Related Commercial Resources

CHAPTER 52

FIRE AND SMOKE MANAGEMENT

Fire Management ..................................................................... 52.1 Zone Smoke Control ............................................................... 52.11
Smoke Movement ..................................................................... 52.2 Computer Analysis for Pressurization Systems...................... 52.12
Smoke Management ................................................................. 52.4 Smoke Management in Large Spaces..................................... 52.12
Smoke Control System Design ................................................. 52.6 Tenability Systems ................................................................... 52.17
Fire and Smoke Dampers ......................................................... 52.8 Acceptance Testing................................................................. 52.17
Pressurized Stairwells.............................................................. 52.9 Special Inspector.................................................................... 52.17
Elevators ................................................................................ 52.11 Extraordinary Incidents ......................................................... 52.18

I N building fires, smoke often flows to locations remote from the

fire, threatening life and damaging property. Stairwells and ele-
vators frequently fill with smoke, thereby blocking or inhibiting
tree for fire protection. Building occupants and managers have the
primary role in preventing fire ignition. The building design team
may incorporate features into the building to assist the occupants
evacuation. Smoke causes the most deaths in fires. Smoke includes and managers in this effort. Because it is impossible to prevent fire
airborne solid and liquid particles and gases produced when a mate- ignition completely, managing fire effect has become significant in
rial undergoes pyrolysis or combustion, together with air that is fire protection design. Examples include compartmentation, sup-
entrained or otherwise mixed into the mass. pression, control of construction materials, exit systems, and smoke
The idea of using pressurization to prevent smoke infiltration of management. The Fire Protection Handbook (NFPA 2003) and
stairwells began to attract attention in the late 1960s. This concept Smoke Movement and Control in High-Rise Buildings (Tamura
was followed by the idea of the pressure sandwich (i.e., venting or 1994) contain detailed fire safety information.
exhausting the fire floor and pressurizing the surrounding floors). Historically, fire safety professionals have considered the HVAC
Frequently, a building’s ventilation system is used for this purpose. system a potentially dangerous penetration of natural building
Smoke control systems use fans to pressurize appropriate areas to membranes (walls, floors, etc.) that can readily transport smoke
limit smoke movement in fire situations. Smoke management sys- and fire. For this reason, HVAC has traditionally been shut down
tems include pressurization and all other methods that can be used when fire is discovered; this prevents fans from forcing smoke
singly or in combination to modify smoke movement. flow, but does not prevent ducted smoke movement caused by
This chapter discusses fire protection and smoke control systems buoyancy, stack effect, or wind. To solve the problem of smoke
in buildings as they relate to HVAC. For a more complete discus- movement, methods of smoke control have been developed; how-
sion, refer to Principles of Smoke Management (Klote and Milke ever, smoke control should be viewed as only one part of the over-
2002). National Fire Protection Association (NFPA) Standard 204 all building fire protection system.
provides information about venting large industrial and storage
buildings. For further information, refer to NFPA Standards 92A
and 92B. FIRE MANAGEMENT
The objective of fire safety is to provide some degree of protec-
tion for a building’s occupants, the building and property inside it, Although most of this chapter discusses smoke management, fire
and neighboring buildings. Various forms of analysis have been management at HVAC penetrations is an additional concern for the
used to quantify protection. Specific life safety objectives differ HVAC engineer. The most efficient way to limit fire damage is
with occupancy; for example, nursing home requirements are dif- through compartmentation. Fire-rated assemblies, such as the floor
ferent from those for office buildings. or walls, keep the fire in a given area for a specific period. However,
Two basic approaches to fire protection are (1) to prevent fire fire can easily pass through openings for plumbing, HVAC duct-
ignition and (2) to manage fire effects. Figure 1 shows a decision work, communication cables, or other services. Therefore, fire stop
systems are installed to maintain the rating of the fire-rated assem-
bly. The rating of a fire stop system depends on the number, size,
Fig. 1 Simplified Fire Protection Decision Tree and type of penetrations, and the construction assembly in which it
is installed.
Performance of the entire fire stop system, which includes the
construction assembly with its penetrations, is tested under fire con-
ditions by recognized independent testing laboratories. ASTM
Standard E814 and UL Standard 1479 describe ways to determine
performance of through-penetration fire stopping (TPFS).
TPFS is required by building codes under certain circumstances
for specific construction types and occupancies. In the United
States, model building codes require that most penetrations meet the
ASTM E814 test standard. TPFS classifications are published by
testing laboratories. Each classification is proprietary, and each ap-
plies to use with a specific set of conditions, so numerous types are
usually required on any given project.
Fig. 1 Simplified Fire Protection Decision Tree The construction manager and general contractor, not the archi-
tects and engineers, make work assignments. Sometimes they
The preparation of this chapter is assigned to TC 5.6, Control of Fire and assign fire stopping to the discipline making the penetration; other
Smoke. times, they assign it to a specialty fire-stopping subcontractor. The

52.1
Copyright © 2007, ASHRAE
52.2 2007 ASHRAE Handbook—HVAC Applications (SI)

Construction Specifications Institute (CSI) assigns fire-stopping shaft and the outside. A positive pressure difference indicates that
specifications to Division 7, which shaft pressure is higher than the outside pressure, and a negative
pressure difference indicates the opposite. Figure 3 illustrates air
• Encourages continuity of fire-stopping products on the project by
movement in buildings caused by both normal and reverse stack
consolidating their requirements (e.g., TPFS, expansion joint fire
effect.
stopping, floor-to-wall joint fire stopping, etc.)
Figure 4 can be used to determine the pressure difference caused
• Maintains flexibility of work assignments for the general contrac-
by stack effect. For normal stack effect, ∆p/h is positive, and the
tor and construction manager
pressure difference is positive above the neutral plane and negative
• Encourages prebid discussions between the contractor and sub-
below it. For reverse stack effect, ∆p/h is negative, and the pressure
contractors regarding appropriate work assignments
difference is negative above the neutral plane and positive below it.
In unusually tight buildings with exterior stairwells, Klote (1980)
SMOKE MOVEMENT observed reverse stack effect even with low outside air temperatures.
A smoke control system must be designed so that it is not over- In this situation, the exterior stairwell temperature is considerably
powered by the driving forces that cause smoke movement, which lower than the building temperature. The stairwell represents the
include stack effect, buoyancy, expansion, wind, and the HVAC sys- cold column of air, and other shafts within the building represent the
tem. During a fire, smoke is generally moved by a combination of warm columns of air.
these forces. If leakage paths are uniform with height, the neutral plane is near
the midheight of the building. However, when the leakage paths are
Stack Effect not uniform, the location of the neutral plane can vary considerably,
When it is cold outside, air tends to move upward within building as in the case of vented shafts. McGuire and Tamura (1975) provide
shafts (e.g., stairwells, elevator shafts, dumbwaiter shafts, mechan- methods for calculating the location of the neutral plane for some
ical shafts, mail chutes). This normal stack effect occurs because vented conditions.
air in the building is warmer and less dense than outside air. Normal Smoke movement from a building fire can be dominated by stack
stack effect is large when outside temperatures are low, especially in effect. In a building with normal stack effect, the existing air currents
tall buildings. However, normal stack effect can exist even in a one- (as shown in Figure 3) can move smoke considerable distances from
story building. the fire origin. If the fire is below the neutral plane, smoke moves
When outside air is warmer than building air, there is a natural with building air into and up the shafts. This upward smoke flow is
tendency for downward airflow, or reverse stack effect, in shafts.
At standard atmospheric pressure, the pressure difference caused by Fig. 3 Air Movement Caused by Normal and Reverse Stack
either normal or reverse stack effect is expressed as Effect
⎛1 1⎞
∆p = 3460 ⎜----- – ---- ⎟ h (1)
⎝ To Ti ⎠
where
∆p = pressure difference, Pa
To = absolute temperature of outside air, K
Ti = absolute temperature of air inside shaft, K
h = distance above neutral plane, m

For a building 60 m tall with a neutral plane at midheight, an out-

side temperature of −18°C (255 K), and an inside temperature of Fig. 3 Air Movement Caused by Normal and
21°C (294 K), the maximum pressure difference from stack effect is Reverse Stack Effect
54 Pa. This means that, at the top of the building, pressure inside a Fig. 4 Pressure Difference Caused by Stack Effect
shaft is 54 Pa greater than the outside pressure. At the base of the
building, pressure inside a shaft is 54 Pa lower than the outside pres-
sure. Figure 2 diagrams the pressure difference between a building

Fig. 2 Pressure Difference Between a Building Shaft and the

Outside Caused by Normal Stack Effect

Fig. 2 Pressure Difference Between a Building Shaft

and the Outside Caused by Normal Stack Effect Fig. 4 Pressure Difference Caused by Stack Effect
Fire and Smoke Management 52.3

enhanced by buoyancy forces due to the temperature of the smoke. If the fire compartment temperature is 700°C (993 K), the pressure
Once above the neutral plane, smoke flows from the shafts into the difference 10 m above the neutral plane is 83 Pa. This is a large fire,
upper floors of the building. If leakage between floors is negligible, and the pressures it produces are beyond present smoke control
floors below the neutral plane (except the fire floor) remain rela- methods. However, the example illustrates the extent to which
tively smoke-free until more smoke is produced than can be handled Equation (2) can be applied.
by stack effect flows. In sprinkler-controlled fires, the temperature in the fire room
Smoke from a fire located above the neutral plane is carried by remains at that of the surroundings except for a short time before
building airflow to the outside through exterior openings in the sprinkler activation. Sprinklers are activated by the ceiling jet, a thin
building. If leakage between floors is negligible, all floors other (50 to 100 mm) layer of hot gas under the ceiling. The maximum
than the fire floor remain relatively smoke-free until more smoke is temperature of the ceiling jet depends on the location of the fire, the
produced than can be handled by stack effect flows. When leakage activation temperature of the sprinkler, and the thermal lag of the
between floors is considerable, smoke flows to the floor above the sprinkler heat-responsive element. For most residential and com-
fire floor. mercial applications, the ceiling jet is between 80 and 150°C. In
Air currents caused by reverse stack effect (see Figure 3) tend to Equation (2), Tf is the average temperature of the fire compartment.
move relatively cool smoke down. In the case of hot smoke, buoy- For a sprinkler-controlled fire,
ancy forces can cause smoke to flow upward, even during reverse
stack effect conditions. T s ( H – H j ) + Tj Hj
T f = -------------------------------------------- (3)
Buoyancy H
High-temperature smoke has buoyancy because of its reduced where
density. At sea level, the pressure difference between a fire compart- H = floor-to-ceiling height, m
ment and its surroundings can be expressed as follows: Hj = thickness of ceiling jet, m
Tj = absolute temperature of ceiling jet, K
∆p = 3460 ⎛ ----- – ---- ⎞ h
1 1
(2)
⎝T T ⎠ For example, for H = 2.5 m, Hj = 0.1 m, Ts = 20 + 273 = 293 K,
s f
and Tj = 150 + 273 = 423 K,
where Tf = [293(2.5 − 0.1) + 423 × 0.1]/2.5 = 298 K or 25°C
∆p = pressure difference, Pa
Ts = absolute temperature of surroundings, K In Equation (2), this results in a pressure difference of 0.5 Pa, which
Tf = average absolute temperature of fire compartment, K is insignificant for smoke control applications.
h = distance above neutral plane, m
The pressure difference caused by buoyancy can be obtained Expansion
from Figure 5 for surroundings at 20°C (293 K). The neutral plane Energy released by a fire can also move smoke by expansion. In
is the plane of equal hydrostatic pressure between the fire compart- a fire compartment with only one opening to the building, building
ment and its surroundings. For a fire with a fire compartment air will flow in, and hot smoke will flow out. Neglecting the added
temperature at 800°C (1073 K), the pressure difference 1.5 m above mass of the fuel, which is small compared to airflow, the ratio of vol-
the neutral plane is 13 Pa. Fang (1980) studied pressures caused by umetric flows can be expressed as a ratio of absolute temperatures:
room fires during a series of full-scale fire tests. During these tests,
the maximum pressure difference reached was 16 Pa across the burn Q out Tout
----------- = ---------- (4)
room wall at the ceiling. Q in Tin
Much larger pressure differences are possible for tall fire com-
partments where the distance h from the neutral plane can be larger. where
Qout = volumetric flow rate of smoke out of fire compartment, m3/s
Fig. 5 Pressure Difference Caused by Buoyancy Qin = volumetric flow rate of air into fire compartment, m3/s
Tout = absolute temperature of smoke leaving fire compartment, K
Tin = absolute temperature of air into fire compartment, K
For smoke at 700°C (973 K) and entering air at 20°C (293 K), the
ratio of volumetric flows is 3.32. Note that absolute temperatures
are used in the calculation. In such a case, if air enters the compart-
ment at 1.5 m3/s, then smoke flows out at 5.0 m3/s, with the gas
expanding to more than three times its original volume.
For a fire compartment with open doors or windows, the pressure
difference across these openings caused by expansion is negligible.
However, for a tightly sealed fire compartment, the pressure differ-
ences from expansion may be important.
Wind
Wind can have a pronounced effect on smoke movement within a
building. The pressure wind exerts on a surface can be expressed as

