Smoke Control

50
John H. Klote

Introduction systems. This chapter focuses on the conventional

approach, but tenability systems are discussed near
In building fires, smoke often flows to locations the end of the chapter. For an exhaustive treatment
remote from the fire, threatening life and damag- of smoke control including weather design data,
ing property. Research has shown that smoke is design fires, conventional systems, tenability
the major killer in building fires (Harland and systems, and method of analysis see the Handbook
Woolley 1979; [37]). of Smoke Control Engineering [21].
NFPA 92 [29] defines a smoke control system In this chapter the term smoke is used in
as an engineered system that includes all accordance with the NFPA 92 definition that
methods that can be used singly or in combina- states that smoke consists of the airborne solid
tion to modify smoke movement. These methods and liquid particulates and gases evolved when a
are the physical mechanisms of smoke control material undergoes pyrolysis of combustion,
which are discussed later in this chapter. together with the quantity of air that is entrained
Research in the field of smoke control has been or otherwise mixed into the mass. It is important
conducted in Australia, Canada, England, for smoke control purposes that the definition of
France, Japan, the United States, and Germany. smoke includes the air that is mixed with the
This research has consisted of field tests, full- particulates and other gases because smoke con-
scale fire tests, scale model fire tests, and com- trol often involves exhausting smoke which is
puter simulations. Many buildings have been mostly air. Including air as a part of smoke is
built with smoke control systems and numerous also important for tenability systems.
others have been retrofitted for smoke control.
The conventional approach to smoke control
consists of using the physical mechanisms to pre- Physical Mechanisms of Smoke
vent people from coming into contact with smoke Control
to the extent possible. A newer approach consists
of evaluating the effect of some smoke contact The physical mechanisms of smoke control are
with the intent of providing a tenable environment (1) compartmentation, (2) dilution, (3) pressuri-
for occupants. Smoke control systems based on zation, (4) airflow, and (5) buoyancy. For
this newer approach are referred to as tenability centuries, compartmentation has been
recognized as a way of controlling the spread of
fire and smoke. When a person closes the door to
a burning room, smoke flow from the room
J.H. Klote (*)
consultant specializing in fire in smoke control, decreases considerably. Also, the amount of air
Leesburg, VA available to the fire drops off. Today this passive

M.J. Hurley (ed.), SFPE Handbook of Fire Protection Engineering, 1785

DOI 10.1007/978-1-4939-2565-0_50, # Society of Fire Protection Engineers 2016

EDUFIRE.IR Telegram EDUFIRE_IR

 1786 J.H. Klote

smoke protection is recognized in many building The use of dilution to produce or maintain
and fire codes even without a design analysis. tenable conditions in the fire space is not
Engineered smoke control systems that use only recommended because such a system is beyond
passive smoke barriers are a form of tenability the current state of the technology. The air sup-
systems that are discussed later in this chapter. plied to the fire space can increase the burning
Dilution can occur naturally as when smoke rate of the fire resulting in increased smoke pro-
flows away from a fire and mixes with air as it duction. The increased smoke production in the
flows or be forced by fan powered flows. Naturally fire space has the potential to result in untenable
occurring dilution can be analyzed by the methods conditions in the fire space. Because of this fail-
discussed in the section on tenability systems. ure mode the use of dilution in the fire space is
Fan powered dilution can be used to remove generally not recommended.
smoke from the fire space after a fire has been Many smoke control systems use mechanical
extinguished, and it can be used to remove smoke fans to control smoke by pressurization. Pres-
from a space connected to the fire space after the sure difference across a barrier can control
connection has been has been closed. Fan powered smoke movement. The idea is that a pressure
dilution consists of supplying air to the fire space difference is produced across a barrier such that
and either exhausting air (or smoke) from the space the smoke on the low pressure side of the bar-
or providing a path for a path for air (or smoke) to rier is prevented from migrating to the high
flow from the space. This kind of fan powered pressure side, and this is shown in Fig. 50.1a
dilution can be analyzed by the methods discussed for a relatively small amount of smoke. How-
in the section on tenability systems. ever, pressurization can control smoke from a

Fig. 50.1 Pressure

difference across a barrier
can control smoke flow

EDUFIRE.IR Telegram EDUFIRE_IR

 50 Smoke Control 1787

Fig. 50.2 Airflow can

control the flow of smoke

Fig. 50.3 Buoyancy can

be used for smoke control
as in an atrium smoke
exhaust system

large fully developed fire as shown in fire and move the smoke away from occupants
Fig. 50.1b. In a room with a fully developed (Fig. 50.3). This form of smoke control is called
fire, everything that can burn is burning. Pres- atrium smoke management, and it is the subject
surization smoke control systems are discussed of Chap. 51.
later in this chapter.
Airflow has been used extensively to control
smoke flow during fires in subway, railroad and Pressurization Smoke Control
highway tunnels (Fig. 50.2). Airflow can also be Systems
used to control smoke between atria and spaces
connected to atria. A number of empirical Commonly used pressurization smoke control
equations have been developed to calculate the systems are (1) stairwell pressurization, (2) eleva-
air velocity needed for specific applications. tor pressurization, and (3) zoned smoke control.
Because the large amounts of air are needed for The concept of stairwell pressurization is to sup-
this method can supply additional oxygen to the ply air to the stairwell with the intent of
fire, the use of airflow to control smoke needs to maintaining tenable conditions in stairwell. The
be done with caution. idea of elevator pressurization is to supply air to
Atria smoke exhaust systems rely on the the elevator shaft with the intent of preventing
buoyancy for a smoke plume to form above the smoke flow through elevator shaft to locations

EDUFIRE.IR Telegram EDUFIRE_IR

 1788 J.H. Klote

Table 50.1 Activation signals for pressurization smoke

control systems Network Modeling
Sprinkler Manual
Network models are a class of software that can
Smoke Heat Water Pull
simulate the flow of air or water through a com-
System Detector Detector Flow Station
plex system of paths which is called the network.
Stairwell Yes Yes Yes Yesa
pressurization Network modeling for smoke control application
Elevator Yes Yes Yes Yesa dates back to the 1960s, but these early models
pressurization were subject to numerical difficulties and data
Zoned smoke Yes Yes Yes No input was extremely cumbersome and time
control consuming.
a
Manual pull stations are not recommended for activation Network computer models such as CONTAM
of stairwell or elevator pressurization systems that rely of
[36] have become widely used for analysis of
fire floor exhaust
pressurization smoke control systems due to
their robust numerical routines and sophisticated
data input. While CONTAM was developed for
remote from the fire floor. The idea of zoned indoor air quality applications, care was taken to
smoke control is to rely on pressurization or assure that it could be used for smoke control
passive smoke control with the intent of applications. Because CONTAM is a product of
preventing or minimizing smoke movement the US National Institute of Standards and Tech-
beyond the zone where the fire is located. nology (NIST), it can be downloaded from the
The primary purpose of pressurization smoke NIST website at no cost. In this chapter, when
control systems is to maintain a tenable environ- CONTAM is discussed, it should be noted that it
ment in the means of egress. Other purposes of may be possible that other network models could
these systems are to control smoke movement be used.
between fire area and adjacent spaces, provide These models represent a building by a net-
conditions to help fire service, reduce property work of spaces or nodes, each at a specific pres-
damage, and aid in post-fire smoke removal. sure and temperature. The stairwells and other
Table 50.1 lists sources of signals that can be shafts can be modeled by a vertical series of
used to activate pressurization smoke control spaces, one for each floor. Air flows through
systems. Zoned smoke control should not be leakage paths from regions of high pressure to
activated from manual fire alarm pull stations. regions of low pressure. These leakage paths are
For proper operation of zoned smoke control doors and windows that may be opened or
needs, the location of the fire needs to be identified. closed. Leakage can also occur through
A person who has seen a fire may start to leave the partitions, floors, and exterior walls and roofs.
building and some distance away from the fire The airflow through a leakage path is a function
zone realize that he or she should pull the manual of the pressure difference across the
pull station. If zoned smoke control were activated leakage path.
by this manual pull station, the wrong zone would In network models, air from outside the build-
be identified as where the fire was located. Some ing can be introduced by a pressurization system
stairwell pressurization systems and elevator pres- into any level of a shaft or into other building
surization systems rely on fire floor exhaust, and spaces. This allows simulation of stairwell pres-
activation of these systems by manual pull stations surization, elevator shaft pressurization, stairwell
can result in the wrong floor being exhausted. For vestibule pressurization, and pressurization of
this reason, manual pull stations are not any other building space. In addition, any build-
recommended for activation of stairwell or eleva- ing space can be exhausted. This allows analysis
tor pressurization systems that rely of fire floor of zoned smoke control systems where the fire
exhaust. zone is exhausted and other zones are

EDUFIRE.IR Telegram EDUFIRE_IR

 50 Smoke Control 1789

pressurized. The pressures and flows throughout driving force as it would act in the absence of
the building are obtained by solving the conser- any other driving force.
vation equations for the network. This analysis
can include the driving forces of wind, the pres-
surization system, and inside-to-outside temper- Stack Effect
ature difference.
The primary purpose of network simulations When it is cold outside, there is often an upward
is to determine if a particular smoke control movement of air within building shafts such as
system in a particular building is capable of stairwells, elevator shafts, dumbwaiter shafts,
being balanced such that it will perform as mechanical shafts, or mail chutes. This phenom-
intended. Network models are capable of enon is referred to as a normal stack effect as
simulating the pressures and flows throughout shown in Fig. 50.4. The air in the building has a
very large and complex building networks with buoyant force because it is warmer and less dense
high accuracy. than the outside air. This buoyant force causes air
There are many flow paths in buildings to rise within the shafts of buildings, and the
including gaps around closed doors, open doors, pressure difference due to normal stack effect is
construction cracks in walls and floors. These shown in Fig. 50.5. The significance of normal
flow paths can be approximated for a design stack effect is greater for low outside
analysis, and the results of a network model temperatures and for tall shafts.
simulation are approximations. However, these When the outside air is warmer than the build-
approximate results can be useful in identifying ing air, downward airflow frequently happens in
problems with specific smoke control systems. If shafts. This downward airflow is called reverse
such problems are identified, the smoke control stack effect (Fig. 50.5), and the pressure differ-
system can be modified appropriately. A second- ence due to reverse stack effect is shown in
ary purpose of these simulations is to provide Fig. 50.5b. At standard atmospheric pressure,
information to help size the system components the pressure difference due to either normal or
such as supply fans, exhaust fans, and vents. reverse stack effect is


1 1
Δ pSO ¼ 7:63  z
Smoke Movement T O þ 460 T S þ 460


1 1
Δ pSO ¼ 3460  z for SI
A smoke control system must be designed so that T O þ 273 T S þ 273
it is not overpowered by the driving forces that ð50:1Þ
cause smoke movement. For this reason, an
understanding of the fundamental concepts of where
smoke movement is a prerequisite to intelligent ΔpSO ¼ pressure difference from a shaft to the
smoke control design. The driving forces of air outside, in. H2O (Pa),
and smoke movement in building are (1) stack TO ¼ temperature outside,  F ( C),
effect, (2) buoyancy of combustion gases, TS ¼ temperature in the shaft,  F ( C),
(3) expansion of combustion gases, (4) wind, z ¼ distance above the neutral plane, ft (m).
(5) forced ventilation, and (6) elevator piston The neutral plane is a horizontal plane where
effect. Forced ventilation consisting of supply the pressure in the shaft is the same as that
air is used for pressurization smoke control. outdoors. For a shaft 200 ft (60 m) tall, with a
Also, forced ventilation is used in heating, neutral plane at the midheight, an outside tem-
ventilating and air-conditioning (HVAC) perature of 0  F (18  C) and an inside tempera-
systems. Generally, in a fire, smoke movement ture of 70  F (21  C), the maximum pressure
will be caused by a combination of these driving difference due to normal stack effect would be
forces. The following sections discuss of each 0.22 in. H2O (55 Pa). This means that at the top

EDUFIRE.IR Telegram EDUFIRE_IR

 1790 J.H. Klote

Fig. 50.4 Airflow and pressure differences of normal stack effect

Fig. 50.5 Airflow and pressure differences of reverse stack effect

of the shaft, the shaft would have a pressure of Stack effect can have a significant impact on
about 0.22 in. H2O (55 Pa) greater than that smoke flow during building fires. When it is
outside. At the bottom of the shaft, the shaft cold outdoors, the upward airflow in shafts can
would have a pressure of about 0.22 in. H2O be enhanced by the buoyancy of the smoke.
(55 Pa) less than the outside pressure. Figure 50.6 shows smoke flows in a building

EDUFIRE.IR Telegram EDUFIRE_IR

 50 Smoke Control 1791

Fig. 50.6 Smoke movement in a high rise building due to normal stack effect

subjected to normal stack effect. For a fire below

Myth: It is a myth that the
the neutral plane, smoke tends to enter and flow pressure difference due to stack
up shafts, and above the neutral plane this effect is nearly proportional to the
smoke flows from the shaft to building spaces temperature difference between
(Fig. 50.6a). Smoke from a fire above the neutral the building and the outside
Fact: This pressure difference is
plane can flow through cracks and gaps in the
nearly proportional to the
floor to the floor above the fire, but the forces of temperature difference between a
stack effect work to prevent smoke from enter- shaft and the outside
ing shafts as shown in Fig. 50.6b. If the smoke
from a fire above the neutral plane has sufficient
buoyancy to overcome stack effect and flow into Another Meaning of Stack Effect
a shaft, it will flow up the shaft and infiltrate The term stack effect is often used in a different
floors above the fire floor as can be seen in way from that discussed above. Sometimes
Fig. 50.6c. engineers will say that a pressurized stairwell
For a building with shafts of various heights (or elevator) needs to be designed to account
and different shaft temperatures, the flows can for the impact of stack effect. If the stairwell
become very complicated. These flows would (or elevator) is properly pressurized, there is no
not look like those of either Fig. 50.4 or neutral plane, and the flows do not look like those
Fig. 50.5. Each shaft could have its own neutral in Figs. 50.4 or 50.5. Strictly speaking there is no
plane with respect to the outside, and sometimes stack effect in the pressurized stairwell
a shaft may have more than one neutral plane. (or elevator). What is meant when an engineer
Equation (50.1) is not applicable for such com- says “that a pressurized stairwell (or elevator
plicated buildings, but the flows and pressures in shaft) needs to be designed to account for the
such buildings can be analyzed by CONTAM. impact of stack effect” is that it needs to be

EDUFIRE.IR Telegram EDUFIRE_IR

 1792 J.H. Klote

designed to account for the temperature fire tests. During these tests, the maximum pres-
differences that cause stack effect. sure difference reached was 0.064 in. H2O (16 Pa)
across the burn room wall at the ceiling.
Myth: It is a myth that stack effect
is the major factor impacting
stairwell and elevator
pressurization Expansion of Combustion Gases
Fact: Today the impact of stack
effect is a minor factor for most In addition to buoyancy, the energy released by a
pressurized stairwells and fire can cause smoke movement due to expan-
elevators. The pressurization air
for many stairwells and elevators sion. In a fire compartment with only one open-
is untreated outside air that is not ing to the building, building air flows into the fire
heated or cooled. The temperature compartment and hot smoke flows out of the
of these shafts is often nearly the fire compartment. Neglecting the added mass of
same as the outside temperature,
and the impact of stack effect is the fuel (which is small compared to the airflow),
significantly reduced as compared the ratio of volumetric flows can simply be
to shafts pressurized with air expressed as
treated to the building temperature
V out T out þ 460
¼
Buoyancy of Combustion Gases V in T in þ 460
ð50:3Þ
V out T out þ 273
High-temperature smoke from a fire has a buoy- ¼ for SI
V in T in þ 273
ancy force due to its reduced density. The pres-
sure difference between a fire compartment and where
its surroundings can be expressed by an equation Vout ¼ volumetric flow of smoke out of the fire
of the same form as Equation (50.1) with the compartment, cfm (m3/s),
variables updated as shown in Equation (50.2). Vin ¼ volumetric flow of air into the fire com-

 partment, cfm (m3/s),
1 1 Tout ¼ temperature of smoke leaving the fire
Δ pFS ¼ 7:63  z
T O þ 460 T F þ 460 compartment,  F ( C),


1 1 Tin ¼ temperature of air entering the fire com-
Δ pFS ¼ 3460  z for SI partment,  F ( C).
T O þ 273 T F þ 273
ð50:2Þ For fire gas temperature of 2200  F (1260  C),
the gas will expand to about five times its original
where volume. For a fire room with open doors or
ΔpFS ¼ pressure difference from a fire space to windows, the pressure difference across these
the surroundings, in. H2O (Pa), openings due to expansion is negligible because
TO ¼ temperature surroundings,  F ( C), of the large flow areas involved. However, for a
TF ¼ temperature in the fire space,  F ( C), fire space without open doors or windows, the
z ¼ distance above the neutral plane, ft (m). pressure differences due to expansion may be
For a fire, the neutral plane is the horizontal important, provided there is sufficient oxygen to
plane of where the pressure in the fire space is the support combustion for a significant time.
same as that of the surroundings. For a fire with a
fire compartment temperature of 1470  F (800  C)
and surroundings at 68  F (20  C), the pressure Wind
difference 5 ft (1.52 m) above the neutral plane can
be calculated from Equation (50.2) to be 0.052 in. In many instances, wind can have a pronounced
H2O (13 Pa). Fang [7] has studied pressures impact on smoke movement within a building.
caused by room fires during a series of full-scale The pressure that wind exerts on a wall is

EDUFIRE.IR Telegram EDUFIRE_IR

 50 Smoke Control 1793

pw ¼ 0:00645 Cw ρo U 2H Forced Ventilation

1 ð50:4Þ
pw ¼ Cw ρo U 2H for SI The current code requirements for heating,
2 ventilating and air conditioning (HVAC) systems
started with a 1939 report by the National Board
where of Fire Underwriters [28]. The NBFU examined
pw ¼ wind pressure, in H2O (Pa), NFPA fire data from January 1936 to April 1938.
Cw ¼ pressure coefficient, dimensionless, Of 25 fires recorded, 19 involved combustion of
ρo ¼ outside air density, lb/ft3 (kg/m3), parts of the air moving system. In five cases of no
UH ¼ velocity at wall height H, mph (m/s). fire in the HVAC system, smoke was distributed
The pressure coefficients, Cw, depends on by the system. Modern HVAC systems are built
wind direction, building geometry and local of materials intended to withstand fires. Also
obstructions to the wind. The pressure modern HVAC systems either shut down in the
coefficients are in range of 0.8 to 0.8, with event of a fire or go into a smoke control mode of
positive values for windward walls and negative operation. This mode of operation is called zoned
values for leeward walls. smoke control, and it is discussed later.
Wind is often measured at airports, and the
standard height for measuring velocity and direc-
tion is 33 ft (10 m). The Chap. 2 of the Handbook Elevator Piston Effect
of Smoke Control Engineering lists design wind
speeds for many locations in the US, Canada, and The transient pressures and flows produced when
other countries. The local design wind can be an elevator car moves in a shaft are called piston
calculated as follows effect. Figure 50.7 shows the airflows resulting

amet
a from an upward-moving elevator car. Such pis-
δmet H ton effect can pull smoke into a normally
U H ¼ U met ð50:5Þ
Hmet δ pressurized elevator lobby or elevator shaft. In a

where
UH ¼ wind velocity at wall w/height H, mph
(m/s),
Umet ¼ measured velocity, mph (m/s),
Hmet ¼ height of wind measurement, ft (m),
δmet ¼ boundary layer height in the vicinity of
the wind anemometer, ft (m),
amet ¼ wind exponent in the vicinity of the wind
anemometer, dimensionless,
H ¼ height of wall, ft (m),
δ ¼ boundary layer height at wall, ft (m),
a ¼ wind exponent at wall.
A number of approaches have been developed
for categorizing terrain boundary layer and the
wind exponent. For additional information about
wind see Chap. 3 of the Handbook of Smoke
Control Engineering, Shaw and Tamura [33],
Kandola [12–14], Aynsley [3], and Klote
[16]. Some civil engineering texts have useful
information about wind [6, 24, 25, 35]. Fig. 50.7 Airflow due to an upward moving elevator car

EDUFIRE.IR Telegram EDUFIRE_IR

 1794 J.H. Klote

joint US and Canadian project, an analysis of
1=2

1 1 1
piston effect was developed and validated Ae ¼ þ þ ð50:7Þ
A2sr A2ir A2io
[15, 20, 23].
The upper limit of piston effect for an elevator
where
with enclosed lobbies is
Asr ¼ leakage area between shaft and lobby, ft2

 (m2),
1:66  106 ρ As Ae U 2
Δ pu, ir ¼ Aio ¼ leakage area between the building and the
2 Aa Air Cc

2 ð50:6Þ outside, ft2 (m2).
ρ As Ae U Figure 50.8 shows the upper limit of piston
Δ pu, ir ¼ for SI
2 Aa Air Cc effect from the lobby to the building for car
velocities from 100 to 1000 fpm (1 to 5 m/s).
where All elevator velocities are in this range with the
Δpu,si ¼ upper limit pressure difference from the exception of those of extremely tall buildings.
shaft to the building, in H2O (Pa), The pressure differences shown in Fig. 50.8
ρ ¼ air density in hoistway, lb/ft3 (kg/m3), happen for a brief time as the elevator car passes
As ¼ cross-sectional area of shaft, ft2 (m2), a floor.
Air ¼ leakage area between building and lobby,
ft2 (m2),
Aa ¼ free area around the elevator car, ft2 (m2),
Ae ¼ effective area, ft2 (m2), Effective Flow Areas
U ¼ elevator car velocity, fpm (m/s),
Cc ¼ flow coefficient for flow around car, Effective flow areas were essential in the early
dimensionless. days of smoke control design to simplify flow
The flow coefficient, Cc, was determined networks. With computer network models such
experimentally at about 0.94 for a multiple car as CONTAM, there is much less need for net-
hoistway and 0.83 for a single car hoistway. The work simplification. However, the effective flow
free area around the elevator car is the cross- area concept is still used for the following:
sectional area of the shaft less the cross-sectional (1) with the equation approach for analysis of
area of the car. For an elevator with enclosed pressurized stairwells, (2) with analysis of eleva-
lobbies, the effective area is tor piston effect, (3) to reduce data input to

Fig. 50.8 Calculated

upper limit of piston effect
across elevator lobby doors

EDUFIRE.IR Telegram EDUFIRE_IR

 50 Smoke Control 1795

network models, and (4) to solve some problems For the three parallel flow paths in Fig. 50.9a,
without calculations. the effective area is
The various paths of air movement in the
system can be parallel with one another A e ¼ A1 þ A þ A3 ð50:8Þ
(Fig. 50.9a), in series with one another
(Fig. 50.9b), or a combination of parallel and and for any number of flow paths in parallel the
series paths (Fig. 50.9c). The effective flow area effective area is
of a given system of flow paths is the area of a
single opening that results in the same flow as the X
n
Ae ¼ Ai ð50:8aÞ
given system when subjected to the same pres- i¼1
sure difference over the total system of flow
paths. This concept is similar to an effective For the three series flow paths in Fig. 50.9b,
resistance of a system of electrical resistances. the effective area is

Fig. 50.9 Flow paths in parallel, in series, and a combination of both

EDUFIRE.IR Telegram EDUFIRE_IR

 1796 J.H. Klote

1=2
1 1 1
Ae ¼ þ þ ð50:9Þ This shows that for a system of flow
A21 A22 A23
areas in series with on area much smaller
and for any number of paths in series the effec- than the others, the effective flow area is
tive area is slightly less than the smallest area.
!1=2 Part 3: In Fig. 50.9c, what is the effective
Xn
1 flow area if A1, is 0.1 ft2 and all the other
Ae ¼ 2
ð50:9aÞ flow areas are 1 ft2?
i¼1 Ai
A1 ¼ 0.1 ft2 and A2 ¼ A3 ¼ A4 ¼
where A5 ¼ 1 ft2.
Ae ¼ effective flow area, ft2 (m2), Because A2 and A3 are in parallel, the
Ai ¼ flow area of path i, ft2 (m2). effective flow are of A2 and A3 is
The above equations for effective flow areas A23e ¼ A2 þ A3 ¼ 1 þ 1 ¼ 2 ft2 :
are based on having the same flow coefficients Because A4 and A5 are in parallel, the
and temperatures for all the paths in the total effective flow are of A4 and A5 is
system of flow paths. For a system with both A45e ¼ A4 þ A5 ¼ 1 þ 1 ¼ 2 ft2 :
parallel and series paths, and the method of The effective flow area of the system
developing an effective area for the system is to of flow paths in Fig. 50.9c is Ae ¼
combine parallel paths first and then series paths.  1=2  1=2
1
2 þ 2 þ 2
A1
1
A23e
1
A45e
¼ 1
0:1 2 þ 2 þ 2
2
1 1
2
Example 1 illustrates calculation of effective
 
1 ¼1=2
flow areas. ¼ 100 þ 4 þ 4
1
¼ 0:0998 ft 2

The reader can look at Fig. 50.9c and

Example 1. Effective Flow Areas see that A23e and A34e are both much larger
Part 1: In Fig. 50.9a, what is the effective than A1, so that for practical purposes, the
flow area if A1, A2 and A3 are 0.1 ft2? effective flow area of the system is A1.
Because these areas are in parallel,
Ae ¼ A1 þ A2 þ A3 ¼ 1 þ 1 þ 1 ¼ 3 ft2
Part 2: In Fig. 50.9b, what is the effective Symmetry
flow area if A1, is 0.1 ft2 and A2 and A3 are
both 1 ft2? As with effective areas, symmetry was essential
Because these areas are in series, Ae ¼ in the early days of smoke control design to
 1=2  1=2 simplify flow networks, symmetry is also used
1
A2
þ 1
A2
þ 1
A2
¼ 1
0:12
þ 1
12
þ 1
12
with the equation approach for analysis of stair-
1 2 3
well pressurization. Figure 50.10 illustrates the
¼ ð100 þ 1 þ 1Þ¼1=2 ¼ 0:099 ft2 floor plan of a multistory building that can be

