 0.2 ,---,---.-------,--_._-..:::-:,-t.J,..,----:1-~2

--1U02
f--
0

E
Cl
Cl

c
(l)
I
()

Q:;
Q.

8.0 L __ __.L_ __L__ ___L _ _...J___ ___L_ _

)
0.0
0.0 600.0 0.0 600.0
n= 1 Time (s) n= 1 Time (s)
1 L 1- 2
0.2 r---,----,---,-----.--:.._,;.-----'---=- 1.0 --2 L 2- 1

zry
~

~
~
.E .E

0.0 L __ __.L_ __L__ ___L _ _ _ L __ ___L_ _

0.0
0.0 600.0 0.0 600.0
n= 1 Time (s) n= 2 Time (s)

Hospital14

225
Spread of Smoke and Fire Gases via the Ventilation System

200.0 140.0
--1U --1 u

0.0 -20.0
0.0 400.0 0.0 400.0
n= 1 Time (s) n= 1 Time (s)

4.0 .300.0

I .s::
-"'
1!
([)
>--
.Q

0.0 0.0 """""--__J.__ _ _ -.i._..._ _ _c___ __ j

0.0 400.0 0.0 400.0

n= 1 Time (s) n= 1 Time (s)

22.0 0.2
--1 u '1- 2
1-
u

""'E
Q
Q

c
I])

2
~ --1U02

12.0 0.0
0.0 400.0 0.0 400.0
n= 1 Time (s) n= 1 Time (s)
1.0 0.2
--1L 1-2 -- 2 L 2- 1

';;)

6""'
I

"'
~
.E

0.0 >-""""''------'------'-...--'---"-==-d 0.0

o.o n= 1 Time (s)
400.0 0.0
n= (s)
400.0
2 Time

Hospital15

226
 Appendix

150.0
250.0 I--1U I --1 u

0.0 -100.0
0.0 400.0 0.0 400.0
Time (s) n= 1 Time (s)
300.0

0.0 0.0
0.0 400.0 0.0 400.0
n= 1 Time (s) n= Time (s)

22.0 0.3
1-- 1 U I 1- 2
I
1-
u
(i)

I
"'-._
"'-._ I
01
E
0.
0.

"'-._
-:::s
r
I

c
<!)
~
u .E
a;
0.

8.0
r- 1 u 02
0.0
0.0 400.0 0.0 400.0
n= 1 Time (s) n= 1 Time (s)

0.2 0.3
2 L I 2- 1

(i) if)
"'-._ "'-._
01 01
-:::s -:::s

~
.E .E
1 L 1- 2
0.0 0.0
0.0 400.0 0.0 400.0
n= 1 Time (s) n= 2 Time (s)

Hospital 16

227
Spread of Smoke and Fire Gases via the Ventilation System

.300.0 100.0
--1 u

1 u
0.0 ' - - - - . . . . . . l . . - - - - ' - - - - - - - - ' - - - . . . . J -400.0 t _ __ _.......L._ _ _..J___ __ _ t_ _ _ _ ....J

0.0 400.0 0.0 400.0

n= 1 Time (s) n= 1 Time (s)
.300.0

0.0 :.--=---.......L.----'--------'-='----....J
0.0 400.0
n= 1 Time (s)

22.0 0.2 r - - - - - - , - - - - - - , - - - - - , - - - - - - ,
- - 1 U 1-2
f-
u

c
([)
u .E
Q;
Q.

8.0 l _ __ ___j__ _ __j___ _ _.t___ __ _ j 0.0 ~--....J..---...1...--'---1...1..-----'

0.0 400.0 0.0 400.0
n= 1 Time (s) n= 1 Time (s)

0 ..3 r------,..----,----,-----~ 1.0

--1L 1-2 --2L 2-1

.E

0.0 ).£------'------'-------\;====;;;;;;;;;! 0.0 .___ ___.__ _ __.__ _...._.t_____l_ ___.

o.o n= 1 Time (s)

400.0 0.0 400.0
n= 2 T"1me (s)

Hospital17

228
 Appendix

400.0
-- 1 u
60.0
-- 1 u

s ~
0

-
~
()_

0.0 -10.0
0.0 600.0 0.0 600.0
r.= 1 Time (s) n= 1 Time (s)
3.0 200.0
-- 1 u

~-~
.c
(!) I
(!)
·o
>,
_Q

0.0 0.0
0.0 600.0 0.0 600.0
n= 1 Time (s) n= 1 Time (s)

25.0 1.0 I

~ 21
-- 1 u 02 1 u
1-
()

'--... In
'--...
(J1
E
0. 6 I
a. I
'-....

c (!)

