Design approaches for smoke control

in atrium buildings

G 0 Hansell*, BSc, PhD, CEng, MCIBSE, AlFireE

H P Morgan, BSc, CPhys, MlnstP, AlFireE

*Colt International Limited

Fire Research Station

Building Research Establishment
Borehamwood, Herts
WD6 2BL
 .! :
~ .. - __ . ~ ~~ . ~ - .. -

Prices for all available

BRE publications can be
obtained from:
BRE Bookshop
Building Research Establishment
Garston, Watford, WD2 7JR
Telephone: 0923 664444
Fax: 0923 664400

BR 258
ISBN 0 85125 615 5

0 Crown copyright 1994

First published 1994

Applications to reproduce extracts

from the text of this publication
should be madc to the Publications Manager
at the Building Research Establishment, Garston Cover illustration by Bob Stoneman
Contents
Page

1Foreword V

Nomenclature vi

Introduction 1

Chapter 1 General principles of smoke production, movement and control 4

Fire growth and smoke production 4
Pressurisation 5
Depressurisation 6
Throughflow ventilation 6

Chapter 2 Design fire size 7

Chapter 3 Smoke control on the floor of fire origin 9

Within the fire room 9
Flow of hot gases out of the room of origin into the atrium 10
Ventilation of the balcony space 13
Smoke layer temperature 14
Effects of sprinkler systems in smoke reservoirs 14
Flowing layer depth 16
Local deepening 17
Inlet air 17
Minimum number of extraction points 18
Required ventilation rate (powered exhaust) 19
Slit extract 19
False ceilings 20
The use of a plenum chamber above a false ceiling 20

Chapter 4 Smoke ventilation within the atrium 21

Smoke movement in the atrium 21
Channelling screens 22
Entrainment into spill plumes rising through the atrium 23
Fires on the atrium floor 33
Throughflow ventilation - area of natural ventilation required 33
Throughflow ventilation - remaining design procedures 35
Limitations to the use of throughflow ventilation 35

Chapter 5 Design considerations other than throughflow ventilation 37

Void filling 37
Compartment separation 37
DepressurisatiEn ventilation 37
Principles 37
Natural depressurisation 38
Natural depressurisation and wind effects 40
POwered depressurisation 41

Chapter 6 Depressurisatiodsmoke ventilation hybrid designs 42

Principles 42
Design procedures for hybrid systems 42
Mass jZo w based systems 42
Temperature based systems 43

(continued)
...
I11
 Page

Chapter 7 Atrium smoke layer temperature 44

Chapter 8 Additional design factors 46

Atrium roof-mounted sprinkler systems 46
Smoke detection systems in the atrium 46
Pressurisation of stairwells and lobbies 46
Air-conditioned atria 46
Channelling screens and hybrid systems 46
Wind-sensing devices and natural depressurisation 47

Appendix A Case history 48

Appendix B Users guide to BRE spill plume calculations 49

Introduction 49
Scenarios and assumptions 49
Outline of procedure 50
Detailed procedure 50

Nomenclature used in Appendix B 54

Acknowledgements 55

References 55

iv
 I -

1 .

!. . . . . ~.

This Report is the culmination of a long-running collaborative project between the Fire
Research Station of the Building Research Establishment and Colt International Limited
on aspects of smoke movement and its control in atrium buildings. It is based on both the
latest scientific knowledge and practical experience of smoke movement and control
systems, and has been prepared under the overall supervision of the Fire Research
Station.

T h e present Report is intended to serve the designers of smoke control systems for atrium
buildings in the same way that the earlier Building Research Establishment Report,
Design principles f o r smoke ventilation in enclosed shopping centres, has served designers
of smoke ventilation systems in shopping malls. As such, those graphs and tables it
contains which are relevant to a particular design of building can be applied directly to
that building; or the formulae cited can be used to apply the work to a broader range of
circumstances.

The Report does not exclude the options of using alternative methods where they are
appropriate, or of using new techniques (such as computational fluid dynamics) as they
are developed and validated.

A P S Ferguson
Building Regulations Division
Department of the Environment

July 1993

V
Nomenclature
Note: The list of nomenclature used in Appendix B is given on page 54.

Area of the fire (m2)

Area of inlet (measured) (m2)
Area of exhaust ventilator (measured) (m2)
Area of opening into atrium from adjacent fire room (m2)
Specific heat of air (kJkg-IK-')
Coefficient of discharge for a vertical opening
Entrainment coefficient
Coefficient of discharge for an inlet
Wind pressure coefficient acting on an inlet
Wind pressure coefficient acting on the leeward side of building
Wind pressure coefficient acting on an exhaust ventilator
Coefficient of discharge for an exhaust ventilator
Depth of smoke beneath an extraction point (m)
Depth of a smoke layer under a balcony (m)
Depth of a downstand fascia (m)
Diameter of fire (m)
Design depth of a smoke layer in a reservoir (m)
Depth of a flowing smoke layer in a vertical opening (m)
Maximum depth of smoke in an atrium (m)
Acceleration due to gravity (ms-2)
Height of a vertical opening (m)
Height of a vertical opening with no upstand (m)
Height of rise of a thermal line plume from an opening or balcony edge to the smoke layer (m)
Height of the atrium (m)
Height to the ceiling (m)
Channelling screen separation (m)
Mass flow rate (kgs-')
Mass flow rate from the fire (kgs-')
Mass flow rate under a balcony (kgs-')
Mass flow rate entering (leaving) a smoke layer in a reservoir (kgs-I)
Mass flow rate flowing through a vertical opening (kgs-')
Critical exhaust rate (kgs-l)
Number of exhaust points
Perimeter of fire (m)
Heat flux (kW)
Convective heat flux of fire (kW)
Convective heat flux passing through a vertical opening (or under a balcony) (kW)
Convective heat flux per unit fire area (kWm-2)
Absolute temperature of gases (K)
Absolute temperature of gas layer under a balcony (K)
Absolute temperature of gas layer in a reservoir (K)
Absolute ambient temperature (K)

vi
 1
' ..
7.

V Volumetric flow rate of gases (m3s-1)

VI Volumetric flow rate of gases from a reservoir (m3s-I)
"wind Design wind velocity (ms-')
W Width of vertical opening (m)
WB Width of balcony (distance from vertical opening to front edge of balcony) (m)
X Height from the base of the smoke layer to the neutral pressure plane (m)
Y Height from the base of the fire to the smoke layer immediately above (m)
P Coefficient in critical exhaust rate equation (kgrnp3)
A Empirical height of virtual source below a balcony edge (m)
ADB Additional smoke depth due to local deepening (m)
e Temperature rise above ambient of smoky gases ("C)
@B Temperature rise above ambient of smoky gases under a balcony ("C)
01 Temperature rise above ambient of smoky gases in a reservoir ("C)
Temperature rise above ambient of smoky gases in a vertical opening ("C)
P Density of gases (kgm-3)
PO Density of ambient air (kgmW3)

vii
 . .< . .. . .

Introduction
This Report is intended to assist designers of smoke have been described as malls, atria, arcades and
ventilation systems in atrium buildings. Most of the lightwells. The generic term for the building type tends
methods advocated are the outcome of research into to be ‘atrium’ and, by their very nature, atrium buildings
smoke movement and control at the Fire Research often run contrary to the traditional Building
Station (FRS), but also take into account experience Regulations’ approach in terms of horizontal
gained and ideas developed whilst the authors and their compartmentation and vertical separation.
colleagues have discussed many proposed schemes with
interested parties. The primary purpose of the Report is The original atrium was an entrance hall in a Roman
to summarise in a readily usable form the design advice house and was one of the most important rooms in the
available from FRS at the time of its preparation. As building. The concept of this space has evolved
such, it does.not attempt to cover installation, detailed architecturally over the past few hundred years and now
specification of hardware, or aspects of fire safety applies to structures much larger than the typical Roman
engineering other than smoke control. house. Atria today are designed as undivided volumes
within a structure, intending to create visually and
The predominant cause of death in fires in the UK has spacially an ideal external environment -indoors6.
been attributed to the inhalation of smoke and toxic
gases’, and the annual number of fire fatalities in the UK In Roman times the control of any smoke and hot gases
is approximately one thousand. However, the majority that may have issued from a fire in a room adjacent to
of these deaths occur in domestic premises. This implies the atrium was likely to have been a simple matter.
that the life-safety measures required by legislation for Providing there were no adverse wind conditions (due to
most public and commercial buildings have been local topography of adjacent structures), then the smoke
effective on the whole. and heat would undoubtedly vent through the open
portion of the atrium roof known as the ‘compluvium’
Fire safety in buildings must in the UK conform to the (generally used for lighting purposes).
relevant regulations (guidance for England and Wales is
given in Approved Document B2). The principal Modern atrium buildings tend to contain large quantities
objective of these regulations is to safeguard life by: of combustible material and often have open-plan
@ reducing the potential for fire incidence, layouts which increase the risk of the spread of fire. The
populations within such buildings are also greater; hence
Q controlling fire propagation and spread, and there has been a substantial increase in the number of
8 the provision of adequate means of escape for the people to be protected and evacuated in an emergency.
building’s occupants.
Modern atrium buildings are usually designed with the
Means of escape in case of fire was first introduced to atrium as a feature which can be appreciated from
the Building Regulations for England and Wales in 1973. within the adjacent rooms. The room/atrium boundary is
Prior to that date, the powers of control in England and usually either glazed or completely open. Thus when
Wales over means of escape had been contained in other compared to ‘conventional’ buildings, this
legisIation~-‘‘,s. architecturaVaesthetic requirement imposes additional
problems of life safety during a fire, since smoke, hot
Historically, the prevention of fire growth within (or gases and even flames may travel from one (or more)
between) buildings has been achieved by the rooms into the atrium and thence affect areas which
containment of the fire and its products by means of would have remained unaffected in the absence of an
compartmentation and/or separation. The design of atrium.
structural compartmentation and separation has been
largely empirical, and the concepts gradually refined and In conventional multi-storey structures there is always
enhanced in such a way that the Building Regulations the possibility of fire-spread up the outside of the
now cover primarily life safety and the protection of building, with flames issuing from one room and
means of escape. It is necessary to consider four major affecting the floors above. Recent examples of this mode
aspects of buildings - purpose, size, separation and of fire-spread have been an office block in Siio Paulo7
resistance to fire - to promote safe design. and the Villiers building in London. If the escape
facilities from the various rooms are of a suitable
Social and technical changes have led to changes in standard and segregated from other compartments (as
building environments which incorporate new (or required in the UK), there should not (in theory) be any
revived) building forms and the use of innovative serious hazard to life safety in this fire condition. It is
construction techniques and new synthetic materials. only when the means of escape are inadequate or the
The buildings adopting these changes often have parameters dictating their design are violated, that the
included within their design large spaces or voids, often loss of life may occur.
integrated with many of the storeys. These large spaces

1
If a building has an atrium then this fire condition can Where smoke from a fire in a room can spread into the
also occur internally, since there is generally a atrium, with the possibility of rapid further spread
maximisation of the window area and/or open boundary affecting other parts of the building, there will be an
between the rooms and the atrium. Hence there is an extreme threat to safe evacuation of occupants from
increased risk to other levels of the entry of smoke, those parts. Similar threats will occur if there is a serious
toxic gases and possibljl flames from a fire. fire in the atrium space itself. In either case, the threat
to means of escape which are either within the atrium or
Recent experience of fires in atrium buildings in the in spaces open 1.0 the atrium, can develop rapidly unless
has shown the problem of flame travel internally some form of smoke control is used in the atrium in
through the atrium to be minor in comparison to that of order to protect those means of escape. In other words
hot and toxic gases accumulating and building down in a smoke control system in the atrium is essential to
the atrium - spreading throughout the building and make certain that escape is unhindered, by ensuring
affecting escape routes. Thus there appears to be a need that any large quantities of thermally buoyant smoky
for a properly designed smoke control system in atrium gases can be kept separate from people who may still be
buildings. using escape routes, or awaiting their turn for
evacuation. Therefore the role of a smoke control
The ideal option would be to prevent any smoke from a system is principally one of life safety.
room fire entering the atrium at all. An easily
understood way of achieving this is to ensure that the It should however also be remembered that fire-
boundary between the room and the atrium is both fighting becomes both difficult and dangerous in a
imperforate and fire-resisting, and that the atrium base smoke-logged building. It follows that to assist the fire
has only a very restricted use. This option has services, the smoke control system should be capable of
frequently been used, but is widely regarded as being performing its design function for a period of time
architecturally restrictive. Consequently it is not often longer than that required for the public to escape,
favoured by designers. The concept has been labelled allowing a speedier attack on the fire to be made after
the ‘Sterile tube’6. the arrival of the fire service.

Where the boundary between the room and the atrium There has been no readily usable guidance available to
is open, it is sometimes feasible to provide a smoke designers of atrium smoke control systems within the
ventilation system within the room, to maintain smoky UK. There have been a number of purely qualitative
fire gases above the opening to the atrium. papers, as well as papers on work using relatively simple
Unfortunately it is often very difficult, impracticable, or models of smoke movement within atria (see for
extremely expensive to fit a separate smoke extraction example References 10-12). The National Fire
system to each and every room, however small. Protection Association of the USA has recently been
Occasionally circumstances dictate that smoke control developing a Code13which sets out a fire engineering
dedicated to each room in this way is the most viable approach to the design of smoke control for atria
option for protecting the atrium (this can occur, for (termed ‘smoke management’ in the USA). While this
example, when the room layout is of a large area, is code is in many ways very comprehensive and broader
predominantly open-plan and open-fronted). There in purpose than the present Report, some of the
have been several examples of this. Nevertheless it approaches used differ from alternatives with which UK
remains generally true that this option is rarely found to designers are more familiar, or are more approximate
be appropriate for most atrium buildings. than methods currently used by the Fire Research
Station. This particularly applies to smoke entering the
Another possibility is that the atrium should be atrium from adjacent rooms.
pressurised to prevent smoke moving into it from a
room. This is not usually a viable option where the The purpose of this Report is to provide guidance to
opening between the room and the atrium is large (for assist designers of smoke control systems in atrium
example, an open-fronted room or a room whose buildings in line with current knowledge. The guidance
glazing has fallen away in whole or in large part). This is is based on results of research where possible, including
because the air speed needed from the atrium into the as yet unpublislied results of experiments, but also on
room in order to prevent the movement of smoky gases the cumulative experience of design features required
the other way through the same opening, can vary for regulatory purposes of many individual smoke
between about 0.5 ms-l and approximately 4 ms-1 control proposals. Many of these design features have
depending on gas temperature, etc. All of this air must been evolved over a number of years by consensus
be continuously removed from within the fire room in between regulatory authorities, developers and fire
order to maintain the flow. The quantities of air- scientists, rather than by specific research. Such advice
handling plant required will exceed the size of smoke has been included in this Report with the intention of
ventilation systems for many typical atrium room giving the fullest picture possible. It is therefore likely
openings. Note however, that pressurising the atrium that some of this guidance will need to be modified in
may be a viable option where the atrium faqade has the future, as the results of continued research become
only relatively small leakage paths. available.

2
 - _ .

A Code of practiceI4 for atrium buildings is currently

being prepared by the British Standards Institution
(BSI). The aim of this present Report is to provide
guidance only on design principles of smoke control
and it is hoped to support the code rather than to pre-
empt it. The Report cannot cover all the infinite
variations of atrium design. Instead it gives general
principles for the design of efficient systems, with
simplified design procedures for an ideal model of an
atrium, and then further guidance on frequently
encountered practical problems. As the design
procedures are of necessity simplified, it also gives their
limitations so that, when necessary, a more detailed
design by specialists can be carried out.
Figure 1 ‘Stcrile tube‘ atrium

Such an approach may employ field modelling, which

exploits the new techniques of computational fluid
dynamics (CFD) to deduce how, and at what rate,
smoke would fill an enclosure. It does this by avoiding
resort (as far as is currently possible) to experimental
correlations, and returning to first principles to solve
the basic laws of physics for the fluid flow. As a
consequence, with adequate validation, this type of
modelling should have a wide applicability. The use of
a computer is necessary since the technique involves
the solution of tens of thousands of mathematical -glazingStandard
equations for every step forward the simulation makes.

An atrium can be defined as any space penetrating

more than one storey of a building where the space is Figure 2 Closcd atrium
fully or partially covered. Most atria within shopping
centres may be considered as part of the shopping mall
and treated accordingly. A BSI Code of practiceIs
specifically for shopping complexes has been published,
and also a BRE ReportI6 giving advice on smoke
ventilation of enclosed shopping centres. Where atria
have mixed occupancies including shops then reference
should be made to these documents, or specialist
advice sought.

In order for a design to be achieved, it is necessary to

identify the various ‘types’ of atrium that are built.
These can be simply defined as follows:
(a) The ‘sterile tube’ atrium
The atrium is separated from the remainder of the Figure 3 Partially open atrium
building by fire-resisting glazing (FRG). The
atrium space generally has no functional use
other than as a circulation area (Figure 1).
(b) The closed atrium
The atrium is separated from the remainder of the
building by ordinary (non fire-resisting) glass.
The atrium space may well be functional
(cafeterias, restaurants, recreation, etc) (Figure 2).
(c) The partially open atrium
Here some lower levels are open to the atrium and
the remaining levels closed off by glazing (Figure 3).
(d) The fully open atrium
Some of the upper levels or all of the building
levels are open to the atrium (Figure 4). Figure 4 Fully open atrium

3
Fire growth and smoke production temperature. The smoke spreads out radially
underneath the ceiling and forms a layer which
In most instances, a room (compartment) fire may be deepens as the compartment begins to fill.
assumed to burn in either of two ways:
If the compartment is open to the atrium, then the
(a) Fuel-bed control gases flow out immediately they reach the opening.
When the rate of combustion, heat output and fire If the compartment is glazed or the opening is
growth are dependent upon the fuel being burned. below a deep downstand then the smoke steadily
(The 'normal' fire condition found in most single- deepens. As the layer gets deeper there is less
storey buildings whilst the fire is still small enough height for the plume of smoke to rise before it
for successful smoke control.) reaches the smoke layer, hence less air is being
(b) Ventilation control entrained, wit'h the result that the temperature of
Where the rate of combustion, etc is dependent the smoke layer increases with layer depth, even for
upon the quantity of air available to the fire a steady fire. Most fires will continue to grow larger
compartment (assuming any mechanical as the layer deepens, reinforcing this effect.
ventilation systems are inactive). 3 6 mm plate glass may shatter when exposed to gases
The quantity of smoky gases produced (ie the mass as little as 100 K warmer than ambient. Thus once
flow rate of gases) in and from the compartment, and this temperature is passed, there is an increasing
the energy (heat flux) contained therein are different likelihood that the glass will fracture. If the
for both regimes. It is therefore important to identify compartment is sprinklered and the water spray hits
the regime which applies. Hence the mass flow and the glass, the localised heating of the glass by
heat flux within the smoky gases may be determined. radiation from the fire and by the gas layer,
combined with sudden cooling due to the water
It is important to understand the basic mechanisms spray will increase the likelihood of the glass
which control the fire condition. A step-by-step history breaking. The smoke and hot gases will then flow
of a growing fire may be as follows: externally to atmosphere, or enter the atrium, or
both, depending upon the nature of the
1 The fire starts for whatever reason, its rate of compartment and its relative position in the
growth depending upon the materials involved. In building, the size and position of the fire in the
most practical compartments there is sufficient compartment, and the strength of differing glazing
oxygen to support combustion in the first few systems. If the fire can be accidentally or
minutes, and the fire growth and smoke production deliberately vented externally then the threat to
are controlled by the fuel, ie, fuel-bed control. other levels via the atrium is greatly reduced.
2 Smoke from the fire rises in a plume to the ceiling. There will, however, be instances when a fire will
As the plume rises, air is entrained into it, vent all its effluent gas into the atrium, and this is
increasing the volume of smoke and reducing its generally the worst design scenario (Figure 5).