2
p w = 0.5C w ρ oV (5)

where
pw = pressure exerted by wind, Pa
Cw = pressure coefficient, dimensionless
ρo = outside air density, kg/m3
Fig. 5 Pressure Difference Caused by Buoyancy V = wind velocity, m/s
52.4 2007 ASHRAE Handbook—HVAC Applications (SI)

The pressure coefficients Cw are in the range of −0.8 to 0.8, with dard 101, provide specific criteria for construction of smoke barriers
positive values for windward walls and negative values for leeward (including doors) and their smoke dampers. The extent to which
walls. The pressure coefficient depends on building geometry and smoke leaks through such barriers depends on the size and shape of the
varies locally over the wall surface. In general, wind velocity in- leakage paths in the barriers and the pressure difference across the
creases with height from the surface of the earth. Houghton and Car- paths.
ruther (1976), MacDonald (1975), Sachs (1972), and Simiu and
Scanlan (1978) give detailed information concerning wind velocity Dilution Remote from Fire
variations and pressure coefficients. Shaw and Tamura (1977) de- Smoke dilution is sometimes referred to as smoke purging,
veloped specific information about wind data with respect to air in- smoke removal, smoke exhaust, or smoke extraction. Dilution
filtration in buildings. can be used to maintain acceptable gas and particulate concentra-
With a pressure coefficient of 0.8 and air density of 1.20 kg/m3, a tions in a compartment subject to smoke infiltration from an adja-
15 m/s wind produces a pressure on a structure of 100 Pa. The effect cent space. It can be effective if the rate of smoke leakage is small
of wind on air movement within tightly constructed buildings with compared to either the total volume of the safeguarded space or the
all exterior doors and windows closed is slight. However, wind rate of purging air supplied to and removed from the space. Also,
effects can be important for loosely constructed buildings or for dilution can be beneficial to the fire service for removing smoke
buildings with open doors or windows. Usually, the resulting air- after a fire has been extinguished. Sometimes, when doors are
flows are complicated, and computer analysis is required. opened, smoke flows into areas intended to be protected. Ideally, the
Frequently, a window breaks in the fire compartment. If the win- doors are only open for short periods during evacuation. Smoke that
dow is on the leeward side of the building, the negative pressure has entered spaces remote from the fire can be purged by supplying
caused by the wind vents the smoke from the fire compartment. This outside air to dilute the smoke.
reduces smoke movement throughout the building. However, if the The following is a simple analysis of smoke dilution for spaces in
broken window is on the windward side, wind forces the smoke which there is no fire. Assume that at time zero (θ = 0), a compart-
throughout the fire floor and to other floors, which endangers the ment is contaminated with some concentration of smoke and that no
lives of building occupants and hampers fire fighting. Wind-induced more smoke flows into the compartment or is generated within it.
pressure in this situation can be large and can dominate air move- Also, assume that the contaminant is uniformly distributed through-
ment throughout the building. out the space. The concentration of contaminant in the space can be
expressed as
HVAC Systems
Before methods of smoke control were developed, HVAC systems C- = e –at
----- (6)
were shut down when fires were discovered because the systems fre- Co
quently transported smoke during fires.
The dilution rate can be determined from the following equation:
In the early stages of a fire, the HVAC system can aid in fire
detection. When a fire starts in an unoccupied portion of a build-
1 ⎛ Co ⎞
ing, the system can transport the smoke to a space where people a = ---- ln ⎜------⎟ (7)
can smell it and be alerted to the fire. However, as the fire pro- t ⎝ C⎠
gresses, the system transports smoke to every area it serves, thus
endangering life in all those spaces. The system also supplies air to where
the fire space, which aids combustion. Although shutting the sys- Co =
initial concentration of contaminant
tem down prevents it from supplying air to the fire, it does not pre- C concentration of contaminant at time θ
=
a =
dilution rate, air changes per minute
vent smoke movement through the supply and return air ducts, air
t =
time after smoke stops entering space or smoke production has
shafts, and other building openings because of stack effect, buoy- stopped, min
ancy, or wind. e = base of natural logarithm (approximately 2.718)

SMOKE MANAGEMENT Concentrations Co and C must be expressed in the same units, but
can be any units appropriate for the particular contaminant being
In this chapter, smoke management includes all methods that can considered.
be used singly or in combination to modify smoke movement for the McGuire et al. (1970) evaluated the maximum levels of smoke
benefit of occupants or firefighters or for reducing property damage. obscuration from a number of fire tests and a number of proposed cri-
Barriers, smoke vents, and smoke shafts are traditional methods of teria for tolerable levels of smoke obscuration. Based on this evalu-
smoke management. The effectiveness of barriers is limited by the ation, they state that the maximum levels of smoke obscuration are
extent to which they are free of leakage paths. Smoke vents and greater by a factor of 100 than those relating to the limit of tolerance.
smoke shafts are limited by the fact that smoke must be sufficiently Thus, they indicate that a space can be considered “reasonably safe”
buoyant to overcome any other driving forces that could be present. with respect to smoke obscuration if the concentration of contami-
In the last few decades, fans have been used to overcome the limi- nants in the space is less than about 1% of the concentration in the
tations of traditional approaches. Compartmentation, dilution, pres- immediate fire area. This level of dilution increases visibility by
surization, airflow, and buoyancy are used by themselves or in about a factor of 100 (e.g., from 0.15 m to 15 m) and reduces the con-
combination to manage smoke conditions in fire situations. These centrations of toxic smoke components. Toxicity is a more complex
mechanisms are discussed in the following sections. problem, and no parallel statement has been made regarding dilution
needed to obtain a safe atmosphere with respect to toxic gases.
Compartmentation In reality, it is impossible to ensure that the concentration of the
Barriers with sufficient fire endurance to remain effective through- contaminant is uniform throughout the compartment. Because of
out a fire exposure have long been used to protect against fire spread. buoyancy, it is likely that higher concentrations are near the ceiling.
In this approach, walls, partitions, floors, doors, and other barriers Therefore, exhausting smoke near the ceiling and supplying air near
provide some level of smoke protection to spaces remote from the the floor probably dilutes smoke even more quickly than indicated
fire. This section discusses passive compartmentation; using by Equation (7). Supply and exhaust points should be placed to
compartmentation with pressurization is discussed in the section on prevent supply air from blowing into the exhaust inlet, thereby
Pressurization (Smoke Control). Many codes, such as NFPA Stan- short-circuiting the dilution.
Fire and Smoke Management 52.5

Example 1. A space is isolated from a fire by smoke barriers and self-

closing doors, so that no smoke enters the compartment when the doors
Q = 0.839A ∆p (9)
are closed. When a door is opened, smoke flows through the open door- The flow area is frequently the same as the cross-sectional area
way into the space. If the door is closed when the contaminant in the of the flow path. A closed door with a crack area of 0.01 m2 and a
space is 20% of the burn room concentration, what dilution rate is
pressure difference of 2.5 Pa has an air leakage rate of approxi-
required to reduce the concentration to 1% of that in the burn room in
6 min? mately 0.013 m3/s. If the pressure difference across the door is
The time t = 6 min and Co /C = 20. From Equation (7), the dilution increased to 75 Pa, the flow is 0.073 m3/s.
rate is about 0.5 air changes per minute, or 30 air changes per hour. Frequently, in field tests of smoke control systems, pressure dif-
ferences across partitions or closed doors have fluctuated by as
Caution about Dilution near Fire. Many people have unrealis- much as 5 Pa. These fluctuations have generally been attributed to
tic expectations about what dilution can accomplish in the fire wind, although they could have been due to the HVAC system or
space. Neither theoretical nor experimental evidence indicates that some other source. To control smoke movement, the pressure dif-
using a building’s HVAC system for smoke dilution will signifi- ference produced by a smoke control system must be large enough
cantly improve tenable conditions in a fire space. The exception is to overcome pressure fluctuations, stack effect, smoke buoyancy,
an unusual space where the fuel is such that fire size cannot grow and wind pressure. However, the pressure difference should not be
above a specific limit; this occurs in some tunnels and underground so large that the door is difficult to open.
transit situations. Because HVAC systems promote a considerable
degree of air mixing in the spaces they serve and because very large Airflow
quantities of smoke can be produced by building fires, it is generally Airflow has been used extensively to manage smoke from fires in
believed that smoke dilution by an HVAC system in the fire space subway, railroad, and highway tunnels (see Chapter 13). Large air-
does not improve tenable conditions in that space. Thus, any attempt flow rates are needed to control smoke flow, and these flow rates can
to improve hazard conditions in the fire space, or in spaces con- supply additional oxygen to the fire. Because of the need for com-
nected to the fire space by large openings, with smoke purging will plex controls, airflow is not used as extensively in buildings. The
be ineffective. control problem consists of having very small flows when a door is
closed and then significantly increased flows when that door is open.
Pressurization (Smoke Control) Furthermore, it is a major concern that the airflow supplies oxygen
Systems that pressurize an area using mechanical fans are to the fire. This section presents the basics of smoke control by air-
referred to as smoke control in this chapter and in NFPA Standard flow and demonstrates why this technique is rarely recommended.
92A. A pressure difference across a barrier can control smoke Thomas (1970) determined that in a corridor in which there is a
movement, as illustrated in Figure 6. Within the barrier is a door. fire, airflow can almost totally prevent smoke from flowing up-
The high-pressure side of the door can be either a refuge area or an stream of the fire. Molecular diffusion is believed to transfer trace
egress route. The low-pressure side is exposed to smoke from a fire. amounts of smoke, which are not hazardous but which are detect-
Airflow through gaps around the door and through construction able as the smell of smoke upstream. Based on work by Thomas, the
cracks prevents smoke infiltration to the high-pressure side. critical air velocity for most applications can be approximated as
For smoke control analysis, the orifice equation can be used to
estimate the flow through building flow paths: ⎛ q ⎞1⁄ 3
Vk = 0.0292 ⎜------c-⎟ (10)
Q = CA 2∆p ⁄ ρ (8) ⎝W ⎠
where where
Q = volumetric airflow rate, m3/s Vk = critical air velocity to prevent smoke backflow, m/s
C = flow coefficient qc = heat release rate into corridor, W
A = flow area (leakage area), m2 W = corridor width, m
∆p = pressure difference across flow path, Pa This relation can be used when the fire is in the corridor or
ρ = density of air entering flow path, kg/m3 when smoke enters the corridor through an open doorway, air
The flow coefficient depends on the geometry of the flow path, as transfer grille, or other opening. Although critical velocities calcu-
well as on turbulence and friction. In the present context, the flow lated from Equation (10) are general and approximate, they indi-
coefficient is generally 0.6 to 0.7. For ρ = 1.2 kg/m3 and C = 0.65, cate the kind of air velocities required to prevent smoke backflow
Equation (8) can be expressed as from fires of different sizes. For specific applications, other equa-
tions may be more appropriate: for tunnel applications, see Chap-
Fig. 6 Smoke Control System Preventing Smoke Infiltration ter 13; for smoke management in atriums and other large spaces,
to High-Pressure Side of Barrier see Klote and Milke (2002) and NFPA Standard 92B.
Although Equation (10) can be used to estimate the airflow rate
necessary to prevent smoke backflow through an open door, the
oxygen supplied is a concern. Huggett (1980) evaluated the oxygen
consumed in the combustion of numerous natural and synthetic sol-
ids. He found that, for most materials involved in building fires, the
energy released is approximately 13.1 MJ per kilogram of oxygen.
Air is 23.3% oxygen by mass. Thus, if all the oxygen in a kilogram
of air is consumed, 3.0 MJ is liberated. If all the oxygen in 1 m3/s of
air with a density of 1.2 kg/m3 is consumed by fire, 3.6 MW is lib-
erated.
Examples 2 and 3 demonstrate that the air needed to prevent
smoke backflow can support an extremely large fire. Most com-
mercial and residential buildings contain enough fuel (paper, card-
board, furniture, etc.) to support very large fires. Even when the
Fig. 6 Smoke Control System Preventing Smoke Infiltration amount of fuel is normally very small, short-term fuel loads (dur-
to High-Pressure Side of Barrier ing building renovation, material delivery, etc.) can be significant.
52.6 2007 ASHRAE Handbook—HVAC Applications (SI)

Therefore, using airflow for smoke control is not recommended, Fig. 7 Door-Opening Force Caused by Pressure Difference
except when the fire is suppressed or in the rare cases when fuel
can be restricted with confidence.