2 .E
(!)
Q.

0.0 -1.0
0.0 600.0 0.0 600.0
n= 1 Time (s) n= 1 Time (s)
0.2 0.0
- '- 1 L 1- 2 1- 2 L 2.!. 1

~
({)
-------
'--...
!£ !£

.E

0.0
o.o n= 1 Time (s)
600.0
.E

J 0.0
n= 2 Time
~
(s)
I

600.0

Office 18

229
Spread of Smoke and Fire Gases via the Ventilation System

300.0 250.0
--1U 1 u

0
s
~
~
Q)
Q

0.0 -50.0
0.0 800.0 0.0 800.0
n= 1 T1me (s) n= 1 Time (s)
3.0 200.0
-- 1 u

~
.'2' 'i'
-""
~
<!.)
·o
>-
S2

0.0
800.0 0.0 800.0
Time (s) Time (s)

0.1
25.0
-- 1 u 02 1 u 1-2
1-
0
~
'--...._ '--...._
CJ'
E 6
Cl
o_

'--...._

(
c
Q)

2
Q)
Q

0.0 0.0

0.0 800.0 0.0 800.0

n= 1 Time (s) n= 1 Time (s)
0.2 0.0
1 L 1- 2 2 L 2- 1

"'--...._
(j) ~
'--...._
0' CJ'
6 6

.E .E

~
I

0.0 o.o I
0.0 800.0 0.0 800.0
n= 1 Time (s) n= 2 Time (s)

Office 19

230
 Appendix

400.0 140.0
-- 1 u

2
0 'l
-
1
0.0 0.0
0.0 800.0 0.0 800.0
n= 1 Twne (s) n= 1 T"1me (s)
3.0 200.0
-- 1 u

~.'2'
1! l
i>,
D ·o
.2

0.0 0.0
0.0 800.0 0.0 800.0
n= 1 Time (s) n= 1 Time (s)

25.0

1-
0
-- 1 u 02
0.1
.- 1 U I 1- 2

if)
""-
""-
CJl
E
0. 6
Q_
I

""-
I I

!l
c

J
Q)
.E
2
Q)
Q_

0.0
0.0 800.0 0.0 800.0
n= 1 Time (s) Time (s)

0.2 0.0
1 L 1- 2 --2 L 2- 1

(;; [f)

""-
ry ""-
CJl
CS 6
N

.E .E

0.0 0.0
0.0 800.0 0.0 800.0
n= 1 Time (s) n= 2 Time (s)

Office 20

231
Spread of Smoke and Fire Gases via the Ventilation System
40.0
.350.0
-- 1 u

s ~

- i::.'Q.

0.0 0.0
0.0 600.0 0.0 800.0
n= 1 Time (s) Time (s)
.3.0 250.0
-- 1 u

I ~ I
t>-. ·a
.2

0.0 0.0
0.0 600.0 0.0 600.0
n= 1 Time (s) n= 1 T"1me (s)

25.0
1.0
- - 1 u 02 1 u 1- 2
1-
0
'-...._
if)
'-...._
E Ql
D.
D. 6
""'-
c
([)

2
([)
D.

0.0 -10
0.0 800.0 0.0 800.0
n= 1 Time (s) n= 1 Time (s)
0.2 0.0
--1L 1- 2 2 L 2- 1

{j)
'-...._
0'
z Ql
6 6
N

.E .E

0.0
800.0 0.0 800.0
n= 1 Time (s) n= 2 Time (s)

Office 21

232
 Appendix

300.0 __ , u 5=.o

!\ I
0
~
2
-
~
Q.