Fig. 50.10 Floor plan of

building floor illustrating
symmetry

EDUFIRE.IR Telegram EDUFIRE_IR

 50 Smoke Control 1797

divided in half by a plane of symmetry. Flow equation can be expressed in terms of volumetric
areas on one side of the plane of symmetry are flow as shown in Equation (50.12).
equal to corresponding flow areas on the other
side. For a building to be so treated, every floor of rffiffiffiffiffiffiffiffiffi
2Δ p
the building must be such that it can be divided in V ¼ 776 CA
ρ
the same manner by the plane of symmetry. If rffiffiffiffiffiffiffiffiffi ð50:12Þ
wind effects are not considered in the analysis or 2Δ p
V ¼ CA for SI
if the wind direction is parallel to the plane of ρ
symmetry, then the airflow in only one-half of
the building needs to be analyzed. It is not nec- where V is volumetric flow through the path in
essary that the building be geometrically sym- cubic feet per minute, cfm (m3/s).
metric, as shown in Fig. 50.10. A building that is Equations (50.10), (50.11) and (50.12) are
not geometrically symmetric can be symmetric equivalent forms of the same equation, and the
with respect to flow. label “orifice equation” applies to all of them.
The orifice equation gets its name because it is
used to calculate the flow through an orifice. For
these flow equations, the area term is the cross-
sectional area, and the flow coefficient is called
Flow and Pressure Difference the discharge coefficient. A network flow pro-
gram such as CONTAM uses this flow meter
The primary equation used for analysis of pres- terminology. Flow areas and flow coefficients
surization smoke control systems is the orifice for building components are discussed later, and
equation given in Equation (50.10). Idelchik [10] also is a source of flow data.
pffiffiffiffiffiffiffiffiffiffiffiffi Airflow paths must be identified and
m ¼ 12:9 CA 2ρΔ p
pffiffiffiffiffiffiffiffiffiffiffiffi ð50:10Þ evaluated in the design of smoke control systems.
m ¼ CA 2ρΔ p for SI Some leakage paths are obvious, such as cracks
For mass flow at 70  F (21  C) and standard around closed doors, open doors, elevator doors,
atmospheric pressure, the orifice equation windows, and air transfer grilles. Construction
becomes cracks in building walls are less obvious but no
less important.
pffiffiffiffiffiffiffi
msv ¼ 2610CA Δ p The flow area of most large openings, such as
pffiffiffiffiffiffiffi ð50:11Þ open windows, can be calculated easily. How-
msv ¼ 0:839CA Δ p for SI
ever, flow areas of cracks are more difficult to
where evaluate. The area of these leakage paths
m ¼ mass flow through the path, lb/s (kg/s), depends on workmanship (such as how well a
msv ¼ mass flow through the path, scfm (stan- door is fitted or how well weatherstripping is
dard m3/s), installed). A door that is 36 in. by 7 ft (0.9 by
C ¼ flow coefficient, dimensionless, 2.1 m) with an average crack width of 1/8 in.
A ¼ flow area (or leakage area), ft2 (m2), (3.2 mm) has a leakage area of 0.21 ft2
Δp ¼ pressure difference across path, in H2O (0.020 m2). However, if this door is installed
(Pa), with a ¾ in. (19 mm) undercut, the leakage area
ρ ¼ gas density in flow path, lb/ft3 (kg/m3). is 0.32 ft2 (0.30 m2). This is a significant differ-
One standard cubic foot per minute, scfm, ence. The leakage area of elevator doors has been
equals 0.00125 lb per second, and one standard measured in the range of 0.55–0.70 ft2
cubic meter per second (standard m3/s) equals (0.051–0.065 m2) per door.
1.2 kg per second at 70  F (21  C) and standard For many flow paths in buildings, a flow coef-
atmospheric pressure. Alternatively, the orifice ficient of 0.65 is used. The open doors of

EDUFIRE.IR Telegram EDUFIRE_IR

 1798 J.H. Klote

Table 50.2 Flow areas of walls and floors of commercial buildings

Area ratio
Leakage area per unit wall area
Construction element Leakage in2/ft2 ft2/ft2 m2/m2
3 5
Exterior Building Walls (includes construction cracks, Tight 7.2  10 5.0  10 5.0  105
2 4
cracks around windows and doors) Average 2.5  10 1.7  10 1.7  104
2 4
Loose 5.0  10 3.5  10 3.5  104
1 3
Very 1.7  10 1.2  10 1.2  103
Loose
Stairwell Walls (includes construction cracks but not Tight 2.0  103 1.4  105 1.4  105
cracks around windows or doors) Average 1.6  102 1.1  104 1.1  104
Loose 5.0  102 3.5  104 3.5  104
Elevator Shaft Walls (includes construction cracks but Tight 2.6  102 1.8  104 1.8  104
not cracks around doors) Average 1.2  101 8.4  104 8.4  104
Loose 2.6  101 1.8  103 1.8  103
Leakage area per unit floor area
Floors (includes construction cracks and gaps Tight 9.5  104 6.6  106 6.6  106
around penetrations) Average 7.5  103 5.2  105 5.2  105
Loose4 2.4  102 1.7  104 1.7  104
Note: The data in this table are for use with the orifice equation with a flow coefficient of C ¼ 0.65. Floor leakage does
not account for gaps that sometimes exist between the floor and curtain walls

pressurized stairwells commonly have stationary the free area, and the free area is smaller than the
vortices which reduce the flow significantly [5, geometric area (height multiplied by width) of
19]. These vortices are thought to be caused by the vent area. Because the slats in louvers are
the asymmetric flow from the stairs, and station- frequently slanted, calculation of the flow area
ary vortices can be expected at many open doors is further complicated. When available,
in other locations of smoke control systems. For manufacturers’ data regarding free are should
open doors in stairwells, the geometric area of be used. It is generally considered that the free
the opening should be used for the flow area area of a vent with a louver is about half the
along with a flow coefficient of 0.35. geometric area.
Typical leakage areas for walls and floors of The density of air and smoke are expressed by
commercial buildings are tabulated as area ratios the ideal gas law which is
as shown in Table 50.2. These data are based
from field tests performed by the National 144 p
ρ¼
Research Council of Canada (Tamura and RðT þ 460Þ
ð50:13Þ
Wilson 1966; Tamura and Shaw 1976a, 1976b, p
ρ¼ for SI
1978; [38–40]). Considerable data concerning RðT þ 273Þ
leakage through building components are also
provided in the Handbook of Smoke Control where
Engineering. ρ ¼ density, lb/ft3 (kg/m3),
The determination of the flow area of a vent is p ¼ pressure, lb/in2 (Pa),
not always straightforward especially when the R ¼ gas constant, 53.34 ft lbf/lbm/ R (287 J/
vent surface is covered by a louver or screen. For kg K)
vents with louvers, the flow area is referred to as T ¼ temperature,  F ( C).

EDUFIRE.IR Telegram EDUFIRE_IR

 50 Smoke Control 1799

W ¼ door width, ft (m),

Friction Losses in Shafts A ¼ door area, ft2 (m2),
d ¼ distance from doorknob to knob side of
The pressure losses due to friction in ducts, door, ft (m),
stairwells and elevator shafts can be significant Δp ¼ pressure difference, in. H2O (Pa).
when flow rates are high. Tamura and Shaw Equation (50.14) applies when the door
(1976b) [40] and Achakji and Tamura [1] opening force is applied at the knob. Example 2
conducted tests of pressure loss in stairwells. illustrates calculation of the door opening force.
Network computer models such as CONTAM
employ algorithms that use this test data to cal-
culate pressure losses due to friction in ducts, Example 2. Door Opening Force
stairwells and other flow paths. What is the door opening force for a side
hinged swinging door 3 ft wide by 7 ft high
with a door closer that requires 9 lb of force
Door Opening Forces and a pressure difference across it of
0.35 in. H2O? The knob is 3 in. (0.25 ft)
The door opening forces resulting from the from the edge of the door.
pressure differences produced by a smoke con- W ¼ 3 ft; Fdc ¼ 9 lb; A ¼ 3
trol system must be considered. Unreasonably (7) ¼ 21 ft2; d ¼ 0.25 ft; Δp ¼ 0.35 in.
high door opening forces can result in H 2O
occupants having difficulty or being unable to The door-opening force is F¼
5:2ð3Þð21Þð0:35Þ
open doors along the egress route such as into a Fdc þ 5:2WAΔ
2ðWdÞ ¼ 9 þ
p
2ð30:25Þ ¼ 30 lb
stairwell.
The force to open a side hinged swinging door
is shown in Fig. 50.11. The force required to
open such a door when the smoke control system
is operating can be determined using Equation
(50.14). Design Pressure Differences

5:2WAΔ p Pressurization smoke control systems are

F ¼ Fdc þ
2ðW  dÞ designed to operate within a pressure difference
ð50:14Þ range. This range is between the minimum
WAΔ p
F ¼ Fdc þ for SI design pressure difference and the maximum
2ð W  d Þ
design pressure difference. A minimum design
where: pressure difference intended to prevent smoke
F ¼ total door-opening force, lb (N), migration across a barrier of a smoke control
Fdc ¼ door closer force, lb (N), system is generally stipulated by the applicable

Fig. 50.11 Door-opening

force for side hinged door

EDUFIRE.IR Telegram EDUFIRE_IR

 1800 J.H. Klote

building and life safety regulations. A smoke idealized building has no vertical leakage
control system should be designed to maintain through the floors and shafts, and has leakage
this minimum design pressure difference under that is the same from floor to floor.
likely conditions of stack effect and wind. The pressure profiles of stairs in actual
The pressure difference across a barrier must buildings depends on many factors including:
not result in door-opening forces that exceed the (1) the leakage values of the various openings
maximum values stipulated by the applicable (flow paths) through building elements such as
building and life safety regulations. For example, walls and floors, (2) the building floor plans,
in NFPA 101, Life Safety Code [30], this maxi- (3) the size of the elevator shaft or shafts and
mum force is 30 lb (133 N). Calculation of door the number of elevator doors, (4) the presence or
opening forces is discussed above. Acceptable absence of elevator vents, and (5) the leakage
pressurization consists of maintaining pressure through other shafts. There are many possible
differences across the barriers of a smoke control shapes for such pressure profiles in actual
system that are between the minimum and maxi- buildings.
mum design values. For a building with vertical leakage, the flows
through the floors and shafts act to even out to
some extent the highest and lowest pressure
Stairwell Pressurization differences across the stairwell. The profile for
a building with vertical leakage is bounded by
Pressure differences across a stairwell tend to the extremes of the pressure profile of the
vary over the height of the stairwell. Figure 50.12 idealized building. This means that other things
shows two pressure profiles for pressurized being equal, the smallest pressure difference of
stairwells during cold winter months. One profile the idealized analysis will be less than that of the
is for an idealized building, and the other is for a actual building, and that the largest pressure dif-
more realistic building with vertical leakage ference of the idealized analysis will be more
through floors and an elevator shaft. When it is than that of the actual building. This is the reason
cold outside, the pressure differences tend to be that the algebraic equation method discussed
less at the bottom of the stairwell than at the top below is conservative.
as can be seen in Fig. 50.12. When it is hot An algebraic equation method for analysis of
outside, the trend is the opposite. For both winter pressurized stairwells is presented in Chap. 10 to
and summer conditions, the pressure profile for the Handbook of Smoke Control Engineering.
an idealized building is a straight line. An This algebraic equation method is based on

Fig. 50.12 Pressure

profile of a pressurized
stairwell in winter

EDUFIRE.IR Telegram EDUFIRE_IR

 50 Smoke Control 1801

(1) the idealized building, (2) flows calculated by A2SB ðT B þ 460Þ

the orifice equation, (3) effective flow areas, and FR ¼ 1 þ
A2BO ðT S þ 460Þ
(4) symmetry. The algebraic equation method ð50:16Þ
does not account for pressure losses in the stair- A2 ðT B þ 273Þ
FR ¼ 1 þ 2SB for SI
well due to friction, but these losses tend to be ABO ðT S þ 273Þ
small for stairwells when all the stair doors are
closed.
where
Network computer models such as CONTAM
ASB ¼ flow area between the stairwell and the
are capable of analyzing pressurized stairwells
building, ft2 (m2),
much more realistically than the algebraic equa-
ABO ¼ flow area per stairwell between the build-
tion method. CONTAM can simulate the impact
ing and the outside, ft2 (m2),
of a realistic building flow network based on the
TS ¼ temperature in stairwell,  F ( C),
performance of pressurized stairwells. As
TB ¼ temperature in building,  F ( C).
already mentioned, computer network models
The area, ASB, is the total flow area between
can also simulate pressure losses in the stairwell
due to friction. the stairwell and the building, which would typi-
cally include the gaps around all the closed doors
and the leakage paths in the walls. For a stairwell
with an unpressurized vestibule, ASB, is the sum
Height Limit
of the effective flow areas for all floors from the
stairwell to the building.
For some tall stairwells, acceptable pressuriza-
The area, ABO, is on a per stairwell basis
tion may not be possible because of the impact of
because of symmetry considerations. For a build-
the indoor to outdoor temperature differences.
ing with an open floor plan, ABO consists of the
This is more likely with systems with treated
total leakage area of the exterior walls divided by
supply air than those with untreated supply air.
the number of stairwells. For more complex floor
The height limit is the height above which
plans, an effective flow area concept discussed
acceptable pressurization is not possible for an
above needs to be used to calculate ABO.
idealized building. For standard atmospheric
pressure at sea level, the height limit can be
determined by Equation (50.15).
Stairwell Temperature
FR ðΔ pmax  Δ pmin Þ
Hm ¼ 0:131 Today, the supply air for most stairwells in North
1 1
 America is not treated so that pressurized
T O þ 460 T S þ 460
stairwells are hot in the summer and cold in the
FR ðΔ pmax  Δ pmin Þ winter. In many applications, the use of untreated
H m ¼ 2:89  104 for SI
1 1 supply air can be justified for the following

T O þ 273 T S þ 273 reasons: (1) fire drills are usually held in the
ð50:15Þ spring or fall when the outside temperature usu-
ally is mild, and (2) during a fire emergency
where being exposed to nearly outdoor temperatures
Hm ¼ height limit, ft (m), seems reasonable considering occupants are
FR ¼ flow area factor (dimensionless), often traveling to the safety of the outdoors.
Δpmax ¼ maximum design pressure difference, When pressurization air is untreated in cold
in. H2O (Pa), climates, there is a concern about the water freez-
Δpmin ¼ minimum design pressure difference, ing in sprinkler and standpipe risers in stairwells.
in. H2O (Pa). To prevent such freezing, listed heat tracing
The flow area factor is systems can be used on the risers.. Alternately,

EDUFIRE.IR Telegram EDUFIRE_IR

 1802 J.H. Klote

pressurization air can be treated to a minimum (2) is intended to maintain acceptable pressuriza-
temperature in the range of 45–50  F (7–10  C) tion with all the doors closed, and (3) has no
to prevent such freezing of water in the riser. features to prevent loss of pressure when stair
Using heat trace systems and untreated pressuri- doors are opened. As discussed later, a
zation air has the advantage of minimizing the compensated stairwell system has features
impact of stack effect, but using air treated to a intended to prevent pressure loss when stair
minimum temperature has the advantage of doors are opened, but such systems can be rather
minimizing the potential of freezing water on complex with regard to their design, installation
the stairwell floor during firefighting. and operation.
When pressurization air is untreated, the stair- Figure 50.13a is an example of a simple build-
well temperature can be expressed as ing. The algebraic equation method can be used
to size the supply fans for a simple building.
T S ¼ T O þ ηðT B  T O Þ ð50:17Þ Some engineers have developed their own rules
where of thumb that are appropriate for certain kinds of
TS ¼ temperature in the stairwell,  F ( C), stairwell pressurization systems in some
TO ¼ temperature outdoors,  F ( C), buildings. Rules of thumb are generally in the
TB ¼ temperature in the building,  F ( C), range of 300–550 cfm (0.14–0.26 m3/s) per floor.
η ¼ heat transfer factor (dimensionless). Engineers determining a rule of thumb for stair-
There has been little research on the heat well pressurization should take into account the
transfer factor. It is believed that the heat transfer building specifications and the anticipated qual-
factor is in the range of 0.05–0.15. In the absence ity of construction. Of course, experienced
of better data for a specific application, a heat engineers develop rules of thumb including an
transfer factor of 0.15 is suggested as being allowance to avoid the expense of replacing fans,
conservative regarding the impact of stack motors and electrical wiring in the event that the
effect. stairwell would be somewhat more leaky than
For untreated supply air, it takes a few anticipated. Example 3 illustrates calculations
minutes for the temperature in the stairwell to for a simple stairwell system in a simple
stabilize near that of the outdoors. During this building.
stabilization, excessive pressure differences
could be produced. To prevent this, supply air
Example 3. Simple Stairwell Pressurization
can gradually be increased so that when the
in a Simple Building
stairwell temperature is near that of the building
there is insufficient flow to cause excessive The stairwells in a 20 story open plan office
pressurization. If needed, the temperature stabi- are to be pressurized, and the stairwells are
lization can be evaluated by a heat transfer the only pressurization smoke control
analysis. systems in the building. The building has
two stairwells that serve all floors. This
building can be considered simple because
Simple and Complicated Buildings the stairwells are all the same height and
the floors are very similar from floor to
For simple stairwell pressurization systems in floor. The winter design temperature is
simple buildings, some designers may know TO ¼ 10  F, and the building temperature
from experience that the pressurized stairwell is TB ¼ 70  F. The minimum and maxi-
will work as intended, and the fans can be sized mum design pressure differences are Δpmin
by simple calculations. A simple stairwell pres- ¼ 0.10 in. H2O and Δpmax ¼ 0.35 in.
surization system is one that: (1) has air supplied H2O. The floor-to-floor height is 10 ft,
at a constant (or nearly so) volumetric flow rate, and building height is 200 ft. For a typical

(continued)

EDUFIRE.IR Telegram EDUFIRE_IR

 50 Smoke Control 1803

Fig. 50.13 Simple and complicated buildings with respect to stairwell pressurization

(continued) the building height, the stairwells can be

floor, the flow area between the stairwell pressurized.
and the building is ASB ¼ 0.34 ft2, and the Part 2: The stairwells are pressurized
flow area per stairwell between the build- with treated air at 70  F, can this stairwell
ing and the outside is ABO ¼ 0.30 ft2. be pressurized?
Part 1: The stairwells are pressurized The flow area factor is FR ¼ 1þ
with untreated outside air, can this stair- A2SB ðT B þ460Þ ð70þ460Þ
2

A2BO ðT S þ460Þ
¼ 1 þ 0:34
0:302 ð70þ460Þ
¼ 2:28 :
well be pressurized?
Using a heat transfer factor of η ¼ 0.15, The height limit is Hm ¼ 0.131
FR ðΔ pmax  Δ pmin Þ 2:28ð0:35  0:10Þ
the stairwell temperature is TS ¼ TO + ¼ 0:131 ¼ 310 ft:
10þ460  70þ460
1 1
1
 T S þ460
1
η(TB  TO) ¼ 10 + 0.15(70  10) ¼ 19  F. T O þ460

The flow area factor is FR ¼ 1þ This height limit is much less than that
A2SB ðT B þ460Þ ð70þ460Þ
2 of Part 1. As with part 1, the stairwells can
A2BO ðT S þ460Þ
¼ 1 þ 0:34
0:302 ð19þ460Þ
¼ 2:42 : be pressurized, because the stairwells are
The height limit is Hm ¼ 0.131 in a simple building and the height limit is
FR ðΔ pmax  Δ pmin Þ 2:42ð0:35  0:10Þ greater than the building height.
¼ 0:131 ¼ 1980 ft:
10þ460  19þ460
1 1
1
T O þ460  T S þ460
1
Part 3: Each stairwell is to be
Because the stairwells are in a simple pressurized with one fan. Choose the
building and the height limit is greater than capacity of the fan.

(continued)
EDUFIRE.IR Telegram EDUFIRE_IR
 1804 J.H. Klote

pressurization, propeller fans should have tops

(continued) that shield the fan from wind effects. Wall-
Based on experience with similar mounted propeller fans should not be used
construction and buildings, the design because they can be adversely impacted by the
engineer chooses 420 cfm per floor for wind unless a wind analysis indicates otherwise.
this application. Because the stairwell is With a bottom injection system such as
20 stories, the fan capacity is 8400 cfm. illustrated in Fig. 50.14b, some of the supply air
Each stairwell needs an 8400 cfm fan. can short circuit the system by flowing directly
out of the opened exterior bottom doorway
reducing system effectiveness. The bottom door-
For buildings that are relatively complicated, way is expected to be open as occupants egress
computer-based network analysis of the the building through the stairwell. Simulation of
pressurized stairwells is often needed to deter- such detailed fluid flow is typically beyond the
mine if the stairwell systems are capable of being capability of network models such as CONTAM,
balanced to perform as intended. For stairwells but it can be simulated with more sophisticated
pressurized with untreated air, building complex- computations fluid dynamics, CFD, computer
ity often has more impact than stack effect. Even models. It is recommended that bottom injection
buildings that are not especially tall are consid- systems be analyzed using CFD to determine the
ered complicated when the floors plans vary sig- extent to which supply air flows out of an open
nificantly from floor to floor. exterior door. Alternatively, the air can be
The building indicated in Fig. 50.13b consists introduced into the stairwell at least one floor
of underground parking levels, general hotel above or below the exterior doors.
floors, guest room floors and a penthouse. It can For tall stairwells, single injection systems
be difficult to maintain acceptable pressurization can fail when a few doors near the air injection
of stairwells that extend from the parking levels point are open simultaneously. Much of the pres-
to the penthouse, because the plans of these surization air can be lost through these open
floors are so different. Complicated buildings doors, and the system will then fail to maintain
should probably be analyzed with a network positive pressures across doors further from the
analysis model such as CONTAM. Wind effects injection point compromising the effectiveness
add complexity to building when there are many of the overall stairwell pressurization system.
openings to the outdoors (operable windows, To reduce the potential for such failure, multiple
balconies with doors that open, etc.). For compli- injection systems can be used. Multiple injection
cated buildings with many openings to the out- systems can consist of one fan supplying air
doors, analysis of the pressurized stairwell through a duct located in a shaft as shown in
systems with a computer network model is Fig. 50.14c. Other arrangements of multiple
needed. injection systems eliminate the need for a shaft
by using more than one fan as shown in
Fig. 50.14d.
Single and Multiple Injection There has been no research on this subject, but
the consensus is that single injection systems for
A single injection system has pressurization air stairwell heights more than 100 ft (30.5 m) need
supplied at one location. Air can be supplied at a design analysis using computer network
the top of the stairwell, the bottom, or at a loca- models. For multiple injection systems supplying
tion in between. Figure 50.14a and b illustrate air through a duct in a shaft, injection points are
top and bottom injection systems. When roof- usually one to three floors apart. Multiple injec-
mounted propeller fans are used for stairwell tion systems that have a separate fan at each

EDUFIRE.IR Telegram EDUFIRE_IR

 50 Smoke Control 1805

Fig. 50.14 Some single

and multiple injection
stairwell pressurization
systems

injection point can have injection points much a stairwell, and vestibules have the potential to
further apart. For systems with two injection reduce the probability of an open-door connec-
points, one at the top and another near the bot- tion existing between the stairwell and the
tom, a computer network model analysis is building.
recommended for stairwell heights more than An evacuation analysis can be performed to
200 ft (61 m). determine the extent to which both vestibule
doors and stairway doors are likely to be opened
simultaneously. For densely populated buildings,
Vestibules it is expected that on many floors both vestibule
doors and stairway doors would be opened simul-
Pressurized stairwells with vestibules are occa- taneously. Therefore, vestibules may provide lit-
sionally used. The vestibules can be: (1) unpres- tle benefit of an extra barrier for densely
surized, (2) pressurized, (3) ventilated, or (4) a populated buildings.
combination of pressurized and ventilated. The algebraic equation method of analysis can
Vestibules provide an additional barrier around be used to analyze a pressurized stairwell with an

EDUFIRE.IR Telegram EDUFIRE_IR

 1806 J.H. Klote

unpressurized vestibule. The pressure differences into the stairwell was diluted by the large amount
and flows of stairwell systems with any kind of of air supplied to the stairwell. In light of this
vestibules can be analyzed by computer network finding, ASHRAE is sponsoring a research proj-
model. It is possible to evaluate the benefits of ect to study the need for compensated stair
ventilated vestibules using tenability analysis. systems.
Many kinds of compensated stairwell pressur-
ization systems have been used, but the most
System with Fire Floor Exhaust common are (1) the open exterior door system
and (2) the variable air volume (VAV) system.
System employing fire floor exhaust can achieve The open exterior door system has “constant-
acceptable pressurization of tall stairwells in supply” airflow, and an exterior stairwell door
very complex buildings. A relatively small that opens automatically upon system activation.
amount of air is supplied to the stairs, and the This system is sometimes called the Canadian
fire floor is exhausted such that acceptable pres- system because it originated in Canada, and it
surization is maintained on the fire floor where it has been used extensively there. The supply air
is needed. It is common to also exhaust one or rate is not actually constant, but it varies to some
two floors above and below the fire floor. Fire extent with the pressure across the fan. For cen-
floor exhaust is a form of zoned smoke control, trifugal fans this variation in flow is generally
and stairwell pressurization with such zoned small. However, the term “constant-supply” is
smoke control is discussed later. used to differentiate this system from the systems
where the supply air rate is designed to
intentionally change.
Stairwells and Open Doors By keeping the exterior stairwell door open
during system operation, the Canadian system
When any stair door is opened in a simple stair- eliminates the major source of pressure
well pressurization system, the pressure differ- fluctuations. This system is simple and relatively
ence drops significantly. When all doors are inexpensive, but there are many locations where
closed suddenly in such a simple system, the opening exterior doors automatically raises
pressure difference increases significantly. A issues of building security. For complex
compensated stairwell pressurization system is buildings, it is recommended that this system be
one that adjusts for changing conditions either evaluated using a computer network model to
by modulating supply airflow or by relieving assure that it operates as intended.
excess pressure. The intent of a compensated With the VAV system, the flow rate of supply
system is to maintain acceptable pressurization air to the stairwell is adjusted to account for
when doors are opening and closing. opening and closing of doors. Tamura (1990)
In the United States, most building and life conducted research on VAV systems at the
safety regulations do not require pressurized National Research Council of Canada. It was
stairwells to be compensated, and such stairwells found that the pressure drops when doors are
are designed to maintain pressurization only opened, and it took about 3–7 min for the pres-
when all the stair doors are closed. Traditionally, sure to recover to the initial value. When all the
some engineers believed that pressurized stair open doors in a VAV system are closed,
stairwells need to be compensated, but an inci- there is a pressure spike. In Tamura’s research,
dental finding of a study by Klote [18] casts the spike had a peak of 0.728 in. H2O (181 Pa).
doubt on this opinion. For two simulations in This spike only lasted about 30 or 40 s, but the
this study with a closed stair door on the fire peak was much more than any reasonable maxi-
floor and some other stair doors open, the stair- mum design pressure difference. Such peaks are
well remained tenable. The reason the stairwell a concern. A person encountering such a peak,
remained tenable was that the smoke that leaked would probably not be able to open the stair door,