Area A,
Perimeter P

Figure 5 Smoke entering an atrium from a fuel-bed-controlled fire in an adjacent room

4
 . ._

There is so much mixing of ambient air into the though the only surface available for air
plume that, except close to the fire itself, the hot entrainment is the width of the opening (as
smoky gases can be regarded as consisting of opposed to the fire perimeter for a fuel-bed-
warmed air when calculating the quantity (mass controlled fire). This condition is known as the
flow rate) being produced in the compartment. 'fully-involved, large-opening fire'".
Initially this mass flow rate of smoke will be 8 There are many factors which determine the
controlled by the fuel-bed, as mentioned above. prevailing condition, including the type and
However, the geometry of the opening on to the disposition of the fuel, the dimension of the
atrium has a crucial effect. As the fire grows large in enclosure and the dimensions of the ventilation
comparison to the area of the opening, the air opening. They can however be reduced to two
supply to the fire is 'throttled', causing it to burn principal parameters for most compartments:
inefficiently.
A,VH This is the area of the opening into the
This leads to the situation where the inability of the atrium A,", multiplied by the square root
compartment to vent the gases effectively due to the of its height H.
restricted area available causes the layer to deepen
Ar The area of the fire.
further which, combined with the increasing fire
area, causes the layer temperature to rise. Once the Note: Both the fully-involved large-opening fire
layer temperature reaches approximately 600 "C, and ventilation-controlled fire conditions will
then in most compartments the downward radiation almost certainly produce flames from the
from the gas layer is sufficient to cause auto-ignition opening into the atrium.
of the remaining combustible materials in the
9 The presence of sprinklers will usually serve to
compartment (Figure 6). Where there is sufficient
prevent fire growth proceeding to full involvement,
fuel within the compartment for the entire
and will usually maintain the fire in a fuel-bed-
compartment to become involved, the layer
controlled state where extinguishment is not
temperature will rapidly rise to flame temperature,
achieved.
very approximately 1200 K (930 "C). The rate of
burning, heat output and mass Qpw leaving the
compartment are now strongly dependent upon the
geometry of the opening, ie ventilation control
Air is introduced into an escape route (usually a
(Figure 7).
stairway) at a rate sufficient to hold back any smoke
The transition from the fuel-bed-controlled fire trying to pass on to that route. The pressure difference
with a layer at 600 "C to the ventilation-controlled across any small opening on to the route must be large
condition is very rapid, and may take only seconds. enough to offset adverse pressures caused by wind,
This condition is often known as 'flashover'. building stack effect and fire buoyancy. It must also be
low enough to allow the escape doors to be opened
There may be an intermediate situation where the
with relative ease. The air supply must also be large
compartment has flashed over or the fire simply
enough to produce a velocity sufficient to hold
grown to encompass the entire width of the
back smoke at any large opening on to the pressurised
compartment, but where the quantity of air now
space. Experience of pressurisation designs suggests
rq9uired to maintain combustion is adequate, even
that the technique is well-suited to the protection of

T+ 873K

/ I

Heat radiation

Figure 6 T h c onset of flashover

5
stairways used as escape routes in tall buildings, Air mixes into the fire plume as it rises, giving a larger
though it can also be useful in other circumstances. volume of smoky gases. These flow outwards below
the ceiling until they reach a barrier (eg the walls, or a
downstand). The gases then form a deepening layer,
Depressurisation whose buoyancy can drive them through natural
This is a special case of pressurisation, where gases are ventilators (or alternatively smoky gases can be
removed from the smoke-affected space in a way that removed using fans). For any given size of fire, an
maintains the desired pressure differences and/or air equilibrium can be reached where the quantity of
speeds across leakage openings between that space gases being removed equals the quantity entering the
and adjacent spaces. Note that depressurisation does layer in the fire plume - no significant mixing of air
not protect the smoke-affected space in any way; occurs upwards into the base of the buoyant smoke
instead it protects the adjacent spaces. In the layer. Sufficient air must enter the space below the
circumstances of an atrium it is sometimes possible to layer to replace the gases being removed from the
use the buoyancy of the smoky gases themselves to layer, otherwise the smoke ventilation system will not
create the desired depressurisation effects. This is work.
explained in more detail in Chapter 5.

Throughflow ventilation
Smoke ventilation (throughflow ventilation) is used
when the fire is in the same space as the people,
contents or escape routes being protected, without it
filling that space. The intention is to keep the smoke in
the upper reaches of the building, leaving clean air
near the floor to allow people to move freely. This
stratification or layering of the smoke is made possible
by the buoyancy of hot smoky gases produced by the
fire, and it follows that to be most successful the high-
level smoke layer must remain warm. Smoke
ventilation is therefore only suitable for atria where
fires can cause smoke to enter the atrium space. Such
fires can either be fuel-bed-controlled fires at the base
of the atrium, or fires in adjacent spaces (rooms) which
allow smoky gases to enter the atrium. Much of this
Report is concerned with the calculation of design
parameters for smoke ventilation systems tailored to
the circumstances found in various types of atria. First
though, it is worth reviewing the underlying principles
of smoke ventilation and the general approach needed
for successful design.

Figure 7 A fully-involved ventilation-controlled fire

6
 .. - . . - -
, ,, :,
. . . .
.-. ~ .__,_--_,.-

Chapter 2
Design fire size
The calculation of the quantity of smoke and heat fires may potentially be adopted. In the present work
produced by a fire requires a knowledge of its size in the following, in terms of fire area and convective heat
terms of area, perimeter and heat flux developed per flux, are used to illustrate the calculation procedures
unit area or from the fire as a whole. When designing adopted:
smoke ventilation or depressurisation systems, the
(a) RetailI6 (sprinklered shops)
mass flow rate and heat flux developed in the room are
10 m2, 12 m perimeter.
major parameters in the calculation of the system
requirements, changes in which can substantially affect (b) Offices” (sprinklered) Offices”(unsprink1ered)
all of the subsequent smoke flow conditions. 16 m2, 14 m perimeter. 47 m2, 24 m perimeter.
(c) Hotel bedrooms2”
The preferred choice of design fire would be a time-
Floor area of the largest bedroom.
dependent growing fire, to which the means of escape
and evacuation time for the particular building
It should be noted that the design fire size for (a) was
occupancy could be related, allowing the increasing
originally chosen by the Home Office and Scottish
threat to occupants to be calculated as time progresses.
Home and Health Department, and that for those of
Unfortunately there is no available research, at the
(b) and (c) there is no official ‘approved’ choice.
time of preparing this Report, which allows assessment
of the probability distribution describing the variation
After considering the heat losses to the structure of the
of fire growth curves for areas typically associated with
room, and any losses to sprinkler sprays, etc, the
atria. Clearly, one does not want an ‘average’ fire for
commonly used heat outputs are approximatelyl’:
safety design, since typically half of all fires would
grow faster. It is much simpler to assess the maximum Sprinklered office 1 MW
size a fire can reasonably be expected to reach during Unsprinklered office 6 MW
the escape period, and to design the system to cope
with that. Such assessments can sometimes be based Hotel bedroom* 1 MW
upon available statistics on fire damaged areas, but
may have to depend upon experienced judgement Gases flowing into the atrium from a fire deep inside a
based on the anticipated fire load where a more large-area office with operating sprinklers may be
rigorous approach is not feasible. cooler than is assumed in the ‘sprinklered office’
design fire above. The mass flow rate of gases entering
‘Work on design guidance for smoke ventilation the final reservoir will be less than would be calculated
systems in shopping centres used the principle of using the value given above. Even for this scenario
selecting a fixed size of fire that would cater for almost therefore, the above value should err on the side of
all of the fire sizes likely to be found in that class of safety. Designers wishing to take sprinkler cooling in
occupancy, and then deducing a pessimistic heat the fire compartment more rigorously into account
output from that fire16,18. This procedure has been should adopt a fully fire-engineered approach
adopted for occupancies other than retail which are appropriate to their specific circumstances, for
also commonly associated with atrium buildings - example by using the methods described in the section
offices and hotel bedrooms1y*20. ‘Effects of sprinkler systems in smoke reservoirs’
(page 14) to assess the effect of sprinkler cooling on
The’procedure has no time dependency and does not the outflowing gases.
reveal any information regarding the growth
characteristics of the fire. It is therefore usual to The hotel bedroom fire represents a fully-involved
assume that the fire is at ‘steady-state’. This unsprinklered fire2”. Where sprinklers are present this
assumption allows the smoke control system to cater will be clearly unrealistic and a value of 500 kW (for a
for all fires within the accepted design fire size, and by 6 m perimeter fire) may be more appropriate for
not considering the growth phase of the fire, designers wishing to adopt a fire-engineering approach
introduces a significant margin of safety to the system to a design.
design.
This Report will however assume the fully-involved
It follows from the foregoing that there is a strongly value for design purposes as this will introduce a large
subjective element in assessing what fire size is safety margin to the design, in particular to the
acceptably infrequent for safety design purposes. calculation of the mass flow rate leaving the room
through the opening.
Various design fires have been suggested for
occupancies associated with atria. A wide range of * Assuming a floor area of approximatcly 20 m2

7
The use of the bedroom floor for the hotel bedroom accommodate this fire condition, the destructive power
design fire reflects the situation where there are no of a fully-involved office room fire is such that smoke
sprinklers present. Unpublished research on control systems cannot usually be designed to
sprinklered bed fires (P G Smith and J V Murrell, Fire satisfactorily protect means of escape in this situation,
Research Station; private communication, 1986), except for fires in small rooms. An accurate
where the low heat output per unit area was assessment of the mass flow rate and heat flux from a
comparable to values for hotel bedrooms, suggests that room fire will allow the potential for flashover to be
the much lower fuel load (compared to an office) estimated, and thence whether additional
expected in a hotel bedroom utilising conventional precautionary measures are required, eg sprinklers.
sprinklers should make it possible for the smoky gases This Report will only provide guidance for the design
to be cooled sufficiently to be retained within the room of smoke control systems for a fuel-bed-controlled fire
of origin (assuming the window is not open). The in an office, and a fully-involved fire in a hotel
operation of sprinklers is likely to cool any smoke bedroom.
from a fire and suppress that fire to such an extent that
the glazing to the bedroom will probably remain intact. Should a different design fire be considered for
This is particularly true for double-glazed windows. whatever reason, the equations, figures, etc given here
may no longer apply, and advice should be sought
The same research indicates that the use of from experts.
conventional sprinklers in a residential environment
may not however allow conditions within the room to
remain tenable, and it may be inferred that the
presence of an open window to the room could
produce hazardous conditions in the atrium, at least
above the floor of fire origin. Since there is no
statistical data available on fires in sprinklered hotel
bedrooms in the UK, any choice of design fire size will
be subjective. Should a designer wish to examine the
effect of a plume emanating from an open window in a
sprinklered hotel bedroom, it would not seem
unreasonable to use a value of 6 m perimeter
(equivalent to a single bed) with a convective heat
output of around 500 kW as the design fire.

Research into the use of fast-response sprinklers in a

residential environment21,22has clearly shown that at
the time of operation of these sprinklers the conditions
inside the rooms were still tenable, ie there was no life-
safety risk from the smoke, even with excessive ceiling
level temperatures. This indicates that for any gases
flowing into the atrium (eg through an open window)
the further entrainment induced by the rising smoke
plume will ensure that conditions within the atrium
must be tenable, regardless of the smoke temperature
or smoke production rate in the room. While it is
possible that this may also be true for cellular offices
employing fast-response sprinklers, there is no
evidence (experimental or empirical) to validate this,
and so, to err on the side of safety, this Report will
regard sprinklered offices employing fast-response
sprinklers in the same way as offices using
conventional sprinklers. Further research and
statistical data are desirable in this area.

As mentioned in the Introduction the retail atrium is

considered separately in the Report on covered
shopping complexesI6, and will not be considered
further in this Report. The fire sizes on the previous
page (excluding (a)) will be those used throughout.
Furthermore, when considering an unsprinklered
office occupancy, there exists the potential for
flashover to occur and for the entire floor to become
involved in fire. Even if the building geometry can

8
 >, '
. ..;
i' . . , . * _. " _. . . ,, . , . h -...,-.- LC. ~. ... .

Chapter 3
Smoke control on the floor of fire origin
Within the fire room Exhaust fromf
ComDartment
In any situation involving the potential movement of
smoke into escape routes, it is always preferable to
control the smoke in the fire room and hence prevent
its passage to otherwise unaffected areas. Ventilation
of the fire room may be achieved by either a dedicated
smoke exhaust system or by adapting and boosting an
air-conditioning or ventilating system. If the
compartment is open to the atrium, then it must have
either a downstand barrier to create a reservoir within
the compartment, or a high-powered exhaust slot at
the boundary edge to achieve a similar effect (Figure 8).

The minimum height of the smoke layer base in the *Volumetric flow rate sufficiently great to prevent
room must be compatible with the openings on to the smoke spillage beneath downstand for height
of rise Y
atrium, with the layer depth being no lower than the
soffit of the opening (Figure 9). Where no downstand Figure 9 Plume height and layer depth with a downstand
exists and an exhaust slot is used instead, the exhaust
capacity provided will need to be compatible with the
layer depth (Figure 10). See the section on exhaust
slots ('Slit extract') on page 19. Having established the clear layer height in the room,
the mass flow rate of smoke can then be calculated.
Recent work by Hansel123drawing on work by
Zukowski et a124and Quintiere et a125has shown that
the rate of air entrainment into a plume of smoke
(a) Use of a downstand to create a smoke reservoir
Exhaust from
rising above a fire ( M , ) may be obtained by using the
compartment equation:
Mr = C, P Y 3'2 kgs-' ...(1)

where C, = 0.188 for large-space rooms such as

auditoria, stadia, large open-plan offices,
atrium floors, etc where the ceiling is
well above the fire.
C, = 0.210 for large-space rooms, such as
open-plan offices, where the ceiling is
close to the fire.
Note: As the two values are approximately similar and
the demarcation between them uncertain, then the
value for all large-space rooms is taken to be 0.188 for
(b) Use of a 'slot exhaust' to prevent smoke entering the atrium the purposes of design.
Exhaust from Boundary edge
compartment exhaust slot
C, = 0.337 for small-space rooms such as unit
shops, cellular offices, hotel bedrooms*,
etc with ventilation openings
predominantly to one side of the fire (eg
from an office window in one wall only).
Most small rooms will therefore take
this value.
P = Perimeter of the fire (m).
Y = Height from the base of the fire to t h e
smoke layer (m)

Figure 8 Smoke ventilation within a compartmcnt * Prior to flashover or full involvement

9
 Exhaust from*
compartment
Boundary e d g g
exhaust slot Flow of hot gases out of the room of
origin into the atrium
The mass flow rate of smoky gases passing through a
vertical opening ( M , ) may be found from23:

L L

U
* Total volumetric flow rate sufficiently great
t o prevent smoke spillage beyond the exhaust where W = Width of opening (m)
slot for height of rise Y
h = Height of the opening above the floor (m)
Figure 10 Plume height and layer depth with a slot exhaust Cd = Effective coefficient of discharge for the
opening
Equation 1has been validated experimentally26for
values of Y up to 10 times U A f ,for fires in large Where the smoke flow directly approaches a ‘spill
spaces, for values of the heat release rate between edge’ with no downstand (eg where the ceiling is flush
200 and 750 kWm-*. with the top of the opening), Cd = 1.0. For other
scenarios the following procedure may be adopted:
There is no information available to show how
Equation 1 (or any current alternatives) should be Where the smoke flows beyond a downstand or lower
modified to allow for the effects of sprinkler spray ceiling level in the form of a plume of height Dd
interactions. Consequently, it is used here unmodified. (Figure 13(a) arid 13(b)), it has been shown23that the
height of rise of the plume has an effect on the rate of
The quantity of smoke entering a ceiling reservoir or flow of smoke leaving the opening. This effect can be
flowing layer given by Equation 1 is shown graphically expressed as a modification to the coefficient of
in Figures 11 and 12 for both cellular and open-plan discharge as follows:
offices (C, = 0.337 and C, = 0.188 respectively) and for
sprinklered and unsprinklered offices ( P = 14 m and Cd = 0.65 I ...(a
P = 24 m respectively). For further discussion on the
criteria for selecting a value of C,, see the section --0
4
‘Flow of hot gases out of the room of origin into the
atrium’ that follows.

Whilst Figures 11 and 12 show the mass flow

production curves for cellular offices, many such 32

configurations will not in practice have a fixed wall 30

construction with a good enough fire resistance, or I28-
have a large enough opening to sustain the VI

26-
replacement air supply needed for such large fires (see VI

the section ‘Inlet air’ on page 17). f 24-

0
h 22- )pen -plan offices

Figure 12 also has a ‘cut-off‘ below which the ; 20-

:,=0.188

temperature of the gas layer will exceed 600 “C and o

r
18-
flashover of the room will almost certainly have
occurred. The mechanism of flashover may well start
to occur prior to this critical point, and gas en
temperatures in excess of 500 “C may be considered a 6 12

conservative lower limit for flashover potential27.The

‘danger-zone’ is shown shaded on Figure 12.

Mass flow rates should be above this shaded zone for

the smoke control systems to operate safely.

Height of smoke base (m)

Figure 11 Rate of production of hot smoky gases -

sprinklered offices
10
 :ellular offices
where D , = Flowing layer depth in the plane of the :,=0,337
opening (m).
Dd = Depth of downstand or height of rise of
plume beyond the opening (m).
Where Dd 2 1.0 then Dd may be taken as 1.0 for most

401
openings of practical interest. For a plain opening with
no downstand obstruction (Figure 14), Dd can be
considered as the rise of the plume beyond the balcony )pen-pla Jffices
edge. The flowing layer depth (0,)may be found from: $ 35 :e = 0.188
Y
>

E
6 30-
.*.(4) r
r

A simple procedure for calculating the mass flow rate,

etc is as follows: Increasing
potential for
(a) Set C, to 0.65. flashover
(b) Calculate mass flow rate from:

Mw=
C, P W h 312 15t // I
[ w213+;[ g3
] 3/2

Height of smoke base (m)

Figure 12 Rate of production of hot smoky gases -

unsprinklcred offices
(c) Calculate buoyant layer depth from:
213
Figure 13 Flow out of an opening
c,
D w = ~ 2w
[ " ] m

(d) Calculate discharge coefficient from:

Cd=0.65 [y]
Dd
DW+
113

(e) Use the new value of Cdand repeat from step (b),
until the difference between the currently
calculated value of Mw(Mw(n))and the
previously calculated value of Mw(Mw(n-l)) is
less than 0.1 Y0.
4- U
ie:
(a) With downstand and projecting balcony
Mw(n) - Mw(n - 1)
x 100 < 0.1
MW(n)

This procedure usually converges after about five

iterations and will therefore quickly yield
M,, C, and D,.