-- 1 u
0.0
0.0 800.0 0.0 800.0
n= 1 Time (s) n= 1 T"1me (s)

250.0
3.0
-- 1

I "'
.c
I
'o
"'>,
2

0.0
800.0 0.0 800.0
n= 1 Time (s) n= 1 Time (s)

25.0 0.0
-- 1 u 02 --1U 1-2
f-
0

E
0.
0.

I
c
"'u .E
G>
0.

0.0 L___ _ _L __ _ _L __ _ _ j _ __ __ _ j
0.0 .!...~~~.b.~~~.b.~~~.b.~~.J.-...J
0.0 800.0 0.0 800.0
n= 1 Time (s) n= 1 Time (s)

o. 3 r----,..-----,------.--:-1--:L-...,.1--2"' 1.0 j__ 2 L I 2- 1

.E .E

0.0 0.0 .J...~..J--I....b.~~~.b.~~~.t....~~.......J

0.0 800.0
0.0
n= 1 Time (s)
800.0 n= 2 Time (s)

Office 22

233
Spread of Smoke and Fire Gases via the Ventilation System

300.0 30.0
--1 u

1 u
0.0 -5.0
o.o 800.0 0.0
(s)
800.0
n= 1 T1me (s) n= 1 Time
3.0 300.0
--1 u

t.'2'
(]) I
.c

:u ·a
>-.
..Q

0.0 0.0
o.o 800.0 0.0
(s)
800.0
n= 1 Time (s) n= 1 T'1me

25.0
1.0

f-
-- 1 u 02 --1 u 1- 2
u
'--.._
Ul
'--.._
E en
D.
D. 6

"""c
(])

2
(])
D.

0.0 -1.0
0.0 800.0 0.0 800.0
n= 1 Time (s) n= 1 1'1me (s)
0.0
0.2
-- 1 L 1- 2 2 L 2- 1

Ul Ul
'--.._ '--.._
ry en
6 6

>
.E

0.0
800.0 0.0 800.0
n= 1 Time (s) n= 2 Time (s)

Office 23

234
 Appendix

r
250.0 100.0

0
I
11.

l Q_

-- 1 u
0.0 -400.0

0.0 800.0 0.0 800.0

n= 1 Time (s) n= 1 T1me (s)
3.0
--1 u
300.0
-- 1 u

~
~
ry

~
Q) 0
>--
.2

0.0
0.0 800.0 800.0
n= 1 Time (s) n= 1 Time (s)

25.0 1.0
--1U02 -- 1 u 1- 2
f--
0

-........ --;;;-
-........
ry
E 6
0.
0.

-........

~
~ .E
Q)
0.

0.0 -1.0
0.0 800.0 0.0 800.0
n= 1 Time (s) n= 1 Time (s)
0.3 1.0
1 L 1- 2 2 L 2- 1

--;;;- --;;;-
-........ -........
ry ry
6 6

.E

0.0 0.0
0.0 800.0 0.0 800.0
n= 1 Time (s) n= 2 Time (s)

Office 24

235
Spread of Smoke and Fire Gases via the Ventilation System

250.0 50.0
~
--1U
V
0
I
2 0..

- 1 ~
0..

-- 1 u
0.0 -300.0
0.0 800.0 0.0 800.0
n= 1 Time (s) n= 1 T"1me (s)
3.0 400.0
--1 u

~.'2'
.L
([) i
0:; ·o
>,
.Q

0.0 0.0
0.0 800.0 0.0 800.0
n= 1 Time (s) n= 1 Time (s)

25.0
_'_ 1 u 021 1.0
--1 u 1-2
f-
u
"-... en
"-...
E
0.
0..

"-...
j 6
0"'

c
([)
0
v
0..