EDUFIRE.IR Telegram EDUFIRE_IR

 50 Smoke Control 1807

but they could open it a minute or so later multiple car shafts with car velocities less than
provided they knew enough to try. It is possible 1000 fpm (5 m/s), piston effect should not
that a person encountering such a peak would adversely impact the performance of elevator
think the stair door was locked, and he or she pressurization. For elevators in single car shafts
might not try to open it again. with car velocities less than 500 fpm (2.5 m/s),
Wind can have a serious impact on VAV stair piston effect should not adversely impact the
pressurization systems. During design analysis of performance of elevator pressurization.
some of these systems, some engineers have Design of pressurized elevator shafts is much
encountered very high pressure differences dur- more complicated than design of pressurized
ing some wind conditions. For example, when an stairwells, but there are a number of approaches
exterior door is opened during the design wind that can deal with this complexity. The reasons
speed, a compensated stair system may supply so for this complexity are: (1) often the building
much air that the pressure difference across some envelope is not capable of effectively handling
stair doors may exceed the maximum design the large airflow resulting from both elevator and
value. It is possible to exceed this design value stairwell pressurization, and (2) open exterior
by as much as 100 %. During such an occurrence, doors on the ground floor can cause high pressure
it would be impossible or extremely difficult for differences across the elevator shaft at the ground
occupants to enter the stairwell. For this reason, floor.
it is recommended that design analysis of VAV Usually a number of exterior doors on the
compensated stairwell pressurization systems ground floor are open during a building fire.
include computer network model simulations During a fire, the fire service opens a number of
under wind conditions. exterior doors or keeps these doors open while
fighting the fire. Occupants also open exterior
doors during evacuation. The shaft pressurization
Elevator Shaft Pressurization system needs to operate as intended with these
exterior doors open.
The elevator pressurization systems discussed in Generally a computer network model analysis
this section are intended to prevent smoke from is needed to determine if pressurized elevators
flowing from the fire floor through an elevator and pressurized stairwells in a particular building
shaft and threatening life on floors away from the are capable of being balanced to perform as
fire floor. This section does not address smoke intended. While it may be theoretically possible
control for elevator evacuation, but elevator to use only a rule of thumb to design these
evacuation is discussed in Chap. 12 of the Hand- systems, a computer network model analysis is
book of Smoke Control Engineering. Usually strongly recommended.
pressurized elevator shafts are in buildings that The elevator pressurization systems discussed
have pressurized stairwells, and the focus of this here are: (1) the basic system, (2) the exterior
section is on both of these pressurization systems vent (EV) system, (3) the floor exhaust
operating together. In the rare situation where (FE) system, and (4) the ground floor lobby
pressurized elevator shafts are the only pressuri- (GFL) system. As mentioned above, these
zation smoke control system in a building, the systems are for use in buildings with pressurized
information in this section may also be useful. stairwells. The results of 36 computer network
The information discussed earlier about piston model simulations using CONTAM were used to
effect can be used to evaluate the impact of study the performance of an elevator shaft pres-
piston effect on the performance of pressurized surization system for a 14-story building
elevator systems. The piston effect produces a illustrated in Fig. 50.15. Further details of this
pressure spike when a car passes a particular analysis are presented in Chap. 11 of the Hand-
floor, and this happens for only a few seconds book of Smoke Control Engineering. The follow-
during the run of an elevator car. For elevators in ing discussion about elevator pressurization

EDUFIRE.IR Telegram EDUFIRE_IR

 1808 J.H. Klote

Table 50.3 Pressure differences criteria for elevator

pressurization simulationsa
Minimum Maximum
System in. H2O Pa in. H2O Pa
Pressurized elevators 0.10 25 0.25 62
Pressurized stairwells 0.10 25 0.35 87
a
The above criteria are for the elevator simulations
discussed Chap. 11 of the Handbook of Smoke Control
Engineering, and some projects may have different
criteria depending on code requirements and requirements
of specific applications

Table 50.4 Flow areas and flow coefficients of doors

used for elevator pressurization simulationsa

Flow Flow area

Flow path Path nameb coefficient ft2 m2
Single door DOOR-SC 0.65 0.25 0.023
(closed)
Single door DOOR-SO 0.35 21 2.0
(opened)
Double door DOOR-DC 0.65 0.48 0.045
(closed)
Double door DOOR-DO 0.35 42 3.9
(opened)
Elevator door DOOR-EC 0.65 0.65 0.06
(closed)
Elevator door DOOR-EO 0.65 6 0.56
(opened)
a
The values in this table were chosen for the elevator
simulations discussed Chap. 11 of the Handbook of
Smoke Control Engineering. The flow areas and flow
coefficients appropriate for a design analysis of a specific
building may be different
b
The path name is an identifier used in the CONTAM
simulations

coefficients used for these simulations are listed

in Tables 50.4 and 50.5. For the CONTAM
simulations of the 14-story building, supply air
was injected only at the top of the elevator shafts,
Fig. 50.15 Floor plans of the example 14 story open plan
office building for elevator pressurization study but about half the supply air was injected at the
top of the stairs and the rest at the second floor.

systems is based on these 36 simulations. For

these simulations, the pressure difference criteria Basic System
listed in Table 50.3 were used, and these criteria
are consistent with pressure differences In the basic system, each stairwell and elevator
requirements in the International Building Code shaft has one or more dedicated fans that supply
(ICC 2012). The leakage values and flow pressurization air. For reasons mentioned above,

EDUFIRE.IR Telegram EDUFIRE_IR

 50 Smoke Control 1809

Table 50.5 Flow areas and flow coefficients of leakages used for elevator pressurization simulations

Flow area
Flow path Leakage classification Path name Flow coefficient ft2 per ft2 of wall m2 per m2 of wall
Exterior walls Tight WALL-EXT 0.65 0.50  104 0.50  104
Average 0.17  103 0.17  103
Loose 0.35  103 0.35  103
Very loose 0.12  102 0.12  102
Interior walls Loose WALL 0.65 0.35  103 0.35  103
Floor (or roof) Tight FLOOR 0.65 0.66  105 0.66  105
Average 0.52  104 0.52  104
Loose 0.17  103 0.17  103
ft2 per ft of wall m2 per m of wall
Curtain wall gap Tight FLOORW 0.65 0.002 0.00061
Loose 0.02 0.0061
See notes on Table 50.4

Fig. 50.16 Elevator pressure differences for basic elevator pressurization system

the basic system also includes stairwell pressuri- with less leaky exterior walls. It can be seen on
zation, and the stair subsystems are not Fig. 50.16 that for leaky exterior walls, the pres-
compensated systems. In most buildings the sure difference across the elevator doors on the
basic system does not result in successful pres- ground floor is about 0.5 in. H2O (75 Pa). For
surization, and the other systems discussed below exterior walls of average leakage the pressure
consist of the basic system plus features to difference across the elevator doors on floor 2 is
improve performance. about 0.35 in. H2O (52 Pa), and at the ground
When the 14-story building contained very floor it is about 1.9 in. H2O (280 Pa). These
leaky exterior walls, the CONTAM simulations values exceed the maximum criteria used for
showed that the basic pressurization system elevator doors, which is 0.25 in. H2O (62 Pa) as
would perform well, but this was not the case indicated in Table 50.3. For average and leaky

EDUFIRE.IR Telegram EDUFIRE_IR

 1810 J.H. Klote

exterior walls, there is insufficient leakage in the

building envelope to accommodate the large
amount of pressurization air supplied to the
shafts.
With very leaky exteriors walls, it can be seen
on Fig. 50.16 that the basic system meets the
pressure difference criteria identified in
Table 50.3. Air was supplied to each elevator
shaft at 27,700 cfm (13.1 m3/s), and air was
supplied to each stairwell at 6560 cfm
(3.09 m3/s). With very leaky exteriors walls,
there is enough wall leakage area to accommo-
date this large amount of pressurization air. For
the few buildings that have very leaky building
envelopes, the basic system can be a simple way
to pressurize elevators and stairwells. For less
leaky buildings, the systems discussed present
other options.

Exterior Vent (EV) System

The idea of the EV system is to use vents in the

exterior walls to increase the leakiness of the
building envelope such that successful pressuri-
zation can be achieved. The vents are usually
closed, but they open when the pressurization
system is activated. The vents should be located
in a manner to minimize adverse wind effects,
and the supply intakes need to be located away
from the vents to minimize the potential for
smoke migration into the supply air. These Fig. 50.17 Typical floor plans of buildings with the
exterior vent (EV) system
vents may need fire dampers depending on build-
ing and fire code requirements.
Figure 50.17a shows a typical floor of the when the vents are located in the exterior walls,
example 14-story building with vents in the exte- but this can be overcome by use of ducted vents.
rior walls. For the example building, the vents Figure 50.17b shows a ducted EV system that
can be sized to assure the design criteria are met. can be used for an office building with permanent
The vents were sized such that the amount of corridors. The ducted EV system can be used
pressurization used for the basic system produce for other occupancies such as hotels and
acceptable pressurization with the EV system in condominiums. Any duct penetrations of a fire
the example building. rated wall will need to be firestopped in accor-
The example building has open office plan. dance with applicable building and fire
For buildings with corridors, the simple EV regulations. For a building where the floors can
approach of Fig. 50.17a is not appropriate. The be either open plan or divided by tenant installed
flow resistance of corridor walls and other walls partitions, an EV system can be achieved by wall
has a negative impact on system performance vents above a suspended ceiling and one or more

EDUFIRE.IR Telegram EDUFIRE_IR

 50 Smoke Control 1811

air transfer grills in the ceiling of the elevator

lobby.
With open exterior doors, it is not necessary to
have exterior vents on the ground floor. Because
the EV system may not be able to achieve accept-
able pressurization with some or all the exterior
doors closed, it may be necessary to have some of
the exterior doors open automatically on system
activation. The number of exterior doors that
need to be opened automatically can be
evaluated by the use of a computer network
model.
In Figure 50.17a and b, the vents are in all four
exterior walls with the intent of minimizing any
adverse impact of the wind. It is suggested that
the vent area be proportional to the area of the
exterior walls. If fewer vents are used, it is
suggested that wind effects be evaluated with
the use of a computer network model.

Floor Exhaust (FE) System

The FE system deals with the building envelope

issue by reducing the amount of supply air used.
In the FE system, a relatively small amount of air
is supplied to the elevator shafts and the
stairwells, and the fire floor is exhausted such
that acceptable pressurization is maintained on
the fire floor where it is needed. It is common to
also exhaust one or two floors above and below
the fire floor. Fig. 50.18 Typical floor plans of buildings with the floor
exhaust (FE) system
The FE system is a kind of zoned smoke
control. As discussed later, exhausting air from
the fire floor and some floors above and below For the 14-story example building, the FE
the fire floor has a beneficial impact on shaft system is shown in Fig. 50.18a. For the
pressurization. Often this system can achieve simulations of the example building, each eleva-
successful pressurization in tall and very com- tor shaft needed 15,100 cfm (7.14 m3/s), and
plex buildings. each stairwell needed 3800 cfm (1.79 m3/s).
Typically the exhaust is through a shaft with a The floor exhaust needed from the floors ranged
fan located in a mechanical floor or on the roof, from 4800 to 5400 cfm (2.28–2.55 m3/s). For a
and dampers between the shaft and the floors are building with many interior partitions, the
closed on all floors when the system is not exhaust can be from the corridor that the
operating. On system activation, the dampers elevators and stairwells open onto, and this is
open on the floors to be exhausted. The outlet shown in Fig. 50.18b.
of the exhaust fan needs to be located away from As with the EV system, some of the exterior
the inlets the supply fans to minimize the poten- doors on the ground floor may need to open
tial for smoke feedback into supply air. automatically upon system activation, and the

EDUFIRE.IR Telegram EDUFIRE_IR

 1812 J.H. Klote

number of such doors needed can be determined life on floors away from the fire floor. In the GFL
with the use of a computer network model. system, the enclosed lobby on the ground floor
protects the elevator from smoke from a fire on
the ground floor. For this reason, the minimum
Ground Floor Lobby (GFL) System elevator pressure difference criterion of
Table 50.3 does not apply to the ground floor
The GFL system has an enclosed elevator lobby for a GFL system. The other criteria of Table 50.3
on the ground floor to reduce the tendency of apply. Table 50.6 identifies the criteria that were
open exterior doors to cause high pressure used for the GFL system simulations. For the
differences across the elevator shaft at the ground GFL system of the simulations discussed below,
floor. The GFL system often has a vent between successful pressurization consists of meeting the
the enclosed lobby and the building with the criteria identified in Table 50.6.
intent of preventing excessive pressure For fires in high-rise buildings, frequently the
differences across the lobby doors. The lobby fire service uses the elevators for rescue and for
doors are the doors between the enclosed lobby mobilization of firefighting equipment. When
and the building. ground floor lobby doors are opened, the pressure
The pressure difference across the lobby door difference may exceed the maximum pressure
and the elevator door depend on the area of the difference. If this can happen for a particular
vent. There is no established criterion for the design, the fire service should be contacted to
maximum pressure difference across the lobby determine if this is acceptable to them.
doors, but the pressure should not be so high as to
prevent the doors from remaining closed. This
value depends on the specific doors and hard-
ware. For discussion here, a maximum pressure
difference for the lobby doors was chosen as 0.35
in H2O (87 Pa), but this value can be much
different for specific applications. The vent
should have a fire damper and a control damper
in series. The control damper can be used to
adjust the flow area of the vent so it can be
balanced during commissioning. Figure 50.19
shows the ground floor of the example building
with a GFL system.
As stated above, the intent of the elevator
pressurization systems discussed in this chapter
is to prevent smoke from flowing from the fire Fig. 50.19 Ground floor of a building with the ground
floor through an elevator shaft and threatening floor lobby (GFL) system

Table 50.6 Pressure differences criteria for GFL elevator pressurization simulationsa
Criteria Minimum Maximum
Location Number in H2O Pa in H2O Pa
Pressurized elevators on ground floor 1 NA NA 0.25 62
Pressurized elevators on other floors 2 0.10 25 0.25 62
Pressurized stairwells on all floors 3 0.10 25 0.35 87
Ground floor elevator lobby door 4 NA NA 0.35 87
a
These pressure differences are with stairwell doors closed, the elevator doors closed, and the ground floor lobby door
closed. The above criteria are for the GFL simulations discussed in this chapter, and some projects may have different
criteria depending on code requirements and requirements of specific applications

EDUFIRE.IR Telegram EDUFIRE_IR

 50 Smoke Control 1813

The floor-to-floor leakage can have a signifi- can be part of a floor as in Fig. 50.20c. A
cant impact on the performance of a GFL system. surrounding zone can be one floor as in
This leakage consists of the leakage of the floor Fig. 50.20a and b, and it can be part of a floor
and that of the curtain wall gap (Table 50.5). as in Fig. 50.20c. A surrounding zone can also be
a number of floors as shown in Fig. 50.20d.
The traditional approach to zoned smoke
Zoned Smoke Control control is to exhaust the smoke zone and to
pressurize the surrounding zones, but many
The traditional approach for HVAC systems is to other approaches have been used. The methods
shut them down during building fires, but HVAC that can be used to treat the smoke zone are:
systems can be designed to operate in a smoke (1) fan powered exhaust, (2), passive smoke
control mode during building fires. Zoned smoke control using smoke barriers (3) exterior wall
control consists of exhausting the zone of the fire vents, or (4) smoke shafts. Fan powered smoke
and possibly pressurizing the surrounding zones. exhaust is the most common method, and pas-
For reasons discussed later in this chapter, sive smoke control using smoke barriers may be
pressurizing the surrounding zones is not satisfactory when fan powered exhaust is not
recommended for zoned smoke control systems practical. Exterior wall vents and smoke shafts
in tall buildings. For zoned smoke control are not commonly used, but they are discussed
systems that rely on smoke exhaust only, the in Chap. 13 of the Handbook of Smoke Control
zoned smoke control can complement the perfor- Engineering.
mance of stairwell pressurization in tall and com- The methods that can be used for the zones
plex buildings. In addition to using the HVAC surrounding the smoke zone are: (1) fan powered
system, dedicated equipment can be used for pressurization or (2) passive smoke control using
zoned smoke control. smoke barriers. Fan powered pressurization of
In zoned smoke control, a building is divided the surrounding zones has a negative impact on
into a number of zones each separated from the stairwell pressurization as discussed below. For
others by barriers. In the event of a fire, the zone the rest of this section, fan powered pressuriza-
with the fire is called the smoke zone, and the tion will be called pressurization, and fan
others are called the non-smoke zones. The zones powered exhaust will be called exhaust.
that border on the smoke zone are called the When the floors of a building are divided into
surrounding zones. Passive smoke protection or many rooms with normally closed doors, these
pressurization smoke protection is used to limit floors do not lend themselves to the traditional
the extent of smoke spread beyond the smoke concept of zoned smoke control. This can also be
zone. It is beyond the capability of smoke control said of wings of a building that are divided into
to make conditions tenable in the smoke zone, many rooms with normally closed doors. For
and it is intended that occupants evacuate the such applications, a form of zoned smoke control
smoke zone as soon as possible. can be used that relies on a combination of corri-
Smoke arrangements of smoke control zones dor exhaust and passive smoke control using
are shown in Fig. 50.20. In this Figure the smoke smoke barriers. Figure 50.21 shows a floor of a
zone is indicated by a minus sign, and the condominium building that can be considered a
surrounding zones are indicated by a plus sign. form of a smoke zone. The floor has corridor
Often the smoke zone is one floor of the building exhaust, and the other spaces rely on passive
(Fig. 50.20a). A common approach is to make the smoke protection of the corridor walls and ceil-
smoke zone be the fire floor plus the floor directly ing floor assembly of the other spaces. This
above and below the fire floor as shown in passive protection tends to minimize smoke
Fig. 50.20b. In a relatively low sprawling build- flow through the ceiling floor assembly during a
ing made of a number of wings, the smoke zone building fire.

EDUFIRE.IR Telegram EDUFIRE_IR

 1814 J.H. Klote

Fig. 50.20 Smoke

arrangements of smoke
control zones

Interaction with Pressurized Stairs and one floor below. However, the same kind of
interactions can happen with smoke zones and
The interaction of zoned smoke control with surrounding zones that are more than one floor.
pressurized stairwells can have a significant Zoned smoke control using both exhaust and
impact on the pressure differences across the stair- pressurization is shown in Fig. Fig. 50.22a, and
well doors. The following discussion is about pressure differences, ΔpSB, from the stairwell to
smoke zones that are comprised of one floor and the building are shown in Fig. 50.22b and
surrounding zones consisting of one floor above Fig. 50.22c. Exhaust of the smoke zone increases

EDUFIRE.IR Telegram EDUFIRE_IR

 50 Smoke Control 1815

the pressure difference across pressurized stair-

well doors on the floor or floors of the smoke
zone. Pressurization of the surrounding zones
decreases the pressure difference across
pressurized stairwell doors on these floors.
The pressure difference, ΔpSB, can be positive
on all the floors as shown in Fig. 50.22b. This
pressure difference can be negative on some
floors as shown in Fig. 50.22c. Negative pressur-
ization can happen on the floors that are
pressurized, and this negative pressurization has
the potential for the significant failure mode
discussed below.
In Fig. 50.20a, smoke should be prevented from
reaching the floor above the smoke zone, and
negative stairwell pressurization should not com-
promise tenability of the stairwell. The effective-
ness of this depends on proper identification of the
fire floor. Properly maintained fire detection and
alarm systems are very good at identifying the
location of a fire, but issues can arise. In some
fire scenarios, the first smoke detector to activate
Fig. 50.21 Example of corridor exhaust and passive has been a floor or so above the fire floor. This can
smoke control be attributed to any of the following: (1) smoke

Fig. 50.22 Interaction between pressurized stairwells and zoned smoke control using both exhaust and pressurization

EDUFIRE.IR Telegram EDUFIRE_IR

 1816 J.H. Klote

Fig. 50.23 Interaction

between pressurized
stairwells and zoned smoke
control using only exhaust

flowing through a complex route to a floor above stairwells is especially important. Alternatively
the fire, (2) smoke detectors not working properly an analysis of this failure mode could be
on the fire floor, and (3) signals from smoke performed that includes factors such as evacua-
detectors being misidentified. tion time, emergency response time, and proba-
Regardless of the reason, when a fire floor is bility using the Firefighter’s Smoke Control
incorrectly identified, the smoke zone is Station (FSCS) for corrective action.
incorrectly chosen. In this situation, the failure The zoned smoke control shown in Fig. 50.23
mode is that inadvertent pressurization of the fire does not have this failure mode. The zoned
floor can push smoke into the stairwells. An smoke control system of Fig. 50.23 consists of a
additional concern is that if this failure mode three story smoke zone that is exhausted and the
happens, it will probably happen to all the surrounding zones rely on passive smoke protec-
stairwells serving the fire floor. This failure tion. The exhaust acts to increase ΔpSB for the
mode is more of a concern for tall buildings three floors of the smoke zone (Fig. 50.23b).
because: (1) acceptable pressurization is more Because this system does not have pressurization
difficult in taller buildings than in shorter ones, of surrounding zones, ΔpSB is not reduced for
and (2) stairwell smoke protection is more surrounding zones, and this eliminates the failure
important in taller buildings than in shorter mode discussed above.
ones. Occupant density is another factor regard- In Fig. 50.23a the fire floor is shaded, and the
ing the importance of stairwell smoke protection. smoke zone consists of the fire floor and the
In this context, a tall building might be thought of floors directly above and below. It is expected
as one having a minimum of about 10 stories. that there will be some smoke flow to the floor
Because of this failure mode, it is recommended above the fire floor, and there may be some
that zoned smoke control using systems smoke flow to the floor below the fire floor.
employing both exhaust and pressurization not This smoke flow is restricted by the floor-ceiling
be used for tall buildings where protection of the assembly. A floor-ceiling assembly is a passive

EDUFIRE.IR Telegram EDUFIRE_IR

 50 Smoke Control 1817

smoke barrier that has significant resistance to be included in the fire scenario. The scenario may
smoke flow. Even a floor-ceiling assembly not also include specifics about the fuel, ignition of
constructed as a passive smoke barrier has con- multiple fuel packages, and the effect of an fire
siderable resistance to smoke flow provided that suppression activities. The selection of the fire
the only openings through it are construction scenario can be based on a combination of pro-
cracks and small cracks around pipe and conduit fessional judgment, fire dynamics, historical fire
penetrations. This means that there will be some data, or code requirements. An analysis of a
amount of time for occupants of the floors smoke control system is likely to include the
directly above and below the fire floor to evacu- consideration of a number of fire scenarios.
ate those floors. Further, the small amount of
smoke on these floors should act to convince Smoke Transport Smoke can flow far from a
occupants of the serious nature of the fire such fire and threaten life. The major driving forces
that pre-movement time before evacuation will that cause smoke movement are naturally occur-
be significantly reduced. ring stack effect, buoyancy of combustion gases,
expansion of combustion gases, wind effect, fan
powered ventilation systems, and elevator piston
Tenability Systems effect.

As previously stated, the conventional smoke Tenability Tenability calculations estimate the
control systems discussed above are based on life hazard of a scenario. Tenability calculations
the approach of preventing occupants from com- address one or more of the following: exposure to
ing into contact with smoke. These conventional toxic gases, exposure to heat, exposure to thermal
systems have some level of smoke contact with radiation, and visibility through smoke. The
the occupant at times when stair doors open for exposures are time-integrated doses of toxic
occupant entry or due to natural fluctuations in gases, heat, and thermal radiation. The conserva-
building pressures. Provided that the tive approach generally used for tenability systems
contaminates are sufficiently diluted, such is to make the tenability calculations as if an occu-
smoke contact is usually considered to be of little pant were to remain at each location under consid-
concern. Tenability systems are designed with eration throughout the duration of the fire scenario.
the intent of providing a tenable environment
for occupants who are exposed to some concen-
tration of smoke. Smoke Transport Calculations

Smoke transport analysis can be done with net-

Analysis Components work models or computational fluid dynamic
(CFD) models. Network models have already
Analysis of tenability systems requires consider- been discussed. The idea of a CFD model is to
ation of the following components: (1) fire sce- divide a space of interest into a large number of
nario, (2) smoke transport mechanisms, and volumetric cells, and to solve the governing
(3) tenability thresholds. equations to calculate the flow, temperature,
and concentrations of fire products at each cell.
Fire Scenario A fire scenario can be thought of The CFD models are appropriate for analysis
as the outline of events and conditions that are of smoke flow in large spaces such as atria, malls,
critical to determining the outcome of alternate and arenas; and the network models are appro-
designs. In addition to the fire location and heat priate for analysis of smoke control systems that
release rate (HRR), the fire scenario includes the involve all or a large part of a large building. Fire
status of the doors, the HVAC systems, and the dynamics simulator (FDS) is a CFD model that is
smoke management system, and other systems. in the public domain and was developed at the
Species (O2, N2, CO, CO2, etc.) generation can National Institute of Standards and Technology

EDUFIRE.IR Telegram EDUFIRE_IR

 1818 J.H. Klote

(NIST) specifically for fire applications [26, This equation is for uniform time intervals as
27]. FDS is available from NIST at no cost. calculated by computer models, and it evaluates
FDS can calculate temperatures, concentrations the FED for the exposure time at the end of
of gases, and visibility; this is a significant aid in interval i (exposure time is nΔt). An FED
the tenability analysis. greater than or equal to one indicates fatality.
Computer network models, such as The concentration is in units mass of the material
CONTAM which is available in the public burned per unit volume. The lethal exposure
domain, were previously discussed for conven- dose, LCt50, is the product of the LC50 and the
tional smoke control systems, but they can also exposure time. The LC50 is the concentration of
be used for smoke transport calculations. airborne combustion products that is lethal to
CONTAM has been used to analyze a number 50 % of the subjects exposed for a specified
of tenability systems ([8, 22], Ferreira 2002; [17, time. The mass concentration of material burned,
18]). CONTAM can simulate the transport of Ci, can be obtained from the smoke transport
contaminants, including the products of combus- calculations.
tion. However, CONTAM cannot simulate heat For any time interval, the visibility can be
transfer, so it cannot calculate the temperatures. calculated from
The user needs to supply the temperatures of
K
spaces to CONTAM, and zone fire models have Si ¼ ð50:19Þ
been used to generate such temperature informa- 2:303δm Ci
tion. The zone fire model, CFAST, has been used where
for this purpose [11, 31, 32]. CFAST is available Si ¼ visibility, ft (m),
from NIST at no cost. K ¼ proportionality constant,
δm ¼ mass optical density, ft2/lb (m2/g),
Tenability Calculations Ci ¼ mass concentration of fuel burned lb/ft3
(g/m3).
There are a number of models that can be used to The proportionality constant, K, is 8 for
evaluate exposure to smoke. For most smoke illuminated signs and 3 for non-illuminated
control applications when smoke is diluted to signs. For objects in reflected light (walls, hand
meet visibility criteria, exposure to it is not life rails, stairs, etc.), a value of K ¼ 3 is normally
threatening. The fractional effective dose (FED) used. Because Ci varies from location to location
model is the simplest model for evaluating expo- in fires, the visibility calculated from the above
sure to smoke, and it can be used to check that equation is considered to be the visibility at the
smoke is not life threatening. point for which Ci is calculated. For example, if
The FED can be used to obtain an approxi- the calculated visibility was 20 ft (6.1 m), it
mate of the effects of exposure to toxic gases. would mean that a person could see 20 ft
(6.1 m) through smoke where Ci was uniform.
X
n
Alternatively visibility can be calculated
Ci Δti
along a path through non-uniform smoke. There
i¼1
FED ¼ ð50:18Þ are many applications where non-uniform smoke
LCt50
happens such as smoke on a balcony in an atrium,
where smoke in a tunnel, and smoke in a hotel corridor.
FED ¼ fractional effective dose, dimensionless, For example, Fig. 50.24 shows a person looking
C,i ¼ mass concentration of material burned at at an exit sign through non-uniform smoke. The
the end of time interval i, lb/ft3 (g/m3), smoke near the exit sign could exceed criteria for
Δti ¼ time interval i, min (min), visibility at a point, but that does not mean that
LCt50 ¼ lethal exposure dose from test data, the person could not see the sign. Calculation of
lb ft3 min (g m-3 min). the visibility along the path between the person
n ¼ number of discrete concentration time pairs. and the sign can evaluate if the sign can be seen.