Figures 15 and 16 give t h e mass flow values in

graphical form for various opening heights and widths,
using t h e above procedure. A ceiling and projecting
balcony 4 m above the floor have been assumed. I t
should be noted that the shaded areas on the graphs
represent t h e onset of flashover (calculated using M,
and Q, appropriate to the example illustrated and a I
layer temperature of approximately 550 "C), and
values of mass flow lying within this band should be 7
(b) With high balcony
regarded with caution.
11
 ---
I ---- - -
.-
-- -
i i.

lDi'
Figure 14 Plain opening Cd= 0.8

2 5 2 h= 4 ' 0 m
h = 3.5m
h

i
W
.20
h = 3.0m
Y
i 15 h= 2.5m

10 h= 2.0m

h=4,0m

h= 3.5m

h= 3.0m
10
h= 2.5m

h= 2.0m

I I I I 1 1 1
0 5 10 15 20 25 30 35 40
W (m)

Figure 15 Mass flow rate through a vertical opening -open-plan offices

35t 3 5 7 7 h= 4'0m

--
I
W
v)
301
25

Y
2

0
0 5 10 15 20 2
-0
5t 5 10 15 20 25
W (m) W (m)
Sprinklered Unsprinklered

Figure 16 Mass flow rate through a vertical opening - cellular offices

12
 The demarcation between a cellular room and an positioned across the balcony to limit the size of this
open-plan layout is determined by the ability of the reservoir to ensure that the smoke retains its
incoming air to flow into the rising plume from all buoyancy. Each reservoir should be limited to an area
sides. The narrower the room becomes, the less easily' not exceeding 1300 m2, with a maximum length of
the air can flow behind the plume. In this Report 60 m by analogy with shopping maIlsI6.The screens
cellular offices are considered to be those in which the around the balconies will, in general, be fairly close to
maximum room dimension is less than or equal to potential fire compartments (eg offices). Being close,
five times the diameter of the design fire size, and the smoke issuing from such a compartment will deepen
incoming air can only enter from one direction locally on meeting a transverse barrier. The depth of
(Figure 17). This demarcation dimension was chosen these screens should take into account local deepening
arbitrarily and has no theoretical derivation. Research (see page 17). Smoke removed from these lower level
in this area is highly desirable. reservoirs should usually be ducted to outside the
building but can be ducted into the ceiling reservoir of
Width< 5 x Df the atrium (Figure 20). The mass flow rate of smoke to
be exhausted from the atrium roof will then be that
calculated for the under-balcony condition28.
Restricted air flow Cellular room
to fire and plume
Early experiments with smoke flow in shopping malls29
and unpublished workI7 at FRS (also N R Marshall,
Fire Research Station; private communication, 1984)
have shown that the smoke flowing from a room with a
deep downstand and then under a balcony beyond the
opening becomes turbulent with increasing mixing of
air. This subsequent evidence suggests that for the
purpose of engineering design the mass flow rate of
smoke entering the balcony reservoir (MB) can be
taken to be approximately double the amount given by
Equation 2, ie:

' ..

Opening 4
...(5 )
Figure 17 Limiting size of a cellular room

Ventilation of the balcony space

Common
If the smoke cannot be contained within the room of balcony
space
origin because:
I
1

0 the rooms have demountable partitions, Atrium

w
I
0 insufficient replacement air can be provided, or
Common
0 the engineering implications are too costly or balcony
difficult to apply, space

then the smoke and hot gases will be able to travel

from the room of origin into the space beyond,
including the atrium.
Figure 18 Schematic section of an atrium with balconies
Some atria are designed with balconies around the
perimeter of the void, serving all the rooms at that
level (Figure 18). Figure 19 illustrates in schematic
form an atrium with floors (two levels only are shown U
in both the figures) which have balconies that leave a Exhaust from
considerable area for pedestrians. On each level there balcony reservoir
is a large area situated below each balcony. If screens
(activated by smoke detectors or as permanent
features) are hung down from the balcony edges, the
region below each balcony can be turned into a ceiling
reservoir. This is similar to the procedure used in
multi-storey shopping complexesI6.

This balcony reservoir can then be provided with its

own extraction system. Other screens can be Figure 19 An under-balcony smoke reservoir

13
 0 alternative escape routes,
0 shorter escape paths along the balcony, and
the installation of sprinklers to cool the gases
further.

Effects of sprinkler systems in

smoke reservoirs
Offices, shops, assembly, industrial and storage or
other non-residential purpose groups are now
expected to have sprinklers if they have a floor more
than 30 metres above ground level. Multi-storey
buildings in the assembly, shop, industrial or storage
purpose groups will also be fitted with sprinklers if
individual uncompartmented floors exceed a given
size. Sprinklers may also be required in other
circumstances for insurance purposes.

The action of a sprinkler system in an office on the

cooling of gases flowing from that office to the atrium
I is accounted for in the derivation of the 1 MW heat
Figure 20 Under-balcony smoke reservoir venting into an atrium output17.Where the smoke layer is contained wholly
smoke reservoir
within the room of origin by a smoke control system
and has a large area, the sprinklers will cool the smoke
layer further. Similarly, where smoke is collected
Entrainment into smoke flows from compartments is within a balcony reservoir adjacent to sprinklered
being studied23.The purpose of this is to determine offices, operation of sprinklers under the balconies will
more accurately the influence of such factors as lead to increased heat loss reducing the buoyancy of
compartment opening geometry, the presence of a smoke, which in turn can contribute to a progressive
downstand fascia and balconyldownstand loss of visibility under the smoky layer. However,
combinations. It follows that Equation 5 may be gases sufficiently hot enough to set off sprinklers will
superseded in due course. remain initially as a thermally buoyant layer under the
balcony ceiling, and will not be pulled out of the layer
Smoke layer temperature by the sprinkler sprays.

The mean temperature rise of the smoke layer above When the fire occurs in an office, the operation of
ambient (0)can be calculated from: sprinklers under the balcony will not assist in
controlling it. If too many sprinklers operated under
QW the balcony, sprinklers in the office could become less
__ "C ...(6) effective as the available water supply approached its
Mc limits.

where Qw= Heat output at window or exit point (kW) It follows therefore that sprinklers need only be
M = The mass flow of smoke (eg M , or MB)(kgs- ') installed in a smoke reservoir i f

c = Speciticheat capacity of the gases (kJkg-'K-') 0 the smoke layer temperature is likely to exceed
200 "C and thus produce sufficient radiation to
Tables 1 and 2 give the temperature rise (0)for a impede escape, or
1 MW and 6 MW fire, taking into account the cooling 0 if there is the likelihood of sufficient
processes mentioned in Chapter 2. combustibles being present to pose a significant
threat of excessive fire-spread.
In unsprinklered fire situations a high smoke layer
temperature will result in intense heat radiation which A powered extract system, to a reasonable
may cause difficulties for people escaping along a approximation. removes a fixed volume of smoke
balcony beneath the smoke layer, especially if the irrespective of temperature. Therefore if the extent of
balconies form a major escape route. The maximum sprinkler coolirig is overestimated, the system could
smoke layer temperature which will allow safe be underdesigned.
evacuation without undue stress is in the order of
200 "C. If this gas temperature (or lower) cannot be A system using natural ventilators depends on the
achieved then consideration should be given to: buoyancy of the hot gases to expel smoke through the

14
Table 1 Volume flow rate and temperature of gases from a 1 MW
fire (including cooling within room of origin)

Mass flow rate Temperature of gases Volume rate of extraction

(Mass rate of extraction) above ambient . (at maximum temperature)
(kgs-9 ("C) (m's-')

4 250 6.0
6 I67 8.0
8 125 9.5
10 100 11.0
12 83 12.5
15 67 15.0
20 50 19.5
25 42 22.5
30 33 27.5
35 28 32.0
40 25 36.0
50 20 44.5
60 17 53.0

ventilators. In this case the system would be temperature beyond the radius of operation of the
underdesigned if the sprinkler cooling were sprinklers. This radius is generally not known.
underestimated.
In the absence of better information, it may be
The heat loss from smoky gases to sprinklers is reasonable to assume that no more sprinklers will
currently the subject of research, although data operate than are assumed when calculating the design
suitable for design application are not yet available. of sprinkler systems and their water supply (eg 18 heads
Nevertheless, an approximate estimate can be for Ordinary Hazard Group 3).
obtained as follows:
For powered extract systems the cooling effect of
If the smoke passing a sprinkler is hotter than the sprinklers can be ignored in determining the volume
sprinkler operating temperature, that sprinkler will extract rate required. This will err on the side of safety.
eventually be set off and its spray will cool the smoke. Alternatively, this further cooling and the consequent
If the smoke is still hot enough the next sprinkler will contraction of smoky gases can be approximately
operate, cooling the smoke further. A stage will be estimated on the basis of an average value between the
reached when the smoke temperature is insufficient to sprinkler operating temperature and the calculated
set off further sprinklers. The smoke layer initial smoke temperature. Where the fan exhaust .
temperature can thereafter be assumed to be openings are sufficiently well separated it can be
approximately equal to the sprinkler operating assumed that one opening may be close to the fire, and

Table 2 Volume flow rate and temperature of gases from a 6 MW

fire (including cooling within room of origin)

Mass flow rate Temperature of gases Volume rate of extraction

(Mass rate of extraction) above ambient (at maximum temperature)
(kgs-9 ("C) (1113s- ')

10 600 25.5
12 500 27.0
15 400 29.5
20 333 32.0
25 240 38.0
30 200 41 .5
35 171 46.5
40 150 50.5
50 120 59.0
60 100 67.5
75 80 80.0
90 67 92.5
I10 54 107.0
130 46 123.5
150 40 140.0
200 30 181.0
300 20 263.0
400 15 345.0

15
 . -... . ~.,_.. . .-

will extract gases at the full initial temperature given shown in Table 3 for layer temperatures 0B2 65 OCl7.
by Equation 6. The other openings in these This ignores the effects of cooling (See the section
circumstances can be assumed to be outside the zone ‘Smoke layer temperature’ on page 14). Each depth
of operating sprinklers, and will extract gases at the shown in this table is the minimum possible regardless
sprinklers’ effective operating temperature. of the smoke extraction method employed
downstream; consequently it represents the minimum
The number of potential ‘hot’ and ‘cool’ intakes must depth for that reservoir.
be assessed when calculating the average temperature
of extracted gases. The depth must be measured below the lowest
transverse downstand obstacle to the flow (eg
If the sprinkler operating temperature is above about structural beams or ductwork) rather than the true
, 140 “C, or above the calculated smoke layer ceiling. Where such structures exist and are an
temperature, then sprinkler cooling can be ignored for appreciable fraction of the overall layer depth, the
natural ventilators. depth below the obstacle should be found using
Table 3(b) rather than 3(a).
Note that the effect of sprinkler cooling is to reduce
the heat flux (Q,) without significantly changing the
mass flux. It follows that once a new value of 0 has
been estimated, the new heat flux can be found using Table 3 Minimum reservoir depths or minimum
Equation 6. channelling screen depths required under
balconies for both 1 MW and 6 MW
convective heat output
Flowing layer depth
Smoke entering a ceiling reservoir will flow from the (a) Unimpeded Itlow
point of entry towards the exhaust points. This flow is
driven by the buoyancy of the smoke. Even if there is a Mass flow rate Width of reservoir W ,
very large ventilation area downstream (eg if the entering or channelling screen width L (m)*
reservoir -
ceiling downstream were to be removed), this flowing (kgs-’1 4 6 8 10 12 15
layer would still have a depth related to the width
available under the remaining ceiling (which can now 10 1.1 0.8 0.7 0.6 0.5 0.5
be considered a balcony), the temperature of the 15 1.4 1.1 0.9 0.8 0.7 0.6
smoke and the mass flow rate of smoke. Work by 20 1.7 1.3 1.1 0.9 0.8 0.7
25 2.0 1.5 1.2 1.1 1.0 0.8
Morgan30 has shown that this depth can be calculated 30 2.3 1.7 1.4 1.2 1.1 0.9
for unidirectional flow as follows:
40 2.8 2.2 1.8 1.5 1.4 1.2
50 3.4 2.6 2.1 1.8 1.6 1.4
70 4.5 3.4 2.8 2.4 2.2 1.9
...(7) 90 5.6 4.3 3.5 3.1 2.7 2.3
110 6.7 5.1 4.2 3.6 3.3 2.8
130 7.8 6.0 4.9 4.2 3.8 3.2
150 9.0 6.8 5.6 4.9 4.3 3.7
where D B= Flowing smoke layer depth under
the balcony (m)
MB= Mass flow rate under the balcony (kgs-I) (b) Impeded flow - deep downstand
WB = Balcony channel width (m) Mass flow rate Width of reservoir W ,
entering or channelling screen width L (m)*
c d = Coefficient of discharge reservoir -
I
(kgs-’) 4 6 8 10 12 15
Note: Values of Cd will vary for differing flow 10 1.8 1.4 1.2 1.0 0.9 0.8
geometries. However, for the purpose of engineering 15 2.3 1.8 1.5 1.3 1.1 1.0
design c d can be taken to be 0.6 if a deep downstand is 20 2.8 2.2 1.8 1.5 1.4 1.2
present at right angles to the flow, or 1.0 in the absence 25 3.3 2.5 2.1 1.8 1.6 1.4
of a downstand. 30 3.8 2.9 2.4 2.1 1.8 1.6
40 4.7 3.6 3.0 2.6 2.3 2.0
At the time of writing, values of Cd for intermediate 50 5.7 4.3 3.6 3.1 2.7 2.4
70 7.5 5.8 4.8 4.1 3.6 3.1
depth downstands cannot be stated with confidence for
90 9.4 7.2 6.0 5.1 4.5 3.9
the wide range of geometries to be found in practice. It 110 11.2 8.6 7.1 6.1 5.4 4.7
is suggested that either of the extreme values should
130 13.1 10.0 8.2 7.1 6.3 5.4
be adopted in seeking a conservative design approach. 150 15.0 11.5 9.5 8.2 7.2 6.2

The resulting values of layer depth for different balcony * For bi-directional flow of smoky gases this should be twice the
reservoir widths and mass flow rates of smoke are actual reservoir width

16
 .( . .. - - ... ...
.. 1,

Local deepening
Where a buoyant layer of hot smoke flows along
beneath a ceiling and meets a transverse barrier, it ...(8)
deepens locally against that barrier3' and, as the gases,
are brought to a halt, the kinetic energy of the
approaching layer is converted to buoyant potential
energy against the barrier.

When designing a smoke ventilation system for atria,

in which the balconies are acting as reservoirs, it is where ADB= the additional deepening at the
often necessary to control the path of smoke flow transverse barrier (m)
using downstand smoke curtains. These are typically H , = the floor-to-ceiling height (m)
installed around the edge of the voids to prevent
smoke flowing up through the voids. If the void edge is D B= the established flowing layer depth (m)
I
close to the room this local deepening could cause WB = distance between the opening and the
smoke to underspill the smoke curtain and flow up transverse barrier (ie balcony depth) (m)
through the void, possibly affecting escape from other
storeys. Clearly, the void edge screens must be deep
enough to contain not only the established layer, but Inlet air
also the additional local deepening outside the room There must be adequate replacement air for the
on fire. efficient operation of a smoke ventilation system.
When ventilating compartments directly, if the faiade
The extent of local deepening can be found from is normally sealed then facilities should be provided
Figure 21. The depth of the established layer (DB in for the necessary quantity of replacement air to be
Figure 21) under the balcony immediately downstream supplied to the fire room automatically. This
of the local deepening must first be found using the requirement often makes the provision of smoke
design procedure given in the preceding sections. ventilation to the room of origin prohibitive or .
Usually this means in the channel formed between the undesirable. The provision of replacement air to a
void edge screen and the room faqade. The additional system employing balcony reservoirs is far easier,
depth ADBcan then be found by inspection of Figure 21, provided the balconies are open to the atrium.
allowing the necessary minimum overall depth (DB +
ADB) of the void edge screen to be found. If the area available for inlet becomes too restricted,
incoming airflow through escape doors may be at too
It can be shown that the following scale-independent high a velocity for easy escape. Such air inflows
formula can be used to approximate Figure 21: through doors in public buildings could hinder

I I I
1.0 1.5 2.0
Dg fm)

Figure 21 Local deepening at a transverse barrier

17
 . ... -

escapees. Recent research32into the ability of people an opening (A J M Heselden, Fire Research Station;
to move through an exit against an opposing airflow private communication, 1976), beyond which air will
has shown that movement is not impeded for airspeeds be drawn through the layer. This critical exhaust rate
below 5 ms-', and is not seriously impeded below (MCRIT)may be found from:
10 ms-' (although some discomfort was reported at
these higher airspeeds). This suggests that inflow
airspeeds should not usually exceed 5 ms-I.
where p= 1.3 for a vent near a wall ( k g ~ n - ~ )
Other values may be appropriate for other or p= 1.8 for a vent distant from a wall (kgm-')
circumstances. For example, in buildings where the
population is largely familiar with the escape routes; g = Acceleration due to gravity (ms-2)
where the incoming air is entering the fire room D = Depth of smoke layer below the
directly, or where (in the instance of the inlet air being extraction point (m)
supplied via the atrium) the major escape routes are
away from the atrium; then a less onerous parameter To = Absolute ambient temperature (K)
can be applied. Current advice regarding 0 = Excess temperature of smoke layer ("C)
pressurisation system design33recommends a
maximum pressure drop across a door of 60 Pa. This T = To + 0 (K)
accords to a face velocity across a rectangular inlet
opening of about 6 ms-I. The pressure drop criterion The required number of extract vents (N) is then given
may be increased if the population of the building is by:
adult and physically fit, to perhaps 100 Pa (8 ms-I). M
N2-- ...( 10)
A fan-driven inlet air supply may be employed, but MCRIT
can give problems when mechanical extraction is used
(the building will usually be fairly well sealed in such where M = The mass flow rate entering the layer
circumstances). This is because the warmed air taken (ie Mf o r M s ) (kgs-I)
out will have a greater volume than the inlet air. As
the fire grows and declines, the mismatch in volume Table 4 Minimlim number of extraction points
between the inlet air and the extracted fire-warmed air needed in a smoke reservoir
will also change. This can result in significant pressure
differences appearing across any doors on the escape (a) 1 MW heat output
routes. For this reason simple 'push-pull' systems
should be avoided. Total mass
rate of Depth of layer below extraction point (m)
extraction
Minimum number of extraction points (kgs-') 0.5 0.75 1.0 1.25 1.5 1.75 2.0

The number of extraction points within the reservoir is 9 21-28 8-11 4-5 3-3 2-2 1-2 1-1
important since, for any specified layer depth, there is 12 2940 11-15 6-8 3-5 2-3 2-2 1-2
a maximum rate at which smoky gases can enter any 15 39-54 14-20 7-10 4-6 3-4 2-3 2-2
individual extraction point. Any further attempt to 18 50-68 18-25 9-13 5-7 4-5 3-3 2-3
increase the rate of extraction through that point 20 57-79 21-29 10-14 6-8 4-6 3-4 2-3
merely serves to draw air into the orifice from below 25 77-1 07 28-39 14-19 8-11 5-7 4-5 3 4
30 99-137 36-50 18-25 10-14 7-9 5-6 4-5
the smoke layer. This is sometimes known as 'plug-
40 149-205 54-75 27-37 15-21 10-14 7-9 5-7
holing'. It follows that, for efficient extraction, the
number of extraction points must be chosen to ensure
that no air is drawn up in this way. Table 4, which is (b) 6 MW heat output
based on experimental subsequently modified
( A J M Heselden, Fire Research Station; private Total mass
communication, 1976), lists the minimum numbers rate of Depth of layer below extraction point (m)
needed for different reservoir conditions and for a extraction -
(kgs-I) 0.5 0.75 1.0 1.25 1.5 1.75 2.0
variety of mass flow rates being extracted from the
ventilators in the reservoir. Table 4 strictly applies to 12 26-35 10-13 5-7 3-4 2-3 2-2 1-2
ventilators which are small compared to the layer 15 31-43 12-16 6-8 4-5 2-3 2-2 1-2
depth below the ventilators (eg where the diameter is 18 37-51 14-19 7-9 4-6 3-4 2-3 2-2
much less than the depth of the layer). 20 41-56 15-21 8-10 5-6 3 4 2-3 2-2
2.5 51-70 19-26 9-13 6-8 4-5 3-4 2-3
Where sprinklers are installed and additional cooling 30 62-85 23-31 11-15 7-9 . 4-6 3-4 2-3
of the smoke layer needs to be accounted for, the 40 85-118 31-43 15-21 9-12 6-8 4-6 3-4
number of extraction points required will differ from Note: In both tables the first number of each pair denotes extraction
those shown in Table 4. The number can be points wcll away from the walls, and thc second is for those
determined by calculating the critical exhaust rate for close to the walls.