Example 2. What airflow at a doorway is needed to stop smoke backflow

from a room fully involved in fire, and how large a fire can this airflow
support?
A room fully involved in fire can have an energy release rate on the
order of 2.4 MW. Assume the door is 0.9 m wide and 2.1 m high. From
Equation (10), Vk = 0.0292(2.4 × 106/0.9)1/3 = 4.0 m/s. A flow through
the doorway of 4.0 × 0.9 × 2.1 = 7.6 m3/s is needed to prevent smoke
from backflowing into the area.
If all the oxygen in this airflow is consumed in the fire, the heat lib-
erated is 7.6 m3/s × 3.6 MW per m3/s of air = 27 MW. This is over 10
times more than the heat generated by the fully involved room fire and
indicates why airflow is generally not recommended for smoke control
in buildings.

Example 3. What airflow is needed to stop smoke backflow from a waste-

basket fire, and how large a fire can this airflow support?
A wastebasket fire can have an energy release rate on the order of
150 kW. As in Example 2, Vk = 0.0292(150 × 103/0.9)1/3 = 1.6 m/s. A
flow through the doorway of 1.6 × 0.9 × 2.1 = 3.0 m3/s is needed to pre-
vent smoke backflow.
If all the oxygen in this airflow is consumed in the fire, the heat lib- Fig. 7 Door-Opening Force Caused by Pressure Difference
erated is 3 m3/s × 3.6 MW per m3/s of air = 10.8 MW. This is still many
times greater than the fully involved room fire and further indicates Flow Areas
why airflow is generally not recommended for smoke control in build-
ings. In designing smoke control systems, airflow paths must be iden-
tified and evaluated. Some leakage paths are obvious, such as cracks
Buoyancy around closed doors, open doors, elevator doors, windows, and air
transfer grilles. Construction cracks in building walls are less obvi-
The buoyancy of hot combustion gases is used in both fan- ous, but they are equally important.
powered and non-fan-powered venting systems. Fan-powered The flow area of most large openings, such as open windows, can
venting for large spaces is commonly used for atriums and cov- be calculated easily. However, flow areas of cracks are more difficult
ered shopping malls, and non-fan-powered venting is commonly to evaluate. The area of these leakage paths depends on such features
used for large industrial and storage buildings. There is a concern as workmanship, door fit, and weatherstripping. A 0.9 by 2.1 m door
that sprinkler flow will cool the smoke, reducing buoyancy and
with an average crack width of 3 mm has a leakage area of 0.018 m2.
thus the system effectiveness. Research is needed in this area. However, if this door is installed with a 20 mm undercut, the leakage
Refer to Klote and Milke (2002) and NFPA Standards 92B and area is 0.033 m2, a significant difference. The leakage area of eleva-
204 for detailed design information about these systems. tor doors is in the range of 0.051 to 0.065 m2 per door.
For open stairwell doorways, Cresci (1973) found complex flow
SMOKE CONTROL SYSTEM DESIGN patterns; the resulting flow through open doorways was considerably
Door-Opening Forces below that calculated using the doorway’s geometric area as the flow
area in Equation (8). Based on this research, it is recommended that
The door-opening forces resulting from the pressure differences the design flow area of an open stairwell doorway be half the geomet-
produced by a smoke control system must be considered. Unreason- ric area (door height × width) of the doorway. An alternative for open
ably high door-opening forces can make it difficult or impossible for stairwell doorways is to use the geometric area as the flow area and
occupants to open doors to refuge areas or escape routes. use a reduced flow coefficient. Because it does not allow the direct use
The force required to open a door is the sum of the forces to over- of Equation (8), this approach is not used here.
come the pressure difference across the door and to overcome the Typical leakage areas for walls and floors of commercial build-
door closer. This can be expressed as ings are tabulated as area ratios in Table 1. These data are based on
WA∆p a relatively small number of tests performed by the National
F = Fdc + ---------------------- (11) Research Council of Canada (Shaw et al. 1993; Tamura and Shaw
2 (W – d ) 1976a, 1976b, 1978; Tamura and Wilson 1966). Actual leakage areas
where depend primarily on workmanship rather than on construction mate-
F = total door-opening force, N rials, and in some cases, the flow areas in particular buildings may
Fdc = force to overcome door closer, N vary from the values listed. Data concerning air leakage through
W = door width, m building components are also provided in Chapter 27 of the 2005
A = door area, m2 ASHRAE Handbook—Fundamentals.
∆p = pressure difference across door, Pa Because a vent surface is usually covered by a louver and screen, a
d = distance from doorknob to edge of knob side of door, m vent’s flow area is less than its area (vent height × width). Calculation
This relation assumes that the door-opening force is applied at is further complicated because the louver slats are frequently slanted.
the knob. Door-opening force Fp caused by pressure difference can Manufacturer’s data should be used for specific information.
be determined from Figure 7 for a value of d = 75 mm. The force to
overcome the door closer is usually greater than 13 N and, in some Effective Flow Areas
cases, can be as great as 90 N. For a door that is 2.1 m high and 0.9 m The concept of effective flow areas is useful for analyzing smoke
wide and subject to a pressure difference of 75 Pa, the total door- control systems. The paths in the system can be in parallel with one
opening force is 130 N, if the force to overcome the door closer is another, in series, or a combination of parallel and series. The effec-
53 N. tive area of a system of flow areas is the area that gives the same
Fire and Smoke Management 52.7

Table 1 Typical Leakage Areas for Walls and Floors of Fig. 8 Leakage Paths in Series
Commercial Buildings
Wall
Construction Element Tightness Area Ratio
A/Aw
Exterior building wallsa Tight 0.50 × 10−4
(includes construction cracks and Average 0.17 × 10−3
cracks around windows and doors) Loose 0.35 × 10−3
Very Loose 0.12 × 10−2
Stairwell wallsa Tight 0.14 × 10−4
(includes construction cracks but not Average 0.11 × 10−3
cracks around windows or doors) Loose 0.35 × 10−3
Elevator shaft wallsa Tight 0.18 × 10−3
(includes construction cracks but Average 0.84 × 10−3 Fig. 8 Leakage Paths in Series
not cracks around doors) Loose 0.18 × 10−2
Fig. 9 Leakage Paths in Parallel
A/Af
Floorsb Tight 0.66 × 10−5
(includes construction cracks and Average 0.52 × 10−4
gaps around penetrations) Loose 0.17 × 10−3
A = leakage area; Aw = wall area; Af = floor area
Leakage areas evaluated at a75 Pa; b25 Pa.

flow as the system when it is subjected to the same pressure differ-

ence over the total system of flow paths. This is similar to the effec-
tive resistance of a system of electrical resistances. The effective
flow area Ae for parallel leakage areas is the sum of the individual
leakage paths:
n
Ae = ∑ Ai (12)
i=1

where n is the number of flow areas Ai in parallel. Fig. 9 Leakage Paths in Parallel
For example, the effective area Ae for the three parallel leakage
areas in Figure 8 is A1 A2
Ae = ------------------------ = 0.0199 m 2
Ae = A1 + A2 + A3 (13) A 12 + A 22

If A1 is 0.10 m2 and A2 and A3 are each 0.05 m2, then the effective Example 5 illustrates that when two paths are in series, and one
flow area Ae is 0.20 m2. is much larger than the other, the effective flow area is approxi-
The general rule for any number of leakage areas in series is mately equal to the smaller area.
n – 0.5
Developing an effective area for a system of both parallel and
Ae = ∑ -----12- (14)
series paths requires combining groups of parallel paths and series
paths systematically. The system illustrated in Figure 10 is analyzed
i=1 A i
as an example. The figure shows that A2 and A3 are in parallel; there-
where n is the number of leakage areas Ai in series. fore, their effective area is
Three leakage areas in series from a pressurized space are illus-
trated in Figure 9. The effective flow area of these paths is ( A23 )e = A 2 + A 3
⎛1 1 1 ⎞
– 0.5
Areas A4, A5, and A6 are also in parallel, so their effective area is
A e = ⎜ -----2- + -----2- + -----2- ⎟ (15)
⎝ A1 A2 A3 ⎠
( A 456 )e = A 4 + A 5 + A 6
In smoke control analysis, there are frequently only two paths in
series, and the effective leakage area is These two effective areas are in series with A1. Therefore, the
effective flow area of the system is given by
A1 A2
A e = ------------------------ (16) – 0.5
A 12 + A 22 1 1 1
A e = -----2- + ---------------2 + -----------------2-
A1 ( A ) (A )
Example 4. Calculate the effective leakage area of two equal flow paths in 23 e 456 e
series. Let A = A1 = A2 = 0.02 m2. From Equation (16),
Example 6. Calculate the effective area of the system in Figure 10, if the
A2 leakage areas are A1 = A2 = A3 = 0.02 m2 and A4 = A5 = A6 = 0.01 m2.
Ae = --------------- = 0.014 m 2
2A 2 (A23)e = 0.04 m2
Example 5. Calculate the effective flow area of two flow paths in series, (A456)e = 0.03 m2
where A1 = 0.02 m2 and A2 = 0.2 m2. From Equation (16), Ae = 0.015 m2
52.8 2007 ASHRAE Handbook—HVAC Applications (SI)

Fig. 10 Combination of Leakage Paths in Parallel and Series Fig. 11 Smoke Control System Damper Recommendation

Fig. 11 Smoke Control System Damper Recommendation

Fig. 10 Combination of Leakage Paths in Parallel and Series

The number of doors that may be open simultaneously depends

largely on building occupancy. For example, in a densely populated
Design Weather Data building, it is likely that all doors will be open during evacuation.
Little weather information has been developed specifically for However, if a staged evacuation plan or refuge area concept is incor-
smoke control system design. Design temperatures for heating and porated in the building fire emergency plan, or if the building is
cooling found in Chapter 28 of the 2005 ASHRAE Handbook—Fun- sparsely occupied, only a few of the doors may be open during a fire.
damentals may be used. Extreme temperatures can be considerably
lower than the winter design temperatures. For example, the 99% FIRE AND SMOKE DAMPERS
design temperature for Tallahassee, Florida, is −2.1°C, but the lowest
temperature observed there was −19°C (NOAA 1979). Openings for ducts in walls and floors with fire resistance ratings
Temperatures are generally below the design values for short should be protected by fire dampers and ceiling dampers, as
periods, and because of the thermal lag of building materials, these required by local codes. Air transfer openings should also be pro-
short intervals of low temperature usually do not cause problems tected. These dampers should be classified and labeled in accor-
with heating. However, there is no time lag for a smoke control sys- dance with Underwriters Laboratories (UL) Standard 555. Figure
tem; it is therefore subjected to all the extreme forces of stack effect 11 shows recommended damper positions for smoke control.
that exist the moment it operates. If the outside temperature is below A smoke damper can be used for either traditional smoke man-
the system’s winter design temperature, stack effect problems may agement (smoke containment) or smoke control. In smoke man-
result. A similar situation can occur with summer design tempera- agement, a smoke damper inhibits passage of smoke under the
tures and reverse stack effect. forces of buoyancy, stack effect, and wind. However, smoke damp-
Extreme wind data for smoke management design are listed in ers are only one of many elements (partitions, floors, doors)
Chapter 28 of the 2005 ASHRAE Handbook—Fundamentals. intended to inhibit smoke flow. In smoke management applications,
the leakage characteristics of smoke dampers should be selected to
Design Pressure Differences be appropriate with the leakage of the other system elements.
Both the maximum and minimum allowable pressure differences In a smoke control system, a smoke damper inhibits the passage
across the boundaries of smoke control should be considered. The of air that may or may not contain smoke. A damper does not need
maximum allowable pressure difference should not cause excessive low leakage characteristics when outside (fresh) air is on the high-
door-opening forces. pressure side of the damper, as is the case for dampers that shut off
The minimum allowable pressure difference across a boundary supply air from a smoke zone or that shut off exhaust air from a non-
of a smoke control system might be the difference such that no smoke zone. In these cases, moderate leakage of smoke-free air
smoke leakage occurs during building evacuation. In this case, the through the damper does not adversely affect control of smoke move-
smoke control system must produce sufficient pressure differences ment. It is best to design smoke control systems so that only
to overcome forces of wind, stack effect, or buoyancy of hot smoke. smoke-free air is on the high-pressure side of a closed smoke damper.
Pressure differences caused by wind and stack effect can be large in Smoke dampers should be classified and listed in accordance
the event of a broken window in the fire compartment. Evaluation of with UL Standard 555S for temperature, leakage, and operating
these pressure differences depends on evacuation time, rate of fire velocity. The velocity rating of a smoke damper is the velocity at
growth, building configuration, and the presence of a fire suppres- which the actuator will open and close the damper.
sion system. NFPA Standard 92A suggests values of minimum and At locations requiring both smoke and fire dampers, combina-
maximum design pressure difference. tion dampers meeting the requirements of both UL Standards 555
and 555S can be used. The combination fire/smoke dampers must
Open Doors close when they reach their UL Standard 555S temperature rating to
Another design concern is the number of doors that could be maintain the integrity of the firewall.
opened simultaneously when the smoke control system is operat- Fire, ceiling, and smoke dampers should be installed in accor-
ing. A design that allows all doors to be open simultaneously may dance with the manufacturers’ instructions. NFPA Standard 90A
ensure that the system always works, but often adds to system cost. gives general guidelines on locations requiring these dampers.
Fire and Smoke Management 52.9