EDUFIRE.IR Telegram EDUFIRE_IR

 50 Smoke Control 1819

Fig. 50.24 Visibility

through non-uniform
smoke

Visibility along a path can also be calculated Δt ¼ time interval, min (min),
from percent obscuration as Ti ¼ temperature of air in interval i,  F ( C).
Incapacitation due to heat exposure would be
KL
S¼ ð50:20Þ expected for FIth greater than or equal to one. If
loge ð1  λ=100Þ contact with gases does not result in incapacita-
where tion due to heat exposure, thermal radiation from
S ¼ visibility, ft (m), those gases would not result in incapacitation for
K ¼ proportionality constant, the same exposure time. Generally, exposure to
L ¼ length of path, ft (m), thermal radiation is not an issue for most smoke
λ ¼ percent obscuration, dimensionless. control applications.
If the visibility is greater than or equal to the
length of the path (S > L ) an object can be seen
over the path. When the path length is the same Commissioning and Testing
as the visibility (L ¼ S), an object at the end of
the path can barely be seen by a person with Commissioning is the means to demonstrate to
average eyesight. If the object were any farther an owner and other project stakeholders that the
away such a person could not see it. Percent installed smoke control system meets the smoke
obscuration, λ, can be calculated by FDS. control system design for the project.
Generally, contact with dry air of Commissioning is the process for verifying and
temperatures greater than 250  F (121  C) can documenting that the performance of facilities,
be expected to result in skin burns. Also, contact systems, and assemblies meets defined fire safety
with dry air at a temperature less than approxi- objectives and criteria. Commissioning refers to
mately 250  F (121  C) leads to hyperthermia. the process of examining, comparing, testing,
For hyperthermia, heat exposure can be and documenting the installation and perfor-
estimated from mance of a smoke control system to ensure that
X it functions according to an approved design.
n
Δt
FIth ¼ Special inspections are a means that an
i¼1
expð5:67  0:0152T i Þ Authority Having Jurisdiction (AHJ) uses to
Xn
Δt
FIth ¼ for SI determine that a smoke control system meets
i¼1
exp ð 5:185  0:0373T i Þ the applicable code requirements and
ð50:21Þ regulations. The International Building Code
(IBC) has requirements for a special inspection
where and describes the qualifications required for a
FIth ¼ total cumulative dose (dimensionless), special inspector (ICC 2012).

EDUFIRE.IR Telegram EDUFIRE_IR

 1820 J.H. Klote

Commissioning Process Testing and balancing refers to the process

where the as-built performance of smoke control
The commissioning process begins at the start of systems is tested in the field and compared to the
the project and continues throughout the project. required design conditions. Adjustments to the
ASHRAE Guideline 5 provides methods for installed system, such as refining the supply air-
verifying and documenting that the performance flow rates, are made to ensure that the smoke
of smoke control systems conforms with respect control system is functioning as intended in the
to the intent of the design [2]. For smoke control approved design documentation.
systems, an AHJ such as a building official or fire System performance testing is the phase
marshal typically enforces a combination of where the code-specified performance
building codes, fire codes, and local standards. parameters appropriate to the smoke control
The intent of the smoke control system design are measured. For example, building
commissioning testing is to determine that the codes require that a minimum pressure difference
system meets the Owner’s Project Requirements exist between a pressurized stairwell and other
(OPR), including code requirements and zones in the building, and that door opening force
inspections by the AHJ throughout the delivery must not exceed a specified amount. In this case,
of the project performance testing would focus on measuring
In order to achieve successful commissioning the pressure difference across stairwell doors and
of a system, a number of different people will door opening forces. Some common parameters
typically be involved in the process. In addition measured during smoke control system perfor-
to the building owner and AHJ, the system mance testing are: (1) exhaust/supply airflow
designer, general contractor, subcontractors, fire quantities, (2) airflow velocities at atrium or
protection engineering consultants, and test and other large open space perimeters, (3) door-
balance technicians can be involved. At the end opening forces, and (4) pressure differences
of the testing, documentation is provided that the between zones.
system is working properly according to the
design. Smoke Bomb Tests Not
Recommended: Chemical smoke
Commissioning activities can occur at multi- from smoke bombs (also called
ple stages during the construction process. Duct smoke candles) is not
inspections, duct leakage testing, and barrier recommended for any
inspections are activities that typically occur performance testing because it
lacks the buoyancy of hot smoke
early in the construction process when the ducts from a real building fire. Smoke
and barriers are readily visible. Component test- near a flaming fire has a
ing, including air flow measurement, can occur at temperature in the range of
a mid-point in construction where power is 1000–2000  F (540–1100  C).
Heating chemical smoke to such
provided to individual devices, but central moni- temperatures to emulate smoke
toring and control has not yet been provided. from a real fire is not
Sequence of operations and final performance recommended unless precautions
testing typically occurs when construction is are taken to protect life and
property
nearly complete, often just before the building
is intended to obtain its permits and open to the
public.
Periodic Testing

Commissioning Testing After a smoke control system has been

commissioned, testing must still be performed
Commonly, testing and balancing is required periodically so that the system is in the proper
before formal acceptance testing to achieve the operating condition in the event of a fire. Periodic
expected performance of all the components. testing needs to be performed over the life of a

EDUFIRE.IR Telegram EDUFIRE_IR

 50 Smoke Control 1821

building to determine that the installed smoke ASB Flow area between the stairwell and the
control systems are capable of operating as building, ft2 (m2),
designed. Periodic testing includes: (1) manual Asr Leakage area between shaft and lobby,
testing involving ongoing inspection and mainte- ft2 (m2),
nance and (2) automatic testing to determine that C Flow coefficient, dimensionless,
integral equipment is functional and operational. C,i Mass concentration of material burned
Automatic testing is often performed at a higher at the end of time interval i, lb/ft3
frequency than manual testing. Continued (g/m3),
inspection and testing helps so that adjustments Cc Flow coefficient for flow around car,
and repairs can be made to account for unfore- dimensionless.
seen changes to the building or failure of Cw Pressure coefficient, dimensionless,
components. d Distance from doorknob to knob side of
Until recently, smoke control system reliabil- door, ft (m),
ity has been somewhat compromised because F Total door-opening force, lb (N),
periodic testing was limited to manual testing. Fdc Door closer force, lb (N),
Inspections performed years after FED Fractional effective dose,
commissioning showed that some smoke control dimensionless,
systems were inoperable, turned off, or made FIth Total cumulative dose (dimensionless),
ineffective due to modifications to equipment or FR Flow area factor (dimensionless),
the building. It is expected that the reliability of g Acceleration due to gravity, ft/s2
smoke control systems will be significantly (m/s2),
improved by the use of automatic weekly self- H Height of wall, ft (m),
testing of system components, afforded by listed Hm Height limit, ft (m),
equipment carrying the appropriate product Hmet Height of wind measurement, ft (m),
designations. K Proportionality constant,
L Length of path, ft (m),
LCt50 Lethal exposure dose from test data, lb
ft3 min (g m3 min).
Nomenclature m Mass flow through the path, lb/s (kg/s),
msv Mass flow through the path, scfm (stan-
A Flow area, ft2 (m2); or door area, ft2 dard m3/s),
(m2), n Number of discrete concentration time
a Wind exponent at wall. pairs.
Aa Free area around the elevator car, ft2 p Pressure, lb/in2 (Pa),
(m2), patm Absolute atmospheric pressure, lb/ft2
ABO Flow area per stairwell between the (Pa),
building and the outside, ft2 (m2), pw Wind pressure, in H2O (Pa),
Ae Effective flow area, ft2 (m2), R Gas constant, 53.34 ft lbf/lbm/ R
Ai Flow area of path i, ft2 (m2). (287 J/kg K)
Aio Leakage area between the building and S Visibility, ft (m),
the outside, ft2 (m2). T Temperature,  F ( C).
Air Leakage area between building and TB Temperature in the building,  F ( C),
lobby, ft2 (m2), TF Temperature in the fire space,  F ( C),
amet Wind exponent in the vicinity of the Ti Temperature of air in interval i,  F ( C).
wind anemometer, dimensionless, Tin Temperature of air entering the fire
As Cross-sectional area of shaft, ft2 (m2), compartment,  F ( C).

EDUFIRE.IR Telegram EDUFIRE_IR

 1822 J.H. Klote

TO Temperature outside,  F ( C); tempera- 2. ASHRAE 2012. Guide for the Commissioning Pro-
ture surroundings,  F ( C), cess of Smoke Control Systems, ASHRAE, Atlanta,
GA.
Tout Temperature of smoke leaving the fire 3. Aynsley, R. M. 1989. The Estimation of Wind
compartment,  F ( C), Pressures at Ventilation Inlets and Outlets on
TS Temperature in the shaft,  F ( C), Buildings, ASHRAE Transactions, Vol. 95, Part 2.
U Elevator car velocity, fpm (m/s), 4. Berl, W.G. and Halpin, B.M. 1980. “Human Fatalities
from Unwanted Fires,” Johns Hopkins APL Techno-
UH Velocity at the upwind wall of height logical Digest, 1, p. 129.
H, mph (m/s), 5. Cresci, R. J. 1973. Smoke and Fire Control in High-
Umet Measured velocity, mph (m/s), Rise Office Buildings - Part II: Analysis of Stair
V Volumetric flow through the path, cfm Pressurization Systems, Symposium on Experience
and Applications on Smoke and Fire Control at the
(m3/s). ASHRAE Annual Meeting, June 1973, Louisville,
Vin Volumetric flow of air into the fire KY, Atlanta, GA, pp. 16-23.
compartment, cfm (m3/s), 6. Dyrbye, C. and Hansen, S. O. 1997. Wind Loads on
Vout Volumetric flow of smoke out of the Structures, Wiley, New York, NY.
7. Fang, J.B. 1980. Static Pressures Produced by Room
fire compartment, cfm (m3/s), Fires, NBSIR 80-1984, National Bureau of Standards,
W Door width, ft (m), Washington.
z Distance above the neutral plane, ft 8. Ferreira, M.J. 1998. Analysis of Smoke Control Sys-
(m). tem Design Using a Computer-based Airflow Analy-
sis, Pacific Rim Conference, Society of Fire
δ Boundary layer height at wall, ft (m), Protection Engineers, Bethesda, MD.
ρ Density, lb/ft3 (kg/m3), 9. ICC 2012. International Building Code, International
η Heat transfer factor (dimensionless). Code Council, Country Club Hills, IL.
λ Percent obscuration, dimensionless. 10. Idelchik, I. E. 1986. Handbook of Hydraulic Resis-
tance, 2nd Edition, Hemisphere Publishing,
δm Mass optical density, ft2/lb (m2/g), New York.
δmet Boundary layer height in the vicinity of 11. Jones W. W., et al. 2009. CFAST – Consolidated
the wind anemometer, ft (m), Model of Fire Growth and Smoke Transport (Version
ρo Outside air density, lb/ft3 (kg/m3), 6) – Technical Reference Guide. NIST Special Publi-
cation 1026, National Institute of Standards and Tech-
Δp Pressure difference, in. H2O (Pa). nology, Gaithersburg, MD.
Δpmax Maximum design pressure difference, 12. Kandola, B. S. 1986a. A Wind Tunnel Building
in. H2O (Pa), Model for the Investigation of Smoke Movement
Δpmin Minimum design pressure difference, Problems, Fire Safety Journal, Vol. 10, No. 3.
13. Kandola, B. S. 1986b. Comparison of Wind Tunnel
in. H2O (Pa). Pressure Measurements and Smoke Movement Com-
ΔpSF Pressure difference from a fire space to puter Predictions inside a Five-storey Model building,
the surroundings, in. H2O (Pa), Fire Safety Journal, Vol. 10, No. 3.
ΔpSO Pressure difference from a shaft to the 14. Kandola, B. S. 1986c. The Effects of Simulated Pres-
sure and Outside Wind on the Internal Pressure Dis-
outside, in. H2O (Pa), tribution in a Five-storey Building, Fire Safety
Δpu,si Upper limit pressure difference from Journal, Vol. 10, No. 3.
the shaft to the building, in H2O (Pa), 15. Klote, J. H. 1988. An Analysis of the Influence of Piston
Δt Time interval, min (min), Effect on Elevator Smoke Control, NBSIR-88-3751,
National Bureau of Standards, Gaithersburg, MD.
16. Klote, J. H. 1995. Design of Smoke Control Systems
for Elevator Fire Evacuation Including Wind Effects,
2nd Symposium on Elevators, Fire, and Accessibility,
Baltimore, MD April 19-21, 1995, ASME, New York.
References 17. Klote, J. H. 2004a. Hazards Due to Smoke Migration
through Elevator Shafts – Vol. I: Analysis and Dis-
1. Achakji, G. Y. and Tamura, G. T. 1988. Pressure Drop cussion. Final Report. NIST GCR 04-864-I, National
Characteristics of Typical Stairshafts in High-Rise Institute of Standards and Technology, Gaithersburg,
Buildings, ASHRAE Transactions, Part 1. MD.

EDUFIRE.IR Telegram EDUFIRE_IR

 50 Smoke Control 1823

18. Klote, J. H. 2004b. Tenability and Open Doors in 32. Peacock, R. D., et al. 2008b. CFAST – Consolidated
Pressurized Stairwells, ASHRAE Transactions, Vol. Model of Fire Growth and Smoke Transport (Version
110. Part 1. 6) – User’s Guide. NIST Special Publication 1041,
19. Klote, J. H. and Bodart, X. 1985. Validation of Net- National Institute of Standards and Technology,
work Models for Smoke Control Analysis, ASHRAE Gaithersburg, MD.
Transactions, Vol. 91, Part 2b. 33. Shaw, C. T. and Tamura, G. T. 1977. The Calculation
20. Klote, J. H. and Tamura, G. T. 1986. Elevator Piston of Air Infiltration Rates Caused by Wind and Stack
Effect and the Smoke Problem, Fire Safety Journal, Action for Tall Buildings, ASHRAE Transactions,
Vol. 11 No. 3. Vol. 83, Part 2.
21. Klote, J. H. et al. 2012. Handbook of Smoke Control 34. Shaw, C. Y., Reardon, J. T. and Cheung, M. S. 1993.
Engineering. ASHRAE, Atlanta, GA. Changes in Air Leakage Levels of Six Canadian
22. Klote, J. H., Nelson, H. E., Deal, S. and Levin, B. M. Office Buildings, ASHRAE Journal, American Soci-
1992. Staging Areas for Persons with Mobility ety of Heating, Refrigerating and Air Conditioning
Limitations, NISTIR 4770, National Institute of Engineers, Atlanta, GA, Vol. 35, No. 2.
Standards and Technology, Gaithersburg, MD. 35. Simiu, E. and Scanlan, R. H. 1996. Wind Effects on
23. Klote, J.H. and Tamura G.T., 1987. “Experiments of Structures: Fundamentals and Application to Design,
Piston Effect on Elevator Smoke Control,” ASHRAE 3rd Ed., Wiley, New York, NY.
Transactions, 93, 2, pp. 2217–2228. 36. Walton, G. N., Dols, W. S. 2005. CONTAM 2.4 User
24. Liu, H. 1991. Wind Engineering – A Handbook Guide and Program Documentation, NISTIR 7251,
for Structural Engineers, Prentice Hall, Englewood, NJ. Revised December 14, 2010. National Institute of
25. MacDonald, A. J. 1975. Wind Loading on Buildings, Standards and Technology, Gaithersburg, MD.
Wiley, New York NY. 37. Harland, W. A., Woolley, W. D. 1979. Fire fatality
26. McGrattan, K. B. et al. 2008a. Fire Dynamics Simu- study – University of Glasgow. Building Research
lator (Version 5) User’s Guide, NIST Special Publi- Establishment, Borehamwood Information Paper IP
cation 1019-5, National Institute of Standards and 18/79.
Technology, Gaithersburg, MD. 38. Tamura, G.T., and A.G. Wilson. 1966. Pressure
27. McGrattan, K. B. et al. 2008b. Fire Dynamics Simu- differences for a nine-story building as a result of
lator (Version 5) Technical Reference Guide, Volume chimney effect and ventilation system operation.
1: Mathematical Model, NIST Special Publication ASHRAE Transactions, 72(1).
1018-5, National Institute of Standards and Technol- 39. Tamura, G.T., and C.Y. Shaw. 1976a. Studies on
ogy, Gaithersburg, MD. exterior wall air tightness and air infiltration of tall
28. NBFU 1939. Smoke Hazards of Air-Conditioning buildings. ASHRAE Transactions, 82(1).
Systems, NFPA Quarterly, Vol. 33, No. 2. 40. Tamura, G.T., and C.Y. Shaw. 1976b. Air leakage
29. NFPA 2012a. Standard for Smoke Control Systems, data for the design of elevator and stair shaft pressuri-
NFPA 92, National Fire Protection Association, zation systems. ASHRAE Transactions, 82(2).
Quincy, MA 41. Tamura, G.T., and C.Y. Shaw. 1978. Experimental
30. NFPA 2012b. Life Safety Code, NFPA 101, National studies of mechanical venting for smoke control in
Fire Protection Association, Quincy, MA. tall office buildings. ASHRAE Transactions, 86(1).
31. Peacock, R. D., et al. 2008a. CFAST – Consolidated 42. Frrreira, M. J. 2002. Use of Multizone Modeling for
Model of Fire Growth and Smoke Transport (Version High-Rise Smoke Control Design. ASHRAE
6) – Software Development and Model Evaluation Transactions, 108(2).
Guide. NIST Special Publication 1086, National
Institute of Standards and Technology, Gaithersburg,
MD. John H. Klote is consultant specializing in fire in smoke
control in Leesburg, VA

EDUFIRE.IR Telegram EDUFIRE_IR

 Smoke Control by Mechanical Exhaust
or Natural Venting 51
James A. Milke

Introduction systems for all of these various large-volume

spaces are the same. However, differences in the
Smoke management in large-volume spaces, smoke control system designs for the variety of
such as atria and covered malls, poses separate large-volume spaces may be found. Differences in
and distinct challenges from well-compartmented designs are a result of differences in fire scenarios
spaces. In particular, smoke control strategies and design goals, reflecting the range of building
using pressure differences and physical barriers uses and operations and the nature of who or what
described by Klote in Chap. 50, and NFPA may be exposed to the smoke. Given the
92, Standard for Smoke-Control Systems [1], are similarities in engineering principles affecting
infeasible. Without physical barriers, smoke smoke control system design, the term atrium
propagation is unimpeded, spreading easily will be used throughout this chapter to refer to
throughout the entire space. The tall ceiling all types of large-volume spaces.
heights in many large-volume spaces pose addi- The discussion presented in this chapter is
tional challenges because of the production of divided into two sections. First, conditions within
substantial quantities of smoke and delayed the atrium prior to actuation of a smoke control
detection times. However, on the positive side, system are discussed. As part of this discussion,
the combination of large-volume space and tall the smoke filling process is described along with
ceiling height permit the smoke to become the time required for actuation of a smoke control
diluted and cooled as it spreads vertically and system. The second part of the chapter includes a
horizontally, thereby reducing the level of hazard description of conditions within the atrium after
posed by the smoke. Even so, there is still a need actuation of the smoke control system.
to ensure that dangerous concentrations of smoke As a preface to any discussion on smoke con-
are prevented in large-volume spaces. trol systems, a definition of smoke must be
In addition to atria and covered malls, there established (NFPA 92, Standard for Smoke Con-
are many other examples of large-volume spaces, trol Systems [1], Section 3.3.13):
including convention centers, airport terminals, The airborne solid and liquid particulates and
sports arenas, and warehouses where smoke gases evolved when a material undergoes pyrolysis
management is of concern. The engineering or combustion, together with the quantity of air
that is entrained or otherwise mixed into the mass.
principles governing the design of smoke control
Although only the combustion products are
visible and potentially toxic, what is visually
J.A. Milke (*)
University of Maryland, Department of Fire Protection observed as smoke is a mixture of the combus-
Engineering, University of Maryland, College Park, tion products and the entrained air. Air is
MD, 20742, USA

M.J. Hurley (ed.), SFPE Handbook of Fire Protection Engineering, 1824

DOI 10.1007/978-1-4939-2565-0_51, # Society of Fire Protection Engineers 2016

EDUFIRE.IR Telegram EDUFIRE_IR

 51 Smoke Control by Mechanical Exhaust or Natural Venting 1825

entrained along the entire height of the smoke smoke and come into contact with combustion
plume below a smoke layer. Proportionally, the gases and smoke particles may become stained or
smoke is mostly entrained air. In the space emit an odor of smoke. Exposure to smoke can
between the base and tip of the flames, most of damage electronic equipment, especially if resto-
the entrained air is not consumed in the combus- ration activities are not initiated promptly after
tion process and only dilutes the combustion the fire [5].
products. Entraining air into the smoke plume Following a fire, a building or portion thereof
increases the mass flow in the plume to increase may be closed due to restoration, threatening
the quantity of smoke produced. However, the mission continuity. This results in loss of revenue
entrained air also dilutes the smoke to decrease for the building owner, temporary unemploy-
the concentration of combustion gases and cool ment of workers in the building, and loss of
the smoke. In some cases, the smoke may be service of the facility to the community, among
sufficiently diluted to mitigate the associated other outcomes.
hazards.

Smoke Layer Interface Position

Hazard Parameters
The smoke layer interface position is located a
Smoke can adversely affect building distance, z, above the top of the fuel, as indicated
occupants, fire brigade members, property in Fig. 51.1. This parameter is used to assess the
(including the building structure and contents), danger of people or objects being immersed in a
and mission continuity. Typically, the threat to smoke layer. Sole use of this parameter to assess
people or objects is posed when they come into hazard level is conservative by considering any
contact with smoke for a sufficient period concentration of smoke to be unacceptable. For
of time. people, even though the physiological effects due
People who become exposed to smoke are to being submerged in “light” smoke levels may
generally harmed as a result of the exposure to be minor, the psychological effects and extended
toxic gases or elevated temperature. The toxic evacuation time may be appreciable. Being
effects of smoke on people are described in Purser surrounded by smoke of any nature may decrease
(see Chap. 63) and Klote et al. [2]. In addition, the speed of evacuation, perhaps until the smoke
smoke may reduce visibility. A reduction of visi- is no longer relatively benign. In terms of prop-
bility may cause people to become disoriented erty protection issues, any smoke may be unac-
and can in turn increase the amount of time they ceptable because of smoke staining or smoke
are exposed to the smoke [3]. A reduction of corrosivity.
visibility may also increase the susceptibility of
building occupants to trip over obstructions or
even fall over balcony railings [4]. Light Obscuration
Building components can be affected by the
elevated temperature due to smoke. Building As with the smoke layer depth parameter, light
components heated by smoke are considered in obscuration is not lethal by itself. Associated
fire resistance analyses. In addition, building with an increase in light obscuration is a reduc-
contents may be affected by exposure to the tion in visibility, which is likely to yield a longer
elevated temperatures, corrosive gases, or partic- evacuation time and extend exposure to the
ulate matter. Contents exposed to heated smoke toxins in smoke. In some documented fires, evac-
may be melted, distorted, or charred, depending uation has been terminated due to a lack of suffi-
on the temperature of the smoke and the degree cient visibility [6–8]. A fire fighter’s injury in an
of exposure. Contents that are submerged in atrium fire was attributed to a significant

EDUFIRE.IR Telegram EDUFIRE_IR

 1826 J.A. Milke

Fig. 51.1 Axisymmetric Axisymmetric plume

plume [1]

H
z

z1

reduction in visibility due to light obscuration 3. Separation from communicating spaces

[4]. The fire fighter fell from an upper balcony Several approaches are available to achieve
because he could not see the edge. smoke management goals in an atrium (e.g.,
Limiting values from 0.23 to 1.2 m
1 have limit the fire size, provide physical barriers, and
been suggested for the extinction coefficient provide mechanical or natural ventilation).
[6–8]. Alternatively, a critical limit may be Selection of the best smoke management
based on a preferred minimum visibility distance approach for a particular atrium should consider
to a particular target. For example, a limit of light the use, size, and arrangement of the associated
obscuration can be suggested such that occupants spaces.
can see an illuminated exit sign across a room or Limiting the fire size can be accomplished by
at the end of a corridor [3, 9]. controlling the type, quantity, and arrangement
of fuel. In addition, the fire size can be controlled
through an automatic suppression system.
Temperature and Gas Specie Physical barriers limit smoke spread to adja-
Concentration cent spaces. The ability of a physical barrier to
limit smoke spread is dependent on the leakage
The final two parameters, elevated temperature of the barrier and pressure difference across the
of the smoke layer and gas specie concentration barrier. The barrier needs to withstand the expo-
(such as CO, CO2, and HCN), can be directly sure to smoke and an elevated temperature envi-
related to the potential for harm (see Chap. 63). ronment. In an atrium with a tall ceiling, the
Critical limits for these two parameters can be temperature of the smoke layer in the atrium
suggested based on toxicity studies. may be only slightly above ambient temperatures
in the space.
Mechanical or natural ventilation may be
Smoke Management Approaches provided to remove smoke from the atrium.
Removing smoke from the atrium can be intended
The design of a smoke control system for an to limit the accumulation of heat and smoke
atrium is influenced by the following three within the atrium or arrest the descent of the
characteristics of the atrium: smoke layer. Mechanical ventilation can be
1. Geometric shape and dimensions provided to oppose smoke movement induced by
2. Relative location within the building the fire to restrict smoke spread to communicating

EDUFIRE.IR Telegram EDUFIRE_IR

 51 Smoke Control by Mechanical Exhaust or Natural Venting 1827

spaces. Gravity vents may be provided to remove smoke poses to contents, property, and mission
smoke, though their performance can be continuity are described elsewhere [2–4, 10].
compromised by environmental factors.