18
Where very large or physically extensive ventilators Such a system is likely to work best with further
are used (eg a long intake grill in the side of a extraction distributed within the fire room, which for a
horizontal duct) an alternative method is possible. For sprinklered room may possibly be provided by the
this case, Table 3(a) or 3(b) can be used with the normal ventilation extraction system (the normal
‘width of reservoir’ being taken as the total horizontal ventilation input and recirculation of air being
accessible perimeter of all the ventilators within the stopped) or, for an unsprinklered room, by a partial
reservoir (eg the total length of intake grilles in the smoke exhaust system. Whilst this system is designed
example above) and the ‘minimum reservoir depth’ to prevent smoke entering the atrium void, it will not
corresponds to the depth of the smoke layer beneath necessarily maintain a clear layer within the room
the top edge of the intake orifice. In practice, for a itself, and any balcony may become ‘fogged’. The
given mass flow rate and layer depth, Table 3(a), 3(b) extraction should be provided very close to the
or Equation 7 can be used to find the minimum value opening from a continuous slit which may be situated
of accessible perimeter. in the plane of the false ceiling.

Intermediate size intakes (ie where the ventilator size It has been shown that powered exhaust from a slit at
is comparable to layer depth) cannot be treated so right angles to a layer flow could completely prevent
simply and it is recommended that Table 4 be used smoke passing that slit, provided that the extraction
since it errs on the side of safety. rate at the slit was at least 5/3 times the flow in the
horizontal layer flowing towards the slit
(H G H Wraight, Fire Research Station; private
Required ventilation rate communication, 1984). This allows a useful general
(powered exhaust) method for sizing such an extract:
A powered smoke exhaust system consists of fans and
associated ductwork designed to remove the mass flow (a) First calculate the flow rate of gases approaching
rate ,of smoke entering the smoke reservoir, and to be the opening (or gap in the balcony edge screens)
capable of withstanding the anticipated smoke using as appropriate the following sections:
temperatures.

The controls and wiring should of course be protected,

to maintain the electrical supply to the fans during a fire. I
I
I
Balcony reservoir
The mass flow rate of smoke determined from the

U
screen
previous sections can be converted to the Drop screen
corresponding volumetric flow rate and temperature,
using Tables 1 or 2 or the following equation for
selection of the appropriate fans:

v,=
M T
~ 1313s-1 ...(11)
I
I
ni
I1 Slit , Atrium void

where
PO To

VI = Volumetric exhaust rate required in the

Fire room

I /I
reservoir (m3s-I)
Drop screen
M = M for M B determined from the
previous section (kgs-I) Balcony reservoir
screen
T = e + T, (K) Plan
po = Density of ambient air (kgmP3)

’//////,///////////////////////////,
Slit extract
When removing the smoke from a common balcony
reservoir, and there is no possibility of using
Fire room
downstand screens to prevent the passage of smoke
into the atrium, a slit extract system may be employed
over the length of the flow path to supplement the
underbalcony exhaust system and replace the screens
(Figure 22). Similarly, a slit extract system can be used
Section
across a room’s openings to prevent any outflow of
smoke. Figure 22 Slit extraction

19
 .- - L J .
‘Flow of hot gases out of the room of origin into false ceiling will ensure that the smoke passes through
the atrium’ (page 10) at least one sprirtkler spray en route to the extract.
‘Ventilation of the balcony space’ (page 13), and
The plenum chamber should not be larger in area than
‘Smoke layer temperature’ (page 14). its associated smoke reservoir. Larger chambers
should be subdivided by smoke screens extending the
(b) Multiply this mass flow rate by 5/3.
full height of the chamber and below the false ceiling
(c) Using the known layer convective heat flux (and to form a complete smoke reservoir below. The
allowing for sprinkler cooling, referring if minimum number of openings through the false ceiling
appropriate to ‘Effects of sprinkler systems in required within a single subdivision can be found from
smoke reservoirs’ on page 14), calculate the Table 4. The total area of such openings per reservoir
volumetric exhaust rate required from the slit, should be decided by consideration of the design
using Equation 6 to calculate the mean gas pressure differences between chamber and smoke
temperature drawn through the fan, and layer, and of the flow impedance of the openings
Equation 11 to calculate the required fan capacity. concerned. A system of reasonably wide (perhaps one-
or two-metre) slots surrounding a region of false
ceiling could perhaps be used instead of screens below
False ceilings the false ceiling.
Where there is an unbroken false ceiling in the fire
room or balcony it must be treated as the top of the
smoke layer. If the false ceiling is porous to smoke, ie
if it has an appreciable free area, any smoke screens
forming the smoke reservoir must be continued above
the ceiling. If the proportion of free area is large
enough, the reservoir and its screens may even be
totally above the false ceiling. The permeable ceiling
ought not to interfere appreciably with the flow of
smoke from the fire to the smoke ventilation openings
above the false ceiling.

It has been shown e ~ p e r i m e n t a l l ythat

~ ~ a minimum
free area of 25% can be used as a rule of thumb value
for allowing safe escape. For balcony reservoirs cool
smoke can be expected to affect some nearby rooms
under some circumstances, but would not significantly
hinder safe escape. Free areas of less than 25% are
possible in some circumstances, but expert advice
should be sought where this possibility is felt desirable.

The use of a plenum chamber above a

false ceiling
Some designs have been seen in which the space above
the mainly solid false ceiling in a roof or above a
balcony is used for the extraction of air for normal
ventilation purposes. A fan extracting air from this
space (effectively a plenum chamber) reduces its
. pressure and so draws air from the space below
through a number of openings in the false ceiling. In
the event of a fire, a fan of suitably larger capacity
starts up and draws smoky gases into the chamber in a
similar way.

A potentially valuable bonus of such a system in a

sprinklered building is that the sprinklers which are
normally required in the space above the false ceiling
will cool the smoky gases before they reach the fan.
Furthermore, it can be desirable to leave the false
ceiling below the extraction points ‘solid’ (ie not able
to pass smoke) to prevent air being drawn up through
the smoke layer. A sufficiently extensive area of solid

’ 20
Chapter 4
Smoke ventilation within the atrium
Smoke movement in the atrium this results in the most entrainment in the rising smoke
plume and lience the largest quantity of smoky gas
When the smoke and heat cannot, for various reasons, entering the buoyant layer.
be confined and removed from the room of origin or
associated balcony space, the use of 'throughflow' or The fire condition in the compartment (the design fire)
steady-state ventilation from the atrium itself is usually should be specified, and the mass flux leaving through
considered. the compartment opening and any entrainment under
the projecting balcony or canopy can be calculated as
This form of smoke control is that most readily described in the first three sections of Chapter 3.
understood by most as 'smoke ventilation' and is based
upon a defined buoyant smoke layer being established As the smoke flows through the room opening into the
at some point within the structure, with a 'clean' layer atrium space it will either:
of air beneath. The mass flow of gases entering this
layer is equivalent to that flowing out through the 0 rotate upwards around the top edge of the opening
exhaust system (Figure 23). and pass directly into the atrium space as a plume, or
0 flow under a horizontal projection such as a balcony
The base of such a layer is usually at a height chosen beyond the opening, pass to the edge of the
for safety reasons or to avoid breaching the practical projection and rise upwards into the atrium space
'cut-off' limits outlined in the section 'Limitations to as a plume.
the use of throughflow ventilation' on page 35. Once
the height of this layer base is chosen for a lowest-level
Such plumes are often referred to as 'spill' plumes, or
fire, the height above the top of the opening (or above
as 'thermal' line plumes. The term 'line' denotes that
the edge of the projecting canopy or balcony over the
the base of the plume immediately following rotation
opening where relevant) must be established where
is long and relatively narrow.
the fire is in an adjacent room.
Line plumes may take one of two forms:
Note that when the fire is on the floor of the atrium
and is directly below the smoke layer that forms under 1 Adhered plumes, where the smoky gases project
the atrium ceiling, entrainment into the rising plume is directly from a compartment opening, and the
different to entrainment into spill plumes. This special plume attaches to the vertical surface above the ,

case of fires on the atrium floor is discussed later opening whilst rising upwards. This will also occur
on page 33. when there is a vertical surface immediately above
any rotation point into the void. The surface of the
In general however, the worst condition to be catered plume in contact with the ambient atmosphere in
for is a fire in an adjacent room on the lowest level, as the atrium will cause additional air to be entrained

4 - Inlet
.........................
...................... .
........................
................
..........................
___Y.....

Figure 23 Throughflow ventilation of the atrium

'r I
I

'21
 into it (Figure 24(a)). This type of plume is also sufficient energy to become stagnant, and will then rise
known variously as a single-sided, attached or wall into the atrium space as a very long line plume
plume. (Figure 25(a)). This results in large quantities of air
being entrained and hence a very large mass flow rate
2 Free plumes, where the smoky gases project into of smoke entering the layer in the atrium roof.
space beyond a horizontal projection, eg a balcony,
thus allowing the forming plume to rise upwards
This excessive entrainment can be reduced by
unhindered. This creates a large surface area for
restricting the sideways travel of the smoke under the
entrainment on both sides of the plume along its
balcony and hence reducing the length of the line
spill width (Figure 24(b)), for which reason they are
plume. The devices used to achieve this are commonly
also known as double-sided plumes.
known as channelling screens, and literally ‘channel’
the smoke from the exit of the room to the balcony
The degree of entrainment into the rising plume, and edge (Figure 25(b)). This concept is used in smoke
hence the total quantity of gases entering the smoke control systems in multi-storey shopping centresI6.
I layer forming under the ceiling of the atrium space, is
The minimum depth required for a pair of these
screens to channel all the smoke is dependent on their
separation at the void edge ( L ) .Some values for 1 MW
and 6 MW fires in offices are given in Table 3(a) for a
balcony with no downstand obstruction at the void
edge (unimpeded flow), and Table 3(b) for a balcony
with a downstand at the void edge, which is deep
relative to the approaching layer (impeded flow).
Alternatively the minimum channelling screen depth
may be calculated using Equation 7, where WB will be
the channelling screen separation width L. Whichever
Reductions in the mass flow rate of smoke entering the procedure is chosen, the resulting depth is the
smoke layer can usually be effected by changes to (c) maximum flowing smoke layer depth and hence
and (d). In practice the height of rise of the plume is minimum screen depth. Good practice suggests that a
usually chosen to permit safe evacuation, leaving only safety margin should be considered. An additional
a dependency on the length of the line plume. depth of 0.25 m would not seem unreasonable.

Screens may be fixed or may descend upon smoke

Channelling screens detection. As described above, the final mass flow
When the atrium has a plane f a p d e with no horizontal entering the layer is a function of four initial
projections, the length of the plume is determined by parameters, one of which is the plume width at the
the width of the opening through which the smoke is balcony edge. The narrower the plume at its base, the
passing. When, however, smoke is able to flow less the mass flow entering the layer.
unrestricted under a horizontal projection, eg a
balcony, it will flow forwards towards the balcony Thus the closer the screens may be installed to each
edge, and laterally sideways. It will continue to flow other, the more the smoke base may be allowed to rise
sideways until it meets an obstruction or loses for the same heat flux.

(a) Adhered plume (b) Free plume

Figure 24 Line plumes within the atrium

22
Figure 25(a) Smoke spreading sideways beneath a projecting Figure 25(b) Smoke confined to a compact spill plume by
canopy or balcony channelling screens

These screens must, of course, meet the wall of a Entrainment into spill plumes rising
compartment where it meets the balcony. Any screen through the atrium
fixed midway across a compartment opening will serve
no purpose, since smoke will flow on both sides Calculations of entrainment into the smoke flows
simultaneously. rotating around the openinghalcony edge and into the
subsequent rising plume follow the procedures of
Recent research37suggests that channelling screens Morgan and M a r ~ h a l lfor~ ~free
? ~ plumes,
~ using the
may be unnecessary if the balcony projects no more modifications introduced by Morgan and Hanselli7.
than 1.5 m beyond the fire room. This research has Entrainment into smoke flows rotating into a rising
also shown that balconies which are shallow (<2 m) adhered plume can be calculated using the similar
will cause the rising plume to curl inwards towards the method of Morgan and Hansell17, although it should be
structure (Figures 26(a) and 26(b)). If there are higher noted that the entrainment constant appropriate to an
ba1conies.a vortex will be created between them, adhered line plume is about half that for a free
smoke-logging the balcony levels above the fire floor. ' . of 0.077 has been used for the
p l ~ m e ~ ~A. ~value
present calculations37.

'
This calculation method is outlined in detail in
Appendix B. The algorithm described can be used
either directly, or as the basis for a computer program.

(a) Deep balcony projection (b) Shallow balcony projection

Figure 26 The effect of balcony depth on plume trajectory

23
 i
-- - - - _ _ - - -

Once the desired height of the layer base (hb)has been L = Length of void edge past which gases spill (m)
chosen, the opening width established, or the
channelling screens separation L (and hence also c = Specific heat of air (kJkg-'K-l)
channelling screen depth using Tables 3(a), 3(b) or To = Absolute ambient temperature (K)
Equation 7) chosen on practicality grounds, eg such
that the screens contact the walls separating the A = Empirical height of virtual source below
rooms, the mass flow rate of smoke entering the layer void edge (m)
forming the ceiling space of the atrium can be found.
hb = Height of rise of thermal plume above
void edge (m)
Results of calculations of smoke production due to
entrainment into the rising plume are shown
graphically in Figures 27-58. Heat outputs of 1 MW It should be realised that the derivation of Equation 12
and 6 MW are considered. Downstand fascia depths of limits its application to scenarios where smoky gases
0 m, 0.5 m, 1.0 m and 1.5 m are used with an overall issue directly from the compartment on fire, with a
compartment height of around 4 m. Opening widths or balcony projecting beyond. With appropriate changes
channelling screen separations of 5 m, 10 m, 20 m and to the value of A to cater for changes in room/opening
40 m are shown for both cellular and open-plan offices. geometry, and hence the mass flow under the balcony
A choice is provided for either adhered o r free plumes. (see the second and third sections of Chapter 3), the
Equation 12 and Figures 43-58 should give broadly
similar results, since although both methods use
The results given in Figures 27-58 are representative of
different empirical approaches, these constants are
typical room/atrium geometries, with a nominal slab-
obtained by fitting to the same data36.It should also be
to-slab height of 4 m. In practice the room opening
realised that Equation 12 only describes a free or
geometry, the presence and absence of a deep
double-sided plume and cannot be adopted for an
downstand fascia and a balcony beyond, and different
adhered or single-sided plume. For a free plume, the
floor-to-floor heights will affect the mass flow rate of
appropriate changes to A may be inferred by
smoke. For example, some atria may have upper levels
inspection of the graphs and comparison with the
set back above the rooms on the storey below with no
calculated results, eg for a free plume from an open-
balcony projection beyond the lower room front. In
plan office with a 1.0 m downstand, A = 0.83H.
such designs the room walls themselves act as
channelling screens, ie in this case the width of the
room opening W is identical with the line plume length An analysis by Morgan4' for a shopping centre
L. Where such rooms have a downstand fascia, the geometry suggests that for practical purposes it can be
plume rise (hb)should be measured from the bottom of taken that A = 0.3 times the height of the compartment
the downstand. Figures 27-58 can again be used to opening H, ie A =: 0.3H, but note that this result strictly
estimate the entrainment into the plume, but a more applies for the specific experimental geometry and for
precise calculation for this case is feasible, using the the entrainment above the electric convector heaters
procedures of Appendix B. in the 'fire compartment' used to simulate the fire.

This fire engineering approach is of necessity more Recent research3' has suggested that the entrainment
complicated and needs individual consideration. into a high-temperature spill plume might be lower
than into a thermal plume. The effect is not sufficiently
well studied to allow quantitative advice to be given,
An alternative method of calculating the entrainment
beyond the statement that the effect becomes apparent
into the line plume is due to Thomas40. This treats the
for values of eB(or of 8 , where the compartment
plume in a 'far plume' approximation apparently rising
opening has no projecting canopy) greater than
from a line source of zero thickness some distance
approximately 300 "C. Where the entrained mass is the
below the void edge.
critical design parameter (eg for estimating the
capacity of powered smoke exhaust ventilators) it is
The relevant formula is: recommended that the same calculation procedures be
followed as for lower temperature thermal plumes,
since this will resiilt in an overestimate of the exhaust
gQwL2 capacity and an underestimate of the smoke-
M I= 0.58~ yib + A)[ 1 + ...(12)
reservoir's layer temperature - giving an extra margin
L
of safety. It is expected that all the calculation
procedures for spill plumes described in this Report
where M I = Mass flow of smoky gases entering the will give sufficiently erroneous results for flame
smoke layer at height hb (kgs-I) plumes (eg typically for eB(or e,) greater than about
= Density of warm gases at height hb (kgmP3) 550 "C) that they should not be employed. At the time
p
of writing it is not clear what calculation procedures
Q, = Convective heat flux in gases (kW) can be adopted for flame spill plumes.