The supply and return/smoke dampers should be a minimum of For tall stairwells, single-injection systems can fail when a few
Class II leakage at 120°C. The return air damper should be a mini- doors are open near the air supply injection point, especially in bottom-
mum of Class I leakage at 120°C to prevent recirculation of smoke injection systems when a ground-level stairwell door is open.
exhaust. The operating velocity of the dampers should be evaluated For tall stairwells, supply air can be supplied at a number of loca-
when the dampers are in smoke control mode. To minimize velocity tions over the height of the stairwell. Figures 12 and 13 show two
build-up, only zones adjacent to the fire need to be pressurized. examples of multiple-injection systems that can be used to over-
The exhaust ductwork and fan must be designed to handle the come the limitations of single-injection systems. In these figures, the
temperature of the exhaust smoke. This temperature can be lowered supply duct is shown in a separate shaft. However, systems have been
by making the smoke control zones large or by pressurizing only the built that eliminated the expense of a separate duct shaft by locating
zones adjacent to the fire zone and leaving all the other zones oper- the supply duct in the stairwell itself. In such a case, care must be
ating normally. taken that the duct does not obstruct orderly building evacuation.
Fans Used to Exhaust Smoke Stairwell Compartmentation
Understanding building code requirements for high-temperature Compartmentation of the stairwell into a number of sections is
fans in smoke control systems is important for both designers, who one alternative to multiple injection (Figure 14). When the doors
must select fans that can operate satisfactorily at elevated tempera- between compartments are open, the effect of compartmentation is
tures, and manufacturers, who can then design suitable off-the-shelf lost. For this reason, compartmentation is inappropriate for densely
fans rather than customizing fans for each application. Only fans populated buildings where total building evacuation by the stairwell
designed for use under elevated temperatures should be used in is planned in the event of fire. However, when a staged evacuation
smoke management applications; other types may fail, or their per- plan is used and the system is designed to operate successfully with
formance may change because of component deformation or altered the maximum number of doors between compartments open, com-
clearances among components. Also, some smoke exhaust applica- partmentation can effectively pressurize tall stairwells.
tions (e.g., transit tunnels) require that smoke-handling fans reverse
direction repeatedly on demand. Until recently, standards did not
address reversibility or airflow performances of high-temperature Fig. 12 Stairwell Pressurization by Multiple Injection with
fans at ambient and elevated temperatures. To allow manufacturers Fan Located at Ground Level
to provide suitable off-the-shelf products, a standard method of
test (MOT) and ratings scale have been developed.
ANSI/ASHRAE Standard 149 provides testing laboratories with
standard testing methods for fan characteristics specific to smoke
exhaust functions, including (1) aerodynamic performance, (2) oper-
ation at specified elevated temperature, (3) reversal, and (4) damper
performance (for dampers included with the fan).
AMCA Publication 212 establishes ratings to allow consistent
comparison among catalog test data. Model code requirements for
elevated temperature and duration of operation are charted on a graph,
which is divided into several fan performance groups. Manufacturers
can request that laboratories test fans according to ANSI/ASHRAE
Standard 149; those data can then be incorporated into catalogs for
off-the-shelf products according to AMCA Publication 212 ratings,
allowing designers to select the most appropriate models and perfor-
mances for their specific applications. This allows designers and code
officials to compare different manufacturers’ products more easily,
and enhances confidence that products will perform as intended; it Fig. 12 Stairwell Pressurization by Multiple Injection with
also allows manufacturers to provide more cost-efficient off-the-shelf Fan Located at Ground Level
products rather than custom-designing fans for each application.
Fig. 13 Stairwell Pressurization by Multiple Injection
PRESSURIZED STAIRWELLS with Roof-Mounted Fan

Many pressurized stairwells have been designed and built to pro-

vide a smoke-free escape route in the event of a building fire. They
also provide a smoke-free staging area for firefighters. On the fire
floor, a pressurized stairwell must maintain a positive pressure dif-
ference across a closed stairwell door to prevent smoke infiltration.
During building fires, some stairwell doors are opened intermit-
tently during evacuation and fire fighting, and some doors may even
be blocked open. Ideally, when the stairwell door is opened on the
fire floor, airflow through the door should be sufficient to prevent
smoke backflow. Designing a system to achieve this goal is difficult
because of the many combinations of open stairwell doors and
weather conditions affecting airflow.
Stairwell pressurization systems may be single- or multiple-
injection systems. A single-injection system supplies pressurized
air to the stairwell at one location, usually at the top. Associated
with this system is the potential for smoke to enter the stairwell
through the pressurization fan intake. Therefore, automatic shut- Fig. 13 Stairwell Pressurization by Multiple Injection with
down during such an event should be considered. Roof-Mounted Fan
52.10 2007 ASHRAE Handbook—HVAC Applications (SI)

Fig. 14 Compartmentation of Pressurized Stairwell Outside design temperature To = 263 K; stairwell temperature Ts =
294 K; maximum design pressure differences when all stairwell doors
are closed is 87 Pa; the minimum allowable pressure difference is
13 Pa.
Using the leakage ratio for an exterior building wall of average
tightness from Table 1, Abo = 560(0.21 × 10−3) = 0.118 m2. Using the
leakage ratio for a stairwell wall of average tightness from Table 1, the
leakage area of the stairwell wall is 52(0.11 × 10−3) = 0.006 m2. The
value of Asb equals the leakage area of the stairwell wall plus the gaps
around the closed doors: Asb = 0.006 + 0.024 = 0.030 m2. The tempera-
ture factor B is calculated at 1.39 Pa/m. The pressure difference at the
stairwell bottom is selected as ∆ psbb = 20 Pa to provide an extra degree
of protection above the minimum allowable value of 13 Pa. The pres-
sure difference ∆ psbt is calculated from Equation (17) at 84.6 Pa, using
y = 15(3.3) = 49.5 m. Thus, ∆ psbt does not exceed the maximum allow-
able pressure. The flow rate of pressurization air is calculated from
Equation (18) at 2.7 m3/s.

The flow rate depends strongly on the leakage area around the
closed doors and on the leakage area in the stairwell walls. In prac-
tice, these areas are difficult to evaluate and even more difficult to
control. If flow area Asb in Example 7 were 0.050 m2 rather than
Fig. 14 Compartmentation of Pressurized Stairwell 0.030 m2, Equation (18) would give a flow rate of pressurization air
of 3.1 m3/s. A fan with a sheave allows adjustment of supply air to
Stairwell Analysis offset for variations in actual leakage from the values used in design
This section presents an analysis for a pressurized stairwell in a calculations.
building without vertical leakage. This method closely approximates
the performance of pressurized stairwells in buildings without eleva- Stairwell Pressurization and Open Doors
tors. It is also useful for buildings with vertical leakage because it The simple pressurization system discussed in the previous sec-
yields conservative results. For evaluating vertical leakage through tion has two limitations regarding open doors. First, when a stair-
the building or with open stairwell doors, computer analysis is rec- well door to the outside and building doors are open, the simple
ommended. The analysis is for buildings where the leakage areas are system cannot provide enough airflow through building doorways
the same for each floor of the building and where the only significant to prevent smoke backflow. Second, when stairwell doors are open,
driving forces are the stairwell pressurization system and the indoor- pressure difference across the closed doors can drop to low levels.
outdoor temperature difference. Two systems used to overcome these problems are overpressure
The pressure difference ∆psb between the stairwell and the build- relief (Tamura 1990) and supply fan bypass.
ing can be expressed as Overpressure Relief. The total airflow rate is selected to provide
By the minimum air velocity when a specific number of doors are open.
∆psb = ∆psbb + -------------------------------------- (17) When all the doors are closed, part of this air is relieved through a
1 + ( Asb ⁄ Abo ) 2 vent to prevent excessive pressure build-up, which could cause
excessive door-opening forces. This excess air should be vented
where from the stairwell to the street-level floor. Fire and relief dampers
∆ psbb = pressure difference between stairwell and building at stairwell should be the low-leakage type. Stairwell doors should have gasket
bottom, Pa
seals at sides and top, leaving the bottom gap open for relief.
B = 3460(1/To − 1/Ts) at sea level standard pressure
y = distance above stairwell bottom, m Barometric dampers that close when pressure drops below a
Asb = flow area between stairwell and building (per floor), m2 specified value can minimize air loss through the vent when doors
Abo = flow area between building and outside (per floor), m2 are open. Figure 15 illustrates a pressurized stairwell with overpres-
To = temperature of outside air, K sure relief vents to the building at each floor. In systems with vents
Ts = temperature of stairwell air, K between stairwell and building, the vents typically have a fire
damper in series with the barometric damper. To conserve energy,
For a stairwell with no leakage directly to the outside, the flow these fire dampers are normally closed, but they open when the pres-
rate of pressurization air is surization system is activated. This arrangement also reduces the
possibility of the annoying damper chatter that frequently occurs
⎛∆p 3 ⁄ 2 – ∆p 3 ⁄ 2 ⎞
with barometric dampers.
Q = 0.559NAsb ⎜------------------------------------ ⎟
sbt sbb
(18)
⎜ ∆p sbt – ∆p sbb ⎟ An exhaust duct can provide overpressure relief in a pressurized
⎝ ⎠ stairwell. The system is designed so that the normal resistance of a
where nonpowered exhaust duct maintains pressure differences within the
Q = volumetric flow rate, m3/s design limits.
N = number of floors Exhaust fans can also relieve excess pressure when all stairwell
∆ psbt = pressure difference from stairwell to building at stairwell top, Pa doors are closed. An exhaust fan should be controlled by a differen-
tial pressure sensor, so that it will not operate when the pressure dif-
Example 7. Each story of a 15-story stairwell is 3.3 m high. The stairwell ference between stairwell and building falls below a specified level.
has a single-leaf door at each floor leading to the occupant space and This control should prevent the fan from pulling smoke into the
one ground-level door to the outside. The exterior of the building has a
wall area of 560 m2 per floor. The exterior building walls and stairwell
stairwell when a number of open doors have reduced stairwell pres-
walls are of average leakiness. The stairwell wall area is 52 m2 per surization. The exhaust fan should be specifically sized so that the
floor. The area of the gap around each stairwell door to the building is pressurization system performs within design limits. A wind shield
0.024 m2. The exterior door is well gasketed, and its leakage can be is recommended because an exhaust fan can be adversely affected
neglected when it is closed. by the wind.
Fire and Smoke Management 52.11

Fig. 15 Stairwell Pressurization with Vents to Fig. 16 Stairwell Pressurization with Bypass
Building at Each Floor Around Supply Fan