Physical Scale Models

Analytical Approach
Physical scale models provide a representation of
Numerous tools are available to aid in the design a space, though in a reduced scale. Physical scale
and evaluate the adequacy of a smoke control models are especially useful in examining atria
system. The selection of a particular tool is with irregular shapes or numerous projections. A
dependent on the accuracy needed for the analy- review of applying physical scale models as a
sis and the applicability of the analytical design aid for atrium smoke control systems was
tools given the characteristics of the large space provided by Milke and Klote [11].
and selected fire scenarios. The principal Quintiere provided a review of scaling
characteristics that affect applicability are relationships based on preserving the Froude num-
• Geometry of the large space: variation of hor- ber [12]. The Froude number, Fr, is defined as v/gl.
izontal cross-sectional area, sloped versus flat The scaling relations are
ceiling Temperature:
• Transient aspects: unsteady versus steady heat Tm ¼ TF ð51:1Þ
release rate, constant versus transient opera-
tion of smoke control system Geometric position:
• Fire development: heat release rate as a func-

lm
tion of time (for example, constant, power- xm ¼ x F ð51:2Þ
lF
law relationship with time, tn)
• Environmental effects: stack effect, wind Pressure:

• Interacting systems: other smoke control lm
Δ pm ¼ Δ pF ð51:3Þ
systems, HVAC, other exhaust systems (for lF
example in laboratories) Velocity:
The range of design tools available to assess
1=2
lm
the performance of smoke control system designs vm ¼ v F ð51:4Þ
lF
can be grouped into the following categories:
• Zone model (algebraic equation based) Time:

1=2
• Zone model (computer based) lm
• Field model tm ¼ t F ð51:5Þ
lF
• Physical scale model
The intent of an engineering analysis of Convectiveheat release:
smoke conditions in an atrium is to express the
5=2
lm
level of hazard in terms of physically based Q_ c, m ¼ Q_ c, f ð51:6Þ
lF
parameters, for example, smoke layer interface
position, temperature, gas concentration (such as Volumetric flow rate:

5=2
carbon monoxide), and light obscuration. The lm
magnitude of each of these parameters can be V fan, m ¼ V fan, F ð51:7Þ
lF
predicted based on engineering principles. In
addition to being predictable, critical threshold Experiments based on Froude modeling may
values are available for the hazard parameters in be done with air at atmospheric pressure. Froude
order to properly assess the severity of the threat modeling does not preserve the Reynolds num-
(See Chap. 63). This chapter will concentrate on ber. However, appropriate selection of the size of
the life hazards posed by smoke. The hazards the physical scale model can ensure that fully

EDUFIRE.IR Telegram EDUFIRE_IR

 1828 J.A. Milke

developed flow is achieved to minimize the

consequences of not preserving the Reynolds
number. Because the smoke behavior in only
certain areas of the scaled atrium may be of
interest, fully developed flow only needs to be
achieved in these particular areas. Often a physi-
cal scale model with a critical dimension of at
least 0.3 m in any areas of interest will be suffi-
cient to achieve fully developed, turbulent flow.
As an example, in most shopping malls and atria,
the critical dimension in question would be the
floor-to-ceiling height of one of the balconies.
In addition, Froude modeling does not pre-
serve the dimensionless parameters concerning Fig. 51.2 Small-scale model of atrium
heat transfer. Generally, this limitation has little
effect because the temperature is the same for the to formulate a small-scale model, the basic
physical scale model and the full-scale facility. parameters for the scale model are
Froude modeling does not apply to locations with • Height: 3.8-m-tall model (1/8 scale)
high temperature and low Reynolds numbers • Fire size: 28 kW
(e.g., near the flame). However, Froude modeling • Fan capacity: 0.78 m3/s
provides useful information about smoke trans-
port away from the fire.
Some surface effects can be preserved by Analytical Models
scaling the thermal properties of the construction
materials for the model. The thermal properties Two categories of analytical models are zone and
can be scaled by field models. A description of field models is
Thermal properties: outside the scope of this chapter. Zone models

0:9 divide each compartment into a limited number
    lm
kρcp w, m
¼ kρcp w, F ð51:8Þ of control volumes, typically an upper and a lower
lF
zone. Inherent in the zone approach is the assump-
Because scaling thermal properties have only tion of uniform properties throughout each zone.
a secondary effect on fluid flow, considerations In spaces with a large floor area, this assumption
of convenient construction and flow visualization may be tenuous. Nonetheless, calculations
may require that some or all surface materials in associated with the zone model approach are rela-
the model are different from those selected based tively easy to perform and are often accepted for
on thermal property scaling. engineering purposes. Calculations following the
zone model approach may be in the form of alge-
Example 1 A physical scale model is proposed braic equations or a computer algorithm.
to determine the equilibrium smoke layer posi- The zone model approach assumes that smoke
tion for the atrium depicted in Fig. 51.2. Because from a fire is buoyant, rises to the ceiling, and
the horizontal cross-sectional area varies with forms a smoke layer. The buoyant nature of
height, algebraic equation and computer-based smoke is due to the decreased density of the
zone models are of limited value. The overall heated smoke. As smoke rises in a plume, air is
height of the atrium being studied is 30.5 m and entrained to increase the mass flow rate in the
the design fire is steady with a heat release rate of plume. A decrease in the velocity and tempera-
5 MW. An exhaust fan capacity of 142 m3/s is ture of the smoke plume results from the increase
proposed. By applying the scaling relationships in the plume mass flow rate, as dictated by

EDUFIRE.IR Telegram EDUFIRE_IR

 51 Smoke Control by Mechanical Exhaust or Natural Venting 1829

conservation of momentum and energy. In addi- immediately. However, in large spaces with tall
tion, the entrained air dilutes the combustion ceilings, the time required to form a smoke layer
products in the plume. The entire smoke layer may be appreciable. The delay in forming a layer
is assumed to have uniform characteristics. As is attributable to the transport lag of the smoke.
smoke is supplied to the smoke layer from the The smoke filling period continues until the
plume, the interface between the smoke layer and properly sized smoke exhaust fans are actuated.
lower clear air zone descends. The additional
smoke supplied by the plume also results in an
Transport Lag
increase in the smoke layer temperature, carbon
monoxide concentration, and light obscuration.
The transport lag is composed of the time for a
Being a simplification, the zone model
smoke plume to reach the ceiling (plume trans-
approach may not be applicable in some
port lag) and the time for the ceiling jet to reach
situations. One example includes a scenario
the bounding enclosure (ceiling jet transport lag).
with operating sprinklers, which may cool the
These two time periods are depicted in Fig. 51.3.
layer and also entrain smoke from the upper
Correlations for the plume and ceiling jet
layer into the water spray pattern descending
transport lag are available in the literature for
into the lower zone. Another example consists
both steady and t2 fires [13, 14]. Because virtu-
of the case where smoke does not reach the ceiling
ally all fires have a growth period before
as a result of a loss of buoyancy, where the pre-fire
reaching a steady phase, the transport lag
temperature near the ceiling of the atrium is
correlations for steady fires have little relevance.
greater than that near the floor. This situation is
Correlations for the plume transport lag for
discussed in more detail later in this chapter. A
steady and t2-fires are
third situation involves an atrium with a large
Steady files:
cross-sectional area where the horizontal variation
in conditions from one portion of the atrium to
tpl ¼ 0:67H4=3 =Q_
1=3
ð51:9Þ
another is important to the analyst. Where
localized conditions associated with the smoke t2 fires:
plume or smoke layer need to be assessed, field
models are more appropriate than zone models. tpl ¼ 0:1H4=5 t2=5
g ð51:10Þ
Two categories of fire scenarios for smoke
management design in atria include (1) fires Estimates of the plume transport lag from
located in the atrium, and (2) fires located in a Equations 51.9 and 51.10 are provided in
space adjacent and open to the atrium. This chap- Fig. 51.4. As indicated in the figure, even the
ter concentrates only on fires within the atrium shortest plume transport lag for t2 fires,
space. Methods to estimate conditions in any of associated with the fast t2 fire, is greater than
the adjacent spaces, resulting from fires that for a modest-size steady fire.
originating in the atrium or from fires in other Comparable correlations for the ceiling jet
adjacent spaces, are addressed elsewhere [2]. transport lag for steady and t2 fires are
Steady files:
r 11=6
Smoke Filling Period tc j ¼ ð51:11Þ
1:2Q_ H1=2
1=3

A smoke layer is formed once the smoke plume t2 fires:

reaches the ceiling and the ceiling jet spreads 2=5
0:72rtg
horizontally to reach the bounding walls of the tc j ¼ ð51:12Þ
H 1=5
space. Subsequently, the smoke layer starts to
descend in the space. In relatively small spaces A comparison of the ceiling jet transport lag
with low ceilings, the smoke layer forms almost for a modest-size steady fire and t2 fires is

EDUFIRE.IR Telegram EDUFIRE_IR

 1830 J.A. Milke

Fig. 51.3 Plume and

ceiling jet transport lag Ceiling jet transport lag

Plume
transport
lag

Fig. 51.4 Plume Slow t 2

20
transport lag

15 Medium t 2
Time (s)

Fast t 2
10
0.5 MW

0
0 6 12 18 24 30
Height (m)

presented in Fig. 51.5. Again, the transport lag height of the plume where entrainment occurs,
associated with the steady fire is much less than i.e., the distance from the top of the fuel to the
that associated with any of the t2 fires. smoke layer.
Many zone models do not account for trans- Both empirical correlations and theoretically
port lag. In low-height spaces with small based methods are available to address
compartments, this is likely to be inconsequen- conditions during the smoke filling period using
tial. In tall spaces with large cross-sectional hor- a zone model approach [15]. Theoretically based
izontal areas, the lag may be important. In such methods use statements of conservation of mass
cases, only models that incorporate transport lag and energy to determine the volume of the upper
are to be selected. layer. Conservation of mass accounts for the
smoke mass supplied from the plume to
the smoke layer along with any smoke leaving
Smoke Layer Interface Position the zone through ventilation openings. Conserva-
tion of energy is applied to address the energy
Once the smoke layer has formed, the initial rate being supplied by the plume along with heat
of descent of the layer is very rapid, slowing as losses from the layer.
the layer descends. This is attributable to the rate Generally, the predicted smoke layer interface
of smoke production being dependent on the position determined by the two analytical

EDUFIRE.IR Telegram EDUFIRE_IR

 51 Smoke Control by Mechanical Exhaust or Natural Venting 1831

Fig. 51.5 Ceiling jet Slow t 2, 5 m

transport lag 200

160
Med. t 2, 5 m
Slow t 2, 30 m
120 Fast t 2, 5 m

Time (s)
Med. t 2, 30 m

80 Fast t 2, 30 m

40
0.5 MW, 5 m
0.5 MW, 30 m
0
0 6 12 18 24 30
Radius (m)

methods differs. The smoke layer is comprised of spaces. In the experimental efforts, the smoke
the uppermost portion of the layer in which the layer interface position was established by a
conditions are relatively uniform at any eleva- variety of means, including visual observations
tion. Below that section is a transition zone, and measurements of temperature change, car-
where the conditions decrease until they reach bon dioxide concentration, or light obscuration.
the bottom edge of the layer and are at their The correlations are simple expressions with
minimum value. The predictions from the empir- easily acquired input and minimal computations.
ical correlations relate to the position of the bot- The correlations provide conservative estimates
tom edge of the transition zone as determined in of the smoke layer interface position (i.e.,
an experimental program. In the theoretically predicting the lower edge of the transition zone
based correlations, all of the smoke is considered of the smoke layer which may include only
to be in one layer with uniform properties. Com- ‘wisps’ of smoke) [16]. The correlations are
bination of the transition zone and the upper applicable to simplified cases related to the fire
portion into one uniform zone effectively results and geometry of the space. Fire scenarios must
in the transition zone being compressed so as to be steady state or, if growing, follow a t2 profile.
have the same properties as the upper portion. As The assumed geometrical configuration is a
such, the theoretically based correlations relate to space of uniform cross-sectional area (i.e., rect-
a thinner smoke layer than the empirical angular or right cylindrical solids). In addition to
approach. the noted simplifications, second-order
parameters such as environmental factors (e.g.,
stack effect, wind) and the effect of HVAC
Empirical Correlations systems are neglected.

Empirical correlations have been developed by Steady Fires The position of the smoke layer
Heskestad to determine the smoke layer interface interface for steady fires can be estimated using
position as a function of time for steady and t2 Equation 51.13 [16, 17]. Equation 51.13 is based
fires. These correlations, included in NFPA on experimental data from fires in large-volume
92 [1], are based on experimental data in large spaces with A/H2 of 0.9–14 [18–20].

EDUFIRE.IR Telegram EDUFIRE_IR

 1832 J.A. Milke

Fig. 51.6 Comparisons of 1

smoke layer position—

Normalized smoke layer position (z/H )

experimental data versus 0.9
predictions 0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0 10 20 30 40 50 60 70 80 90
(tQ1/3H –4/3)/(A/H 2)(kW1/3–s/m4/3)

BRI NRCC Heskestad

NFPA 92B EQN 13 FMD

!
tQ_ H
4=3
1=3 is obtained for short time durations where the
z
¼ 1:11
0:28 ln ð51:13Þ resulting z/H is greater than 1.0 (otherwise z/
H A=H 2
H > 1 would literally mean that the smoke
layer interface is above the ceiling). The lower
Where z/H  0.2. limit for z/H of 0.2 relates to the lowest level
Equation 51.13 is presented in non-dimensional where data were taken in any of the referenced
form. The quantity tQ_ H
4=3 represents the
1=3 experiments.
normalized time from ignition. The significance
of the normalized time parameter is to t2fires Equation 51.14 provides a correlation of
indicate that the same relative smoke layer posi- the time-dependent smoke layer interface posi-
tion occurs for a long duration, low heat release tion for fires following a t2-type profile
rate fire in a tall ceiling height atrium, as for a short [16]. Equation 51.14 is also based on experimen-
duration, large fire in an atrium with a short tal data in spaces with shape factors ranging from
ceiling height. Different atrium geometries are 0.9 to 14 [20, 23].
accounted for by the non-dimensional shape z h  
3=5 i
1:45
factor, (A/H2) [18, 19]. ¼ 0:91 tt
2=5
g H
4=5 A=H 2
H
The limits noted for A/H2 reflect the range of ð51:14Þ
shape factors for the facilities in which the
experiments were performed [18, 19]. Examples Equations 51.13 and 51.14 both assume that
of atria within the noted range include atria with the fire is located near the center of the atrium
a cross-sectional area of 10,000 m2 and a height floor, remote from any walls. Smoke production
of 105 m (A/H2 ¼ 0.9) or a height of 27 m (A/ is greatest for the centered configuration and
H2 ¼ 14). Comparisons of the predictions from thereby represents the worst-case condition.
Equation 51.13 to experimental data from fires in
tall spaces are provided in Fig. 51.6 [20–22]. Example 2 For a fast, t2 fire in an atrium with a
Transport lag, or the initial time period to cross-sectional area of 800 m2 and height of
form a smoke layer, is implicitly included in 20 m, determine the position of the smoke layer
Equation 51.13. Evidence of this characteristic interface after 120 s.

EDUFIRE.IR Telegram EDUFIRE_IR

 51 Smoke Control by Mechanical Exhaust or Natural Venting 1833

Solution Applying Equation 51.14 with A/H2 ¼ dV u

ðρhÞu ¼ Q_ c þ mh
_ 1 ð51:17Þ
2.0 and tg ¼ 150 s, z/H is 0.95 or z ¼ 19 m. dt

Example 3 For a fast, t2 fire in an atrium with a Given the previously assumed conditions,
cross-sectional area of 800 m2 and height of ρh is a constant. Substituting the volumetric
20 m, determine the time for the smoke layer flow rate for the mass flow rate and simplifying,
interface to reach 15 m above floor level.
dV u Q_ c
¼ þ V_ ð51:18Þ
dt ρh
Solution Re-expressing Equation 51.14 to solve
for t, The growth rate of the upper layer indicated in
  Equation 51.18 is dependent on two terms:
t ¼ 0:94t2=5
g H
4=5
A=H 2 3=5
ðz=H Þ
0:69 ð51:15Þ (1) the volume supplied by the plume and
(2) the expansion of the volume due to heating.
Applying Equation 51.15 with A/H2 ¼ 2.0
For the case of an atrium with a constant cross-
and tg ¼ 150 s, t is 140 s.
sectional area, A,
Reviewing the results from Examples 2 and
3, the smoke layer barely descends below the dV u dzu
¼A ð51:19Þ
ceiling in the first 120 s. This is indicative of dt dt
the lag time required for the plume to reach the
ceiling and to form a layer. Then, after only As long as the smoke layer interface is well
another 20 s, the smoke layer descends 4 m, above the flaming region (see discussion later in
demonstrating the rapid initial descent rate of this chapter), the plume mass entrainment rate
the smoke layer interface. The rapid descent is can be estimated from [27].
attributable to the significant quantity of smoke dV u m_
¼ ¼ kv Q_ z5=3
1=3
produced during the early stage of a fire in a ð51:20Þ
dt ρ
tall ceiling space when the height available
for entrainment is at its largest value. The Several simplifications can be made for large
predicted trend of rapid filling during the clear heights (i.e., clear heights in excess of
early stage of a fire has been reported by eye- 10 m). The clear height is the distance from the
witness accounts from four fires in atria [4, top of the fuel to the bottom of the smoke layer.
24–26]. The magnitude of the second term is much less
than the first. Generally, z is much greater than zo.
In addition, the volume increase of the upper
Theoretically Based Approach layer supplied by the plume is appreciably
greater than that due to expansion. With
Conservation of mass and energy can be applied these simplifications and by substituting
to provide an estimate for the position of the Equations 51.19 and 51.20 into Equation 51.18,
theoretical smoke layer interface. Equation 51.16 an expression for dzu/dt can be formulated
expresses the conservation of mass, mu, for the
upper smoke layer, assuming no exhaust from the _ 5=3
dzu kv Qz
layer. ¼ ð51:21Þ
dt A
dmu In Equation 51.21, kv is the volumetric
¼ m_ ð51:16Þ
dt entrainment constant, defined as [36].
Approximating the smoke as an ideal gas with kv ¼ 0:076=ρ
properties of heated air, and assuming that the
ambient pressure and specific heat are constant, The convective heat release fraction is the
the expression for conservation of energy for the ratio of the convective heat release rate to the
smoke layer is total heat release rate and is typically assumed

EDUFIRE.IR Telegram EDUFIRE_IR

 1834 J.A. Milke

to be on the order of 0.7–0.8. Throughout is equal to the mass entrainment rate by the
this chapter, a value of 0.7 is selected for the plume. Algebraic equations are available to esti-
convective heat release fraction [1]. Assuming a mate the properties of the smoke layer, including
plume entrainment constant of 0.076 kg 1. Position of smoke layer interface
kW
1/3m
5/3s
1 and the density of ambient air 2. Temperature of smoke layer
as 1.2 kg/m3, the volumetric entrainment con- 3. Light obscuration in smoke layer and
stant is 0.064 m4/3 kW
1/3 s
1. 4. Gas concentration in smoke layer
An expression for the smoke layer position
resulting from a steady fire as a function of time
can be obtained by integrating Equation 51.9:
" #
3=2 Equilibrium Smoke Layer Interface
2kv tQ_
1=3
z Position
¼ 1þ   ð51:22Þ
H 3 A=H2 H 4=3
The exhaust rate necessary to arrest the descent
2
Alternatively, for a t fire of the smoke layer can be estimated based on
knowledge of the mass entrainment rate into the
"  2=3 #
3=2
z 4kv t t=tg plume. The mass entrainment rate depends on the
¼ 1þ  ð51:23Þ configuration of the plume. Plume configurations
H A=H2 H 4=3
reviewed in this chapter are
A comparison of the predictions from 1. Axisymmetric plume
Equations 51.13 and 51.22 is provided in 2. Wall plume
Fig. 51.6. One principal difference relates to the 3. Corner plume
time delay for the smoke layer to form, i.e., 4. Balcony spill plume
transport lag. Transport lag is included implicitly
in Equation 51.13. Equation 51.22 assumes that a Axisymmetric Plume Axisymmetric plumes
smoke layer forms immediately. The transport are formed from fires involving fuel packages
lag can be accounted for separately [13]. remote from any walls (i.e., near the center of
the atrium floor). Being remote from any walls,
Example 4 For a fast, t2 fire in an atrium with a air is entrained around all of the plume perimeter
cross-sectional area of 800 m2 and height of along the entire clear height of the plume. The
20 m, determine the position of the smoke layer functional relationship of the mass entrainment
interface after 120 s. rate to the heat release rate and clear height
is [28].
Solution Applying Equation 51.23 with A/  1=3 
m_ ¼ f Q_ c z5=3 ð51:24Þ
H2 ¼ 2.0 and tg ¼ 150 s, z/H is 0.72 or
z ¼ 14.4 m.
One set of equations for the mass entrainment
rate was originally derived by Heskestad
[27]. One of the equations in the pair developed
Vented Period by Heskestad applies to estimating the entrain-
ment in the flaming portion of the plume and
If a smoke control system has the capability to another deals with the overall plume, including
exhaust smoke, the descent of the smoke layer flaming portion and upper portion where flames
can be arrested if the volumetric rate of smoke are absent.
exhaust from the smoke layer equals the volu- The limiting height is defined as the height of
metric rate of smoke supplied to the layer. the continuous flaming region, (i.e., where flames
Neglecting the effect of expansion, the layer are present 50 % of the time). The limiting height
descent is stopped when the mass exhaust rate may be estimated as [27].

EDUFIRE.IR Telegram EDUFIRE_IR

 51 Smoke Control by Mechanical Exhaust or Natural Venting 1835

z f ¼ 0:166Q_ c
2=5 the type of materials involved in the fire, other
ð51:25Þ
than indirectly in terms of the heat release rate.
For clear heights less than the limiting height, This is due to the mass rate of entrained air being
i.e., where flames extend into the smoke layer, much greater than the mass rate of combustion
the entrainment rate is estimated using products generated, which is true as long as suf-
Equation 51.26 ficient air is available for combustion. As a result
of the fire being approximated as a point source
m_ ¼ 0:032Q_ c z
3=5
ð51:26Þ in the entrainment equations, the shape or form
of the fuel is not of primary importance. Thus,
For clear heights greater than the limiting the parameters associated with a detailed
height, i.e., where the flaming region ends prior description of the fuel package are relegated to
to reaching the smoke layer, the entrainment rate a level of secondary importance.
is estimated using Equation 51.27: In both Equations 51.26 and 51.27, the mass
entrainment rate is dependent on the clear height,
m_ ¼ 0:071Q_ c z5=3 þ 0:0018Q_ c
1=3
ð51:27Þ where the mass entrainment rate increases with
increasing values of the clear height. During the
Equation 51.27 is a simplified version of the
early stages of the fire, the clear height has its
original expression developed by Heskestad (see
maximum value thereby providing the maximum
Chap. 13, with zo from the original expression set
smoke production rate. This is especially true if
equal to zero. The validity of neglecting zo in
the flame height is well below the smoke layer,
Equation 51.27 is based on the observation that
where the smoke production rate is proportional
zo is typically small, compared to z [2]. The loca-
to z5/3.
tion of the virtual origin of an assumed point
In most engineering applications, the smoke
source can be estimated as [27].
production (or exhaust) rate is expressed in terms
zo ¼ 0:083Q_
1:02d o of a volumetric rate rather than a mass rate. In
2=5
ð51:28Þ
order to accommodate this preference, the rela-
For noncircular fuels, an equivalent diameter tionship between the volumetric rate and mass
needs to be defined. The definition of an equiva- rate is expressed as Equation 51.29.
lent diameter is based on a circle that has an area
equal to the floor area covered by the fuel. Con- m_
V_ ¼ ð51:29Þ
sidering a wide range of diameters and heat ρ
release rates associated with a variety of typical
fuel packages, the virtual origin ranges from 0.5 Assuming smoke to have the same properties
to
5 m. Negative values are obtained when the as air, the density of smoke may be evaluated as
second term is greater than the first (i.e., for fuel the density of air at the temperature of the smoke
commodities with modest heat release rates layer [3]. Graphs relating the volumetric smoke
spread over a large area). production rate to the clear height for selected
Originally, Equations 51.26 and 51.27 were total heat release rates ranging from 1000 to
developed to describe plumes from horizontal, 10,000 kW are provided in Fig. 51.7.
circular flammable liquid pool fires. However,
these equations have been shown to be applicable Example 5 A fire has a total heat release rate of
to more complex fuels, as long as the limiting 5000 kW and is located at the center of the atrium
height is greater than the diameter of the fuel, and floor. The smoke layer interface is 35 m above
the fire only involves the surface of the material the floor. Determine the mass and volumetric
(i.e., is not deep-seated) [27]. rates of smoke being supplied by the plume to
The mass rate of smoke production estimated the smoke layer (i.e., at the location of the smoke
by Equations 51.26 and 51.27 is independent of layer interface).

EDUFIRE.IR Telegram EDUFIRE_IR

 1836 J.A. Milke

Fig. 51.7 Smoke 800

production rate for
axisymmetric plumes 700

600

Volumetric rate (m3/s)

500

400

300

200

100

0
0 5 10 15 20 25 30 35 40 45 50
Clear height (m)

1000 kW 2500 kW 5000 kW 10,000 kW

Solution First, the limiting height is evaluated not cancel out the impact of considering twice
using Equation 51.25 to determine the applicable the fire size as the entrainment is proportional to
equation for the mass rate of entrainment, assum- the one-third power of the heat release rate).
ing the convective heat release fraction is 0.7, Similarly, a plume generated by a fire located
zf ¼ 4.3 m. Because z > zf, Equation 51.27 is the near a corner of a room is referred to as a corner
applicable equation for determining the mass rate plume (see Fig. 51.8). Using the concept of a
of smoke production. Neglecting zo, the mass reflection, the smoke production rate from corner
smoke production rate is 410 kg/s. The plumes, where the intersecting walls form a 90
associated volumetric rate (from Equation 51.29, angle, is estimated as one-quarter of that from a
assuming 20  C and 1 atm pressure) is 340 m3/s. fire that is four times as large.