24
 N=40

N = 20 Y=20

Y-10
N =10

Y= 5
NE 5

I I I I l l l l I
0 1 2 3 4 5 6 7 8 9 1 0 0 1 2 3 4 5 6 7 8 9 1
Height of rise ( m ) Height of rise (in)

, Figure 27 Adhered plume from open-plan sprinklered office Figure 29 Adhered plume from open-plan sprinklered office
Heat output: 1 MW Heat output: 1 MW
Downstand depth at opening: 0 m Downstand depth at opening: 1.0 m

l3I
120

/
V.40
120 -
-

w=20

d=10

J= 5

Height of rise ( m )

Figure 28 Adhered plume from open-plan sprinklered office Figure 30 Adhered plume from open-plan sprinklered office
Heat output: 1 MW Heat output: 1 MW
Downstand depth at opening: 0.5 m Downstand depth at opening: 1.5 rn

25
 460 -

420 -
I
-

W=40 461
420

380 -
~
W=40

340 -

W=20 x
L

7 300- 5cn 300- iN. 20

cn - Y -
Y
m
VI 0
VI

2 260-
r r
260-
- Y.10 m
m
c c w.10
m ?
-2 220- v. 5 2
-
220- JV. 5
r - r -
VI VI

r" 180- 5 180-

-

140 - 140 -
-

100 L 100 -

60 /U
1 -
60 -
L l I I I I I 1 I
0 2 4 6 8 10 12 14 16 18 0 2 4 6 8 10 12 14 16 18 2
Height of rise (m)
Height of rise (rn)

Figure 31 Adhered plume from open-plan unsprinklered office Figure 33 Adhered plume from open-plan unsprinklered office
Heat output: 6 MW Heat output: 6 MW
Downstand depth at opening: 0 m Downstand depth at opening: 1.0 m

V-40
N.40

360

320

v. 20 m
Y 280
Y=20

240
v=10
v=10
U- 5
Y= 5

601
0
'
2 4
'
6 8
' '
10
'
12
I
14
'
16 18 2
Height of rise lrn) Height of rise (m)

Figure 32 Adhered plume from open-plan unsprinklered office Figure 34 Adhered plume from open-plan unsprinklered office
Heat output: 6 MW Heat output: 6 MW
Downstand depth at opening: 0.5 m Downstand depth at opening: 1.5 m

26
 W=40

t W=40

'I20
l'F

i-

w=20
100
ll0I
N-20
.-- -

2 90-
VI
VI
-
0
w=10 6 80-
a,
c
m - Y=10

c 70-
VI
-
w=5 2 Y=5
60 -
-
50 -

Height of rise [m) Height of rise (m)

Figure 35 Adhered plume from cellular sprinklered office Figure 37 Adhered plume from cellular sprinklered office
Heat output: 1 MW Heat output: 1 MW
Downstand depth at opening: 0 m Downstand depth at opening: 1.0 m

'I-

0 1 2 3 4 5 6 7 8 9 10
Height of rise (m) Height of rise (m)

Figure 36 Adhcrcd plume from cellular sprinklcrcd office Figure 38 Adhcrcd plume from ccllular sprinklcrcd office
Heat output: 1 MW Heat output: 1 MW
Downstand depth at opening: 0.5 m Downstand depth at opening: I .5 m

27
 480 - -
- W.40 460 -

440 - -
420 -

400 -
380 -

360 -

w-20 340 -
I

7 320- - N = 20
YI
m
5
L

-L 300-
a
YI m
Y -
Bm 280- YI
r
w- 3
10 % 260-
m
" r w = 10
EI 240- -
.+a
-
0
-
iN-5
E w= 5
r
VI g 220-
2 200-
=
YI
-

r" 180-

160 - -

- 140 -

120 - -
-

80 -

0 2 4 6 8 10 12 14 16 18 20
Height of rise (m) Height of rise (m)

Figure 39 Adhered plume from cellular unsprinklered office Figure 4 Adhered plume from cellular unsprinklered office
Heat output: 6 MW Heat output: 6 MW
Downstand depth at opening: 0 m Downstand depth at opening: 1.0 m

N= 40

/ N =40

N = 20

Y=20

N = 10

N= 5 M.: 10

Y.5

0 2 4 6 8 10 12 14 16 18
Height of rise (m)

Figure 40 Adhered plume from cellular unsprinklered office Figure 42 Adhered plume from cellular unsprinklered office
Heat output: 6 MW Heat output: 6 MW
Downstand depth at opening: 0.5 m Downstand depth at opening: 1.5 m

28
 L=40

240 - 240 -

220 - 220-
L:20

200 - .= 20 200 -

180 - 180 -

= 160- . I 10

-
U)
.I 10 w
Y
-
U)

140-
W . I 5
r
=5 a
*
e 120-
-2
r
-
P
r" 100-
-
80 -

Height of rise (m) Height of rise (m)

Figure 43 Free plume from open-plan sprinklered office Figure 45 Free plume from open-plan sprinklered office
Heat output: 1 MW Heat output: 1 MW
Downstand depth at opening: 0 m Downstand depth at opening: I .O m

L-40

L .20

L=10

L= 5

I I I I I I I I I
0 1 2 3 4 5 6 7 8 9 1 0
Height of rise (m) Height of rise (m)

Figure 44 Free plume from open-plan sprinklered office Figure 46 Free plume from open-plan sprinklered office
Heat output: 1 MW Heat output: 1 MW
Downstand depth at opening: 0.5 m Downstand depth at opening: 1.5 m

29
 - = 10
- I10
550 -
- LE5
.=5
500 -
-
450 -
-
400 -
-
c

& 350-
Y
VI -
-
VI
m
m

al
c
300-
-
m
2
E
250-
VI
-
s 200 -
-
150-

0 2 4 6 8 10 12 14 16 18 :
Height of rise (m)

Figure 47 Free plume from open-plan unsprinklered office Figure 49 Free: plume from open-plan unsprinklered office
Heat output: 6 MW Heat output: 6 MW
Downstand depth at opening: 0 m Downstand depth at opening: 1.0 m

600-
. = 10

-110
550 -
.=5
-

500 - _ =5

450 -
-
400 -

--
VI
-
2 350-
VI
al
-
m
0
300-
al
c
m -

4
r
U)
250-

-
s
200 -

150-

100 -
-

50'
0
I
2
'
4
'
6
I
8
I
10
'
12
"
14 16
'
18 20
Height of rise (m) Height of rise (m)

Figure 48 Free plume from open-plan unsprinklered office Figure 50 Free plume from open-plan unsprinklered office
Heat output: 6 MW Hear output: 6 MW
Downstand depth at opening: 0.5 m Downstand depth at opening: 1.5 m

30
 .= 2 0

10
.=10

.=5

.I 5

t
Height of rise (m) Height of rise (m)

Figure 51 Free plume from cellular sprinklered office Figure 53 Free plume from cellular sprinklered office
Heat output: 1 MW Heat output: 1 MW
Downstand depth at opening: 0 m Downstand depth at opening: 1.0 m

. = 20

.= 10

.=10

- .= 5
-
VI
m VI ..5
1 m
VI 1
t a,
VI

m mm
r

a
.-
a
E m
L

=2
m
P
E
VI

s f

t
-1

0 1 2 3 4 5 6 7 8 9 1 0 0 1 2 3 4 5 6 7 8 9 1 0
Height of rise (m) Height of rise (m)

Figure 52 Free plume from cellular sprinklered office Figure 54 Free plume from cellular sprinklered office
Heat output: I MW Heat output: I M W
Downstand depth at opening: 0.5 m Downstancl depth at opening: 1.5 m

31
 .= 10 L=40 L = 20
. .10
600-
.=5 ~

550 -
.I5

500 -

450 -

400 -

50; ; 6 b 1;
Height of rise
1;
(m)
lk 16 18 : 0
I
2
I
4
I
6
I
8
I
10 12
I I
14 16
I I
18 2
Height of rise (rn)

Figure 55 Free plume from cellular unsprinklered office Figure 57 Free plume from cellular unsprinklered office
Heat output: 6 MW He;3t output: 6 MW
Downstand depth at opening: 0 m Downstand depth at opening: 1.0 rn

i. 10
L=5

.=5

50
"0 2 4 6 8 10 12 14 16 18 0 2 4 6 8 10 12 14 16 18 2
Height of rise (rn) Height of rise (m)

Figure 56 Free plume from cellular unsprinklered office Figure 58 Frec plume from cellular unsprinklered office
Heat output: 6 MW Heat output: 6 MW
Downstand depth at opening: 0.5 m Downstand depth at opening: 1.5 m

32
 . ,.
,- I
$ 4
1

_\.- I- . *
,
?

Fires on the atrium floor

This relatively simple case can be treated in the same
way as a fire in a single-storey space, where the plume
can rise unhindered from the fire directly into the base
of the layer. The design fire can be specified in terms
of area ( A f )and perimeter (P), based on expert where A, = Measured throat area of ventilators (m2)
assessment of the fire load at the atrium floor (which
can vary from trees to cars, from furniture to A i = Total area of all inlets (m2)
exhibitions). If known, the calorific value of the likely C, = Coefficient of discharge (usually
fuel can be used to estimate the heat flux in the rising between 0.5 and 0.7)
gases (Qf).Examples of known heat fluxes may be:
Ci = Entry coefficient for inlets (typically
(a) A group of four easy chairs clustered together, about 0.6)
forming a perimeter of around 6 m, with a heat
flux of 2 MW. MI = Mass flow rate of smoke to be extracted
(kgs - '1
(b) A sprinklered office environment (providing the
sprinklers can operate over the fire area) with a p, = Ambient air density (kgmP3)
total convective heat flux (qf) of about 115 kWmP2
of fire. g = Acceleration due to gravity (ms-*)

Note: if the atrium ceiling is high, special D B = Depth of smoke beneath ventilator (m)
provisions may have to be made to ensure €4 = Temperature rise of smoke layer above
effective sprinkler operation. ambient ("C)
(c) An unsprinklered office environment with a total Tl = Absolute temperature of smoke layer (K)
convective heat flux (qf) of about 185 kWmP2of fire.
To = Absolute temperature of ambient air (K)
(d) A vehicle (car) with a fire perimeter of 12 m and a
total convective heat flux (qf) of 2.5 MW.

If the heat flux is not known for the predicted fuel

load, a convective heat flux qf of 0.5 MW per m2 of fire
area is a usefully pessimistic rule of thumb covering
P=12m
many cases.

The mass flow rate in the plume as it enters the smoke

layer may be established from Figure 59 or Equation 1.
This procedure can be used for Y < 10.0 d A f . For
larger values of Y it would be better to seek specialist
advice on the use of 'small fire' plume theories.

Throughflow ventilation - area of

natural ventilation required
A natural ventilation system uses the buoyancy of the P= 6 m
smoke to provide the driving force for extraction. The
rate of extraction is largely dependent upon the depth
and temperature of the smoke. The advantage of a
natural ventilation system is that it is very simple and
reliable, and can cope with a wide.range of fire
conditions. Should for any reason the fire grow larger
than the design fire size, a greater depth and
temperature of smoke leads to an increased extraction
.rate, so to an extent a natural ventilation system has a
self-compensating mechanism.

The precise relationship between the mass flow rate I I I I I I I I I I

extracted, the ventilator area, the inlet area and the 1 2 3 4 5 6 7 8 9 1 0 1 1
Height of smoke base (m)
smoke layer is42:
Figure 59 Rate of production of hot smoky gases from a fire
on the atrium floor

33
Table 5 gives the minimum free area of ventilation overpressure ancl suction have been identified for all
required, ignoring the effect of any inlet restriction (ie possible wind directions, design of ventilators or fans
assuming an infinite area of inlet ventilation). can proceed as before.

The effect of limited fresh air inlets can be allowed for, A powered extract system should be used where
using the following approximations: positive wind pressures are likely to be a problem, or
where it is necessary to extract smoke via an extensive
0 If the inlet area to the atrium is twice the exhaust
ductwork system.
ventilation area given by Table 5, the indicated
ventilation area and the inlet area should both be
increased by approximately 10%.
If the inlet area is equal to the exhaust ventilation
area, the indicated ventilation area and the inlet Table 5 Minimum total ventilation area (m2)
needed
area should both be increased by approximately 35%. for a smoke reservoir
(from Equation 13 with C, = 0.6)
0 If the inlet area is half the exhaust ventilation area,
the indicated ventilation area and the inlet area (a)Q = 1 M W
should both be increased by approximately 125%.
Mass flow rate Smoke depth beneath ventilators (m)
When natural ventilators are used for smoke (exhaust rate) ~

extraction, it is important that they are positioned (kgs-I) 1.5 2 3 4 5 7 10

where they will not be adversely affected by external 4 2.1 1.8 1.5 1.3 1.1 1.0 0.8
wind conditions. A positive wind pressure can be much 6 3.2 2.8 2.3 2.0 1.7 1.5 1.2
greater than the pressure head developed by a smoke 8 4.5 3.9 3.2 2.7 2.4 2.1 1.7
layer. Should this occur the ventilator may act as an 10 5.9 5.1 4.1 3.6 3.2 2.7 2.3
inlet rather than as an extract. However, if sited in an 12 7.4 6.4 5.2 4.5 4.0 3.4 2.9
area of negative wind pressure, the resultant suction 15 9.9 8.5 7.0 6.0 5.4 4.6 3.8
force on a natural ventilator would assist smoke 20 14.5 12.5 10.2 8.9 7.9 6.7 5.6
25 19.6 17.0 13.9 12.0 10.8 9.1 7.6
extraction. 30 25.3 21.9 17.9 15.5 13.9 11.7 9.8
35 31.4 27.2 22.2 19.2 17.2 14.5 12.2
Tall buildings or taller areas of the same building (such
40 37.9 32.9 26.8 23.2 20.8 17.6 14.7
as rooftop plant rooms, etc) can create a positive wind 50 52.2 45.2 36.9 32.0 28.6 24.2 20.2
pressure on nearby lower roofs. Steeply pitched roofs, 60 67.9 58.8 48.0 41.5 37.2 31.4 26.3
ie roofs over 30" pitch, may also have a positive wind
pressure on the windward slope.
(b)Q = 6 M W
A suggestion sometimes advanced for offsetting wind
overpressure is to increase the total area of natural Mass flow rate Smoke depth beneath ventilators (m)
(exhaust rate) -
ventilation per reservoir. Since the overpressure is, by (kgs-I) 1.5 2 3 4 5 7 10
definition, force per unit area, this will usually not
work and indeed could exacerbate the problem by 10 5.5 4.7 3.9 3.3 3.0 2.5 2.1
allowing even greater quantities of air to be driven 12 6.4 5.5 4.5 3.9 3.5 2.9 2.5
through the ventilator to mix into the smoke. 15 7.8 6.7 5.5 4.8 4.3 3.6 3.0
20 10.2 8.9 7.2 6.3 5.6 4.7 4.0
25 12.9 11.1 9.2 7.9 7.0 6.0 5.0
In some cases it may be possible to retain natural
30 15.6 13.5 11.1 9.6 8.6 7.2 6.1
ventilation openings in a vertical plane by arranging
35 18.6 16.1 13.1 11.4 10.2 8.6 7.2
them to face inwards to either a region sheltered from 40 21.6 18.7 15.3 13.2 11.8 10.0 8.4
wind action, or where the wind will always produce a 50 28.2 24.4 19.9 17.2 15.4 13.0 10.9
suction. In other cases the erection of suitably 60 35.2 30.5 24.9 21.6 19.3 16.3 13.6
designed screens or wind baffles (outside the vertical 75 46.7 40.4 33.0 28.6 25.6 21.6 18.1
wall or window holding the ventilators) can overcome 90 59.2 51.2 41.8 36.2 32.4 27.4 22.9
wind interference and may even be able to convert an 110 77.2 66.9 54.6 47.3 42.3 35.8 29.9
overpressure into a suction. There is also the possiblity 130 96.8 83.8 68.5 59.3 53.0 44.8 37.5
150 117.8 102.0 83.3 72.1 64.5 54.5 45.6
of selectively opening ventilators in response to signals
from a wind direction sensor. Expert advice should be 200 175.9 152.3 124.4 107.7 96.3 81.4 68.1
300 313.1 271.1 221.4 191.7 171.5 144.9 121.2
sought for such designs. 400 474.2 335.3 290.4 259.7
410.7 219.5 183.7

Due to the complexity of wind-induced air flow over Note: To account for the restriction imposed by the inlet area, add
some atrium buildings and the surrounding buildings, the following to both inlet and exhaust ventilation areas:
it may sometimes be desirable to carry out boundary 10% if the inlet area is twice the ventilation area,
layer wind tunnel studies to establish the wind 35% if the inlet area is equal to the ventilation area, and
pressure over the building's envelope. Once areas of 125% if the inlet area is half the ventilation area.

34
 ;.
y-: . . .. . A .. .
{, .* ;I
I

Throughflow ventilation - remaining smoke. This problem of early stratification can to some
extent be overcome by providing smoke detectors at
design procedures many heights within the atrium or located to ensure
Other design calculations are essentially the same as detection of smoke close to the fire. Once a forming
described in the previous chapter, eg mean smoke smoke stratum is detected and the smoke ventilation
layer temperature (page 14), flowing layer depth system caused to operate, the hottest (and therefore
(page 16), inlet air (page 17), minimum number of highest) gases will be removed first, allowing any
extraction points (page 18) and required ventilation cooler strata to rise to take their place. Hence smoky
rate of powered exhaust ventilators (page 19). gases will reach the ventilators and the smoke
ventilation system should settle into its ‘design’ state.
The timescale for this process is uncertain and hence
Limitations to the use of throughflow early detection of smoke in these circumstances
ventilation is essential.
As may be seen from the graphs in Figures 27-58, the A further problem which could be encountered may be
mass flow rate generated by the entrainment into the more problematical during cooler weather. Atria with
rising plume is very large, and hence the plume cools large areas of external glazing will present a large surface
quickly with height. This large increase in mass flow area to the smoke layer, which can lead to considerable
with increases in height tends to suggest there may be heat losses from it. Most throughflow smoke control
some cut-off point in the rise of the plume, above which systems are designed with an arbitrary limitation to the
it might become economically impracticable in terms of ceiling reservoir of between 1000-3000 m2
a smoke control system. Experience suggests that this is (see for example References 16,36 and 42), one reason
often true for flows larger than 150-200 kgs-I. being to prevent excessive energy loss from the buoyant
smoke layer. Many atria cannot physically or
Another effective limit may occur if the temperature of architecturally adopt such reservoir formations and, if
the smoky gas layer forming in the roof void is too low. larger than the areas mentioned above, will cause
If internal day-to-day heat gains (solar, plant, etc) are additional energy to be lost from the layer.
allowed to accumulate within the atrium roofspace (eg
passive solar atria) then high-level air temperatures This energy loss will increase with distance, the further
within the atrium may be very high. Roofspace the smoke has to travel from the fire source, and will
temperatures have been recorded at or above 50 “C. manifest itself as a loss of buoyancy within the flowing
Smoke spreading into an atrium during the incipient layer. This in turn can cause the layer to deepen
stages of a fire will naturally be very cool, and the beyond the desired design depth, perhaps
entrainment processes will draw in the surrounding considerably so.
ambient air as the plume rises. In most instances this
ambient air will be at or near 20 “C ( due either to Cool smoke will also be sensitive to airflow
ventilation or air-conditioning), producing a plume movements, such as air currents (draughts) due to
temperature which may be considerably lower than the ventilation, air conditioning or weather conditions.
air within the roofspace. Furthermore, experimental evidence3’ has shown that
excessive air movement (such as that which may occur
Unless the hot air can be removed sufficiently quickly, due to an arbitrary air change rate) into a cool but
it will result in the initial smoke layer forming at a otherwise stable smoke layer can cause it to become
point lower down in the building than may be unstable, spreading further throughout the building.
desirable. This process is known as early (or The formation of a smoke layer depends upon
premature) stratification (Figure 60). As the fire is buoyancy for the maintenance of stability. Smoke
probably growing, the plume temperature will layers which have temperatures (and hence densities)
progressively rise with time. This may result in hotter approaching that of the incoming replacement air
smoke ‘punching’ its way through the cooler smoke supply will have a tendency to ‘mix’ with this air, rather
layer and forming another warmer layer above. The than ‘float’ above it. This process is known as dilution
process may continue until the smoke ‘strata’ have ventilation and is frequently used in industry to reduce
become sufficiently mixed to rise up as a single bulk of contamination levels in buildings (eg welding shops).

Figure 60 Early (or premature) stratification

35
The mechanisms involved in dilution ventilation can
easily induce downward mixing of a smoke layer to the
extent that, with sufficient air movement, complete
smoke-logging of an atrium can occur. It follows
therefore that the atrium smoke layer should be at a
temperature compatible with stable stratification.