Fig. 15 Stairwell Pressurization with Vents to Fig. 16 Stairwell Pressurization with Bypass Around
Building at Each Floor Supply Fan

An alternative method of venting a stairwell is through an auto- major problem was maintaining pressurization with open building
matically opening stairwell door to the outside at ground level. doors, especially doors on the ground floor. Of the systems evalu-
Under normal conditions, this door would be closed and, in most ated, only one with a supply fan bypass with feedback control main-
cases, locked for security reasons. Provisions are needed to prevent tained adequate pressurization with any combination of open or
this lock from conflicting with the automatic operation of the sys- closed doors. There are probably other systems capable of providing
tem. Possible adverse wind effects are also a concern with a system adequate smoke control; the procedure used by Klote and Tamura
that uses an open outside door as a vent. Occasionally, high local can be viewed as an example of a method of evaluating the perfor-
wind velocities develop near the exterior stairwell door; such winds mance of a system to determine whether it suits the particular char-
are difficult to estimate without expensive modeling. Nearby acteristics of a building under construction.
obstructions can act as windbreaks or wind shields. Transient pressures caused by piston effect when an elevator car
Supply Fan Bypass. In this system, the supply fan is sized to moves in a shaft have been a concern in elevator smoke control. Pis-
provide at least the minimum air velocity when the design number ton effect is not a concern for slow-moving cars in multiple-car
of doors are open. Figure 16 illustrates such a system. The flow rate shafts, but can be considerable for fast cars in single-car shafts.
of air into the stairwell is varied by modulating bypass dampers,
which are controlled by one or more static pressure sensors that
ZONE SMOKE CONTROL
sense the pressure difference between the stairwell and the building.
When all the stairwell doors are closed, the pressure difference Klote (1990) conducted a series of tests on full-scale fires that
increases and the bypass damper opens to increase the bypass air demonstrated that zone smoke control can restrict smoke movement
and decrease the flow of supply air to the stairwell. In this manner, to the zone where a fire starts.
excessive stairwell pressures and excessive pressure differences Pressurized stairwells are intended to prevent smoke infiltration
between the stairwell and the building are prevented. into stairwells. However, in a building with only stairwell pressur-
ization, smoke can flow through cracks in floors and partitions and
ELEVATORS through shafts to damage property and threaten life at locations
Elevator smoke control systems intended for use by firefighters remote from the fire. Zone smoke control is intended to limit this
should keep elevator cars, elevator shafts, and elevator machinery smoke movement.
rooms smoke-free. Small amounts of smoke in these spaces are A building is divided into a number of smoke control zones, each
acceptable, provided that the smoke is nontoxic and that operation separated from the others by partitions, floors, and doors that can be
of elevator equipment is not affected. Elevator smoke control sys- closed to inhibit smoke movement. In the event of a fire, pressure
tems intended for fire evacuation of people unable to self-rescue or differences and airflows produced by mechanical fans limit spread
other building occupants should also keep elevator lobbies smoke- of smoke from the zone in which the fire started. The concentration
free or nearly smoke-free. Obstacles to fire evacuation by eleva- of smoke in this zone goes unchecked; thus, in zone smoke control
tors include systems, occupants should evacuate the smoke zone as soon as pos-
sible after fire detection.
• Logistics of evacuation
A smoke control zone can consist of one floor, more than one
• Reliability of electrical power
floor, or part of a floor. Sprinkler zones and smoke control zones
• Jamming of elevator doors
should be coordinated so that sprinkler water flow activates the
• Fire and smoke protection zone’s smoke control system. Some arrangements of smoke control
All these obstacles, except smoke protection, can be addressed by zones are illustrated in Figure 17. All the nonsmoke zones in the
existing technology (Klote 1984). building may be pressurized. The term pressure sandwich describes
Klote and Tamura (1986) studied conceptual elevator smoke cases where only zones adjacent to the smoke zone are pressurized,
control systems for evacuation of people unable to self-rescue. The as in Figures 17B and 17D.
52.12 2007 ASHRAE Handbook—HVAC Applications (SI)

Fig. 17 Some Arrangements of Smoke Control Zones the performance of smoke control systems. However, ASCOS was
intended as a research tool for application to 10- and 20-story
buildings. Not surprisingly, convergence failures have been en-
countered with applications to much larger buildings.
Wray and Yuill (1993) evaluated several flow algorithms to find
the most appropriate one for analysis of smoke control systems. They
selected the AIRNET flow routine developed by Walton (1989) as
the best algorithm based on computational speed and use of com-
puter memory. None of the algorithms from this study takes advan-
tage of the repetitive nature of building flow networks, so data entry
is difficult. However, Walton and Dols (2005) developed CONTAM,
a public domain program with an improved version of the AIR-
NET flow routine and an easier method of input. CONTAM can be
downloaded free of charge from http://www.bfrl.nist.gov/IAQ
analysis/CONTAM.
These models are appropriate for analyzing systems that use
pressurization to control smoke flow. For systems that rely on
buoyancy of hot smoke (such as atrium smoke exhaust), zone fire
models are appropriate. The concepts behind zone fire modeling
are discussed by Bukowski (1991), Jones (1983), and Mitler
(1985). Some frequently used zone models are ASET (Cooper
1985), CCFM (Cooper and Forney 1987), and CFAST (Peacock
et al. 1993). Milke and Mowrer (1994) enhanced the CCFM model
for atrium applications.

SMOKE MANAGEMENT IN LARGE SPACES

In recent years, atrium buildings have become commonplace.
Other large, open spaces include enclosed shopping malls, arcades,
sports arenas, exhibition halls, and airplane hangars. For simplic-
ity, the term atrium is used in this chapter in a generic sense to
mean any of these large spaces. Traditional fire protection by com-
partmentation is not applicable to these large-volume spaces.
Most atrium smoke management systems are designed to prevent
exposure of occupants to smoke during evacuation; this is the
Fig. 17 Some Arrangements of Smoke Control Zones approach described in this section. An alternative goal is to maintain
tenable conditions even when occupants have some contact with
Zone smoke control is intended to limit smoke movement to the smoke, as discussed in the section on Tenability Systems.
smoke zone by the use of pressurization. Pressure differences in The following approaches can be used to manage smoke in
the desired direction across the barriers of a smoke zone can be atriums:
achieved by supplying outside (fresh) air to nonsmoke zones, by
venting the smoke zone, or by a combination of these methods. • Smoke filling. This approach allows smoke to fill the atrium
Venting smoke from a smoke zone prevents significant over- space while occupants evacuate the atrium. It applies only to
pressure from thermal expansion of gases caused by the fire. This spaces where the smoke-filling time is sufficient for both decision
venting can be accomplished by exterior wall vents, smoke making and evacuation. Nelson and Mowrer (2002), Chapter 4 of
shafts, and mechanical venting (exhausting). However, venting Klote and Milke (2002), and Proulx (2002) have information on
only slightly reduces smoke concentration in the smoke zone. people movement during evacuation. The filling time can be esti-
mated either by zone fire models or by filling equations [e.g.,
COMPUTER ANALYSIS FOR Equation (21)].
PRESSURIZATION SYSTEMS • Unsteady clear height with upper layer exhaust. This approach
Because of the complex airflow in buildings, network com- exhausts smoke from the top of the atrium at a rate such that occu-
puter programs were developed to model the airflow with pres- pants have sufficient time for decision making and evacuation. It
surization systems. These models represent rooms and shafts by requires analysis of people movement and fire model analysis of
nodes; airflow is from nodes of high pressure to nodes of lower smoke filling.
pressure. Some programs calculate steady-state airflow and pres- • Steady clear height with upper layer exhaust. This approach
sures throughout a building (Sander 1974; Sander and Tamura exhausts smoke from the top of the atrium to achieve a steady
1973). Other programs go beyond this to calculate the smoke clear height for a steady fire (Figure 18). A calculation method is
concentrations that would be produced throughout a building in presented in the section on Steady Clear Height with Upper Layer
the event of a fire (Evers and Waterhouse 1978; Rilling 1978; Exhaust.
Wakamatsu 1977; Yoshida et al. 1979).
The ASCOS program was developed specifically for analyzing Design Fires
pressurization smoke control systems (Klote 1982). ASCOS was The design fire has a major effect on the atrium smoke manage-
the most widely used program for smoke control analysis (Said ment system. Fire size is expressed in terms of rate of heat release.
1988), and has been validated against field data from flow experi- Fire growth is the rate of change of the heat release rate and is
ments at an eight-story tower in Champs sur Marne, France (Klote sometimes expressed as a growth constant that identifies the time
and Bodart 1985). ASCOS and the other network models have required for the fire to attain a particular rate of heat release.
been used extensively for design and for parametric analysis of Designs may be based on either steady fires or unsteady fires.
Fire and Smoke Management 52.13

Fig. 18 Smoke Exhaust to Maintain Steady Clear Height Fig. 19 Room Fire and Zone Fire Model Idealization

Fig. 18 Smoke Exhaust to Maintain Steady Clear Height

Table 2 Steady Design Fire Sizes for Atriums

kW
Minimum fire for fuel-restricted atrium 2 000
Minimum fire for atrium with combustibles 5 000
Large fires 25 000

Fires are by nature unsteady, but the steady fire is a very useful
idealization. Steady fires have a constant heat release rate. In many Fig. 19 Room Fire and Zone Fire Model Idealization
applications, using a steady design fire leads to straightforward and
conservative design.
Table 3 Typical Fire Growth Times
Morgan (1979) suggests 500 kW/m2 as a typical rate of heat
release per unit floor area for mercantile occupancies. Fang and t-Squared Fires Growth Time tg , s
Breese (1980) found about the same rate of heat release for resi- Slowa 600
dential occupancies. Law (1982) and Morgan and Hansell (1987) Mediuma 300
suggest a heat release rate per unit floor area for office buildings
Fasta 150
of 225 kW/m2.
Ultrafastb 75
In many atriums, fuel loading is severely restricted with the
a Constants based on data from NFPA Standards 92B and 204.
intent of restricting fire size. Such atriums are characterized by inte- b Constant based on data from Nelson (1987).
rior finishes of metal, brick, stone, or gypsum board and furnished
with objects made of similar materials plus plants. Even in such a
fuel-restricted atrium, many combustible objects are present for These unsteady fires are called t-squared fires; typical growth
short periods. Packing materials, holiday decorations, displays, times are listed in Table 3.
construction materials, and furniture being moved into another part
of the building are a few examples of transient fuels. Zone Fire Models
In this chapter, a heat release rate per floor area of 225 kW/m2 is
used for a fuel-restricted atrium, and 500 kW/m2 is used for atri- Atrium smoke management design is based on the zone fire model
ums containing furniture, wood, or other combustible materials. concept. This concept has been applied to several computer models
Transient fuels must not be overlooked when selecting a design used for atrium smoke management design analysis, including the
fire. Klote and Milke (2002) suggest incorporating transient fuels Harvard Code (Mitler and Emmons 1981), ASET (Cooper 1985), the
in a design fire by considering the fire occurring over 9.3 m2 of BRI Model (Tanaka 1983), CCFM (Cooper and Forney 1987), and
floor space. This results in a design fire of 2100 kW for fuel- CFAST (Peacock et al. 1993). The University of Maryland modified
restricted atriums. In an atrium with combustibles, the design fire CCFM specifically for atrium smoke management design (Milke and
would be 4600 kW. However, the area involved in fire may be Mowrer 1994). Although each of these models has unique features,
much greater; flame spread considerations must be taken into they all share the same basic two-zone model concept.
account (Klote and Milke 2002; NFPA Standard 92B). A large For more information about zone models, see Mitler (1984),
atrium fire of 25 000 kW would involve an area of 50 m2 at Mitler and Rockett (1986), and Quintiere (1989). The ASET-B
500 kW/m. Table 2 lists some steady design fires. model (Walton 1985) is a good starting point for learning about
Unsteady fires are often characterized by the following equation: zone models.
⎛t⎞
2 Zone models were developed for room fires. In a room fire, hot
q = 1055 ⎜-----⎟ (19) gases rise above the fire, forming a plume. As the plume rises, it
⎝ tg ⎠ entrains air from the room so that the diameter and mass flow rate
of the plume increase with elevation. Accordingly, plume tempera-
where ture decreases with elevation. Fire gases from the plume flow up to
q = heat release rate of fire, kW the ceiling and form a hot stratified layer under the ceiling. Hot
t = time, s gases can flow through openings in walls to other spaces; this flow
52.14 2007 ASHRAE Handbook—HVAC Applications (SI)

is referred to as a door jet, which is similar to a plume, except that Fig. 20 Smoke Layer Interface
it flows through an opening in a wall.
Figure 19A is a sketch of a room fire. Zone modeling is an
idealization of the room fire conditions, as illustrated in Figure
19B. For this idealization, the temperatures of the hot upper and
lower layers of the room are uniform. The height of the disconti-
nuity between these layers is the same everywhere. The dynamic
effects on pressure are considered negligible, so pressures are
treated as hydrostatic. Other properties are considered uniform
for each layer. Algebraic equations are used to calculate the mass
flows caused by plumes and door jets.
Many computer zone models allow exhaust from the upper layer,
which is essential for simulating atrium smoke exhaust systems.
Heat transfer is estimated by methods ranging from a simple allow-
ance as a fraction of the heat released by the fire to a complicated
simulation including the effects of conduction, convection, and
radiation.