0.0 -1.0
0.0 800.0 0.0 800.0
n= 1 Time (s) n= 1 Time (s)
1.0 1.0
--1 L 1- 2 --'2 L 2- 1

V; V;
"-... "-...
0"' 0"'
6 6
I

"'
.E

0.0
800.0 0.0 800.0
n= 1 Time (s) n= 2 Time (s)

Office 25

236
 Appendix

300.0
--1 u
200.0

s
0
,

-
__ )

0.0
0.0
n= Time (s)
600.0
-OO~J'------'-~1.
0.0
n= 1 Time (s)
600.0

400.0 3.0
--1

·a

0.0 JL__ _ _ _ j _ __ _~_....__ _ _- - J

0.0 600.0 600.0

T1me (s) n= 1 Time (s)

0.0 t=====""'-___.,=======~
0.0
n= 1 Time (s)
600.0
I
0.2 1.0
--1L 1-2 2 L 2- 1

~~
.E .E

0.0 ' - - - - - - ' - - - - - l . . - " " " " ' ' - _ L _ - - - - - - ' 0.0 '------'
0.0 600.0 0.0 600.0
n= 1 Time (s) n= 2 Time (s)

Hotel26

237
Spread of Smoke and Fire Gases via the Ventilation System

300.0 200.0

0
Q s
-
~
Q

--1 u
0.0
0.0 500.0 0.0 500.0
n= 1 Time (s) n= 1 Time (s)
4.0 1000.0
--1U

~
.Q'
1! I
(l) 'o
>,
_Q

0.0
500.0 0.0 500.0
n= 1 Time (s) n= 1 Time (s)

25.0 1.0
-- 1 u 02 1 u 1-2
1-
u
......... (/)
.........
ry
E
Q 6
Q

.........

c
(l)

f
2 .E
(l)
Q

5.0 0.0
0.0 600.0 0.0 600.0
n= 1 Time (s) n= 1 rime (s)
0.3
1.0
--1 L 1- 2 --'- 2 L 2- 1

(;) (/)
......... .........
CJ1 ry
6 6
.'-

>
.E

0.0
0.0
n= 1 Time (s)
600.0
0.0
0.0
n= 2 Time
\ (s)
600.0

Hotel27

238
 Appendix

350.0 1000.0
--1 u

--tu

l
600.0 0.0 600.0
n= 1 Time (s) n= 1 Time (s)

4D ,--------,-------,---_-_-..:::--:,---,u:-:1
800.0

0.0 OD
0.0 600.0 0.0 600.0
n= 1 Time (s) n= 1 Time (s)

25.0 1.0
I
--1 --1 u 1-2
f-
0

"'--.
"l "'
(/)

"'--.
Cl'
E

I
Q_ 6
Q_

"'--.

c
~
~
Q_
I
.E n
0.0 0.0 I
0.0 600.0 0.0 600.0
n= 1 Time (s) n= 1 Time (s)

1.0 2.0
1 L 1- 2 -- 2 L 2- 1

if;' if)
"'--. "'--.
6
01
£

>
.E .E
I

0.0 o.o I
0.0 600.0 0.0 600.0
n= 1 Time (s) n= 2 Time (s)

Hotel28

239
Spread of Smoke and Fire Gases via the Ventilation System

300.0 I 1000.0

¥ I
"
~
[J)
[J)
Q)

0..

0.0
~~,)
.. -
1 u

0.0 600.0
n= 1 Time (s) 0.0 600.0
n= 1 Time (s)
1200.0 4.0 , - - - - - - - - , - - - - - - , - - _------:-1-:-:lU

·a

0.0 0.0
0.0 600.0 0.0 600.0
n= 1 Time (s) n= 1 Time (s)

22.0 1.0

f-
0
(;)
""'E ""'
CJl
0. ~
0.

""'
c
Q)

2
Q)
0.
--1 u 1-2
6.0 0.0
0.0 600.0 0.0 600.0
n= 1 Time (s) n= 1 T"1me (s)
2.0 3.0
--1 L 1- 2 2 L 2- 1

(;)
~
6
CJl ""'
~
CJl

.E

0.0
600.0 0.0 600.0
n= 1 Time (s) n= 2 Time (s)

Hotel29

240