There is little information available on the'

destabilisation of cool buoyant iayers, so a precise
limiting temperature beyond which the above effects
will lessen cannot be given. Further research is
desirable in this area:Experience and experimental
observation however indicate that these effects may be
severe in terms of smoke control, perhaps leading to
smoke spreading to otherwise unaffected escape
routes.

In the absence of the necessary experimental data, as a

result of practical experience this Report will adopt a
temperature band of between 15-20 "C above ambient
as the critical layer temperature below which
undesirable effects may occur. This temperature rise
should be regarded as that which the layer will have
after suffering heat losses to the structure containing it
(see Chapter 7).

The practical limitations to the use of throughflow

ventilation are therefore a maximum mass flow rate of
150-200 kgs-' ,and/or a minimum smoke layer
temperature of 15-20 "C above ambient.

Which limit is reached first will depend upon the

situation being considered, ie on the type of fire, the
construction of the compartment, the geometry of the
atrium, etc.

Experience (and Figures 27-58) suggests that one or

other limit is usually reached when the height of rise
above the fire room opening exceeds 8-12 m. It
follows that it does not usually appear to be
practicable to design a throughflow ventilation system
requiring more than three to four storeys (sometimes
less) to be kept free of smoke, regardless of whether it
is powered or natural smoke ventilation.

This limitation would appear to pose a serious threat

to the design of interesting atria, and indeed building
codes and fire safety standards in the USA are
apparently reflecting this43.The large, open atrium
designs for which American architects have become
renowned are no longer widely regarded as acceptable,
and new atrium designs are sometimes restricted to a
maximum of three open (non fire-separated) floor
levels, one of which must be the ground floor of the
atrium. However, various methods exist whereby this
limitation may be overcome to a certain extent.

36
 -1 - - -_ -- 1

Chapter 5
Design considerations other than throughflow ventilation
Void filling material contained upon the atrium floor. There can
be no areas ’of public movement within the atrium
Some atria provide large available volumes in which space other than at ground-floor level.
any smoke from a fire could be contained, such that
smoke control/ventilation may be unnecessary. Since there is the potential for the atrium to be wholly
full of smoke, the facade should be well sealed. If the
This approach is usually based upon the assumption gases in the atrium become hot, as they often may
that a fire will grow at a predictable rate, and that the locally to the fire, the facade materials and
quantity of smoke generated can be contained safely in construction, and the sealing techniques used must be
the roof void during the evacuation period without able to withstand these higher temperatures.
prejudicing the evacuation of occupants of the space.
Such atria may be fitted with means of removing
This relies upon quantitative predictions of both fire smoke for fire service use. These systems are often
growth and personnel escape times. Fire growth is provided on an arbitrary design basis, usually
difficult to predict during the very early stages of comprising an air change rate if powered ventilation is
development and can therefore, at best, be only a used, or a percentage of the atrium floor if natural
rough estimate. Similarly, actual times needed for ventilation is used. These systems are purely for fire
evacuation are also extremely difficult to determine. service use only, for clearing of residual smoke
has shown that escape periods in multi-storey (usually post-extinction) and must not be regarded as
buildings can vary from 10 minutes for a 15-storey life-safety systems.
building, where escapees were ‘caught’ on the
thirteenth floor for 5 minutes before being able to
descend, to 31 minutes for a 21-storey structure where Depressurisation ventilation
escapees were ‘caught’ on the twelth floor for Principles
20 minutes. A similar exercise in a public complex in Greater architectural freedom becomes possible if the
the UK (St David’s Centre, Cardiff) has shown a total atrium facade need not be sealed, but can be allowed
evacuation time in excess of 30 minutes. to be leaky, even if the upper atrium is filled with
smoke. Examples of such ‘leaky faqade’ designs might
It should also be remembered that in the UK it is include:
customary for the fire service to search buildings for
trapped or lost escapees. There will be some designs Hotel bedrooms having doors on to ‘decorative’
where the evacuation times will be shorter than the balconies overlooking the atrium (ie not access or
time for smoke to endanger the escape routes. It escape routes), small enough to be evacuated
follows that the smoke control option of ‘doing through the doors in a few seconds.
nothing’ should not be ruled out completely, but Where unsealed windows are used for simplicity
should only be accepted when supported by fire and cheapness.
engineering calculations embodying appropriate
safety margins. Where small ventilation openings allow air to
circulate between the accommodation spaces and
the atrium. Clearly there must be no escape routes
Compartmegt separation open to the upper atrium.
One approach that may be considered for the
protection of the atrium from fires in adjacent rooms If such doors and other such leakage paths do not have
(or vice versa) is the concept of the ‘sterile tube’, tight seals, smoke from the atrium may enter many
which is outlined in the Introduction. adjacent rooms o n many levels, causing a loss of
visibility in those rooms and possibly affecting escape
In this instance the atrium is glazed throughout with routes away from the atrium (Figure 61).
fire-resisting glass or its engineered equivalent. Thus
there is no opportunity for hot smoky gases to enter This might happen simultaneously on many floors,
rooms adjacent to the atrium, and the building fire requiring the simultaneous evacuation of all affected
safety precautions revert to those found in the absence floors, thus adding to the pressure of use on escape
of an atrium. The obvious advantage is simplicity. routes elsewhere in the building. This is likely to be a
particular problem where there is a ‘sleeping risk’, eg
The technique has several disadvantages. It is rather atrium hotels. It will also be a problem for firefighters,
restrictive for building designers, as the atrium cannot since they may feel the need to search all
be utilised as a functional space, and generally there accommodation on all the affected floors to ensure
must only be limited quantities of combustible that no-one remains at risk. Such a search would be

37
 ........... 1 . . * ..... ~.,. . .

Figure 61 Smoke..logging in a ‘leaky’ closed atrium

much quicker if all accommodation were kept clear under a positive pressure (defined positive outwards
of smoke. from the atrium). Thus there will be a flow of smoke
from the atrium into rooms above the neutral pressure
Hence smoke must be prevented from passing in plane through any leakage path which may exist.
appreciable quantities through these small leakage
openings. One way of achieving this may be by However, careful manipulation of the neutral pressure
depressurising the a t r i ~ m ~ ~ . ~ ~ . plane can raise it to a safe height above sensitive
levels, where there is little or no threat from the
Natural depressurisation positive pressure above (Figure 64). The pressure in
In any structure with natural ventilation openings at the atrium below the neutral pressure plane will be at
high and low level, and with a quantity of heat trapped a pressure lower than ambient, thus any airflow will be
inside, a ventilation rate will be created due to the from the room into the atrium. Hence the levels below
‘stack effect’. the neutral pressure plane are protected from heat and
smoke contamination.
In order for air to move out through the high-level
opening, the pressure at high level inside must be Appendix A gives a description of a fire that occurred
greater than the external pressure otherwise there in the IMF Building in Washington. It started on the
would be no air movement. Similarly, for air to flow tenth floor of a 13-storey atrium, and by the time the
inwards at low level the pressure at low level inside fire service arrived (16 minutes later) the smoke level
must be less than that outside. Thus there must be a had descended below the tenth floor.
position within the structure where the pressure inside
is equal to that outside. This is known as the ‘neutral An interesting aspect of this fire was that, despite the
pressure plane’ (NPP). Any openings situated at the presence of a natural ventilation system in the roof,
neutral pressure plane will have no airflow through the atrium became completely smoke-logged at one
them, as there will be no pressure differential at point. This apparent failure of the venting system was
that point. attributed to the use of natural ventilation in a ‘tall’
building, where the smoke had insufficient buoyancy
In buildings where a throughflow ventilation system is to reach the vents.
installed, where the inlet area is equal to the exhaust
vent area, then the neutral pressure plane will exist However, the fire occurred on the tenth floor, and for
approximately midway within the smoke layer all practical purposes, when the fire broke out it was
(Figure 62). If the inlet vent area is smaller than the effectively in a three-storey building with a deep
exhaust vent area, then the neutral pressure plane will basement. Natural ventilation works extremely well in
move upwards (Figure 63). ‘shallow’ buildin,gs, and therefore there must have
been some other mechanism in action affecting the
Any openings above the neutral pressure plane will be operation of the ventilation system.

..............
...........
Open ventilators

Figure 62 Neutral pressure plane -

throughflow ventilation

38
 Figure 63 Neutral pressurc planc -
vent largcr than inlet

The atrium had no apparent inlet facility and Equation 14 represents the condition where the atrium
accordingly, instead of the ventilators providing a has a single, dominant inlet leakage path from the
throughflow ventilation effect, the atrium became exterior (eg access doors) but smaller leakage paths
depressurised in the’manner described above. This in between the atrium and accommodation and the
turn prevented smoke from spreading beyond the exterior (Figure 66).
atrium, despite being smoke-logged to ground-floor
level at one stage. With the technique as described above it is quite possible
for the atrium to be entirely filled with smoke (see
The neutral pressure plane will lie somewhere within Appendix A - IMF Building), in which case D,,,, will
the depth of the smoke layer in the atrium depending approach the height of the atrium (H:,),eg D,;,, + H,.
upon factors such as inlethent area ratio, gas
temperatures, wind pressures, etc. It is not, and should It is a straightforward task to calculate the ventilation
not be confused with, the actual base of the smoke requirements for a ‘pure’ depressurisation system
layer. using Equation 14 or Figure 65 where the smoke layer
temperature is known or can be determined as shown
The equation describing the above relationship, in the in Chapter 6.
absence of wind effects is45.46:
If the neutral pressure plane were to descend below
the desired design depth then some of the higher
storeys may become endangered. This can arise from
...(14) an increase in the actual inlet leakage area available,
for example, where the fire brigade have opened
access doors to the atrium to investigate the severity of
the fire. A successful depressurisation design should be
where X = The height from the base of the smoke able to prevent smoke infiltration into adjacent spaces
layer to the desired position of the on the higher floors even in this condition.
NPP (m)
D,,, = Maximum depth of smoke layer from In addition it is possible that the fire may cause
the centre line of the exhaust windows to break on both the external f a p d e and t h e
ventilator (m) atrium f a p d e of the fire room. In this case the broken
areas can act as a ‘dominant’ leakage path from,the
This equation is represented graphically in Figure 65. exterior.

Figure 64 Ncutral pressure planc abovc highest

‘leaky’ storey

39
 storeys above the layer’s base the same
depressurisation principle can be employed, but a
more complicated ‘flow network’ calculation must be
used. This is best left to specialists in the field.

It is difficult to give a simple general rule to identify

when a building can be regarded as having a single
dominant inlet. Nevertheless, it may be sufficient to
adopt a guideline from the related field of ‘air
infiltration’, so that one can assume a dominant inlet if
the total area of all openings below the layer base is
more than twice the total area of all openings above
the layer base (excluding the area of the ventilators
them~elves)~’.

Natural depressurisation and wind effects

The neutral pressure plane is sensitive to the effects of
wind, and ‘adverse’ wind pressures might cause the
NPP to fall to a lower position on the leeward side of
the building, po!jsibly contaminating the topmost
leeward storeys. It follows that the depressurisation
design procedure must take wind force into account.

To assess the efficiency of operation of a

depressurisation system a knowledge of the wind
00
pressure coefficients acting upon a building will be
00
30 - OC
necessary. These are a well established way of relating
60 the wind pressure anywhere on a building to the wind
10 velocity at roof level.
I 1.5 2.0

X Wind pressure coefficients have often been measured

so that structural wind-loading can be calculated.
Figure 65 Solution t o neutral pressure plane equation (14) There is a considerable body of data in existence.
Where complete certainty is required for a novel or
complicated building, wind-tunnel observations using
scale models will yield usable results. In general
Thus all potential inlet leakage paths must be assessed however it should often be possible to obtain
when using Equation 14 or Figure 65. reasonable values for the wind pressure coefficients
needed for smoke control calculations from available
It should be noted that the simple approach set out literature (see Reference 47 for example).
here will be invalid where the leakage paths across the
atrium boundary have appreciable areas on several Figure 67(a) shows the typical three-dimensional
storeys (although all leakage areas below the smoke complicated pat tern of wind pressure coefficients over
layer’s base can be aggregated and regarded as being a tall tower In practice it would be necessary
at the layer’s base for calculation purposes when using to identify the most pessimistic values for each storey,
Equation 14 or Figure 65). Where there are in which case the problem can be simplified to two-
appreciable significant leakage paths on several dimensional as shown in Figure 67(b).

en ventilators

Figure 66 Neutral pressure plane -dominant inlet

40
(a) Three-dimensional distribution for Providing the requirements of Equation 15 are
typical tower block satisfied then natural ventilation will work at all wind
speeds. This implies that the roof ventilation system
should be subjected to suction wind pressures at all
times. However, if it is impossible to employ a natural
ventilator on a particular building, fans can be used
instead.

POwered depressurisation
The necessary capacity is a little harder to calculate,
and the best fan is one which is not affected by wind
pressures on its exhaust. With a fan however, a
maximum wind speed must always be assumed for
design purposes. The required volumetric flow rate
may be calculated from4?

I12

...(17)

where VI = Fan capacity required (m3s-I

(b) Two-dimensional distribution for
V,i,,d = Design wind velocity (ms-I)
typical tower block

A natural smoke control system will be affected by the

wind pressures operating against all the openings in
the structure; thus pressure differentials vary with
wind direction and opening position, and the
throughflow of air will vary with wind velocity.

However when the hole in the roof is replaced by a

fan, the pressure differentials within the building now
have to be changed by mechanically altering the
Figure 67 Wind pressure coefficients
throughflow of air. Therefore the system must be
around buildings
designed with a maximum design wind velocity to
cater for all conditions.

With these data established for any specific building, Further sophistication may be achieved by the use of
the design procedure for checking on the performance an anemometer and by having ‘groups’ of fans, each
of a natural depressurisation system is fairly simple group operating at a different wind velocity. So if the ,

where there is a single dominant opening. wind was light, one group would operate and, if the
wind speed increased, further groups might be
To prevent smoke leakage into the top leeward storeys activated as necessary.
for all wind speeds4?

(A-I)C,v-AC,L+C,i] I 0 ...(15 )

where C,, = Wind pressure coefficient at the vent

CpL=Wind pressure coefficient at the
topmost leeward storey of the building
CPi =Wind pressure coefficient at the inlet

2
To AVCV
and A=- +1 ...(16)
7i

41
Chapter 6
Depressurisatiodsmoke ventilation hybrid designs
Principles Hybrid designs usually follow one of two approaches:

In Chapters 2 to 4 we have indicated how smoke 1 Mass flow based, where the atrium is designed with a
ventilation can only keep a limited number of lower number of open levels above the atrium floor, requiring
storeys clear of smoke below the buoyant smoke layer a plume of a specific height. The maximum number of
formed in the atrium. The technique does in principle, levels will be determined by either the magnitude of the
however, allow those lower storeys to have adjacent mass flow rate entering the layer, or the smoke layer
spaces - and their escape routes - open to the atrium. temperature falling below the minimum value of
The section ‘Depressurisation ventilation’ on page 37 15-20 “C (see Chapter 4).
shows that it is often possible to design a
2 Temperature based, in order to cool a potentially hot
depressurisation system where clean air is drawn
smoke layer by the deliberate entrainment of ambient
through all significant leakage openings on the atrium
air into the rising plume. This may enable the use of
faqade immersed in the smoke layer.
faqade materials that cannot withstand high
temperatures (eg float glass).
Depressurisation does not however protect any large
leakage openings on any storey above the layer base in
the atrium, nor will it protect any escape routes on that Design procedures for hybrid systems
storey open to the atrium. In this context a large opening
is one where the opening in the atrium faqade is larger Massflow based systems (see Figure 68)
than the sum of openings further along the same leakage (a) Determine the height of rise of the smoke plume
path away from the atrium (eg if the atrium faqade required to clear the open levels (hb),with the
opening is larger than openings in the external wall). design fire (from Chapter 2) chosen on the lowest
open level. This will also yield the smoke layer
But it will often be the case that architects will want to depth (D), measured from the centre line of the
maximise use of the atrium space, and an obvious way is ventilator.
to combine the smoke ventilation approach of Chapters 2 (b) From Figures 27-58 or by detailed calculation and
to 4, allowing greater freedom of design on the lowest with the desired channelling screen separation ( L )
storeys, with the lesser freedom of ‘leaky faGades’ or opening width ( W ) ,determine the mass flow
allowed by the depressurisation technique set out in rate ( M , ) entering the base of the layer. If the fire
Chapter 5. In this ‘hybrid’ design the ratio of vent area to is on the atrium floor, determine M I using the
fresh-air inlet area will be determined by Equation 15, section ‘Fires on the atrium floor’ on page 33.
whereas the actual values of these areas must be
consistent with the necessary smoke extraction (c) Calculate the total surface area of the smoke layer
requirement as defined in the two sections on (the atrium surface area in contact with the
throughflow ventilation design in Chapter 4 (pages 33 smoke layer plus the area of the layer base), and
and 35). It should be appreciated that in such a hybrid determine the likely smoke layer temperature,
design the smoke layer temperature in the atrium using Chapter 7. If the smoke layer temperature is
required for the depressurisation calculations is a below 15-20 “C above ambient then the number
natural outcome of the plume entrainment calculations of open levels may need to be reconsidered, or
needed for the smoke extract calculation. Note that some (or all) of the lower levels vented
hybrid designs are similarly possible where powered independently from the atrium, using the
ventilators are used for atrium smoke exhaust. procedures set out in Chapter 3.

Neutral pressure plan

Ai
Figure 68 Principles of hybrid smoke ventilation system -
mass flow based

42
 '...l
* . 1

. . . . . -- .~ .. ... ... .- .. . -. .
'
'-4
I

(d) Set the neutral pressure plane height ( X ) to that (c) From Figures 27-58 or by detailed calculation and
required above the base of the smoke layer, and with the channelling screen separation ( L ) or
determine the value of (A,CvIA,C,)2from opening width (W), determine the height of rise
Equation 14 or Figure 65. ( h b ) to the base of the layer, necessary to give the
required mass flow rate.
(e) With these values of (A,Cv/A,C,)2, DB, M I and
01calculate the ventilation area required from (d) With the design fire at the lowest level and taking
Equation 13. into account the necessary height of rise (hb) for
cooling purposes, determine the maximum smoke
(f) With the values of (A$, /A,CJ2 and A,C, layer depth (Dmax). Set the neutral pressure plane
calculate the quantity of inlet ventilation required. height ( X ) to that required above the base of this
In the event that the actual inlet area available is smoke layer depth, and determine the value of
greater than that required by calculation, then the (A,Cv/A,C,)2from Equation 14 or Figure 65.
ventilation area should be increased to maintain
the ratio of (A,C,/A,C,). (e) With the required value of hb, determine the
shallowest smoke layer depth (Ds),compatible
(g) Using Equations 15 and 16 and the appropriate with the depressurisation concept (this is often the
wind pressure coefficients, check the system second level beneath the NPP).
operation with regard to wind effects.
(f) With this value of (A,Cv/A,C,)2,DB,M Iand €4
(h) In the event that the wind effects may adversely calculate the ventilation area required from
affect the operation of a natural ventilation Equation 13. In the event that the actual inlet area
system, calculate the fan capacity required using available is greater than that required by
Equation 17, with an appropriate value of design calculation, then the ventilation area should be
wind velocity. increased to maintain the ratio of (A$, /A,C,).
(i) Check that the anticipated suction pressure and/or (8) Using Equations 15 and 16 and the appropriate
air inflow velocities d o not in themselves endanger wind pressure coefficients, check the system
the safe use of any escape routes away from the operation with regard to wind effects.
atrium (see the section 'Inlet air' on page 17).
(h) In the event that the wind effects may adversely
Temperature based systems (see Figure 69) affect the operation of a natural ventilation
(a) Decide upon a smoke layer temperature rise (0,) system, calculate the fan capacity required using
compatible with the faqade material employed. Equation 17, with the appropriate value of design
For float glass a temperature rise of 70 "C above wind velocity.
ambient will give a reasonable safety margin to the
(i) Check that the anticipated suction pressure and/or
system design. Toughened glass may be capable of air inflow velocities d o not in themselves endanger
withstanding higher temperature rises (eg 200 "C). the safe use of any escape routes away from the
(b) Calculate the total surface area of the smoke layer atrium (see the section 'Inlet air' on page 17).
(the atrium surface area in contact with the smoke
layer plus the area of the layer base), and
determine the mass flow rate required t o give the
desired temperature rise, using Chapter 7. As a
simplification incorporating a margin of safety,
this step can be omitted and the mass flow rate
calculated using Equation 6.