Atrium Smoke Filling by a Steady Fire Fig. 20 Smoke Layer Interface

The following experimental correlation of the accumulation of
smoke in a space by a steady fire is the steady-filling equation:
4⁄3
- exp ---------- ⎛ 1.11 – ----
z⎞
A H 1
t = ------2- × ----------- (21)
⎛ tq 1 ⁄ 3 H –4 ⁄ 3⎞ 1⁄3 0.28 ⎝ H⎠
H q
---- = 1.11 – 0.28 ln ⎜ ---------------------------
-⎟
z
(20)
H ⎜ 2 ⎟
⎝ A⁄H ⎠ Atrium Smoke Filling by an Unsteady Fire
To analyze atrium smoke filling with an unsteady fire, use zone
where fire or CFD models. Klote and Milke (2002) provide an algebraic
z = height of first indication of smoke above fire, m equation for the smoke-filling time for an atrium with a t-squared
fire, but this equation has limited applicability because t-squared
H = ceiling height above fire, m
fires can be extremely large for the times considered in many smoke-
t = time, s
filling applications.
q = heat release rate from steady fire, kW
A = cross-sectional area of atrium, m2 Steady Clear Height with Upper Layer Exhaust
Equation (20) is conservative in that it estimates the height of the Figure 18 illustrates smoke exhaust from the hot smoke layer
first indication of smoke above the fire rather than the smoke inter- at the top of an atrium to maintain a steady clear height. Smoke
face, as illustrated in Figure 20. In the idealized zone model, the flow into the upper layer from the fire plume depends on the
smoke interface is considered to be a height where there is smoke fire’s heat release rate, clear height, fuel type, and fuel orienta-
above and none below. In actual fires, there is a gradual transition tion. The following is a generalized plume approximation that
zone between the lower cool layer and upper hot layer. The first indi- does not take into account the specifics of the material being
cation of smoke can be thought of as the bottom of the transition burned.
zone. Another factor making Equation (20) conservative is that it is 1⁄3 5⁄3
based on a plume that has no contact with the walls, which would m· = 0.071q c z + 0.0018q c (22)
reduce entrainment of air.
where
Equation (20) is for a constant cross-sectional area with respect to m· = mass flow of plume, kg/s
height. For other atrium shapes, physical modeling or computational qc = convective heat release rate of fire, kW
fluid dynamics (see Chapter 34 of the 2005 ASHRAE Handbook— z = clear height above top of fuel, m
Fundamentals for more information) can be used. Alternatively, a
sensitivity analysis can be made using Equation (20) to set bounds on Clear height z is the distance from the top of the fuel to the interface
the filling time for an atrium of complex shape. The equation is between the “clear” space and the smoke layer. Because a smoke man-
appropriate for A/H 2 from 0.9 to 14 and for values of z greater than agement system generally must protect against fire at any location, it
or equal to 20% of H. A value of z/H greater than 1 means that the is suggested that the top of the fuel be considered at the floor level.
smoke layer under the ceiling has not yet begun to descend. These Equation (22) is not applicable when the mean flame height is
conditions can be expressed as greater than the clear height. An approximate relationship for mean
flame height is
A = Constant with respect to H
2⁄5
z f = 0.166q c (23)
z
0.2 ≤ ---- < 1.0 where zf = mean flame height, m.
H
The convective portion qc of the heat release rate can be ex-
pressed as
A
0.9 ≤ ------2 ≤ 14
H qc = ξ q (24)

When Equation (20) is solved for z/H, z/H is often outside the where ξ is the convective fraction of heat release. The convective
acceptable range. Equation (20) can be solved for time. fraction depends on the material being burned, heat conduction
Fire and Smoke Management 52.15

Fig. 21 Average Plume Temperature Fig. 22 Atrium Exhaust to Maintain Steady Clear Height

Fig. 22 Atrium Exhaust to Maintain Steady Clear Height

Fig. 23 Enlarged Scale for Figure 23

Fig. 21 Average Plume Temperature

through the fuel, and the radiative heat transfer of the flames, but a
value of 0.7 is often used.
The temperature of smoke entering the upper smoke layer is
qc
T p = T a + ---------
· (25)
m cp
where
Tp = plume temperature at clear height, K
Ta = ambient temperature, K
Fig. 23 Enlarged Scale for Figure 22
m· = mass flow of plume, kg/s
qc = convective heat release rate of fire, kW show the exhaust rate needed to maintain a constant clear height for
cp = specific heat of plume gases, kJ/(kg·K) an atrium with negligible heat loss from the smoke layer and negli-
Figure 21 shows plume temperature as a function of height above gible air leakage into the smoke layer from the outside.
the fuel as calculated from Equations (22) and (25). Smoke plumes The major assumptions of the analysis plotted in Figures 22 and
consist primarily of air mixed with combustion products, and the 23 are as follows:
specific heat of plume gases is generally taken to be the same as that • Plume has space to flow to top of atrium without obstructions
of air [cp = 1.00 kJ/(kg·K)]. Equation (22) was developed for • Heat release rate of fire is constant
strongly buoyant plumes. For small temperature differences be- • Clear height is greater than mean flame height
tween the plume and ambient, errors because of low buoyancy could • Smoke layer is adiabatic
be significant. This topic needs study, and, in the absence of better • Plume flow and exhaust are the only significant mass flows into or
data, it is recommended that the plume equations not be used when out of smoke layer (i.e., outside airflow, either as leakage or as
this temperature difference is small (less than 2 K). makeup air, into smoke layer is insignificant)
The density of smoke gases can be calculated from the perfect
gas law: Minimum Smoke Layer Depth
p An atrium smoke management system must be designed with a
ρ = ------- (26) smoke layer deep enough to accommodate a ceiling jet, a radial jet
RT of smoke formed when a plume hits the ceiling. Usual estimates of
where ceiling jet depth are 10 to 20% of the distance between the base of
ρ = density, kg/m3 the fuel and the ceiling (the ceiling jet itself is only about 10% of this
p = absolute pressure, Pa distance, but at the walls the jet reverses and flows under itself).
R = gas constant, J/(kg·K) Generally, the smoke layer depth should be at least 20% of the dis-
T = absolute temperature of smoke gases, K tance between the base of the fuel and the ceiling.
Volumetric flow is expressed as Number of Exhaust Inlets
m· When the flow rate of a smoke exhaust inlet is relatively large,
Q = ----- (27) cold air from the lower layer can be pulled into the smoke exhaust.
ρp
This phenomenon is called plugholing. A number of exhaust air
where inlets may be needed to prevent plugholing. The maximum volu-
metric flow rate that can be exhausted by a single exhaust inlet with-
m· = mass flow of plume or exhaust air, kg/s
Q = volumetric flow of exhaust gases, m3/s out plugholing is calculated by
ρ = density of plume or exhaust gases, kg/m3 1⁄2
5 ⁄ 2⎛ s
T – T o⎞
Atrium exhaust should equal the mass flow of the plume plus any V max = 4.16γd ⎜ -----------------⎟ (28)
leakage flow into the atrium above the clear height. ⎝ T o ⎠
For an atrium with negligible heat loss from the smoke layer and where
negligible air leakage into the smoke layer from the outside, exhaust Vmax = maximum volumetric flow rate without plugholing at Ts, m3/s
equals the plume’s mass flow rate from Equation (22) at the same γ = exhaust location factor, dimensionless
temperature as the plume from Equation (25). Figures 22 and 23 d = depth of smoke layer below lowest point of exhaust inlet, m
52.16 2007 ASHRAE Handbook—HVAC Applications (SI)

Ts = absolute temperature of smoke layer, K Fig. 24 Prestratified Layer of Hot Air under Atrium Ceiling
To = absolute ambient temperature, K and Resulting Temperature Profile
The ratio d/Di should be greater than 2 where Di is the diameter
of the of the exhaust inlet. For exhaust inlets centered no closer than
2Di from the nearest wall, γ = 1 should be used; for less than 2Di,
γ = 0.5 should be used. For exhaust inlets on a wall, use γ = 0.5.
For rectangular exhaust inlets, calculate Di as

2ab-
D i = -----------
a+b

where
a = length of inlet
b = width of inlet
Fig. 24 Prestratified Layer of Hot Air under Atrium
Where multiple inlets are needed to prevent plugholing, the min- Ceiling and Resulting Temperature Profile
imum separation between inlets should be

1⁄2 Fig. 25 Smoke Filling a Prestratified Atrium

S min = 0.065V e

where
Smin = minimum edge-to-edge separation between inlets, m
Ve = volumetric flow rate of one exhaust inlet, m3/s
This approach for calculating Vmax and Smin is consistent with
that of NFPA Standard 92B. A less conservative approach was in
earlier versions of this Standard and Klote and Milke (2002);
research is needed to evaluate these approaches.
Separation Between Inlets
When exhaust at an inlet is near the maximum flow rate Qmax ,
adequate separation between exhaust inlets must be maintained to
minimize interaction between flows near the inlets. One criterion Fig. 25 Smoke Filling a Prestratified Atrium
for separation between inlets is that it be at least the distance from
a single inlet that would result in an arbitrarily small velocity based
on sink flow. Using 0.2 m/s as the arbitrary velocity, the minimum Fig. 26 Beam Detectors Used for Activation of Atrium Smoke
separation distance for inlets located in a wall near the ceiling (or in Management System
the ceiling near the wall) is

S min = 0.32β Q e (29)

where
Smin = minimum separation between inlets, m
Qe = volumetric flow rate, m3/s
β = exhaust location factor, dimensionless

Prestratification and Detection

A layer of hot air often forms under the ceiling of an atrium
because of solar radiation on the atrium roof. Although no studies
have been made of this prestratification layer, building designers Fig. 26 Beam Detectors Used for Activation of Atrium Smoke
indicate that the temperature of such a layer can exceed 50°C. Management System
Temperatures below this layer are controlled by the building’s
heating and cooling system; the temperature can be considered to
increase significantly over a small increase in elevation, as shown Beam smoke detectors can overcome this detection difficulty.
in Figure 24. The analysis of smoke stratification given in NFPA The following approaches can provide prompt detection regardless
Standard 92B is not appropriate for the temperature profile of air temperature under the ceiling when a fire begins:
addressed in this section because it is for a constant temperature
increase per unit elevation. • Upward-Angled Beam to Detect Smoke Layer. One or more
When the average temperature of the plume is lower than that of beams are aimed upward to intersect the smoke layer regardless of
the prestratification layer, the smoke will form a stratified layer the level of smoke stratification. For redundancy, more than one
beneath the prestratification layer, as shown in Figure 25. Average beam smoke detector is recommended. Advantages include not
plume temperatures can be calculated from Equations (22) and (25); needing to locate several horizontal beams, and the minimized
they are plotted in Figure 21, which shows that the average plume risk of false activation by sunlight (a risk with some beam smoke
temperature is usually less than expected temperatures of the hot air detectors) because the receivers are angled downward. Figure 26
layer. Thus, when there is a prestratified layer, smoke cannot be illustrates the upward-angled beam approach.
expected to reach the atrium ceiling, and smoke detectors mounted • Horizontal Beams at Various Levels to Detect Smoke Layer.
on that ceiling cannot be expected to go into alarm. One or more beam detectors are located at roof level, with addi-
Fire and Smoke Management 52.17