Wall and Corner Plumes Fires located near Example 6 A fire located on the floor of an
walls and corners principally entrain air only atrium has a total heat release rate of 5000 kW.
along the surface of the plume away from the The smoke layer interface is 35 m above the
walls or corner. Consequently, the amount of floor. Compare the mass rates of smoke being
smoke production is reduced for these locations, supplied by the plume to the smoke layer, given
compared to the axisymmetric plume remotely an axisymmetric, wall, or corner plume
located from the walls. Using the concept of configuration.
reflection, the smoke production rate from wall
and corner plumes can be estimated [29, 30]. Solution In Example 5, zf ¼ 4.3 m and the
A plume generated by a fire located against a smoke production rate for the axisymmetric
wall only entrains air from approximately half of plume using Equation 51.27 is 410 kg/s. Apply-
its perimeter, as indicated in Fig. 51.8. According ing the same equation for the wall plume, the
to the concept of reflection, the smoke produc- smoke production rate for a fire size of
tion rate is estimated as half of that from a fire 10,000 kW is estimated as 520 kg/s. Dividing
that is twice as large (in terms of heat release that rate by two provides the smoke production
rate) (note: having half of the entrainment does rate for the wall plume (260 kg/s). Similarly, for

EDUFIRE.IR Telegram EDUFIRE_IR

 51 Smoke Control by Mechanical Exhaust or Natural Venting 1837

horizontally under the obstruction, and then turns

Wall plume
and rises vertically. Scenarios with balcony spill
plumes involve smoke rising above a fire, reaching
a ceiling, balcony, or other significant horizontal
projection, then traveling horizontally toward the
edge of the balcony. Characteristics of the
resulting balcony spill plume depend on
characteristics of the fire, width of the spill
plume, and height of the ceiling above the fire. In
addition, the path of horizontal travel from the
plume centerline to the balcony edge is significant.
Several correlations on air entrainment into
balcony spill plumes have been presented in the
literature over several decades. A comprehensive
review of the proposed correlations is provided
Corner plume by Harrison [31], Lougheed et al. [32] and Lim
[33]. The correlations presented in NFPA
92 reflect the results obtained by Lougheed
et al. from large-scale experiments and numerical
simulations. One of the correlations in NFPA
92 has its roots back to Law’s [34] interpretation
of small-scale experimental data obtained by
Morgan and Marshall [35]. This correlation is
presented as:
 
2 1=3
m_ ¼ 0:36 Q_ W_ ðzb þ 0:25H Þ ð51:30Þ

Lougheed et al. found that their large scale

data was well described by this correlation for
Fig. 51.8 Wall and corner plume diagrams
clear heights (z) in excess of 15 m. For lower
heights, Lougheed et al. suggest the following
the case of the corner plume, the smoke produc-
correlation:
tion rate is 170 kg/s (considering one-quarter of
the smoke production rate from a
m_ ¼ 0:59Q_ W 1=5 ðzb þ 0:17W_
1=3 7=15
20,000 kW fire). H
Comparing the smoke production rates for the þ 10:35W 7=15
15Þ ð51:31Þ
three plumes (axisymmetric, wall, and corner
plumes), the smoke production rate is greatest The correlations presented in Equations 51.30
for the axisymmetric plume (410 kg/s) compared and 51.31, as well as others presented by numer-
to 260 and 170 kg/s for the wall and corner ous previous researchers, apply to balcony spill
plumes, respectively. Thus, conservative hazard plumes of a specific configuration. The configu-
assessments should assume an axisymmetric ration considered is depicted in Fig. 51.9. As
plume is developed from a fire that is located illustrated in the figure, the fire is located in a
away from the walls, near the center of the space. communicating space and the smoke flows under
a soffit out from the room of fire origin, then
Balcony Spill Plume A balcony spill plume is under a short horizontal obstruction, i.e., bal-
generated in cases where smoke reaches an inter- cony. The balcony is oriented perpendicular to
mediate obstruction, such as a balcony, travels the opening from the room. Any variations from

EDUFIRE.IR Telegram EDUFIRE_IR

 1838 J.A. Milke

1600.0

1400.0

Volumetric Rate (m3/s) 1200.0

1000.0

800.0

600.0

400.0

200.0

0.0
0 5 10 15 20 25 30 35 40 45 50
Clear height above balcony (m)

1000 kW, W=5 m 1000 kW, W=10 m 5000 kW, W=5 m

5000 kW, W=10 m 10000 kW, W=5 m 10000 kW, W=10 m

Fig. 51.9 Approximation of a balcony spill plume

Fig. 51.10 Smoke production rate predictions for balcony spill plumes (H ¼ 3 m)

this specific configuration have not been Copyright# 2011, National Fire Protection
investigated and thus the balcony spill plume Association. This reprinted material is not the
correlations presented as Equations 51.30 and complete and official position of the NFPA on
51.31 should not be applied for those situations. the referenced subject, which is represented only
Instead, the application of CFD codes or small- by the standard in its entirety.
scale models should be applied to assess those A comparison of the smoke production rate
situations. for axisymmetric and balcony spill plumes is
Predictions of the smoke production rate provided in Fig. 51.11. The results from both
using Equation 51.30 for the balcony spill Equations 51.30 and 51.31 are depicted in
plume are included in Fig. 51.10. The Fig. 51.11 and are the reason for the points of
calculations represented in the figure consider a inflection at a clear height of 15 m. The heat
3-m height to the underside of the balcony. release rate for both fires is a steady state
Reprinted with permission from NFPA 5000 kW, and H is 3 m for the balcony spill
92-2012, Standard for Smoke Control Systems, plume. For short heights, the smoke production

EDUFIRE.IR Telegram EDUFIRE_IR

 51 Smoke Control by Mechanical Exhaust or Natural Venting 1839

Fig. 51.11 Comparison of 800.0

smoke production rate for
axisymmetric and balcony 700.0

Volumetric rate (m3/s)

spill plumes 600.0 Axisymmetric
500.0
BALC, H= 3 m, W= 3 m
400.0
BALC, H= 3 m, W= 5 m
300.0

200.0 BALC, H= 5 m, W= 3 m

100.0 BALC, H= 5 m, W= 5 m
0.0
0 5 10 15 20 25 30 35 40 45 50
Clear height (m)

rate for the balcony spill plume is appreciably vented period. During the filling period, determi-
greater than that for the axisymmetric plume. nation of the smoke layer properties is important
However, with increasing height, the smoke pro- to assess the level of hazard prior to actuation of
duction rates from the two plumes become com- a mechanical smoke control system. During the
parable. Eventually, the two curves intersect, vented period, smoke layer properties are of
suggesting that, at some height, the balcony interest to assess the level of hazard associated
spill plume behaves in the same manner (i.e., with those cases where occupants are exposed to
produces the same amount of smoke) as an axi- smoke (i.e., the highest walking level is
symmetric plume. The point of intersection can submerged in the smoke layer). The smoke
be determined by setting the mass flow in Equa- layer properties of interest include temperature,
tion 51.27 equal to that in Equation 51.30. light obscuration, and gas species concentration.
The width of the plume, W, can be estimated
by considering the presence of any physical ver- Temperature Rise in Smoke Layer The tem-
tical barriers attached to the balcony. The perature of the smoke layer can be determined
barriers act to restrict dispersion of the horizontal based on an energy balance for the volume of the
flow of smoke under the balcony. However, in smoke layer. Energy is supplied to the layer by
the absence of any barriers, an equivalent width the fire. Energy may be lost from the layer to the
can be defined, based on results from visual enclosure (walls, ceiling) of the space. During
observations of the width of the balcony spill the filling period, the resulting expression is [1].
plume at the balcony edge from the set of

ð1
χl ÞQ
small-scale experiments by Morgan and Mar- T ¼ T o exp ð51:33Þ
shall [35]. The definition of an equivalent con- Qo
fined plume width is the width that entrains the Estimates for χl (heat loss fraction from the
same amount of air as an unconfined balcony smoke to enclosure) vary appreciably. Some of
spill plume. The equivalent width is evaluated the design guides suggest assuming that the
using the following expression smoke layer is adiabatic (i.e., setting χl ¼ 0),
L¼wþb ð51:32Þ in order to be conservative [1]. Walton
suggested values for χl between 0.6 and 0.9
for relatively small spaces of near cubic shape
Properties of Smoke Layer [36]. In many of the large spaces with tall
ceiling heights, the temperature rise anticipated
Properties of the smoke layer are of interest both for the smoke layer is relatively modest such
during the filling period of the fire and during the that convection and radiation heat transfer to an

EDUFIRE.IR Telegram EDUFIRE_IR

 1840 J.A. Milke

Fig. 51.12 Temperature 400

rise of smoke layer for
axisymmetric plumes 350

300

Temperature rise (K)

250

200

150

100

50

0
0 5 10 15 20 25 30 35 40 45 50
Clear height (m)

1000 kW 2500 kW 5000 kW 10,000 kW

enclosure will also be modest. Consequently, in specify the heat loss fraction [19, 39]. The adia-
such applications, the adiabatic assumption will batic smoke layer temperature for a range of fire
provide reasonable predictions of the tempera- sizes is presented in Fig. 51.12.
ture rise. However, in low ceiling spaces (under The degree of overestimation can be assessed
approximately 10 m) the temperature may be by comparing the estimated smoke layer temper-
significantly overestimated by applying the adi- ature with the plume centerline temperature. For
abatic assumption. thermodynamic reasons, the smoke layer temper-
Similarly, the equilibrium smoke layer tem- ature cannot exceed the plume centerline temper-
perature during venting can be approximated by ature. The plume centerline temperature, Tc, can
applying an energy balance to the smoke layer. In be evaluated using Equation 51.35 [40]
this case, energy is also lost from the layer due to
T c ¼ 0:08T o Q_ c z
5=3 þ T o
2=3
smoke being exhausted from the atrium. Equa- ð51:35Þ
tion 51.34 can be used to determine the tempera-
ture rise of the smoke layer under adiabatic The volumetric venting rate for other heat
conditions. release rates or temperature rises may be deter-
mined using Equation 51.36 considering that the
ð1
χl ÞQ_ c specific heat is virtually constant for the expected
ΔT ¼ ð51:34Þ
c p m_ temperature range of interest

If the adiabatic assumption is applied, the Q_ c1 V 1 ΔT ad1 T 2

¼ ð51:36Þ
smoke layer temperature will be overestimated, Q_ c2 V 2 T ad2 T 1
providing a conservative estimate of the hazard.
In reality, some heat is lost from the upper smoke As can be observed from Equation 51.36, dou-
layer to the surrounding walls and ceiling. How- bling the volumetric venting rate for the same
ever, no elementary method is available to esti- size fire reduces the temperature rise by approxi-
mate the overall proportion of heat that is lost to mately 50 % (the temperature rise is not precisely
the surroundings [37, 38]. Some zone and field halved, since the absolute temperature of the
computer fire models account for heat losses to smoke layer in both instances is not exactly
the boundary, thereby avoiding the need to the same).

EDUFIRE.IR Telegram EDUFIRE_IR

 51 Smoke Control by Mechanical Exhaust or Natural Venting 1841

Fig. 51.13 Relationship 1.4

between visibility through
smoke and walking speed Nonirritating
1.2
smoke

Walking speed (m/s)

1.0

0.8
Irritating
0.6 smoke

0.4

0.2 Blindfolded person

0
0.00 0.10 0.20 0.30 0.40
Optical density (m)

Light Obscuration The visibility distance operation of sprinklers. The mass optical density
through smoke can be related to the optical den- can vary by orders of magnitude for different
sity per unit pathlength via empirical correlations ventilation conditions.
[41, 42]. The experimental basis for the Although a reduction in visibility is not
correlations consists of tests with humans directly life-threatening, it does reduce the walk-
viewing objects through smoke. However, ing speed of individuals, thereby increasing the
the participants were not directly exposed to the exposure time to toxic gases and elevated
irritating effects of smoke. Consequently, the temperatures. In addition, the reduction in visi-
reported correlations are likely to overestimate bility may lead to an increased susceptibility to
the visibility distance. occupants tripping or falling. The relationship
In addition to the light obscuration quality of between visibility and movement speed is
the smoke, the visibility of an object is dependent indicated in Fig. 51.13.
on the light source for the object being viewed as
well as ambient lighting conditions [42, 43]. Carbon Monoxide Concentration The con-
The optical density of the smoke layer can be centration of gas species contained in the smoke
determined considering that all of the layer can be determined considering that all of
particulates generated by the fire are transported the mass that is supplied to the layer via the
to the layer via the plume and accumulate in the plume accumulates in the layer. No absorption
layer. Such an approach neglects any deposition by the enclosure is assumed. The resulting
of soot on enclosure surfaces, thereby expressions for the smoke filling and vented
overestimating the optical densities. The periods are [16].
expressions for the smoke filling and vented
f iQ
periods are provided as Equations 51.37 and Smoke filling : ϒi ¼
51.38 [16]. ρo χa Hc AðH
zÞ
ð51:39Þ
Dm Q
Smoke filling : D¼ ð51:37Þ
χa Hc AðH
zÞ
f iQ
Vented : ϒi ¼ ð51:40Þ
Dm Q_ _
mχa Hc
Vented : D¼ ð51:38Þ
_
χc ΔH c m=ρ
In order to express the gas species concentra-
The mass optical density is dependent on the tion in units of ppm, Equation 51.41 needs to be
fuel, burning mode, ventilation conditions, and applied

EDUFIRE.IR Telegram EDUFIRE_IR

 1842 J.A. Milke

MW air Dm ¼ 260 m2 =kg

ppmi ¼ γ  106 ð51:41Þ
MW i i H c ¼ 12, 400 kJ=kg
Input for evaluating the gas species concen- Considering smoke layer density, ρ, at the tem-
tration includes the yield fraction and heat of perature of the smoke layer to be 1.17 kg/m3, the
combustion, both of which are fuel dependent optical density is 0.32 m
1 and the associated
parameters. The yield fraction is dependent on visibility is 8.5 m.
the burning mode and oxygen concentration. CO Concentration CO concentration for the
Most of the information tabulated on the yield vented period is estimated using Equations 51.40
fraction, such as that by Khan (see Chap. 36), and 51.41, with the fuel-related properties again
assumes well-ventilated, flaming combustion. evaluated from, Appendix C.
Most of the fires of interest in large spaces will
involve flaming combustion and are likely to be f CO for polyurethane ise 0:030kgco =kgfuel
well ventilated. However, fires in small,
The resulting CO concentration in the smoke
connected spaces may become underventilated.
layer is 31 ppm.
Caution needs to be exercised in properly
identifying ventilation conditions when
predicting these parameters because the yield
fraction can vary by orders of magnitude for Comparison of Mechanical Exhaust
different ventilation conditions. Also, the yield and Natural Venting Designs
fractions noted by Tewarson are relevant
only to cases where sprinklers are not Design Aspects of Mechanical
operating [44]. Venting Systems

Example 7 Estimate the steady-state smoke Most smoke control systems for covered malls
layer properties (temperature, visibility to an and atria in the United States use mechanical
internally illuminated exit sign, and CO concen- venting systems. Mechanical venting systems
tration) during the vented period, given the fol- need to be designed to exhaust the amount of
lowing situation: smoke needed to satisfy design objectives. The
1. The smoke layer interface is maintained 35 m volumetric flow of smoke needs to be adjusted
above floor level. for temperature, using the methods discussed
2. The rate of heat release of the flaming fire is a previously in this chapter.
steady state 5000 kW. Mechanical exhaust systems are relatively
3. The fuel is comprised principally of immune to environmental effects because
polyurethane foam. the energy associated with the fan is able to pro-
vide a sufficient force for smoke movement and,
SOLUTION Smoke Layer Temperature thus, is not relying as much on the buoyancy of
Equation 51.34 can be applied to determine the the smoke or stack effect. Protection from wind
adiabatic smoke layer temperature rise. In Exam- effects can be accommodated by hardware.
ple 5, a mass rate of smoke production of 410 kg/ Response time is a principal limitation for
s was determined. Thus, assuming an adiabatic mechanical exhaust systems. The response
smoke layer, a convective heat release rate frac- time is the sum of the time for detection and the
tion of 0.7 and specific heat of air of 1.0 kJ/kgK, time for the system to reach capacity (which may
the temperature rise is 8.5  C. be up to a minute). This combined time may be
longer than the time for the smoke layer to reach
Visibility Visibility during the vented period is the critical height established by design goals.
estimated using Equation 51.38. Fuel-related Also, because the capacity of a mechanical
parameters are obtained in Chaps. 36, 24. venting system is sized considering a particular

EDUFIRE.IR Telegram EDUFIRE_IR

 51 Smoke Control by Mechanical Exhaust or Natural Venting 1843

size of design fire (see Equations 51.26, 51.27, as part of the continuing discussion in this
51.30 and 51.31), if an actual fire has a greater section.
heat release rate than considered in the design, The engineering principles that apply to vent
the capacity of the mechanical exhaust will not operation addressed in this section consider
be sufficient. the scenarios depicted in Fig. 51.14. Because
In addition, mechanical exhaust systems are smoke filling along the underside of the ceiling
susceptible to plugholing, a situation in which a in a curtained area is similar to that in a compart-
hole is created in the smoke layer below the ment, additional information on compartment
exhaust inlet by a high-capacity exhaust system. fire scenarios is presented in Chap. 33. If the
This results in a reduction in efficiency of the draft curtains are deep enough, they can be
exhaust system because air from beneath thought of as simulating the walls of a single
the smoke layer is being extracted, thereby the compartment.
desired quantity of smoke is not being extracted, The description of engineering principles of
causing the smoke layer to deepen. Plugholing is natural vents will be provided from the perspec-
addressed later in this chapter. tive of a two-layer zone model. The overall
The limitations of mechanical venting building compartment is assumed to have near-
systems can be overcome in some cases by floor inlet vents that are large enough to maintain
providing detection devices that minimize the the area below the smoke layer at outside-
time required for detection and by using several ambient conditions. The upper smoke-layer
small capacity exhaust fans to avoid plugholing. thickness will change with time, but at any
However, despite these measures, it is still possi- instant it is assumed to be uniform in space,
ble that design goals will not be able to be with absolute temperature, T, and density, ρ.
achieved by mechanical venting designs. Thus, Mass and energy are transferred continuously
the feasibility of such goals may need to be to and from the upper and lower layers. Conser-
evaluated. Alternative smoke management vation of energy and mass along with the Ideal
approaches may be sought, for example, Gas Law is applied to the layers, which leads to
providing physical barriers at upper levels to equations that require estimates of components
reduce the required clear height or considering of heat transfer, enthalpy flow, and mass flow to
opposed airflow at openings above the design the layers. Qualitative and some key quantitative
smoke layer interface position. features of these phenomena are described and
presented below. The reader is referred to
Chap. 15, for a general discussion on the topic
Design Aspects of Natural of flow through vents. Considering a vent in a
Venting Systems wall or ceiling, flow is driven through such a vent
mainly by cross-vent hydrostatic pressure
Natural venting removes smoke by taking advan- differences from the high- to the low-pressure
tage of the buoyancy of the smoke. In the United side of the vent. The traditional means of calcu-
States, natural venting systems are primarily found lating vent-flow rates is by using an orifice-type
only in facilities such as industrial or warehouse flow calculation.
structures. Outside of the United States, natural Assuming relatively quiescent conditions in
venting is often utilized in many applications. the areas on both sides of the vent, the pressure
The key advantages of natural venting in each space can be described as the hydrostatic
systems are the self-correcting aspect of the pressure. The mass flow through a vent is derived
vents in case the design fire is inappropriately from Bernoulli’s equation, where the buoyancy
defined and the simplicity of the operation of pressure is related to the dynamic pressure at the
natural vents. These advantages will be described vent:

EDUFIRE.IR Telegram EDUFIRE_IR

 1844 J.A. Milke

Fig. 51.14 Fire in a Ceiling Ceiling

building space with draft vents jet
curtains and ceiling vents
Draft
curtain

Upper layer

Layer Z LAY Plume Z CEIL Z CURT

interface Z FIRE

Distance
below Velocity Vent or
ceiling sprinkler
link


1 2
ρ μ ¼ Δρgd ð51:42Þ  2 1=2 T o ΔT 1=2 Av d 1=2
2 o m_ ¼ 2ρo g r ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
T2 C 2 A2 T o
1 þ Cd2, vA2v T
where d, i i o

ρ ¼ Density of smoke (kg/m3) ð51:45Þ

ρo ¼ Density of ambient air (kg/m3)
Δρ ¼ ρo
ρ (kg/m3) For vents installed in sloping roofs, the design
Relating the mass flow through the vent to the position of the smoke layer should be at least
velocity of the gases below the bottom of the vent. To ensure that
only smoke is exhausted from that vent and not
m_ ¼ ρAv u ð51:43Þ any air from below the layer, the smoke layer
position should be at least 10 % of the vertical
where
distance from the top of the vent (Fig. 51.15).
ṁ ¼ Mass flow rate through vent (kg/s)
Then the distances d and z (recall that clear
Av ¼ Flow area of vent (m2)
height is imbedded in the consideration of ṁ)
Replacing the densities with temperatures
are measured from the center of the vent.
using the ideal gas law:

1=2
 1=2 T o ΔT Makeup Air Supply The effect of the inlet area
m_ ¼ 2ρ2o g Av d1=2 ð51:44Þ on the flow rate through the vent can be assessed
T
by recognizing that the pressure drop across the
As indicated in Equations 51.42–51.44, the inlets associated with the inflow of replacement
capacity of natural vents is related to the pressure air must be subtracted from the buoyancy pres-
difference caused by the buoyancy of the smoke sure causing the gases to flow through the vents.
layer. As such, the flow rate of smoke through the The effect of inlet pressure may be included in
vent increases with increasing smoke-layer tem- Equation 51.45 by replacing Av by an effective
 
perature and depth. vent area A∗ v where

EDUFIRE.IR Telegram EDUFIRE_IR

 51 Smoke Control by Mechanical Exhaust or Natural Venting 1845

Fig. 51.15 Design

position of gas layer
versus vent
Vent center
d Zt
Zb
Bottom of layer
of hot gases

(Zb > Zt /10)

h is effective ceiling height
d is effective depth of layer of hot gases


1 1 1 To hotter smoke will be more buoyant, thereby
¼ þ ð51:46Þ
A∗2
v A2v A2i T increasing the buoyancy force at the vent leading
to an increase in the mass flow rate through
As such, the ratio of the actual vent area to the the vent to reduce the amount of smoke
effective vent area, K, is given as accumulating under the ceiling.
"
2 #1=2
The ability of a vent to perform similarly for
Av Av T two different fire sizes is a significant benefit of
K ¼ * ¼ 1þ ð51:47Þ
Av Ai T o natural vents. Unlike mechanical exhaust, for
natural vents if an error is made such that an
The effect of vent ratio (ratio of outlet to inlet actual fire is greater than the defined “design
areas) on the effectiveness of natural venting is fire,” the natural vents should still able to provide
presented in Fig. 51.16 with a design fire of near-satisfactory performance.
2.5 MW and a ceiling height of 15 m. As
indicated in the figure, with a vent ratio of 0 (hav- Limitations The limitations of natural venting
ing infinite inlet area), the clear height is slightly systems are related to the forces affecting smoke
greater than when the outlet to inlet areas are movement: principally a lack of buoyancy and
equal. Thus, as with mechanical systems, the wind effects. The smoke must be buoyant rela-
inlet area is an important consideration. tive to the ambient environment in order for
One of the principal advantages of natural natural venting systems to be effective. Smoke
venting systems is the relative insensitivity of may lose its buoyancy either due to cooling from
the equilibrium smoke-layer position with the sprinkler operation or dilution from entrained,
fire size, as indicated in Fig. 51.15. The graphs cool air. Because the mass flow is strongly
in Fig. 51.17 indicate that for two different ceil- dependent on the difference in the smoke-layer
ing heights (15 and 30 m), the equilibrium temperature and outdoor temperature, if the
smoke-layer temperature is virtually identical smoke-layer temperature rise is only slightly dif-
for the two significantly different fire sizes. This ferent than the ambient temperature, then the
similarity is due to the bigger fire size producing flow from a vent will also be modest. As such,
a smoke layer with a greater temperature. The in tall spaces with relatively small fire sizes, the

EDUFIRE.IR Telegram EDUFIRE_IR

 1846 J.A. Milke

Fig. 51.16 Effect of vent 90

ratio on natural venting Vent ratio = 0
80 Vent ratio = 0.5
Vent ratio = 1
70

60

Vent area (m2)

50

40

30

20

10

0
0 3 6 9 12 15
Clear height (m)

Fig. 51.17 Effect of vent 70

15 m, 2.5 MW
ratio on natural venting for
two ceiling heights 15 m, 5 MW
60
(15, 30 m) and fire sizes 30 m, 2.5 MW
(2.5 MW, 5 MW) 30 m, 5 MW
50
Vent area (m2)

40

30

20

10

0
0 6 12 18 24 30
Clear height (m)

modest capacity of natural vents may constrain proximity and geometry of neighboring
the ability to achieve design objectives, buildings. Wind effects on buildings are
necessitating that mechanical ventilation addressed in Klote et al. [2].
be used. If the building vents are open and if vent areas
To consider the effect of outside wind are relatively small compared to the building
conditions, pressures on the outside of the build- surface area, then pressures near the vent
ing in the vicinity of the vent need to be assessed. openings will be substantially unchanged from
The pressure on the building depends on the wind the above-mentioned, closed-vent pressure dis-
speed at the elevation of the vent, wind direction tribution, except near any local through-vent
relative to the outside building geometry, and flows that may develop. Also, although the

EDUFIRE.IR Telegram EDUFIRE_IR

 51 Smoke Control by Mechanical Exhaust or Natural Venting 1847

exterior pressures generally vary from vent to vents and within the building. Inlet air will be
vent, they will be relatively uniform for any provided at the high-pressure upwind vent and
particular vent. Under these conditions, a deter- outlet air at the low-pressure downwind vent,
mination of flow rates into and/or out of vents with levels of through-vent flows and of interior
and through the interior of the building is based hydrostatic pressures determined by an appropri-
on an interior building flow analysis, with ate analysis that accounts for conservation of
pressure-specified boundary conditions at the momentum (i.e., Bernoulli’s equation) and mass
open vents. at the exterior vents and at room-to-room vents
within the interior of the building. The changes in
A Single, Open Inlet Vent or Multiple hydrostatic pressures within the rooms of the
Openings at the Same Pressure If there is building, over and above the hydrostatic
only one open inlet vent on the upwind side of pressures that would be present in a quiescent
the building that experiences a relatively high environment, would be somewhere between the
pressure differential above the local hydrostatic wind-induced pressures at the locations of the
pressure or if there are several open vents, all at high-pressure vent and the low-pressure vent.
locations on the outside surface of the building
where pressures are substantially identical, then, Wind-Modified Pressures at Roof Surfaces
the wind will have no effect on the inflow or and Wind-Modified Action of Ceiling
outflow through the vents. Thus, if the air inside Vents Roof surfaces of flat-roofed buildings
the building is uniformly at the outside air tem- tend to have negative, wind-induced pressure
perature and if there is no mechanical ventilation, coefficients, unless the buildings are very long
then the effect of the wind will be simply to bring in a direction parallel with the wind direction.
the interior hydrostatic pressure at the location of Sloping roofs may have pressure coefficients that
the vent(s) to the aerodynamic-flow-specified can be positive or negative, depending on wind
value; the interior of the building will be direction. Therefore, if the interior, wind-induced
“pressurized” as a result of the open vent(s), but hydrostatic pressures are greater than those
there will be no wind-induced interior flows. If associated with a quiescent environment (e.g.,
there is a fire in the room with the open vent (e.g., the result of open vents in the upwind side of
the vent is a broken window), then, in the usual the building), then the flow of smoke through
way, there will be fresh air inflow into the room ceiling vents can be enhanced significantly by
toward the bottom of the vent and buoyant smoke virtue of increased, favorable, cross-vent
outflow toward the top of the vent, all this taking pressures. However, for reduced interior
place at an aerodynamic-flow-specified, elevated pressures (e.g., as a result of open vents on the
hydrostatic pressure within the room. downwind side of the building), the effect of
If the open vent is in a side of the building wind conditions can substantially disrupt the
with a negative wind coefficient (e.g., facing desired smoke-removing action of ceiling vents,
downwind or on roofs near the upwind side), even reducing the direction of the cross-vent
the pressure at the vent will be relatively low, pressures and, as a result, the direction of the
and the local hydrostatic pressure will be reduced flow through the vents (i.e., making the flow
by an amount only on the order of ρou2/2. Again, travel from the outside to the inside).
no wind-induced flow at the vent is expected.