.......
...............
..............
.................
...................
...................
.................
.................
.........
..................
................
..........
...........
...........
iii/ijjiir Centre line of ventilators AvCv
Neutral pressure plyne A-,-
--q---

Figure 69 Principles of hybrid smoke ventilation system -

tcmpcrature based

43
 .^.

Chapter 7
Atrium smoke layer temperature
In most smoke control system designs no account is
taken of the heat losses to the structure. It is assumed
there is conservation of heat and that all of the heat
flux entering a smoke reservoir is contained in, and
remains in, the smoke. Experimental work in the past
has shown that for relatively small smoke reservoirs
with medium to high thermal resistance, or for high
mass flow rates of smoke, this assumption holds good.

When considering atria however, the assumption can

no longer be considered entirely valid. An atrium
generally has a large surface area, which is
predominantly glazed in most cases, thus providing a
good heat sink. There will be a passage of heat energy
from the smoke layer into the structure, and
accordingly the smoke layer will suffer a reduction in
temperature.

Figure 70 shows the heat balance in an atrium. This

model was used to determine the loss of energy from
the smoke layer, based upon 'worst case' assumptions
for the f a ~ a d eThe
~ ~ fasade
. fabric is assumed to be
thin glazing, with no apparent delay in the transfer of
the energy from the layer. The results of using this
model are shown graphically in Figures 71,72 and 73
for 1 MW, 5 MW and 6 MW fires respectively. As can
be seen from the graphs - despite the fact that many
differing atrium geometries were considered, with
different values of external exposure - the resultant
calculation points may be plotted comfortably as single
curves for each value of mass flow rate.

At high values of mass flow rate there is little change

in the atrium smoke layer temperature for wide
variati0ns.h smoke layer surface area. This is due to
the gas flow being the prime mover of energy, and
tends to justify the assumption that loss of heat to the
structure of a building may be ignored for relatively
small contact areas.

Heat carried by
exhaust gases
Heat loss
through roof
Open ventilators
or powered extract units

Figure 70 Heat balance in an atrium

44
140

130

Q=lMW

M i = 100 kgs-'
,o
Figure 71 The atrium layer
0; 1 I
4 6
I I
8 ,b ,h 1L ,b 2b d2 24

Total surface area of smoke layer ( including base ) ( m2 x 1000)

Q =5MW

40 -

I I I I I I I I 1 I Figure 72 The atrium layer

4 6 8 10 12 14 16 18 20 22 :
temperature (Q = 5 MW)
Total surface area of smoke layer (including base) ( m z xl000)

Me = 100kgs-'
6o

40 -

Figure 73 Thc atrium laycr

0 2 8 10 12 14 16 18 20 22
temperature (Q = 6 MW)
Total surface area of smoke layer (including base ) ( m 2 x 1000 )

45
Chapter 8
Additional design factors
Atrium roof-mounted sprinkler systems This increased pressure differential will increase the
air flow through the leakage paths of the lobby, thus
Conventional sprinklers mounted in the roof of an enhancing the efficiency of the pressurisation system in
atrium will only be of potential benefit if there is a fire preventing the passage of smoke into the escape route.
in the atrium floor itself. However, due to the height
that these sprinklers will generally be mounted above
the floor, the fire will be of considerable proportions Air-conditioned atria
before the sprinklers activate and they will therefore In some hot countries it is common practice to totally
be of limited benefit. More importantly perhaps, if a air-condition an atrium, so that its internal ambient
smoke layer is just above the operating temperature of temperature is lower than the external ambient
the sprinklers, it will be reasonably stable (e,2 50 “C). temperature. In this case the use of a natural
The action of all the roof sprinklers operating almost ventilation system would cause a reverse stack effect
simultaneously (as can be inferred from work by (ie vents acting as inlets) and could cause problems of
H i r ~ k l e y may
~ ~ ) rapidly cool the layer and cause it to cool smoke spreading downwards towards the escape
become unstable. This can occur if a fire is not at the doors. Once the cool atrium air has been flushed out
atrium floor, and is therefore highly undesirable. by the warmer ambient air entering through the vents,
the system will reverse its direction of flow.
If there is a likelihood that a fire load will be present
on the atrium floor, the smoke control system designed This problem of an initial downward movement of
for a fire in an adjacent room will generally be able to smoke may be alleviated by the use of smoke detectors
cope with the products of a much larger fire directly in the rooms, rather than the atrium, causing the
under the roof17.Thus with good housekeeping and ventilation system to operate and creating a balance
proper management to restrict the use of fuel between the internal and external temperatures prior
assemblies, such a fire could feasibly exist and burn to the smoke entering the atrium in quantity, or by
itself out before becoming a problem to other areas using powered ventilators.
outside the atrium.
ling screens and hybrid systems
Smoke detection sysftems in the atrium It has been shown in Chapter 4 (‘Channelling screens’,
The operation of any smoke control system in an page 22) that when smoke passes under a balcony to
atrium is generally dependent upon a smoke detector rise into the roof void above, the quantity of smoke
operating. If the atrium is particularly tall, 1
entering the smoke layer in the rising smoke plume
stratification of the smoke layer before reaching the can be reduced by restricting the width of the plume as
ceiling is probable - especially in atria which are air- it passes the balcony edge, by the use of channelling
conditioned in the lower portion only, or have a high screens.
proportion of roof glazing (see the section ‘Limitations
to the use of throughflow ventilation’ on page 35). The need for plume width restriction is necessary for
Thus if the only detection system present is roof- any smoke conlrol design where a clear layer of air is
mounted it may not operate, or at least its operation required above a balcony projection beyond the fire
may be considerably delayed. This problem can be room (eg for escape purposes), and so will apply to a
overcome by the installation of smoke detection to the hybrid smoke control system when the height of rise of
various rooms or by intermediate detection zones at the smoke plume is fixed by escape requirements
different heights in the atrium, possibly using beam (mass flow based systems).
detectors.
As described in the sub-section ‘Temperature based
systems’ on pag,e 43, an alternative use of a hybrid
Pressurisation of stairwells an system is to cool the smoke layer for some purpose (eg
In some atrium buildings there may be a requirement to prevent glazing from cracking) by deliberate
or desire to pressurise the escape stairs and associated entrainment of air into the smoke plume (temperature
lobbies. If the atrium employs a depressurisation or based systems). When designing a natural ventilation
hybrid smoke control system and the glazing between system for this purpose a knowledge of the depth of
the fire room and the atrium has cracked or shattered, the smoke layer in the atrium is necessary to calculate
the pressure within the fire room will, of necessity, be the vent area required. This in turn implies a
lower than the outside ambient pressure. This knowledge of the height of rise of the smoke plume.
reduction in pressure will act as though an extract fan Therefore an estimate of the plume width leaving the
were fitted to the fire room, increasing the pressure room is desirable to determine the height of rise
differential developed across the escape lobby doors. required for cooling purposes, and this estimate should

46
 be reasonably narrow (usually not more than 10-20 m).
However, variations between the mass flow entering,
and that being vented from, the atrium smoke layer
with this type of hybrid system will be immediately
compensated for by a change in the depth of the
smoke layer. Thus the actual width of plume achieved
is irrelevant to the satisfactory operation of the system.
Hence there is no practical advantage in physically
reducing the width of the rising plume for this form of
hybrid system. Therefore in the design of temperature
based hybrid systems, channelling screens are
unnecessary.

Wind-sensing devices and natural

depressurisation
The sub-section 'Natural depressurisation and wind
effects' on page 40 detailed the effects of wind
pressures around an atrium and concluded that roof
vents should operate in areas of high suction pressure
for all wind directions. In certain instances it may be
likely that roof vents may experience an adverse
pressure effect (eg vertically mounted vents), in which
case the ventilation system should be controlled by a
wind direction indicator, and the required amount of
ventilation should operate in a leeward zone only.

47

I
 ry (based on Reference 9)

Building 13-storey square-shaped reinforced concrete office building with penthouse, basement and
4-storey underground garage.
Atrium A centrally situated enclosed courtyard created the atrium. The windows of the offices
facing the atrium were of 6.35 mm plate glass.
Date of fire 13 May 1977.
fire difficult. Thick black smoke issuing from the office
had built down from the roof of the atrium to below
Two ventilation systems recirculated air at the top of the tenth floor.
the atrium, and at its base there was an air handling
unit. Smoke detectors were provided at the fans of the Although the smoke detector had operated, only two
air handling unit and were arranged to shut down the of the six smoke ventilators had opened. The other
fans when the detectors activated. The units could be four had released but the springs had lost sufficient
manually restarted and put on exhaust. The general strength to open them fully. These units had to be
office area was fed by penthouse air handling units manually opened from outside. Smoke however did
that could go into a ‘smoke-purge mode’ if they were not vent effectively and at one stage the atrium was
running when a fire occurred. None of the above completely smoke-logged.
systems was in operation at the time of the fire.
Smoke extractors could not be connected to the smoke
The roof of the atrium was made of clear plastic ventilators and so firemen used large extractors
panels. Six custom-made smoke ventilators were pointed upward from the atrium ground floor to pull
provided in the atrium’s roof and comprised clear fresh air from the front doors and push smoke upward
plastic panels on hinges equipped with springs and and out through the ventilators. No building
release mechanisms. The release device was operated engineering staff were available to advise firemen on
by one smoke detector located in the atrium roof. the HVAC smoke purge capability until much later. It
Fusible links on individual ventilators were also fitted. took 2-3 hours to finally remove the smoke from the
Sprinklers were provided at roof level in the atrium, atrium.
and the building was equipped with manual fire alarm
points and hydrant valves on each floor.
1 The fire was confined to the room of origin by the
he fire closed office door and the wall construction.
At 18.45 h a worker discovered a fire in a small office
2 Windows facing the atrium above the fire floor
(3 m X 4.6 m) on the tenth floor; a plan of this floor is
were cracked by heat but fire and smoke had not
shown in Figure A l . The fire brigade received the
alarm at 19.01 h. On arrival firemen found fire venting penetrated other floors.
from the office window into the atrium. The fire floor 3 The temperature of the gas layer in the atrium was
was hot and smoky and this, coupled with the fact that insufficient to activate the sprinklers in the atrium
the fire involved an inner office, made locating of the roof.
4 Due to an insufficiency of replacement air the
existing ventilation system design was inappropriate
for clearance of smoke from the atrium, and t h e
‘dilution’ ventilation approach used by the fire
brigade took many hours to clear the smoke.
5 If this had been an atrium with balconies providing
access to escape-ways, the smoke may well have
caused serious escape problems from upper floors.
6 Despite the l’act there were unprotected openings
on to the atrium, and that at one point the atrium
was totally smoke-logged, smoke did not migrate
to other parts of the building. This indicates that the
Figure A1 International Monetary Fund Building.
existing ventilation arrangements apparently
Plan of the tenth floor, showing location ‘depressurised’ the atrium.
of the office involved

48
IInndrodnnction Many of the variables used in equations in this
appendix d o not appear in the main body of the
The Fire Research Station has carried out a number Report. To avoid unnecessary complications for the
of studies into the movement of smoke in buildings. reader who does not wish to use this calculation
Part of this work has resulted in the development of a procedure, the appendix is provided with a separate
theory by Morgan and Marshal136to estimate the list of nomenclature (page 54).
amount of air. entrained into free (or double-sided)
thermal spill plumes (see Figure 24(b)). This
calculation method is important for smoke control Scenarios and assnmptioms
design in that it enables the designer to calculate the The calculation method strictly only applies to fire
required fan capacity o r vent area for a smoke scenarios where a horizontally-flowing thermally-
ventilation system for large undivided volume buoyant layer of smoky gases approaches a void,
buildings (eg multilevel shopping malls and atria). A through which those gases then rise. More specifically,
number of studies have since been carried out which the following assumptions are made:
have resulted in the modification of the original
theory to include more recent work on thermally- o This approach flow is assumed to be beneath a flat
buoyant horizontal flowsI7 and adhered (or attached, ceiling (or a downstand) at the edge of the void.
or wall, or single-sided) p l ~ m e s '(see
~ . ~Figure
~ 24(a)).
Q It is channelled by downstands (which may be
This appendix presents the modified theory in a form either walls or channelling screens).
which the designer can follow more easily than in the
0 The flow has flow-lines which are everywhere
earlier research papers.
parallel and which approach the edge of the void at
a right angle.
The calculations can be done using an electronic
calculator having full scientific functions. This o The approach flow is assumed to be fully
however may be time-consuming, particularly where developed.
the designer wishes to look at a number of geometries
or conditions. The calculations can be incorporated in o There is no immersed ceiling jet.
a computer program where frequent calculations are
required. An alternative method t o Figure B1 is given o It is assumed that the velocity of the clear air
later in this appendix (page 53) in order to facilitate below the smoke layer has a velocity much smaller
such programming. than the velocity of the layer itself.

5I

Modified distance above virtual source I ~ ( v I

Figure B l Graphical representation (from Rcfcrcnce 50) of the theoretical

solution for a plume issuing from a rcstrained source ( F < 1)

49
Fortunately these assumptions correspond to many both buoyancy and (the vertical component of)
practical scenarios of interest to designers. velocity which can be described by Gaussian
functions. This source is, of course, virtual. We have
It should further be noted that experimental evidence” followed Lee and E m m ~ n s in’ ~using this source,
suggests that the calculation procedure which is the and indeed in the method of calculating the plume
subject of this guide should not be used for approach above the source.‘Wefollow Lee and Emmons in
flow layer temperatures higher than about 350 “C. calling this source an ‘Equivalent Gaussian source’.
Accurate methods for higher temperatures do not yet
Calculate the key parameters of the Equivalent
exist. The present method significantly overpredicts
Gaussian source by ensuring that the three key
the mixing of air into the rising hot gases for higher
parameters from Stage 3 above keep the same
temperatures.
values.
In practice the designer will have arrived at the key 5 Knowing t h e height above the ceilinghoid edge (for
parameters of the approach flow by some calculation example, this is likely to be chosen to be equal to
procedure independent of the present guide. For the smoke layer base in the reservoir above the
example, Equations 6 and 7 of Reference 17 could be void), calculate the entrainment into the spill
used to calculate the flow of smoky gases passing from plume. This calculation treats the plume as a perfect
a room into an atrium void. Another example is two-dimensional plume having a length equal to the
where a single-storey mall allows smoke to rise width of the channel of the approach flow.
through the void of a two-storey mall: here the flow in
6 Calculate the additional entrainment into the free
the single-storey mall can be calculated in the usual
ends of the plume. This assumes that the bulk of the
way using, for example, Reference 16, which will give
plume is relatively unaffected by these end effects
the designer values for both the mass flow rate and
- reasonable for plume heights typically smaller
the heat flux.
than or comparable to the plume length”.

The calculation proceeds in discrete stages:

Stage 1
-1 The designer must know: Complete all necessary pre-calculations to derive the
key parameters, of the approach flow described in l b
(a) the internal geometry of his building, on this page.
including relevant channel widths, and
(b) at least two of the key parameters of the Stage 2
approach flow. Useful pairs are: Select from the following equationsI7 to determine the
mass flow/heat flux remaining parameters for the approach flow from the
mass flowhean layer temperature initial known parameters:
mass flowkeiling temperature -
heat fluxhean layer temperature Calculate the mean layer temperature (e,)
. heat fluxkeiling temperature
heat flux/layer depth
layer depth/mean layer temperature ...(B 1)
layer depth/ceiling temperature
2 Using the known parameters for the approach flow,
calculate the remaining parameters of the flow. Calculate the mass flow rate (M,) at the opening given
by30:
3 Using the results from the preceding stage,
calculate the entrainment into the flow as it rotates
around the void edge, ie as the smoky gases change
. from a horizontally-moving flow to a vertically- 2 Cd3/* (2g 0,, To)1/2 -
WPO d;I2 KM ...(B2)
M, = -
moving flow. By the end of this stage the key
3 T,,“
parameters of the vertically-moving gases will be
known at the horizontal plane passing through the
ceilinghoid edge. These parameters are the heat where po = 1.22 kgmp3for an ambient temperature
flux, the vertically-moving mass flux, and the kinetic To of 288 K
energy of the gases (this last is based only on the
vertical component of velocity). Cd = 0.6 for opening with a deep downstand
or 1.0 for no downstand
4 The plume at greater heights behaves as if it rises
from an infinitely wide source located in the g = 9.81 ms-2
horizontal plane passing through the ceilinghoid
edge, where that source has horizontal profiles of KM= 1.3 for most typical flowing layers

50
The depth of the layer (d,) at the opening is then If the line plume is single-sided, go to Stage 7 of this
given by30 : procedure, after completion of Stage 3.

3 M ” Tc, Stage 4

~ c ~ ~ ‘wp0(2g
~ K , ] ...(B3)
T ~ ) ~ ’ ~
Calculate the Equivalent Gaussian Source:

First convert Q and My into the corresponding

The mass-weighted average temperature of the gas parameters per unit length of plume (ie divide by the
layer is30 : channel width (W) to give Q, and A ) .

Then solve the following equations36

...(B4)
1
where KQ = 0.95 for most typical flowing layers. ...(BlO)
t=[A+El
Note the importance of knowing whether there is a
downstand running along the edge of the void (and Q, 2-
thus at right angles to the direction of the flow),
because this changes the value of Cd. ...(B11)

Greater accuracy can be achieved by calculating the

values of the profile correction factor KM and KQ using
T0ch A + -
[ ,“.I1
the temperature dependent formulae in Reference 30, where the empirical thermal constant5”(A) = 0.9
although this is usually unnecessary for most practical
designs.
2B
The layer’s characteristic velocity ( v ) is given byI7:
h ...(B 1 2)

v=0.96 -
C d K ~

K Q ” ~ c PO wTo
2[g Qw Tcw
]
I 13

...(B5)

For a deep downstand, where Cd = 0.6,this becomes:

uG =Jr ...(B13)

...(B6) and:

With no downstand at the opening, Cd = 1.0, and:

...(B14)

v = 1.27 [ g Qw Tc,
I/3
..(B7)
c P O WTO
where [”,] and b G are parameters of the
G , UG
Calculate the horizontal flux ( B ) of vertical buoyant
potential e n e r g ~(relative
~ ~ . ~ to
~ the void edge): Equivalent Gaussian Source.