tional detectors at lower levels. Exact beam positioning depends specified, and makeup air conditions are also defined. This allows
on the specific design, but should include beams at the bottom of simulation of fluid flow in considerable detail.
identified unconditioned spaces and at or near the design smoke Although CFD modeling has significant advantages in realisti-
level, with several beams at intermediate positions. cally simulating smoke flow, it is computationally intensive and
• Horizontal Beams to Detect Smoke Plume. Beams are arranged requires a lot of computer memory and time; it is not uncommon for
at a level below the lowest expected stratification level. These a CFD simulation to run for hours and sometimes days. Because
beams must be close enough to each other to ensure intersection CFD is computationally intensive, it produces so many numbers
of the plume; spacing should be based on the width of the beam at that the usual ways to evaluate computer output are inappropriate.
the least elevation above a point of fire potential. Visualization methods have been developed so people can under-
All components of a beam smoke detector must be accessible for stand CFD results.
maintenance. For the arrangement shown in Figure 26, a roof open- Several general-purpose CFD models are commercially avail-
ing (not shown) could provide access for maintenance. able that can be used for atrium smoke control. The U.S. National
Institute of Standards and Technology (NIST) developed the Fire
TENABILITY SYSTEMS Dynamics Simulator (FDS) model (McGrattan 2004; McGrattan
and Forney 2004), and its visualization software called Smokeview
The intent of smoke control systems is to provide a tenable envi- (Forney and McGrattan 2004). The FDS model includes numerous
ronment in the means of egress or other locations during building fire-modeling applications, such as smoke plume flow, fires in
fires. A tenable environment is one in which combustion products, enclosures, a burning townhouse, sprinklered fires, an oil tank fire,
including heat, are limited to a level that is not life threatening. fires in aircraft hangars, rack storage commodity fire, and a brush
Analysis of a tenability system consists of a smoke transport analy- fire advancing toward a house. FDS and Smokeview can be down-
sis and a tenability evaluation. For most applications, smoke trans- loaded free of charge (www.fire.nist.gov/fds).
port calculations are done by either a computational fluid dynamic
(CFD) model or a network model. Large Multicompartmented Buildings
Tenability Evaluation Because of the number of rooms and shafts in large multicom-
Tenability evaluation considers the effects of exposure to toxic gas, partmented buildings, it is often not feasible to use CFD to simulate
heat, and thermal radiation, as well as reduced visibility. Toxic gas, smoke transport. Ideally, the model used for such simulations would
heat, and thermal radiation exposure are direct threats to life. The be able to simulate a fire and mass and heat transfer throughout the
severity of each threat depends on the intensity and length of exposure. building. No practical model can do this, but research is under way
Reduced visibility does not directly threaten life, but it is an indi- to develop such a model. Currently, the network model CONTAM is
rect hazard. Reduced visibility can reduce walking speed. When used to simulate smoke transport in large multicompartmented
occupants and firefighters cannot see very much, they often become buildings. See the section on Computer Analysis for Pressurization
disoriented and cannot get away from the smoke, thus prolonging Systems for more information about CONTAM. Ferreira (2002),
their exposure. Another concern is that a disoriented person can fall Hadjisophocleous et al. (2002), Klote (2002, 2004), and Chapter 9
from an atrium balcony, which can be fatal. of Klote and Milke (2002) provide examples of tenability calcula-
Considerable research has been conducted regarding tenability, tions using CONTAM. Because CONTAM requires that tempera-
and methods are available for calculating exposures to combustion tures be supplied as data input by the user, the temperatures need to
gases and reduced visibility (ISO Standard 13571; Jin 2002; Klote be calculated or estimated by the user.
and Milke 2002; Purser 2002). There is no broad consensus on vis-
ibility criteria, but Jin (2002) suggests 13 m for applications where ACCEPTANCE TESTING
occupants are unfamiliar with the building (e.g., museums, casinos), Regardless of the care, skill, and attention to detail with which a
and 4 m when occupants are familiar with the building (e.g., resi- smoke control system is designed, an acceptance test is needed as
dential applications). The U.K. Chartered Institution of Building assurance that the system, as built, operates as intended.
Service Engineers (CIBSE 2003) suggests 8 to 10 m.
When combustion products from most materials are diluted An acceptance test should be composed of two levels of testing.
enough to meet any of these visibility criteria, the hazards to life The first is functional: an initial check of the system components.
from toxic gases, heat, and thermal radiation are also eliminated for The importance of the initial check has become apparent because
exposures up to 20 min. This means that, for most fires, tenability of problems encountered during tests of smoke control systems:
can be evaluated by calculating visibility, if the hazards of other fans operating backward, fans to which no electrical power was
exposures are checked. supplied, controls that did not work properly, etc.
The second level of testing is of performance, to determine
Atria and Other Large Spaces whether the system performs adequately under all required modes
For atria and other large spaces, it is appropriate to use a CFD of operation. This can consist of measuring pressure differences
model (see Chapter 34 of the 2005 ASHRAE Handbook—Funda- across barriers under various modes of smoke control system oper-
mentals for CFD information) to calculate tenability. CFD has been ation. If airflows through open doors are important, these should be
used for a wide range of applications, including aircraft design, measured. Chemical smoke from smoke candles (sometimes called
automotive design, boiler design, and weather forecasting, and is smoke bombs) is not recommended for performance testing because
extensively used for atrium smoke management systems. For exam- it normally lacks the buoyancy of hot smoke from a real building
ples of CFD simulations of smoke transport in large spaces, see fire. Smoke near a flaming fire has a temperature of 500 to 1100°C.
Hadjisophocleous et al. (1999), Kashef et al. (2002), Klote (2005), Heating chemical smoke to such temperatures to emulate smoke
and Lougheed and Hadjisophocleous (2000). from a real fire is not recommended unless precautions are taken to
The idea of CFD is to divide the space of interest into a large protect life and property. The same comments about buoyancy
number of cells and to solve the governing equations for each cell. apply to tracer gases. Thus, pressure-difference testing is the most
For atrium applications, the number of cells typically ranges from practical performance test. However, chemical smoke can be used to
100,000 to 1,000,000. Obstructions such as walls, balconies, and aid flow visualization.
stairs need to be taken into account. Conditions at the boundaries ASHRAE Guideline 5 covers the commissioning of smoke man-
need to be defined. Exhaust flow at or near the top of the atrium is agement systems.
52.18 2007 ASHRAE Handbook—HVAC Applications (SI)

SPECIAL INSPECTOR Smin =

minimum separation between inlets, m
Some building codes require special inspection and tests of T =
absolute temperature of smoke gases, K
smoke control systems in addition to the ordinary inspection of and t =
time, min or s
test requirements for buildings, structures, and parts of buildings. Ta =
ambient temperature, K
These special inspections and tests should verify the proper com- Tf =
average absolute temperature of fire compartment, K
missioning of the smoke control design in its final installed condi- tg =
growth time, s
tion. Procedures for inspection and testing should be developed by Ti =
absolute temperature of air inside shaft, K
the smoke control system’s special inspector, with approval of the Tin =
absolute temperature of air into fire compartment, K
authorities having jurisdiction. The special inspector needs to Tj =
absolute temperature of ceiling jet, K
understand the principles of smoke control, including code require- To =
absolute temperature of outside air, K
ments. The special inspector should check that the components of Tout =
absolute temperature of smoke leaving fire compartment, K
the system are as specified and that those components are installed Tp =
plume temperature at clear height, K
as intended, as well as whether the smoke control system performs Ts =
absolute temperature of stairwell air or surroundings, K
as intended. V =
wind velocity, m/s
Vk =
critical air velocity to prevent smoke backflow, m/s
W =
corridor or door width, m
EXTRAORDINARY INCIDENTS
y =
distance above stairwell bottom, m
Extraordinary incidents, whether caused by war, terrorism, acci- z =
height of first indication of smoke above fire, or clear height
dent, or natural disaster, can affect immediate human needs such as above top of fuel, m
survival and safety, and also longer-term needs such as air, water, zf = mean flame height, m
food, and shelter. Some buildings are designed with specific fea- Greek
tures intended to make them less susceptible to extraordinary inci-
dents. It is recommended that actuation of systems for fire and β = exhaust location factor, dimensionless
smoke protection be of higher priority than possibly conflicting ∆p = pressure difference, Pa
∆ psb = pressure difference between stairwell and building, Pa
automatic strategies designed to respond to other extraordinary con- ∆ psbb = pressure difference between stairwell and building at stairwell
ditions. bottom, Pa
Some acts of terrorism use fire, and those using bombs often lead ∆ psbt = pressure difference from stairwell to building at stairwell top, Pa
to fires. It is well known that war, terrorist attacks, and natural disas- θ = time
ters have the potential to disrupt utilities and interfere with fire fight- ρ = density, kg/m3
ing, and this often permits any fires that occur to grow unchecked. ρo = outside air density, kg/m3
For these reasons, simultaneous fire and other extraordinary inci- ξ = convective fraction of heat release
dents should be considered likely, and any features intended to mit-
igate extraordinary conditions should be designed accordingly. For REFERENCES
more information, see ASHRAE’s (2003) report, Risk Management AMCA. 2007. Certified ratings program—Product rating manual for
Guidance for Health, Safety and Environmental Security under smoke management fan performance. Publication 212-07. Air Move-
Extraordinary Incidents and Chapter 58 of this volume. ment and Control Association, Arlington Heights, IL.
ASHRAE. 2000. Laboratory methods of testing fans used to exhaust smoke
SYMBOLS in smoke management systems. ANSI/ASHRAE Standard 149-2000
(RA 2005).
A = area, m2 ASHRAE. 1994. Commissioning of smoke management systems. Guide-
a = dilution rate, air changes per minute line 5.
Abo = flow area between building and outside (per floor), m2 ASHRAE. 2003. Risk management guidance for health, safety and environ-
Af = area of floor, m2 mental security under extraordinary incidents. Report, Presidential Ad
Asb = flow area between stairwell and building (per floor), m2 Hoc Committee for Building Health and Safety under Extraordinary
Aw = area of wall, m2 Incidents.
ASTM. 2002. Test method for fire tests of through-penetration fire stops.
B = 3460(1/To – 1/Ts) at sea level standard pressure
Standard E814. American Society for Testing and Materials, West Con-
C = concentration of contaminant at initial time θ; flow coefficient shohocken, PA.
Co = initial concentration of contaminant Bukowski, R.W. 1991. Fire models, the future is now! NFPA Journal 85(2):
cp = specific heat of plume gases, kJ/(kg·K) 60-69.
Cw = pressure coefficient, dimensionless CIBSE. 2003. Fire engineering. CIBSE Guide E, 2nd ed. Chartered Institu-
d = distance from doorknob to edge of knob side of door, m tion of Building Services Engineers, London.
e = base of natural logarithm (approximately 2.718) Cooper, L.Y. 1985. ASET—A computer program for calculating available
F = total door-opening force, N safe egress time. Fire Safety Journal 9:29-45.
Fdc = force to overcome door closer, N Cooper, L.Y. and G.P. Forney. 1987. Fire in a room with a hole: A prototype
application of the consolidated compartment fire model (CCFM) com-
i = floor-to-ceiling height, or ceiling height above fire, m
puter code. Presented at the 1987 Combined Meetings of Eastern Section
Hj = thickness of ceiling jet, m of Combustion Institute and NBS Annual Conference on Fire Research.
h = distance above neutral plane, m Cresci, R.J. 1973. Smoke and fire control in high-rise office buildings—Part
m· = mass flow of plume or exhaust air, kg/s II, Analysis of stair pressurization systems. Symposium on Experience
i = number of floors and Applications on Smoke and Fire Control, ASHRAE Annual Meet-
p = absolute pressure, Pa ing, June.
pw = pressure exerted by wind, Pa Evers, E. and A. Waterhouse. 1978. A computer model for analyzing smoke
Q = volumetric flow rate, m3/s movement in buildings. Building Research Establishment, Fire Research
q = heat release rate of fire, kW Station, Borehamwood, Hertsfordshire, U.K.
Fang, J.B. 1980. Static pressures produced by room fires. NBSIR 80-1984.
qc = heat release rate, W
National Bureau of Standards. Available from National Institute of Stan-
Qe = volumetric flow rate, m3/s dards and Technology, Gaithersburg, MD.
Qin = volumetric flow rate of air into fire compartment, m3/s Fang, J.B. and J.N. Breese. 1980. Fire development in residential basement
Qout = volumetric flow rate of smoke out of fire compartment, m3/s rooms. NBSIR 80-2120. National Bureau of Standards. Available from
R = gas constant, J/(kg·K) National Institute of Standards and Technology, Gaithersburg, MD.
Fire and Smoke Management 52.19