Two Inlet Vents, One on the Upwind Side and Thermal Activation of Vents
One on the Downwind Side of the Building If
there are two inlet vents in the walls of the Convective heating and thermal response of
building, one upwind and one downwind (ignor- near-ceiling-deployed fusible links or other
ing heating and mechanical ventilation), then near-ceiling thermal sensor devices (including
there will be wind-induced flow through the thermoplastic vent covers designed to soften

EDUFIRE.IR Telegram EDUFIRE_IR

 1848 J.A. Milke

and “drop out” at specified actuation design performance). As an example of the latter, con-
temperatures) are determined from the local sider the case in which water spray from a sprin-
time-dependent distributions of ceiling jet veloc- kler system cools the smoke, thereby reducing
ity and temperature. These distributions will the buoyancy of the smoke. The reduced buoy-
depend on vertical distance below the ceiling ancy reduces the mass flow rate through a vent,
and radial distance from the fire-plume axis. thereby resulting in a deeper smoke layer. How-
Once the operating temperature of the thermal ever, the sprinkler discharge can also dramati-
element is reached, the device or devices cally reduce the fire size resulting in a decreased
operated by the element will be actuated. production of smoke.
Characteristics of ceiling jets are described in Numerous research projects have been
Chap. 14. conducted to address the interaction between
The mathematical fire model LAVENT (fus- vents and sprinklers (see Annex A of NFPA
ible-Link-Actuated VENTs) [22–24] was devel- 204 and Beyler and Cooper [46] for a thorough
oped and is available to simulate most of the review of the previous research). The previous
phenomena described above. The LAVENT projects have sought to demonstrate the level of
model can be used to simulate on a time- impact that one system has on the other’s per-
dependent basis and to study parametrically a formance, either by indicating that it is always
wide range of scenarios with natural vents. Full significant, always insignificant, or significant
documentation for LAVENT, including its theo- only if a particular set of conditions is provided.
retical basis [22], a user guide for the computer Projects have also attempted to identify design
code [23], and sample problems using the code changes necessary if both systems are present
are included in Annexes B, C, and D of NFPA (i.e., perhaps larger vent sizes could be installed
204 [45]. In its current form, LAVENT does not in sprinklered buildings to counteract the
account for wind effects, the reduced effective- reduced mass flow of cooled smoke). To date,
ness of vents as a result of limited-area inlet none of the previous projects have been able to
vents, or the presence of mechanical systems provide the conclusive results that provide
[22]. Input data on the thermal response definitive information illustrating the degree of
characteristics of the link will be needed for influence that one system has on the other for
such an analysis. The use of LAVENT has not all situations.
been validated for estimating the response of
“drop out” vents. Past Studies of Combined Vent/Sprinkler
Systems A review of 34 papers evaluated the
validity of generic claims and counterclaims on
Sprinklers and Vents the benefits of combined vent/sprinkler systems
[46]. A listing of these claims and counterclaims
Vents and sprinklers provide different fire safety and a summary of conclusions on their validity
benefits. The level of fire safety in a facility follow.
would be enhanced if both sets of benefits could Claims and Counterclaims [46] In the literature,
be achieved systematically. However, simply claims that have been made in favor of vent/
providing the two technologies following design sprinkler systems can be reduced to the following
rules independent of each other will not neces- three:
sarily lead to a combination of their respective 1. Smoke and heat vents limit the distribution of
benefits. Potential problems may occur as a result products of combustion in the facility whether
of the interaction of the two technologies (i.e., deployed sprinklers are operative or
operation of the smoke and heat vents can mod- inoperative.
ify sprinkler performance and the operation of 2. Smoke and heat vents decrease the number of
the sprinklers can modify smoke and heat vent activated sprinklers.

EDUFIRE.IR Telegram EDUFIRE_IR

 51 Smoke Control by Mechanical Exhaust or Natural Venting 1849

3. Smoke and heat vents assist the fire depart- should be ganged so that the benefit of roof
ment in identifying the location of the fire vents is fully realized.
within the facility and in reducing the need • When deployed with vents and draft curtains,
for manual roof venting. a sprinkler design needs to take full account of
draft curtains as obstructions to ceiling jet
In the literature, claims that have been made flows and sprinkler discharge.
against vent/sprinkler systems can be reduced to • Draft curtains should be placed in aisles rather
the following four: than over storage.
1. Smoke and heat vents cause enhanced burning
rates. Considerations for the Design of Combined
2. Smoke and heat vents delay sprinkler Vent/Sprinkler Systems Taking Beyler and
activation. Cooper’s conclusions into account and drawing
3. Smoke and heat vents increase the number of on current knowledge of basic physical phenom-
activated sprinklers. ena involved in vent/sprinkler interactions, the
4. Smoke and heat vent flow rates are insuffi- design of combined sprinkler/vent systems should
cient to realize any benefit. seek to satisfy the following general criteria:
1. A successful vent design, whether deployed
Validity of Claims for and Against Combined with or without sprinklers, is one that leads to
Vent/Sprinkler Systems After evaluating reports the benefits of improved visibility and safety
of studies of combined vent/sprinkler systems, during a fire by limiting the descent of the
Beyler and Cooper [46] came to the following upper smoke layer to a specified height (i.e.,
conclusions: eye level of occupants and fire fighters).
• Venting does not have a negative effect on 2. When draft-curtain compartmentation is
sprinkler performance. included in the vent design, a significant addi-
• If a fire is directly beneath a vent, activation of tional possible benefit results from the smoke
the first sprinklers may be delayed slightly, being contained within the curtained compart-
but there is no evidence that this delay will ment of fire origin by action of the venting.
have a significant impact on overall sprinkler
performance. Interaction of Sprinkler Spray and Smoke
• Venting does limit the spread of smoke by Layer The action of sprinkler sprays on a
removing smoke from the building near the smoke layer includes a combination of evapora-
source of the fire (within the curtained com- tive cooling and dilution of the smoke. Dilution
partment of fire origin), improving visibility occurs due to entrainment of the relatively cool
for building occupants while evacuating and and uncontaminated lower-layer gases and the
for fire fighters during fire control operations. upper layer by the spray [47–59]. Provided the
• By limiting the spread of smoke and heat, sprinkler spray–reduced smoke temperature and
venting reduces smoke and heat damage to associated loss of buoyancy are not too great, the
the building. effect of evaporative cooling of the smoke, even
• In the event that sprinklers do not operate, if accompanied by moderate sprinkler
venting remains a valuable aid in controlling spray–driven mixing, could be offset by addi-
the fire manually. tional vent capacity. However, even without sig-
• In many fires, current vent design practices, nificant evaporative cooling, sprinkler
for example, those of NFPA 204 [45], are spray–driven mixing action can be so significant
likely to limit the number of vents operated that it leads to a precipitous increase in the vol-
to one and, in successful sprinkler operations, ume of smoke and thus a deeper smoke-layer. If
vents may not operate at all. and when the latter vigorous mixing occurs, then
• Design practices should use methods that even impractically large increases in vent capac-
ensure early operation of vents; vent operation ity are unlikely to lead to any significant

EDUFIRE.IR Telegram EDUFIRE_IR

 1850 J.A. Milke

improvement. The latter phenomenon is com- the relatively cool and uncontaminated lower
monly referred to as smoke logging. As such, a layer. The analysis led to the identification of
vent design that is developed to meet the above six possible modes of sprinkler/layer interaction
general criteria must be based on an analysis that [58, 59]. The mode that prevailed at any time
accounts for and avoids the phenomena of smoke during the development of a particular fire was
logging. found to depend mainly on the thickness and
There is experimental evidence that smoke temperature of the upper smoke layer and on
logging can be controlled by venting [60], and a the momentum, spread angle, and characteristic
preliminary analysis to explain the phenomenon droplet size of the sprinkler spray. In any partic-
has been provided [47, 48]. Thus, it has been ular fire scenario, the action of open vents and/or
reported that “preliminary tests in [a] . . . large- draft-curtain compartmentation could provide
scale mall . . . showed that, under some some control of the thickness and temperature
conditions, [a] . . . smoke layer could be brought of the layer and, therefore, of sprinkler/layer
down by a manually operated sprinkler spray, interactions that prevail.
[and that] smoke logging then occurred rapidly, Of the six above-referenced modes of sprin-
with a high smoke density at low level. However, kler/vent interaction, four were found to be par-
under some conditions, the smoke layer was not ticularly favorable in the sense that they would
disturbed by a sprinkler spray” [48]. maximize the success of a combined sprinkler/
vent design. Thus, with proper vent design the
Computer Simulations of Sprinkler favorable modes could lead to the desired control
Spray–Driven Cooling, Mixing, and Smoke of the smoke-layer depth while minimizing
Logging A computer model could be applied smoke mixing to the lower layer to the point
to address the issue of sprinkler spray–driven that any smoke there is only in a highly dilute
cooling, mixing, and smoke logging. However, state. Thus, for a given set of sprinkler spray
the past experimental studies have not led to an characteristics, if the smoke layer is kept rela-
understanding of the complex phenomena of tively thin and/or not too buoyant (i.e., its tem-
sprinkler spray cooling and sprinkler perature is not too high), then the rates of both
spray–driven smoke transport and mixing that mass and enthalpy flow entrained into the upper-
causes the temperature reduction that could be layer part of the sprinkler’s “spray cone of influ-
used as a basis for such a model. What is known ence” would be relatively insignificant compared
through anecdotal accounts of visual to the corresponding rates associated with the
observations is that spray-driven mixing and fire-plume flow to the upper layer. In the early
transport of an initially stable and growing part of a typical fire scenario and immediately
upper smoke layer can and often does lead to subsequent to one or more rapid-response sprin-
onset of smoke logging, whereby the mixing kler discharges, the condition of a relatively thin
actions of sprinkler sprays are so vigorous as to and not-too-high-temperature upper smoke layer
effectively and continuously mix any newly should be prevalent. As a result, the combined
generated smoke from the fire plume with the action of cooling and momentum exchange in the
smoke already present in the smoke layer to fill spray cone would be strong enough to transport
the entire space. the entrained smoke through the layer interface
Analytic fire-modeling has been employed to and well into the depth of the lower layer to be
assess a generic interaction of a downward- mixed eventually, with negligible consequences,
directed sprinkler spray and a two-layer fire envi- into the rest of the lower-layer gases.
ronment can be used to resolve the above issues In contrast to the above, there were two par-
[59]. The model simulated the action of the sprin- ticularly unfavorable modes of sprinkler/vent
kler spray, including the effects of evaporative interaction that would minimize the likely suc-
cooling and the spray-driven mixing of the cess of a combined sprinkler/vent design. These
elevated-temperature, upper smoke layer and configurations could lead to relatively vigorous

EDUFIRE.IR Telegram EDUFIRE_IR

 51 Smoke Control by Mechanical Exhaust or Natural Venting 1851

mixing between the smoke layer and the lower ganged to open together immediately following
layer, leading to a rapid growth of the upper first sprinkler discharge.
smoke layer and possibly to smoke logging.

Resolving the Problem of Sprinkler Skipping A Consensus Approach to the Design

and Vent Skipping In terms of achieving vent/ of Combined Sprinkler/Vent Systems
sprinkler design objectives, it is important to
identify a possible means of resolving problems
associated with the phenomena known as sprin- Using Mathematical Fire Models to Achieve
kler skipping and vent skipping. Design Objectives The above discussion
If ceiling jet–convected water droplets strike a indicates that effective sprinkler/vent systems
sprinkler link or bulb, then, because of effects of are feasible and that mathematical fire models
evaporative cooling, there will be a significant with a proven capability for simulating sprin-
reduction of its rate of heating, which can lead to kler/smoke interactions can be used as the basis
a significant delay in sprinkler discharge. It is the for a consensus approach to identify and estab-
resulting, unpredictable, and deleterious delay in lish effective sprinkler/vent system designs.
sprinkler discharge that is referred to as sprinkler The capabilities of mathematical models to sim-
skipping. ulate sprinkler/smoke interactions have been
Although research has been conducted to reviewed [62, 63]. The models considered were
characterize the spray from sprinklers, a general those that are complete (i.e., they can simulate
description of all aspects of the spray is beyond both isolated sprinkler/smoke interactions and
the current state of knowledge. Such a descrip- full fire scenarios, where the latter would be
tion would be needed to provide a reliable used to establish the success of sprinkler/vent
model that can be used to predict the designs). Both zone-type and field model–type
phenomenon. simulation approaches were found to be applica-
ble for addressing the problem. In the usual way,
Accounting for Sprinkler Skipping and Vent the two approaches are complementary in the
Skipping in Design In the design of sprinklers sense that the zone model approach is more
without vents, the effects of sprinkler skipping on applicable and appropriate for parametric studies
the ability of a sprinkler system to control a fire and as a practical design tool and the field model
are taken into account by the empirically based approach is more applicable for simulating and
design standard, NFPA 204. In contrast, when studying the details of specific scenarios, for
vents and sprinklers are used together, the ran- example, the discharge sequence of sprinklers
dom and unpredictable effects of vent skipping and the effectiveness of a vent design where
are not accounted for in the design of automatic draft curtains are almost directly above the fire.
vent systems as outlined in NFPA 204. A sprinkler/vent design approach that uses
In terms of combined vent/draft-curtain zone-type fire model simulations might involve
designs, and as an alternative to traditional auto- application of an advanced version of LAVENT
matic, fusible-link-actuated vents, which could that would include the sprinkler/smoke-
involve the problem of vent skipping, a more interaction simulation model [59] discussed ear-
controllable and reliable means of ensuring lier. A successful preliminary implementation of
timely and effective vent action is available. this approach, with a revised prototype model
One generic possibility would involve ganging, called LAVENTS (fusible Link-Actuated
that is, opening together all or most vent units in VENTs and Sprinklers), has already been
the compartment of fire origin [61]. A ganging presented [64]. Applications of the LES (Large
strategy that could be well integrated into a reli- Eddy Simulation) model [65–68], the JASMINE
able, consensus sprinkler/vent design is one in model [69, 70], and others [71, 72] have also
which all vents of the fire compartment are been reported.

EDUFIRE.IR Telegram EDUFIRE_IR

 1852 J.A. Milke

One of the difficulties in applying the above vent design that meets the established design
for design applications is the limited availability objectives.
of input data to describe the initial sprinkler
spray from a wide range of sprinklers and the
response characteristics of the vent.
Special Conditions
A Set of Example Guidelines for Design of a
Consensus Sprinkler/Vent System As a sum- There are some aspects of smoke control system
mary to the above discussion, the following design that involve special attention. These
example guidelines are provided for the design aspects, which affect actuation of active smoke
of a sprinkler/vent system. control systems and the efficiency of exhaust
1. Establish the sprinkler design in the tradi- fans, are the following:
tional way; that is, develop design parameters • Intermediate stratification
using full-scale testing involving effective, • Confined flow
rapid, sprinkler-activation strategies in the • Plugholing
absence of vents (in this context, “rapid” • Makeup air supply
means that the design problem involves an
effectively unconfined ceiling where smoke-
layer buildup is negligible and does not affect Intermediate Stratification
the timing or sequence of early sprinkler
discharge). The upward movement of smoke in the plume is
2. Establish a vent design objective. In cases in dependent on the smoke being buoyant relative
which sprinkler action is expected to control to the surroundings. Delays in activation may be
the fire (i.e., the fire will not exceed a experienced where ceiling-mounted initiating
specified, maximum energy-release rate), the devices are present if the air near the ceiling is
design objective for scenarios as shown in warmer than the rising smoke [2, 73]. Dillon [74]
Fig. 51.13 might be for the vents to maintain reported measurements of the difference in ambi-
indefinitely the smoke from spreading beyond ent temperature from floor to ceiling to be on the
the curtained compartment of fire origin (i.e., order of 50  C in some atria with glazed ceilings.
the smoke-layer interface does not descend A prefire, warm air layer may be created due to a
below the bottom of the draft curtains). If the solar load where the ceiling contains glazing
latter design objective is too ambitious or in materials. In such cases, the smoke will stratify
cases in which sprinkler action is expected below this warm air layer and not reach the
only to slow but not to stop the growth of the ceiling. Early after ignition, the maximum height
fire, then the design objective would be for the to which the smoke plume will rise depends on
vents to maintain the smoke from spreading the convective heat release rate and the ambient
beyond the curtained compartment of fire ori- temperature variation in the open space.
gin for a specified time interval (e.g., the time Algebraic correlations may be applied to
expected for the fire department to respond address two situations (Fig. 51.18):
and initiate an attack on the fire). 1. The temperature of the ambient air is assumed
3. Adopt a practical/achievable strategy of early constant up to a height above which there is
opening of all vents in the compartment of fire discrete increase in temperature associated
origin, e.g., ganged operation of all vents in with a layer of warm air. This situation may
the curtained compartment of fire origin based occur if the upper portion of a mall, atrium, or
on and subsequent to first sprinkler activation. other large space is unoccupied so that the air
4. Using a fire model with a proven capability of in that portion is left unconditioned.
simulating the time-dependent interaction of 2. The ambient interior air within the large space
sprinklers, vents, and draft curtains, develop a has a constant temperature gradient

EDUFIRE.IR Telegram EDUFIRE_IR

 51 Smoke Control by Mechanical Exhaust or Natural Venting 1853

Fig. 51.18 Pre-fire

temperature profiles

Building Step function Linear

with temperature temperature
atrium profile profile

(temperature change per unit height) from F ¼ 0:0277Q_ c

floor level to the ceiling. This case is less
G ¼ 0:0335dT o =dz
likely than the first.
In the first case, where the interior air has a Because dTo/dz is a constant, ΔTo/H may be
discrete temperature change at some elevation substituted for the derivative. Substituting the
above floor level, then the potential for stratifica- simplified expressions for F and G into Equa-
tion can be assessed by determining the tempera- tion 51.48 yields [73].
ture of the plume at the height associated with
zm ¼ 5:54Q_ c ðΔT o =H Þ
3=8
1=4
the lower edge of the warm air layer. Where the ð51:49Þ
plume centerline temperature is equal to the
ambient temperature, the plume is no longer By reformulating Equation 51.49 to solve for
buoyant, loses its ability to rise, and stratifies at _
Q c, a minimum fire size can be determined that is
that height. One correlation for the plume center- just large enough to force the smoke to the ceil-
line temperature was presented previously as ing of an atrium without prematurely stratifying
Equation 51.35. due to the increasing ambient temperature.
In the particular case where the ambient,
pre-fire temperature increases uniformly along Q_ c ¼ 0:00118H5=2 ΔT 3=2
o ð51:50Þ
the entire height, the maximum plume rise can
The results of an analysis of intermediate
be determined from [19].
stratification are presented in Fig. 51.19. In one
zm ¼ 3:79F1=4 G
3=8 ð51:48Þ case, a step function is assumed to provide a
30  C change in temperature 15 m above the
where floor due to the upper portion of the atrium
  being unconditioned. For the other case, a tem-
F ¼ gQ_ c = T o ρo c p perature gradient of 1.5  C/m is arbitrarily
assumed in an atrium with a ceiling height of
G ¼
ðg=ρo Þdρo =dz
20 m. Plume centerline temperatures from two
Assuming standard conditions and that the size fires are graphed based on Equation 51.35.
smoke in the space behaves as an ideal gas, the As indicated in the figure, for the case with the
expressions for F and G are uniform gradient, smoke is expected to stratify

EDUFIRE.IR Telegram EDUFIRE_IR

 1854 J.A. Milke

Fig. 51.19 Indoor air and Plume, 2000 kW

plume temperature profiles
with the potential for 200 Plume, 1000 kW
intermediate stratification
Const. gradient

Temperature (°C)
150 Step function

100

50

0
0 5 10 15 20
Height (m)

approximately 13 or 15 m above the floor, In order to determine the point of contact of

depending on the fire size. For the case involving the plume with the walls, the plume width must
the step function change in temperature, the be expressed as a function of height. The width of
smoke stratifies from both fire sizes at the height the plume has been addressed theoretically and
of the step change in temperature. also experimentally.
If the smoke is expected to stratify at an inter- Based on theory (see Chap. 13), the plume
mediate height below the ceiling, then a device width is expected to be
other than ceiling-mounted detectors (such as
d ¼ 2:4αz ð51:51Þ
projected beam detectors) needs to be considered
to initiate the smoke control system. The beam where α ffi 0:15
detectors should be placed below the height of
stratification to intercept the rising plume. In Thus, d ¼ 0:36z ð51:52Þ
general, once the smoke control system operates,
Experimentally, the plume width is estimated
the warm air layer should be exhausted to permit
by examining photographs [75] or the difference
the smoke to reach the ceiling.
between the plume temperature and ambient
temperature (i.e., temperature excess at various
horizontal distances from the plume centerline)
Plume Width
[30]. Using temperature measurements, the
plume width is defined as the position where
As a plume rises it also widens as a result of the
the temperature excess is one-half of the value
entrainment of additional mass into the plume.
at the centerline.
For tall, narrow spaces, the plume may fill the
Handa and Sugawa [75] developed an empiri-
entire cross section of the atrium prior to
cal correlation of the width of the plume deter-
reaching the ceiling. Above this position, air
mined from photographs of the visual plume
entrainment into the plume is greatly reduced
from wood crib fires
due to the limited amount of air available. In
such situations, initially the bottom of the d ¼ do z1=2 ð51:53Þ
smoke layer may be assumed to be located at
this point of contact. Plume width is also impor- Heskestad [76] noted that the visible plume
tant when determining the location of projected diameter was greater than that determined from
beam detectors intended to intercept the plume. the temperature excess. Consequently, Heskestad

EDUFIRE.IR Telegram EDUFIRE_IR

 51 Smoke Control by Mechanical Exhaust or Natural Venting 1855

Fig. 51.20 Effect of 140

smoke-layer depth and
Temp rise=10°C
temperature rise on 120
maximum exhaust capacity Temp rise=20°C

Exhaust Rate (m3/s)

100 Temp rise=50°C
Temp rise=100°C
80

60

40

20

0
0 1 2 3 4 5
Smoke Layer Depth

estimated the visible plume diameter to be twice the lower layer in addition to smoke. As such,
that determined by the excess temperature less smoke is removed by the exhaust fans and a
approach. Thus, the plume diameter is estimated deeper layer results. Because a simple method to
as estimate the proportion of air drawn in from

1=2 below the smoke layer by the fans is unavailable,
Tc an elementary method of estimating the smoke
d ¼ 0:48 z ð51:54Þ
To layer depth during plugholing is not available. As
such, simple calculations can only be performed
As indicated in Equation 51.35, the plume
to assess the occurrence of plugholing, not the
centerline temperature decreases appreciably
effect.
with increasing height. Thus, for tall spaces, the
The original research on plugholing was done
plume centerline temperature may be close to
for natural vents. Recently, Lougheed and
ambient. For example, at a height of 30 m with
Hadjisophocleous demonstrated that the
a fire size of 5000 kW and To of 293 K, Tc is
plugholing analysis for natural vents was also
312 K. In this case (Tc/To)1/2 in Equation 51.51 is
applicable to mechanical venting [77]. In order
only 1.03. Because of the rapid decline in Tc with
to avoid plugholing, the maximum exhaust
increasing height, for engineering purposes (Tc/
capacity at an extract point is:
To)1/2 can be approximated as being 1.0. Conse-
quently, in many cases the total plume diameter
1=2
_ 5=2 ΔT
may be approximated by considering the plume V max ¼ 4:16γd ð51:55Þ
To
diameter to be approximately one-half of the
height. Where γ is a factor relating to the location of
Considering the variety of analyses for plume the vent. If the vent is in the middle of the space,
width, the plume width is estimated to be γ ¼ 1 [1].
25–50 % of the height above the top of the fuel Results of applying Equation 51.56 are
package, with the 36 % proportion from theory provided in Fig. 51.20 for a range of temperature
being near the middle of the range. rise values of the smoke. Where venting
capacities greater than the maximum limit are
needed to achieve smoke management
Plugholing objectives, multiple extract points need to be
provided to avoid plugholing.
Plugholing occurs when the exhaust capacity at a Assuming an axisymmetric plume, ṁ can be
single point is sufficiently large to draw air from replaced using Equation 51.20, and the smoke

EDUFIRE.IR Telegram EDUFIRE_IR

 1856 J.A. Milke

Fig. 51.21 Ratio of 0.7

smoke-layer depth to clear 2.5 MW
height for single exhaust 0.6
5 MW
point
0.5 10 MW

0.4

d/z
0.3

0.2

0.1

0.0
0 20 40 60 80 100
3/s)
Exhaust Rate (m

layer temperature can be replaced using Equa- mass to the smoke layer, which must be added
tion 51.34 (assuming adiabatic conditions) to to the required capacity of the smoke exhaust to
express the minimum smoke layer depth in prevent an increase in the smoke layer depth. If
terms of the heat release rate and clear height as introduced near the smoke layer interface, the
indicated in Fig. 51.21. For a single extract point, makeup air may increase the amount of mixing
the minimum smoke layer depth is slightly less of clean air with the smoke to further add to the
than 40 % of the clear height. smoke layer.
Makeup air should be provided at a slow
velocity so that the plume, fire, and smoke
Makeup Air Supply layer are not adversely affected. Makeup air
supplied at a rapid velocity near the plume
The makeup air supplied to the atrium should be may deflect the plume to enhance the entrain-
• Uncontaminated ment rate, thereby increasing the rate of smoke
• Introduced below the smoke layer production. In addition, the burning rate of the
• Introduced at a slow velocity fire may be increased by makeup air provided at
• Supplied at a rate less than the required an excessive velocity. Because the entrainment
exhaust rate process induces an air velocity of approxi-
Air that is not contaminated by smoke can be mately 1 m/s, the maximum makeup air veloc-
provided by locating intakes for the makeup air ity in the vicinity of the plume is often
remote from the smoke exhaust discharge, recommended to be 1 m/s. Because of the dif-
preventing smoke feedback. To address the fusion of air once past the diffuser, the makeup
potential for smoke being introduced into the air velocity at the diffuser may be greater than
makeup air supply, a smoke detector should be 1 m/s.
provided to shut down the makeup air supply Finally, the mass rate of makeup air supplied
system. Selection of a smoke detector for this must be less than that being exhausted. Failure to
application should consider the operating follow this guideline may lead to the atrium
conditions, range of temperatures, and installa- being pressurized relative to the communicating
tion within a duct. spaces. Being at a positive pressure, smoke
All makeup air should be provided below the movement will be forced through any unpro-
smoke layer interface. Any makeup air provided tected openings in physical barriers into the com-
above the smoke layer interface merely adds municating spaces.