Stage 5
Calculate the entrainment into the rising plume.
...(B8)
The Source Froude number (F) for the line plume is36:

[mG]
Stage 3 a 112

Calculate the mass flux (My) rising past the void edge17: ...(B15)
F=[t]’” (g bG)1‘2

My =
2
-

3
po W a’ [2g 1 I/ 2

d,:’* + M, ...(B9)
where a = 0.16for double-sidedsOand 0.077 for
single-sided line plumes3’. Calculate the transformed
where the entrainment constant (a’)= 1.1 parameter (VG) for the Equivalent Gaussian Source:

51
 1 Calculate the mass flow per unit plume length (nz,)
Z)G = ...(B16) passing the chosen height36x:
r .
Determine the value of I,(z)G)by using the following
procedure or an alternative method set out in later in
this appendix on page 53.

uGrepresents a value on the vertical axis of Figure B1. Convert to the total mass flow in line plume (ignoring
Look across to the middle solid curve and find the end effects) by multiplying Equation B25 by the
corresponding value of I 1 ( u ~on
) the other axis. channel width (ie rn,W).
Calculate the transformed height parameter of x’ Stage 6
corresponding to the desired plume height (x): Calculate3xthe entrainment 6 M , into the free ends of
the line plume. The width of the line plume (and also
its axial velocity) can be taken as being approximately
...(B17) constant for most of its height as a first order
approximation. and equal to the mean of the values at
the Equivalent Gaussian Source and at the chosen
Next calculate A Il(u): height (x).

The entrainment 6M,.into both ends of the line plume

X’
is then38:
A I,(u) = ...(BlS)
[ P (1 - P ) ]
6 M r = 4 b L c axp, ...(B26)

and Il(u) = Il(uG)+ A Il(u) ...(B 19) where:

-
Determine values of b’, p’ and LC‘ corresponding to the
b = (b(;+ b ) / 2 ...(B27)
calculated value of Il(u) using the following procedure
or the alternative method on page 53. -
U = (U(; +U)/2 ...(B28)
Il(u) represents a value on the horizontal axis of
Figure B1. Using this value find the corresponding Add this to the plume entrainment result from Stage 5
values (from all three curves) for U”, p” and b”. Then to obtain the total mass flow M,of smoky gases rising
use the following equations to determine U’, p’ and b’, past the specified height (x).
where:
ie: M,= w,!,W+ 6M, ...(B29)
...(B20)
It should be noted that where both ends of a plume are
bounded by side walls (eg, as in a shaft) then 6 M , = 0.

Stage 7
Modifications to the above procedure for single-
sided17,37,51
(or adhered) line plumes.
..(B22)
Convert both the Equivalent Gaussian Source and the
plume into a composite of a real and an imaginary half,
Next determine the characteristic half-width (b) of the
so that the centre line of the composite lies along the
line plume36at height x:
vertical wall to which the plume is adhering. This is
done by doubling values for B , M y (and hence A ) , and
b = b’ bG ...(B23) Q from Stage 3 before returning to Stages 4 to 6
above. Note that experiment^^^ show that the value of
Then calculate the axial vertical velocity component a needed in Stages 4 to 6 should change from 0.16
( U ) of the gases at height x: (valid for a free or double-sided plume) to 0.077 for
the adhered plume.
U’ UG
U = ~
...(B24) On completing Stage 6, halve the final value of mass
F flow M , rising past the desired plume height (x).

52
 . I . . .. .~
.. .. __ /I . . . .. .. . . .
..- . .-.. -., . c

Alternative method for determination of I , (UC)

If uG is greater than 1.549 then /,(uG)= (uG- 0.75)/0.9607
If uG is less than or equal to 1.549 and UG is greater
than 1.242 then II(uG)= (uc - 0.843)/0.8594
If uG is less than or equal to 1.242 and uGis greater
than 1.059 then /,(uG)= (uG - 0.9429)/0.6243
If uGis less than 1.069 then /,(uG)= (vG - 1.0)/0.3714

Alternative method for calculating value of

b , p' and U'
(a) Determination of U''
If /,(U) is greater than 1.896 then U" = 1.0
If /,(U) is greater than 0.786 and I , (U) is less than or
equal to 1.896 then U" = 0.0908 I ,(U) + 0.821.
If /,(U) is less than or equal to 0.786 then U" = II (u)0,35

(b) Determination ofp"

If /,(U)is greater than 0.832 then p" = 0.9607 I,(u) +

0.75
If /,(U) is greater than 0.464 and less than or equal to
0.832 then p" = 0.8594 /,(U) + 0.8429
If /,(U) is greater than 0.186 and /,(U) is less than or
equal to 0.464 then p" = 0.6243 /,(U) + 0.9429
If /,(U) is less than or equal to 0.186 then
p" = 0.3714 /,(U) + 1.0

(c) Determination of 6"

If /,(U) is greater than 2.161 then b" = 0.938 /,(U) + 0.82
If /,(U) is less than or equal to 2.161 and /,(U) is
greater than 1.296 then b" = 0.89 /,(U) + 0.95.
If /,(U) is less than or equal to 1.296 and /,(U) is
greater than 0.896 then b" = 0.81 /,(U) + 1.071.
If /,(U) is less than or equal to 0.896 and /,(U) is
greater than 0.65 then b" = 0.619 /,(U) + 1.214.
If /,(U) is less than or equal to 0.65 and /,(U) is greater
than 0.543 then b" = 0.331 Il(u) + 1.414.
If /,(U) is less than or equal to 0.543 and /,(U) is
greater than 0.421 then b" = 0.0627 /,(U) + 1.55.
If /,(U) is less than or equal to 0.421 and /,(U) is
greater than 0.348 then b" = 1.821 - 0.6 /,(U)
If /,(U) is less than or equal to 0.348 then b" = / , ( u ) - O . ~
Now calculate U', p' and b' from Equations B20, B21
and B22 in Stage 5.

53
 _--- __ ._-

Nomenclature used in Appendix B

Note: The list of nomenclature used in the main text of this Report is given on page vi.

A Upward mass flow rate per metre across the horizontal plane through the balcony (kgs-lm-l)
b Characteristic half-width of line plume at height x (m)
b' Dimensionless half-width of line plume
b" Transformed dimensionless half width of line plume
B Potential energy flux per metre of horizontal gas stream below corridor ceiling edge (Wm-l)
Cd Coefficient of discharge
C Specific heat at constant pressure of gas (k.lkg-'"C-')
d Depth of gas stream beneath ceiling (m)
F Source Froude number (for line plume)
g Acceleration due to gravity (mss2)
11 Transformed height (dimensionless)
m Mass flow rate per unit width of gas stream (kgss'm-')
6m Mass per second per metre of air entrained into hot gas stream at corridor ceiling edge (kgm-Is-')
M Mass flow rate of gases (kgs-I)
6M Mass per second of air entrained into free ends of plume (kgs-l)
P' Dimensionless buoyancy on plume axis
P" Transformed dimensionless buoyancy on plume axis
Q Heat flux in the gas (kW)
QO Heat flux per second per unit width of gas flow (kWm-')
T Absolute gas temperature (K)
U Vertical gas velocity at height x (ms-')
Uf Dimensionless vertical gas velocity
Uff Transformed dimensionless vertical gas velocity
V Horizontal velocity component of gas (ms-')
W Width of gas flow (m)
X Height of clear layer above fire compartment/balcony edge (m)
xf I
Dimensionless variable
a Entrainment constant for plume (= 0.077 and 0.16 for single- and double-sided plume respectively)
a' Entrainment constant for air mixing into gases rotating around a horizontal edge
P Gas density (kgmP3)
e Excess temperature of gases above ambient temperature ("C)
KM Profile correction factor for mass flow (approx. 1.3)
KQ Profile correction factor for heat flux (approx. 0.95)
h An empirical thermal plume constant ( h= 0.9)
2) Transformed reciprocal of buoyancy (dimensionless)
5 Function defined in text (Equation B12)
5 Function defined in text (Equation BlO)

Subscripts
0 An ambient property
C Variable evaluated at highest point in a flow (but outside any boundary layer)
G A property of the equivalent Gaussian source
r Base of ceiling smoke reservoir
W Variable evaluated in the horizontal flow at opening
Y Variable evaluated in vertical flow past top of opening

54
 ..
. R. . . . .. I- !.

Acknowledgements References
We would like to thank Mr N R Marshall of the Fire Butcher E G and Parnell A C. S m o k e control in
Research Station for his help in preparing Appendix B. fire safety'design. London, E & F N Spon Limited,
1979.
The Fire Research Station is grateful to the Task
Group of the Chartered Institution of Building Department of the Environment and the Welsh
Services Engineers engaged in preparing a draft Ofice. The Building Regulations 1991. Approved
CIBSE guidance document on smoke control in atria, Document B. Fire safety (1992 edition). London,
for its comments on a draft of of this Report. These HMSO. 1991.
have as far as possible been taken into account.
House of Commons. Public Health Acts 1936 and
1961. London, HMSO.

House of Commons. Factories Act 1962.

London, HMSO.

House of Commons. Offices, Shops and Railway

Premises Act 1963. London, HMSO.

Saxon R. Atrium buildings. Development and

design. London, The Architectural Press, 1983.

The Andraeus Building fire in SBo Paulo, Brazil.

Fire Prevention, 1973,97 (January) 37.

Sharry J A. An atrium fire. Fire Journal, 1973,67

(6) 39-41.

Lathrop J K. Atrium fire proves difficult to

ventilate. Fire Journal, 1979,73 (1) 30-31.

10 Robinson P. Atrium buildings: a fire service view.

Fire Surveyor, 1982,11(4) 42-47.

11 Degenkolb J G. Atriums. The Building Official

and Code Administrator, 1983, XVIl(6) 18-22.

12 Parnell A C and Butcher E G. Smoke movement in

atria. Fire Protection (South Africa), 1984, U(3) 4-6.

13 National Fire Protection Association. S m o k e

management systems in malls, atria and large areas
92B. Quincy MA, NFPA, 1991.

14 British Standards Institution. Fire precautions in

the design, construction and use of buildings.
Part 7. Code of practice for atrium buildings.
British Standard BS 5588:Part 7. London, BSI.
To be published.

15 British Standards Institution. Fire precautions in

the design, construction and use of buildings.
Part 10. Code of practice for shopping complexes.
British Standard BS 5588:Part 10:1991.
London, BSI, 1991.

16 Morgan H P and Gardner J P. Design principles

f o r smoke ventilation in enclosed shopping centres.
Building Research Establishment Report (BRE
Bookshop ref BR186). Garston, BRE, 1990.

55
17 Morgan M P and Hansell G Ca. Atrium buildings: 31 Morgan H P and Marshall N R. The depth of void-
calculating smoke flows in atria for smoke control edge screens in : ,loj-ping malls. Fire Engineers
design. Fire Safety Journal, 1987,U (1) 9-35. Journal, 1989,44 \ 152) 7-9.

18 Morgan M P and Chandler S E. Fire sizes and 32 Bosley K. ‘The effects of wind speed on escai’e
sprinkler effectiveness in shopping complexes and behaviour through emergency exits. Summary
retail premises. Fire Surveyor, 1981,lO ( 5 ) 23-28. Report. F R D C Research Report Number 53.
London, Home Office, 1992.
19 Morgan M P and Hansell G 0. Fire sizes and
sprinkler effectiveness in offices - implications 33 British Standards Institution. Fire precautions in
for smoke control design. Fire Safety Journal, 1985, the design of buildings. Part 4. Smoke control in
8 (3) 187-198. protected t:scape routes using pressurization.
British Standard BS 5588:Part 4:1978.
U) Hansell G 0 and Morgan H B. Fire sizes in hotel London, BSI, 1978.
bedrooms - implications for smoke control
design. Fire Safety Journal, 1985,s (3) 177-186. 34 Spratt D and Heselden A J M. Efficient extraction
of smoke from a thin layer under a ceiling.
21 Vincent B G, Kung PI C and Will E E. Residential Fire Research Station Fire Research Note 1001.
side wall sprinkler fire tests with limited water FRS, Borehamwood, 1974.
supply. Fire Science and Technology, 1988,s (2)
41-53. 35 Marshall RI R, Feng S Q and Morgan H P.
The influence of a perforated false ceiling on the
22 Cote A E. Highlights of a field test of a retrofit performance of smoke ventilation systems.
sprinkler system. Fire Journal, 1983,77 (3) 93-103. Fire Safety Journal, 198.5,8(3) 227-237.

23 Hansell G 0. Heat and mass transfer process 36 Morgan W B and Marshall N R. Smoke hazards in
affecting smoke control in atrium buildings. covered multi-level shopping malls: an
PhD Thesis. London, South Bank University, 1993. experimen tally-based theory for smoke
production. Building Research Establishment
24 Zukowski E E,Kubota T and Cetegen B. Current Paper CP48/75. Garston, BRE, 197.5.
Entrainment in fire plumes. Fire Safety Journal,
1981,3 (2/3) 107. 37 Wansell G 0,Marshall N R and Morgan M P.
Smoke flow experiments in a model atrium.
25 Quintiere d G, Rinkinen W J and Jones W W. Building Research Establishment Occasional Paper
The effects of room openings on fire plumes OP55. Garston, BRE, 1993.
entrainment. Combustion Science and Technology,
1981,26 (5/6) 193-201. 38 Morgan H P and Marshall N R. Smoke control
measures in a covered two-storey shopping mall
26 Hinkley P L. Rates of production of hot gases in having balconies as pedestrian walkways. Building
roof venting experiments. Fire Safety Journal, Research Establishment Current Paper CPl 1/79.
1986,lO (1) 57-65. Garston, BRE, 1979.

27 McCaffrey B J, Quintiere 9 G and Harkleroad M F. 39 Grella J 9 and Faeth G M. Measurements in a two-

Estimating room temperatures and the likelihood dimensional thermal plume along a vertical
of flashover using fire test data correlations. adiabatic wall. Journhl of Fluid Mechanics, 1975,
Fire Technology, 1981,17 (2) 98-119. 71 (4) 701-710.

28 Morgan H P and Marshall N R. Smoke hazards in 40 Thomas P H. On the upward movement of smoke
covered multi-level shopping malls: a method of and related shopping mall problems. Fire Safefy
extracting smoke from each level separately. Journal, 1987,U (3) 191-203.
Building Research Establishment Current Paper
CP19/78. Garston, BRE, 1978. 41 Morgan H P. Comments on ‘A note on smoke
plumes from fires in multi-level shopping malls’.
29 Heselden A J M. Fire problems of pedestrian Fire Safety Journal, 1987,U (1) 83-84.
precincts. Part 1. The smoke production of various
materials. Fire Research Station Fire Research Note 42 Thomas P .H, Hinkley P L, Theobald C R and
856. FRS, Borehamwood, 1971. Simms D B,. Investigations into the flow of hot
gases in roof venting. Fire Research Technical
30 Morgan H P. The horizontal flow of buoyant gases Paper No 5‘. London, HMSO, 1963.
toward an opening. Fire Safety Journal, 1986,11
(3) 193-200.
, -

56
43 Boehmer D 9. Atrium fire engineering - North
American experience. Proceedings of seminrir
‘Atrium Engineering’ held by Environmental
Energy Group of the Institution of Mechanical
Engineers, 26 June 1990. London, I Mech E, 1990.

44 Pads 9. Calculating evacuation times for tall

buildings. Proceedings of symposium ‘Qurmtitative
Methods for Life Safety Analysis’ held by the
Society of Fire Protection Engineers, University of
Maryland, March 1986. Boston MA, SFPE, 1986.

45 Hansell 6 0 and Morgan H P. Smoke control in

atrium buildings using depressurisation. Part 1:
Design principles. Fire Science and Technology,
1990,lO (1 & 2) 11-26.

46 Nansell G 0 and Morgan W P. Smoke control in

atrium buildings using depressurisation. Part 2:
Considerations affecting practical design. Fire
Science ancl Technology, 1990,lO (1 & 2) 27-41.

47 Building Research Establishment. The assessment

of wind loads. Part 8: Internal pressures.
B R E Digest 346. Garston, BRE, 1990.

48 Douglas-Baines W. Effects of velocity distribution

on wind loads and flow pattern on buildings.
Proceedings of Symposium N o 16 ‘Wind Effects
on Buildings and Structures’ held by the National
Physical Laboratory, Tedclington, 26-28 June 1963.
London, HMSO, 1965.

49 Hinkley P L. The effect of smoke venting on the

operation of sprinklers subsequent to the first.
Fire Safety Journal, 1989,14 (4) 221-240.

50 Shao-Lin Lee and Emmons H W. A study of

natural convection above a line fire. Journal of
Fluid Mechanics, 1961,ll (3) 353-368.

51 Marshall N R. Air entrainment into smoke and hot

gases in an open shaft. Fire Safety Journal, 1986,lO
(1) 37-46.

Printed in the UK for HMSO. Dd.8392663, 1/94, CIO, 38938 57

 Also available from BFIE

Design principlesfor smoke ventilation in Smoke control in large stores: an extended

enclosed shopping centres calculation method for slit extraction design
H P Morgan and J P Gardner N R Marshall
Ref BR186 €28 BRE Report 1990 BRE Occasional Paper OP51 (Free from
This Report is intended to assist designers of smoke N R Marshall at BRE’s Fire Research Station)
ventilation systems who are faced with many Extends the existing calculation method for slit
difficulties when evolving a smoke control system that extraction (which is one of three methods used to
will ensure safe escape from fire in an enclosed prevent cool smoky gases flowing in bulk from large
shopping complex. A serious fire in most shops will stores on to enclosed shopping malls) to include
usually penetrate the shop front and cause hot smoky stores having one or more openings of differing
gases to spread quickly along the malls, leading to heights and widths. It also updates the existing
rapid loss of visibility unless preventive measures are method to include the effects of different geometry
taken. Work at FRS has shown that these gases can openings between store and mall.
be confined to a ceiling reservoir formed by screens,
downstands or other features provided that the gases
are then removed by some means and that an equal Experimentsat the Muitifunctioneel
amount of replacement air is allowed to enter the Trainingcentrum, Ghent, on the interaction
main smoke layer. The lateral spread of smoke on the between sprinklers and smoke venting
lower level of a multi-storey mall must be controlled to P L Hinkley, G 0 Hansell, N R Marshall and R Harrison
reduce the size of the smoke plume rising through the Ref 6R224 €25 BFlE Report 1992
upper level. This minimises the quantity of gases Describes a series of experiments carried out to validate
entering the ceiling reservoir. Such systems for a mathematical model of the effect of roof venting on
controlling smoke movement provide a smoke-free the operation of sprinklers (see BR213 below).
zone below the main layer and allow people to
escape. Sprinkler operation and the effect of venting:
studies using a zone model
A simplified approach to smoke-ventilation P L Hinkley
calculations Ref BR213 f30 BRE Report 1992
H P Morgan This Report describes a mathematical zone model
BRE Information Paper lP19/85 f3.50 developed as a joint project by the Fire Research
Outlines the calculations for either fan capacities or Station and Colt International Limited to remedy
vent areas for single-storey, large undivided-volume deficiencies in current knowledge about the interaction
buildings where the fire is directly below the smoke between roof venting and sprinklers.
reservoir. It is not intended for use with more
complicated building geometries such as shopping
malls. Will be of particular value to building control
officers, fire officers and others who have to check
smoke ventilation proposals or give approvals.

Prices current at January 1994 but subject to change

Please send order, enclosing cheque, to:
BRE Bookshop, Bulldlng Research Establishment,
Garston, Watford, WD2 7JR

Cheques payable to ‘DOBBRE‘ I S B N 0-85125-615-5

For enquiries and immediate credit card orders

(AccesdMastercard or Visa):
Telephone: 0923 664444 or Fax: 0923 664400

BR 258