Ferreira, M.J. 2002. Use of multizone modeling for high-rise smoke control Mitler, H.E. 1985. Comparison of several compartment fire models: An
system design. ASHRAE Transactions 108(2):837-846. interim report. NBSIR 85-3233. National Bureau of Standards. Avail-
Forney, G.P. and K. McGrattan. 2004. User’s guide for Smokeview version able from National Institute of Standards and Technology, Gaithersburg,
4—A tool for visualizing fire dynamics simulation data. NIST Special MD.
Publication 1017. National Institute of Standards and Technology, Mitler, H.E. and H.W. Emmons. 1981. Documentation for CFC V, the fifth
Gaithersburg, MD. Harvard computer code. Home Fire Project Technical Report 45. Har-
Hadjisophocleous, G.V., G.D. Lougheed, and S. Cao. 1999. Numerical study vard University, Cambridge, MA.
of the effectiveness of atrium smoke exhaust systems. ASHRAE Trans- Mitler, H.E. and J.A. Rockett. 1986. How accurate is mathematical fire mod-
actions 105(1):699-715. eling? NBSIR 86-3459. National Bureau of Standards. Available from
Hadjisophocleous, G., F. Zhuman, and G. Lougheed. 2002. Computational National Institute of Standards and Technology, Gaithersburg, MD.
and experimental study of smoke flow in the stair shaft of a 10-story Morgan, H.P. 1979. Smoke control methods in enclosed shopping complexes
tower. ASHRAE Transactions 108(1):724-730. of one or more stories: A design summary. Fire Research Station, Bore-
Houghton, E.L. and N.B. Carruther. 1976. Wind forces on buildings and struc- hamwood, Hertsfordshire, U.K.
tures. John Wiley & Sons, New York. Morgan, H.P. and G.O. Hansell. 1987. Atrium buildings: Calculating smoke
Huggett, C. 1980. Estimation of heat release by means of oxygen consumption flows in atria for smoke control design. Fire Safety Journal 12:9-12.
measurements. Fire and Materials 4(2). Nelson, H.E. 1987. An engineering analysis of the early stages of fire devel-
opment—The fire at the Dupont Plaza Hotel and Casino—Dec. 31, 1986.
ISO. 2002. Life-threatening components of fire—Guidelines for the estima-
NISTIR 87-3560. National Institute of Standards and Technology, Gaith-
tion of time available for escape using fire data. ISO/TS Standard 13571.
ersburg, MD.
International Organization for Standardization, Geneva.
Nelson, H.E. and F.W. Mowrer. 2002. SFPE handbook of fire protection
Jin, T. 2002. SFPE handbook of fire protection engineering, Ch. 2-4,Visibil-
engineering, Ch. 3-14, Emergency movement. Society of Fire Protection
ity and human behavior in fire smoke. Society of Fire Protection Engi-
Engineers, Bethesda, MD.
neers, Bethesda, MD.
NFPA. 2002. Installation of air-conditioning and ventilating systems. Stan-
Jones, W.W. 1983. A review of compartment fire models. NBSIR 83-2684. dard 90A-02. National Fire Protection Association, Quincy, MA.
National Bureau of Standards. Available from National Institute of Stan- NFPA. 2006. Recommended practice for smoke-control systems. Standard
dards and Technology, Gaithersburg, MD. 92A. National Fire Protection Association, Quincy, MA.
Kashef, A., N. Benichou, G.D. Lougheed, and C. McCartney. 2002. A com- NFPA. 2005. Standard for smoke management systems in malls, atria,
putational and experimental study of fire growth and smoke movement in and large areas. Standard 92B. National Fire Protection Association,
large spaces. 10th Annual Conference of the Computational Fluid Quincy, MA.
Dynamics Society of Canada, Windsor, Ontario, pp. 1-6. NFPA. 2006. Life safety code®. Standard 101. National Fire Protection Asso-
Klote, J.H. 1980. Stairwell pressurization. ASHRAE Transactions 86(1): ciation, Quincy, MA.
604-673. NFPA. 2002. Guide for smoke and heat venting. Standard 204. National Fire
Klote, J.H. 1982. A computer program for analysis of smoke control systems. Protection Association, Quincy, MA.
NBSIR 82-2512. National Bureau of Standards. Available from National NFPA. 2003. Fire protection handbook, 19th ed. National Fire Protection
Institute of Standards and Technology, Gaithersburg, MD. Association, Quincy, MA.
Klote, J.H. 1984. Smoke control for elevators. ASHRAE Journal 26(4): NOAA. 1979. Temperature extremes in the United States. National Oceanic
23-33. and Atmospheric Administration (U.S.), National Climatic Center,
Klote, J.H. 1990. Fire experiments of zoned smoke control at the Plaza Hotel Asheville, NC.
in Washington, D.C. ASHRAE Transactions 96(2):399-416. Peacock, R.D., G.P. Forney, P. Reneke, R. Portier, and W.W. Jones. 1993.
Klote, J.H. 2002. Smoke management applications of CONTAM. ASHRAE CFAST, the consolidated model of fire growth and smoke transport.
Transactions 108(2):827-836. NIST Technical Note 1299. National Institute of Standards and Technol-
Klote, J.H. 2004. Tenability and open doors in pressurized stairwells. ogy, Gaithersburg, MD.
ASHRAE Transactions 110(1):613-637. Proulx, G. 2002. SFPE handbook of fire protection engineering, Ch. 3-13,
Klote, J.H. 2005. CFD analysis of atrium smoke control at the Newseum. Movement of people: The evacuation timing. Society of Fire Protection
ASHRAE Transactions 111(2):567-574. Engineers, Bethesda, MD.
Purser, D.A. 2002. SFPE handbook of fire protection engineering, Ch. 2-6,
Klote, J.H. and X. Bodart. 1985. Validation of network models for smoke
Toxicity assessment of combustion products. Society of Fire Protection
control analysis. ASHRAE Transactions 91(2B):1134-1145.
Engineers, Bethesda, MD.
Klote, J.H. and J.A. Milke. 2002. Principles of smoke management. Quintiere, J.G. 1989. Fundamentals of enclosure fire “zone” models. Jour-
ASHRAE. nal of Fire Protection Engineering 1(3):99-119.
Klote, J.H. and G.T. Tamura. 1986. Smoke control and fire evacuation by Rilling, J. 1978. Smoke study, 3rd phase: Method of calculating the smoke
elevators. ASHRAE Transactions 92(1). movement between building spaces. Centre Scientifique et Technique du
Law, M. 1982. Air-supported structures: Fire and smoke hazards. Fire Pre- Bâtiment (CSTB), Champs sur Marne, France.
vention 148:24-28. Sachs, P. 1972. Wind forces in engineering. Pergamon, New York.
Lougheed, G.D. and G.V. Hadjisophocleous. 2000. The smoke hazard from Said, M.N.A. 1988. A review of smoke control models. ASHRAE Journal
a fire in high spaces. ASHRAE Transactions 107(1):720-729. 30(4):36-40.
MacDonald, A.J. 1975. Wind loading on buildings. John Wiley & Sons, New Sander, D.M. 1974. FORTRAN IV program to calculate air infiltration in
York. buildings. DBR Computer Program 37. National Research Council,
McGrattan, K. 2004. Fire Dynamics Simulator (version 4)—Technical ref- Ottawa, Canada.
erence guide. NIST Special Publication 1018. National Institute of Stan- Sander, D.M. and G.T. Tamura. 1973. FORTRAN IV program to simulate
dards and Technology, Gaithersburg, MD. air movement in multi-story buildings. DBR Computer Program 35.
McGrattan, K. and G.P. Forney. 2004. Fire Dynamics Simulator (version National Research Council, Ottawa, Canada.
4)—User’s guide. NIST Special Publication 1019. National Institute of Shaw, C.Y., J.T. Reardon, and M.S. Cheung. 1993. Changes in air leakage
Standards and Technology, Gaithersburg, MD. levels of six Canadian office buildings. ASHRAE Journal 35(2):34-36.
Shaw, C.Y. and G.T. Tamura. 1977. The calculation of air infiltration rates
McGuire, J.H. and G.T. Tamura. 1975. Simple analysis of smoke flow prob- caused by wind and stack action for tall buildings. ASHRAE Transactions
lems in high buildings. Fire Technology 11(1):15-22. 83(2):145-158.
McGuire, J.H., G.T. Tamura, and A.G. Wilson. 1970. Factors in controlling Simiu, E. and R.H. Scanlan. 1978. Wind effects on structures: An introduc-
smoke in high buildings. Symposium on Fire Hazards in Buildings, tion to wind engineering. John Wiley & Sons, New York.
ASHRAE Winter Meeting. Tamura, G.T. 1990. Field tests of stair pressurization systems with overpres-
Milke, J.A. and F.W. Mowrer. 1994. Computer-aided design for smoke sure relief. ASHRAE Transactions 96(1):951-958.
management in atria and covered malls. ASHRAE Transactions 100(2): Tamura, G.T. 1994. Smoke movement and control in high-rise buildings.
448-456. National Fire Protection Association, Quincy, MA.
Mitler, H.E. 1984. Zone modeling of forced ventilation fires. Combustion Tamura, G.T. and C.Y. Shaw. 1976a. Studies on exterior wall air tightness and
Science and Technology 39:83-106. air infiltration of tall buildings. ASHRAE Transactions 83(1):122-134.
52.20 2007 ASHRAE Handbook—HVAC Applications (SI)

Tamura, G.T. and C.Y. Shaw. 1976b. Air leakage data for the design of ele- Walton, G.N. 1989. AIRNET—A computer program for building airflow net-
vator and stair shaft pressurization systems. ASHRAE Transactions work modeling. National Institute of Standards and Technology, Gaithers-
83(2):179-190. burg, MD.
Tamura, G.T. and C.Y. Shaw. 1978. Experimental studies of mechanical Walton, W.D. 1985. ASET-B: A room fire program for personal computers.
venting for smoke control in tall office buildings. ASHRAE Transactions NBSIR 85-3144-1. National Bureau of Standards. Available from National
86(1):54-71. Institute of Standards and Technology, Gaithersburg, MD.
Tamura, G.T. and A.G. Wilson. 1966. Pressure differences for a 9-story Walton, G.N. and W.S. Dols. 2005. CONTAM 2.4 user guide and program
building as a result of chimney effect and ventilation system operation. documentation. NISTIR 7251. National Institute of Standards and Tech-
ASHRAE Transactions 72(1):180-189. nology, Gaithersburg, MD.
Tanaka, T. 1983. A model of multiroom fire spread. NBSIR 83-2718. Wray, C.P. and G.K. Yuill. 1993. An evaluation of algorithms for analyzing
National Bureau of Standards. Available from National Institute of Stan- smoke control systems. ASHRAE Transactions 99(1):160-174.
dards and Technology, Gaithersburg, MD.
Thomas, P.H. 1970. Movement of smoke in horizontal corridors against an Yoshida, H., C.Y. Shaw, and G.T. Tamura. 1979. A FORTRAN IV program
airflow. Institution of Fire Engineers Quarterly 30(77):45-53. to calculate smoke concentrations in a multi-story building. DBR Com-
UL. 1999. Fire dampers. ANSI/UL Standard 555-99. Underwriters Labora- puter Program 45. National Research Council, Ottawa, Canada.
tories, Northbrook, IL.
UL. 1999. Smoke dampers. ANSI/UL Standard 555S-99. Underwriters Lab- BIBLIOGRAPHY
oratories, Northbrook, IL.
UL. 2003. Fire tests of through-penetration firestops. ANSI/UL Standard CIBSE. 1995. Relationships for smoke control calculations. TM19:1995.
1479-03. Underwriters Laboratories, Northbrook, IL. Chartered Institution of Building Service Engineers, London.
Wakamatsu, T. 1977. Calculation methods for predicting smoke movement Lougheed, G.D. and G.V. Hadjisophocleous. 1997. Investigation of atrium
in building fires and designing smoke control systems. In Fire Standards smoke exhaust effectiveness. ASHRAE Transactions 103(2):519-533.
and Safety, ASTM STP 614, pp. 168-193. American Society for Testing Sprat, D. and A.J.M. Heselden. 1974. Efficient extraction of smoke from a
and Materials, West Conshohocken, PA. thin layer under a ceiling. Fire Research Note 1001.

Related Commercial Resources