EDUFIRE.IR Telegram EDUFIRE_IR

 51 Smoke Control by Mechanical Exhaust or Natural Venting 1857

attempting to enter the adjoining space. A mini-

Limited Fuel mum airflow velocity at all points of the opening
must be provided in order to prevent smoke
In some cases smoke management objectives migration through the opening. Empirical
may be fulfilled without a dedicated smoke con- correlations to estimate the minimum average
trol system due to the intrinsic qualities of the velocity for the entire opening are available,
atrium. The intrinsic qualities of the atrium based on limited experimental data [78]. The
include parameters, such as the composition and calculated average velocity is greater than the
quantity of fuel and geometry of the atrium. As actual minimum velocity required at an opening
an example, a limited amount of fuel may be to oppose smoke propagation to insure that the
present that is unable to sustain a fire for a suffi- minimum critical velocity is achieved at all
cient period of time to create conditions beyond points, considering the effects of turbulence
the allowable limits. The amount of fuel con- caused by the edges and corners of the opening.
sumed during the time period of interest depends The minimum average velocity to oppose
on whether the fire is steady or unsteady. In the smoke originating in the communicating space
case of a steady fire, the fuel mass consumed in a is evaluated using Equation 51.59.
given period of time is determined as sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
gH ðT s
T o Þ
Q_ t ve ¼ 0:64 ð51:58Þ
m_ f ¼ ð51:56Þ Ts
Hc

Alternatively, for an unsteady, t2 profile fire, Alternatively, if the smoke at the opening is
the fuel mass consumed during a given period of part of a rising plume that is rising along the side
time is given as of the atrium wall, then Equation 51.60 is
applicable.
t3 !1=3
m_ f ¼ 333 ð51:57Þ
H c t2g Q_
ve ¼ 0:057 ð51:59Þ
z
When analyzing the inherent ability of the
atrium to fulfill the smoke management design The opposed airflow velocity should not
goals, the time period should relate either to the exceed 1 m/s. Above that limit, the airflow veloc-
performance of a fire protection system or to the ity may deflect the plume away from the wall,
development of smoke layer conditions in excess making more plume surface area available for
of acceptable levels. For example, in life entrainment. The increased area for entrainment
safety–oriented designs, the time period may be will enhance the smoke generation rate. Conse-
either that required for evacuation, or for untena- quently, the problem of propagation to the com-
ble conditions to be generated, whichever is less. municating space may be solved by an excessive
average velocity; however, other problems may
be created by the increased smoke production
Opposed Airflow rate and a possible increase in the depth of the
smoke layer in the atrium. The volumetric capac-
Opposed airflow refers to systems where airflow ity of the mechanical equipment required to
is provided in a direction opposite to the unde- deliver the necessary velocity for opposed air-
sired direction of smoke movement. Opposed flow can be approximated as
airflow may be used in lieu of physical barriers
to prevent smoke spread from one space to V oa ¼ Ao ve ð51:60Þ
another (i.e., between the communicating space
and the atrium). Opposed airflow limits smoke If several openings are protected with the
flow by countering the momentum of the smoke opposed airflow approach using the same

EDUFIRE.IR Telegram EDUFIRE_IR

 1858 J.A. Milke

mechanical equipment, the cross-sectional area of the floor of the atrium, determine the
should be the sum of the areas for all of the following:
openings. The opposed airflow technique may 1. Minimum airflow velocity required for
be infeasible due to the substantial amount of opposed airflow
airflow capacity required to protect numerous 2. Volumetric rate of air supply for opposed
openings having a large total area. airflow
Where opposed airflow is utilized, the impact 3. Capacity of the exhaust fans in the atrium to
of the volume of air being introduced into the maintain the smoke layer interface at an ele-
space with the fire must be assessed. For exam- vation 25 m above floor level and also to
ple, if the airflow is directed into the atrium and accommodate the additional air from the
smoke exhaust equipment is also provided to opposed airflow approach
maintain a constant position of the smoke layer
interface in the atrium, then all of the additional Solution The minimum opposed airflow veloc-
air used for opposed airflow must also be ity can be determined using Equation 51.60.
exhausted. The additional air can be accounted However, the temperature of the smoke layer,
for by increasing the required mass rate of T, is unknown. Thus, Equation 51.61 must be
exhaust in the atrium by the amount used for applied first. Solving iteratively, T is approxi-
the opposed airflow. The additional air being mately 305 K. The minimum airflow velocity is
exhausted will also affect the qualities of the 0.20 m/s. The volumetric supply capacity for the
smoke layer within the atrium (see opposed airflow strategy for all five openings is
Equations 51.34, 51.38, and 51.40). The smoke 12.5 m3/s. The associated mass flow rate is
layer temperature, Ts(K), can be determined 15.0 kg/s.
using Equation 51.61, based on an analysis Without the opposed airflow, the mass rate of
included elsewhere [3]. smoke exhaust required to maintain the smoke
layer interface height in the atrium at a height of
T ¼ 293 25 m is determined using Equation 51.27 to be
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1

2=3 712Ao H ðT
293Þ 236 kg/s. Thus, the combined mass exhaust rate
þ 0:0018 þ 0:072Q_ c z5=3 þ Q_ T 3=2
c necessary is 251 kg/s. This mass flow rate
ð51:61Þ corresponds to a volumetric rate of 209 m3/s.
As a practical issue, this exhaust rate should
Equation 51.61 must be applied iteratively to be compared to that required to keep the smoke
determine the resulting smoke layer temperature. layer interface above the top of the openings
In cases with large clear heights, the temperature (i.e., 32.5 m above floor level). Based on
of the air used for the opposed airflow strategy Equations 51.27 and 51.29, the required volu-
will be virtually equal to the temperature of the metric exhaust rate is 362 kg/s. Thus, in this
smoke layer to permit the addition of volumetric situation, the combined exhaust rate with the
rates of air rather than mass rates. opposed airflow strategy is less than that
Alternatively, if airflow is directed from the associated with the strategy to keep the smoke
atrium into a communicating space, the commu- layer interface above the opening.
nicating space must also be exhausted, otherwise
the communicating space will become positively
pressurized.
Nomenclature
Example 8 Considering the atrium from Exam-
ple 5. There are five 5-m-wide  2.5-m-high A Cross-sectional area of the atrium (m2)
openings to the communicating space. The bot- Ao Cross-sectional area of opening (m2)
tom of the openings is 30 m above the floor of the b Distance from the store opening to the
atrium. Considering a 5000 kW fire in the center balcony edge (m)

EDUFIRE.IR Telegram EDUFIRE_IR

 51 Smoke Control by Mechanical Exhaust or Natural Venting 1859

CCO Volumetric concentration of carbon ΔTo Prefire temperature change from floor
monoxide (ppm) to ceiling of the ambient air ( C)
cp Specific heat (kJ/kg-K) t Time (s)
D Optical density per unit pathlength tcj Ceiling jet transport lag (s)
(m
1) tg Growth time (s)
Dm Mass optical density (m2/kg) tpl Plume transport lag (s)
d Plume diameter (based on excess V Volumetric flow rate (m3/s)
temperature) (m) Voa Volumetric capacity required for
do Diameter of fire (m) opposed air-flow (m3/s)
fCO Yield fraction of CO (kgCO/kgfuel) Vu Volume of upper layer (m3)
fi Yield fraction of species i (kg of v Characteristic velocity (m/s)
species i per kg of fuel consumed) ve Opposed airflow velocity (m/s)
g Gravitational acceleration (9.8 m/s2) w Width of the balcony opening from the
H Height of ceiling above top of fuel area of origin (m)
surface (m) x Position (m)
Hb Height of balcony above top of fuel YCO Mass fraction of CO (kg of species CO
surface (m) per kg of smoke)
Hc Heat of combustion (kJ/kg) Yi Mass fraction of gas species i (kg of
Hc,conv Convective heat of combustion (kJ/kg) species i per kg of smoke)
h Enthalpy z Clear height, position of smoke layer
K Constant, depending on target being interface above the top of fuel surface
viewed (e.g., ¼ 6 for lighted signs)[3] (m)
k Thermal conductivity (W/mK) zb position of smoke layer interface
kv Volumetric entrainment constant above top of balcony (m)
(0.065 m4/3 kW 
1/3 s
1) zf Limiting height above fuel (m)
L Width of balcony spill plume (m) zm Maximum rise of plume (m)
l Characteristic length (m) zo Virtual origin of plume (m)
MWi Molecular weight of species i (kg) χa Combustion efficiency
MCO Molecular weight of carbon monoxide χl Heat loss fraction from smoke to
(28 kg) enclosure
Mair Molecular weight of air (29 kg) ρ Density (kg/m3)
mu Mass of upper smoke layer (kg)
ṁ Mass entrainment rate in plume (kg/s) Subscripts
mf Mass burning rate (kg/s)
Δp Pressure difference (Pa) F Full-scale building
r Radius (i.e., horizontal distance from m Small-scale model
plume centerline (m) o Ambient air
Q t3
¼ 1055 2
t2 3 for t fires (kJ)
w Wall, ceiling, or floor of enclosure
g

Q¼ Q_ t for steady fires (kJ)

Q o¼ ρocpToA(H–z) (kJ)
Q_ Heat release rate of fire (kW)
Q_ c Convective portion of heat release rate References
of fire (kW)
1. NFPA 92, Standard for Smoke-Control Systems,
Tc Temperature at plume centerline (K) National Fire Protection Association, Quincy, MA
T Temperature (K) (2012).***was 92A
ΔTad Temperature difference between 2. J.H. Klote, J.A. Milke, P.G. Turnbull, A. Kashef, and
M.J. Ferreira, Handbook of Smoke Control Engineer-
smoke layer and ambient air ( C)
ing, ASHRAE, Atlanta (2012).

EDUFIRE.IR Telegram EDUFIRE_IR

 1860 J.A. Milke

3. T. Jin, “Irritating Effects of Fire Smoke on Visibility,” Layer Stratification in Full-Scale Multiroom Fire
Fire Science and Technology, 5, 1 (1985). Scenarios,” Paper 81-HT-9, The American Society
4. J. Morehart, “Sprinklers in the NIH Atrium: How Did of Mechanical Engineers, New York (1981).
They React During the Fire Last May?” Fire Journal, 20. G. Heskestad, Letter to the Editor, Fire Technology,
83, pp. 56–57 (1989). 27, 2, pp. 174–185 (1991).
5. NFPA 75, Standard for the Protection of Information 21. T. Yamana and T. Tanaka, “Smoke Control in Large
Technology Equipment, National Fire Protection Spaces (Part 2—Smoke Control Experiments in a
Association, Quincy, MA (2009). Large-Scale Space),” Fire Science and Technology,
6. R.D. Peacock and E. Braun, “Fire Tests of Amtrak 5, 1, pp. 41–54 (1985).
Passenger Rail Vehicle Interiors,” NBS Technical 22. G.D. Lougheed, Personal Communication, National
Note 1193, National Bureau of Standards, Research Council of Canada (Mar. 20, 1991).
Gaithersburg, MD (1984). 23. S.P. Nowlen, “Enclosure Environment Characteriza-
7. V. Babrauskas, “A Laboratory Flammability Test for tion Testing for the Baseline Validation of Computer
Institutional Mattresses,” Fire Journal, 72, 93, Fire Simulation Codes,” NUREG/ CR-4681, SAND
pp. 35–40 (1981). 86–1296, Sandia National Laboratories, Albuquerque,
8. S.W. Harpe, T.E. Waterman, and W.S. Christian, NM (1987).
“Detector Sensitivity and Siting Requirements for 24. J.A. Sharry, “An Atrium Fire,” Fire Journal, 67, 6, pp.
Dwellings, Phase 2, Part 2 ‘Indiana Dunes Tests,’” 39–41 (1973).
Report No. PB-263882, National Bureau of 25. J. Lathrop, “Atrium Fire Proves Difficult to Venti-
Standards, Gaithersburg, MD (1977). late,” Fire Journal, 73, 1, pp. 30–31 (1979).
9. Milke, J.A., Hugue, D.E., Hoskins, B.L., and Carroll 26. D.M. McGrail, “Denver’s Polo Club Condo Fire:
J.P., “Tenability Analyses in Performance-Based Atrium Turns High-Rise Chimney,” Fire Engineer-
Design,” Fire Protection Engineering, 28, 50-56 ing, pp. 67–74 (1992).
(2005), 27. G. Heskestad, “Engineering Relations for Fire
10. J.L. Bryan, “Damageability of Buildings, Contents, Plumes,” SFPE TR 82-8, Society of Fire Protection
and Personnel from Exposure to Fire,” Fire Safety Engineers, Boston (1982).
Journal, 11, pp. 15–32 (1984). 28. C. Beyler, “Fire Plumes and Ceiling Jets,” Fire Safety
11. J.A. Milke and J.H. Klote, “Smoke Management in Journal, 11, pp. 53–76 (1986).
Large Spaces in Buildings,” Building Control Com- 29. R.L. Alpert and E.J. Ward, “Evaluation of
mission of Victoria and The Broken Hill Proprietary Unsprinklered Fire Hazards,” Fire Safety Journal,
Company Limited, Melbourne, Australia (1998). 7, pp. 127–143 (1984).
12. J.G. Quintiere, “Scaling Applications in Fire 30. F.W. Mowrer and B. Williamson, “Estimating Room
Research,” Fire Safety Journal, 15, pp. 3–29 (1989). Temperatures from Fires along Walls and in Corners,”
13. F.W. Mowrer, “Lag Times Associated with Fire Fire Technology, 23, 2, pp. 133–145 (1987).
Detection and Suppression,” Fire Technology, 31. Harrison, R., “Entrainment of Air Into Thermal Spill
26, 3, pp. 244–265 (1990). Plumes,” PhD Dissertation, University of Canterbury,
14. J.S. Newman, “Principles for Fire Detection,” Fire New Zealand (2009).
Technology, 24, 2, pp. 116–127 (1988). 32. Lougheed, GD., McCartney, C.J. and Gibbs, E., “Bal-
15. J.A. Milke, “Smoke Management for Covered Malls cony Spill Plumes Final Report, RP-1247, ASHRAE,
and Atria,” Fire Technology, 26, 3, pp. 223–243 Atlanta (2007).
(1990). 33. Lim, J.M.K., “Numerical Modeling of Balcony Spill
16. G. Heskestad and M.A. Delichatsios, “Environments Plumes Using Fire Dynamics Simulator (FDS),” MS
of Fire Detectors—Phase I: Effect of Fire Size, Ceil- Thesis, University of Maryland (2010).
ing Height, and Materials,” Vol. I—Measurements 34. M. Law, “A Note on Smoke Plumes from Fires in
(NBS-GCR-77-86), Vol. II—Analysis (NBS-GCR-77- Multi-Level Shopping Malls,” Fire Safety Journal,
95), National Bureau of Standards, Gaithersburg, MD 10, pp. 197–202 (1986).
(1977). 35. H.P. Morgan and N.R. Marshall, “Smoke Control
17. B.R. Morton, Sir Geoffrey Taylor, and J.S. Turner, Measures in a Covered Two-Story Shopping Mall
“Turbulent Gravitational Convection from Having Balconies and Pedestrian Walkways,” BRE
Maintained and Instantaneous Sources,” in CP11/79, Fire Research Station, Borehamwood,
Proceedings of Royal Society A, 234, pp. 1–23, England (1979).
London (1956). 36. W.D. Walton, “ASET-B: A Room Fire Program for
18. G. Mulholland, T. Handa, O. Sugawa, and Personal Computers,” NBSIR 85-3144, National
H. Yamamoto, “Smoke Filling in an Enclosure,” Bureau of Standards, Gaithersburg, MD (1985).
Paper 81-HT-8, The American Society of Mechanical 37. Lincolne Scott Australia Pty Ltd., Jupiters
Engineers, New York (1981). Casino—Report on Atrium Smoke Tests, Lincolne
19. L.Y. Cooper, M. Harkleroad, J. Quintiere, and Scott Australia Pty Ltd., Toowong, Australia
W. Rinkinen, “An Experimental Study of Upper Hot (1986).

EDUFIRE.IR Telegram EDUFIRE_IR

 51 Smoke Control by Mechanical Exhaust or Natural Venting 1861

38. G.O. Hansell and H.P. Morgan, “Smoke Control in 54. L.A. Jackman, Second Report on the Interaction
Atrium Buildings Using Depressurization,” PD 66/88, between Sprinkler Sprays and the Thermally Buoyant
Fire Research Station, Borehamwood, UK (1988). Layers of Gases from Fires, South Bank Polytechnic,
39. R.A. Waters, “Stansted Terminal Building and Early London (1990).
Atrium Studies,” Journal of Fire Protection Engi- 55. L.A. Jackman, Third Report on the Interaction
neering, 1, 2, pp. 63–76 (1989). between Sprinkler Sprays and the Thermally Buoyant
40. G. Heskestad, “Similarity Relations for the Initial Layers of Gases from Fires, South Bank Polytechnic,
Convective Flow Generated by Fire,” Paper 72-WA/ London (1991).
HT-17, American Society of Mechanical Engineers, 56. L.A. Jackman, P.F. Nolan, A.J. Gardiner, and
New York (1972). H.P. Morgan, Mathematical Model of the Interac-
41. T. Jin, “Visibility Through Fire Smoke (Part 2),” tion of Sprinkler Spray Drops with Fire Gases,
Report of the Fire Research Institute of Japan, Nos. South Bank University, London, Swedish Fire
33, 31, Tokyo (1971). Research Board, and National Institute of Standards
42. J.G. Quintiere, “An Assessment of Correlations and Technology (NIST), Fire Suppression Research,
Between Laboratory and Full-Scale Experiments for First International Conference on Fire Suppression
the FAA Aircraft Fire Safety Program, Part 1: Research, May 5–8, 1992, Stockholm, Sweden,
Smoke,” NBSIR 82-2508, National Bureau of (V. Sjolin, D.D. Evans, and N.H. Jason, eds.),
Standards, Gaithersburg, MD (1982). pp. 209–227 (1992).
43. G. Heskestad, “Hazard Evaluation,” submitted to 57. G. Heskestad, “Sprinkler/Hot Layer Interaction,”
NFPA Task Group on Smoke Management of Atria, NIST-GCR-91-590, National Institute of Standards
Covered Malls, and Large Spaces, unpublished man- and Technology (NIST), Gaithersburg, MD (1991).
uscript (1988). 58. L.Y. Cooper, “The Interaction of an Isolated Sprinkler
44. G.D. Lougheed, “Probability of Occurrence and Spray and a Two-Layer Compartment Fire Environ-
Expected Size of Shielded Fires in Sprinklered Build- ment. Phenomena and Model Simulations,” Fire
ing; Phase 2, Full-Scale Fire Tests,” National Safety Journal, 25, 2, pp. 89–107 (1995).
Research Council of Canada, Ottawa (1997). 59. L.Y. Cooper, “The Interaction of an Isolated Sprinkler
45. NFPA 204, Standard for Smoke and Heat Venting, Spray and a Two-Layer Compartment Fire Environ-
National Fore Protection Association, Quincy, MA ment,” International Journal Heat and Mass Transfer,
(2012). 38, 4, pp. 679–690 (1995).
46. C.L. Beyler and L.Y. Cooper, “Interaction of 60. P.L. Hinkley, “Work by the Fire Research Station on
Sprinklers with Smoke Vents,” Fire Technology, the Control of Smoke in Covered Shopping Centers,”
37, 1, pp. 9–36 (2001). BRE Paper CP 83/75, Building Research Establish-
47. M.L. Bullen, “The Effect of a Sprinkler on the Stabil- ment (BRE), Fire Research Station (FRS), London
ity of a Smoke Layer beneath a Ceiling,” Fire (1975).
Research Note No. 1016, Department of the Environ- 61. A.J.M. Heselden, “The Interaction of Sprinklers and
ment and Fire Officers’ Committee, Joint Fire Fire Venting,” Fire Surveyor, 11, 5, pp. 13–28 (1982).
Research Organization, Watford, UK (1974). 62. G. Holmstedt, “Sprinkler and Fire Venting Interac-
48. M.L. Bullen, “The Effect of a Sprinkler on the Stabil- tion,” Literature Survey on Modeling Approaches and
ity of a Smoke Layer beneath a Ceiling,” Fire Tech- Experiments Available and Recommendations for
nology, 13, 1, pp. 21–34 (1977). Further Studies, 92178 AR GH/AB, Swedish Fire
49. R.L. Alpert, “Calculated Interaction of Sprays with Research Board, Stockholm, Sweden (1992).
Large-Scale Buoyant Flows,” Journal of Heat Trans- 63. B. Persson and H. Ingason, “Modeling of Interaction
fer, 106, pp. 310–317 (1984). between Sprinklers and Fire Vents,” SP Report
50. A.J. Gardiner, First Report on the Interaction between 1996:32, Swedish National Testing and Research
Sprinkler Sprays and the Thermally Buoyant Layers of Institute, Sweden (1996).
Gases from Fires, South Bank Polytechnic, London 64. L.Y. Cooper, “LAVENTS—A Computer Program to
(1985). Model the Interaction between Sprinklers, Smoke
51. A.J. Gardiner, Second Report on the Interaction Layers, and Smoke Vents,” presented at the Interna-
between Sprinkler Sprays and the Thermally Buoyant tional Conference on Smoke Ventilation and
Layers of Gases from Fires, South Bank Polytechnic, Sprinklers—Aspects of Combined Use, Fire Research
London (1986). Station, Borehamwood, UK (1992).
52. A.J. Gardiner, Third Report on the Interaction 65. K.B. McGrattan, A. Hamins, D.W. Stroup, “Sprinkler,
between Sprinkler Sprays and the Thermally Buoyant Smoke and Heat Vent, Draft Curtain Interaction—
Layers of Gases from Fires, South Bank Polytechnic, Large-Scale Experiments and Model Development,”
London (1988). NISTIR 6196, National Institute of Standards and
53. A.J. Gardiner, Fourth Report on the Interaction Technology (NIST), Gaithersburg, MD (1998).
between Sprinkler Sprays and the Thermally Buoyant 66. K.B. McGrattan, H.R. Baum, and R.G. Rehm, “Large
Layers of Gases from Fires, South Bank Polytechnic, Eddy Simulations of Smoke Movement,” Fire Safety
London (1988). Journal, 30, pp. 161–178 (1998).

EDUFIRE.IR Telegram EDUFIRE_IR

 1862 J.A. Milke

67. K.B. McGrattan and D.W. Stroup, “Sprinkler, Vent 74. M. Dillon, “Acceptance Testing and Techniques,”
and Draft Curtain Interaction: Experiment and Com- presented at The Roundtable on Fire Safety in Atri-
puter Simulation,” National Institute of Standards and ums—Are the Codes Meeting the Challenge?,
Technology (NIST), Gaithersburg, MD (1997). Washington, DC (1988).
68. K.B. McGrattan and D.W. Stroup, “Large Eddy 75. T. Handa and O. Sugawa, “Flow Behavior of Plume
Simulations of Sprinkler, Vent and Draft Curtain Per- from Growing Fire Source in High Ceiling Enclo-
formance,” Fire Suppression and Detection Research sure,” Journal of Fire and Flammability, 12, 1, pp.
Application, Symposium, Research and Practice: 31–50 (1981).
Bridging the Gap, Proceedings, National Fire Protec- 76. G. Heskestad, “Fire Plume Entrainment and
tion Research Foundation, February 12–14, 1997, Related Problems in Venting of Fire and Smoke
Orlando, FL, pp. 59–68 (1997). from Large Open Spaces,” submitted to NFPA Task
69. H. Touvinen, “Validation of Ceiling Jet Flows in a Group on Smoke Management of Atria, Covered
Large Corridor with Vents Using the CFD Code JAS- Malls, and Large Spaces, unpublished manuscript
MINE,” Fire Technology, 32, 1, pp. 25–49 (1996). (1987).
70. H. Touvinen and L.Y. Cooper, “Validation of Ceiling 77. G.D. Lougheed and G.V. Hadjisophocleous, “Investi-
Jet Flows in a Large Corridor with Vents Using the gation of Atrium Exhaust Effectiveness,” ASHRAE
CFD Code JASMINE: Errata and Additional Transactions, 103, 2 (1997).
Remarks,” Fire Technology, 33, 2, pp. 183–186 (1997). 78. G. Heskestad, “Inflow of Air Required at Wall and
71. R.N. Mawhinney, E.R. Galea, and M.K. Patel, “Euler- Ceiling Apertures to Prevent Escape of Fire Smoke,”
Lagrange Modeling of Fire/Sprinkler Interactions,” FMRC J. I. OQ4E4.RU, Factory Mutual Research
Fire Safety Science—Proceedings of the Fifth Inter- Corporation, Norwood, MA (1989).
national Symposium, International Association of Fire
Safety Science (IAFSS), Australia, p. 1336 (1997).
72. W.K. Chow and A.C. Tang, “Experimental Studies on James A. Milke is a professor and chair in the Depart-
Sprinkler Water Spray Smoke Layer Interaction,” Jour- ment of Fire Protection Engineering at the University of
nal of Applied Fire Science, 4, 3, pp. 171–184 (1994–95). Maryland. His recent research activities have included
73. G. Heskestad, “Note on Maximum Rise of Fire assessing the performance of smoke control systems in
Plumes in Temperature-Stratified Ambients,” Fire fires.
Safety Journal, 15, pp. 271–276 (1989).

EDUFIRE.IR Telegram EDUFIRE_IR

 ‫ویدیو – انیمیشن های اطفا‪ ،‬اعالم حریق‪ ،‬تهویه وتخلیه دود (رایگان)‬
‫برای مشاهده هر یک از ویدیوهای زیر کافیست بر روی عنوان آن آموزش کلیک نمایید تا به صفحه‬
‫ویدیو و آموزش آن عنوان هدایت شوید‪.‬‬

‫انیمیشن ویدیو‬

‫• اطفا حریق آبی‬

‫‪ o‬سیستم اطفا لوله خشک اسپرینکلر‬
‫‪ o‬سیستم اطفا لوله تر اسپرینکلر‬
‫‪ o‬سیستم اطفا پیش عملگر‬
‫‪ o‬سیستم اطفا واترمیست‬
‫‪ o‬سیستم اطفا سیالبی‬
‫• سیستم اطفا فوم‬
‫• تجهیزات اطفا حریق ‪ -‬تجهیزات هشدار دهنده‬
‫• تجهیزات اطفا حریق‪ -‬اسپرینکلر‬
‫• سیستم اطفا آشپزخانه صنعتی‬
‫• سیستم اطفا آیروسل‬
‫• سیستم اطفا دستی‬
‫• سیستم اطفا گازی‬
‫‪ o‬سیستم اطفا )‪FM200 , NOVEC, Inert Gas (IG‬‬
‫‪ o‬سیستم اطفا ‪CO2‬‬
‫• سیستم تهویه و تخلیه دود‬
‫• سیستم اعالم حریق‬
‫‪ o‬آدرسپذیر‬
‫‪ o‬متعارف‬
 ‫آموزش استاندارد‬
‫آموزش استاندارد ‪NFPA 13‬‬ ‫•‬
‫آموزش استاندارد ‪NFPA 14‬‬ ‫•‬
‫آموزش استاندارد ‪NFPA 20‬‬ ‫•‬

‫آموزش نرمافزار‬
‫اتواسپرینک‬ ‫‪o‬‬
‫پایروسیم‬ ‫‪o‬‬
‫پثفایندر‬ ‫‪o‬‬

‫دانلود استاندارد‬
‫ترجمه استاندارد ‪NFPA 30,14,13,10‬‬ ‫‪o‬‬
‫تمامی استانداردهای ‪NFPA & FM‬‬ ‫‪o‬‬
‫ترجمه استاندارد ‪NFPA 1037‬‬ ‫‪o‬‬

‫محصوالت‬
‫اتواسپرینک ‪ 2013‬و ‪ 2019‬به زبان فارسی برای اولین بار در ایران‬ ‫‪o‬‬
‫آالرم کد ‪ 2019‬به زبان فارسی برای اولین بار در ایران‬ ‫‪o‬‬
‫اطفا حریق آبی‬ ‫‪o‬‬
‫پایپنت (ماژول اسپرینکلر)‬ ‫‪o‬‬
‫کانتم‬ ‫‪o‬‬
‫اعالن حریق‬ ‫‪o‬‬

‫دورههای حضوری‬
‫آموزش اتواسپرینک‬ ‫‪o‬‬
‫آموزش آالرمکد‬ ‫‪o‬‬
‫آموزش پایروسیم‬ ‫‪o‬‬
‫آموزش کانتم ‪ +‬اگزاست‬ ‫‪o‬‬
‫آموزش اطفا آبی ‪+‬پمپ ‪ +‬بازدید از کارگاه‬ ‫‪o‬‬
‫آموزش اطفا گازی‬ ‫‪o‬‬
‫آموزش اطفا فوم‬ ‫‪o‬‬
‫آموزش مبحث سوم مقررات ملی‬ ‫‪o‬‬
‫آموزش اعالم حریق ‪F&G‬‬ ‫‪o‬‬
‫آموزش اعالم حریق آدرسپذیر‬ ‫‪o‬‬
‫آموزش اعالم حریق متعارف‬ ‫‪o‬‬
‫دوره آمادگی آزمون آتشنشانی (برق‪ ،‬مکانیک‪ ،‬عمران‪ ،‬معماری)‬ ‫‪o‬‬