Fire Safety Journal, 7 (1984) 107 - 113 107

Smoke Management in Office Buildings

WILLIAM BLAZEK
Fire Protection Engineering, General Services Administration, 18th and F. Street, NW, Washington, DC 20405
(U.S.A.)

SUMMARY SMOKE CONTROL VS. SMOKE REMOVAL

Design considerations for smoke control A smoke control system is n o t synony-

systems have been addressed by numerous mous with a smoke removal system, but both
researchers and practicing engineers. How- are smoke management systems. A smoke
ever, papers discussing the actual observed control system should confine the smoke
performance o f smoke control systems are within a predetermined smoke control zone.
less frequently encountered. The following This smoke will slowly disperse t h r o u g h o u t
three papers address the testing o f smoke the zone and be mechanically exhausted from
control systems from two perspectives. First, the building. No air is mechanically supplied
suggested procedures for testing smoke con- into the zone. The rate at which the contam-
trol systems are described relative to test inated air is exhausted is dependent upon the
methodology, system parameters requiring air infiltration (leakage) into the smoke con-
evaluation, and measurement systems. Sec- trol zone from the surrounding areas. If the
ond, the performance o f actual systems is windows are closed and none are broken
described as observed in tests. during the fire, the air changes per hour
would probably be less than one in a reason-
ably tight building.
INTRODUCTION Smoke removal systems require large vol-
umes of outside air which must be mechan-
A successful smoke management system ically supplied to the smoke zone and mech-
must provide areas of refuge that will main- anically exhausted from the building. This re-
tain life support capabilities for the length sults in rapid dilution and removal of the
of time which m a y be required to accomplish smoke. Large dedicated fans are required to
total building evacuation. This may be accom- provide twelve or more air changes per hour.
plished by c o m p a r t m e n t a t i o n , pressurization To prevent migration of smoke from the
of floors and/or shafts, exhaust, etc. In high- smoke zone to adjacent zones, the capacity
rise office buildings smoke management m a y of the exhaust fans must sufficiently exceed
be extremely difficult to achieve due to the that o f the supply fans to cause air infiltra-
excessive time required to evacuate the build- tion from the surrounding areas. In U.S.
ing. However, if total sprinkler protection is federal buildings the systems are designed to
provided in conjunction with total smoke provide 12 to 20 air changes per hour.
control, evacuation and relocation of occu- Flush systems which utilize the building
pants on most of the nonfire floors will air-handling systems are smoke removal sys-
probably be unnecessary. tems on a much smaller scale (two to six air
Total smoke control cannot be accom- changes per hour). The supply fans provide
plished with only elevator lobby/hoistway 100% outside air and all return air is ex-
and stairwell pressurization. Sufficient pres- hausted. However, there is an important
sure differentials must also be established difference. In the flush mode some smoke
between floors to prevent smoke m o v e m e n t may migrate to the surrounding areas unless
through other floor openings. Smoke control, a negative pressure differential is developed
as used in this paper, includes the above two in the flush zone. This becomes improbable
conditions. since the return air fan capacities are up to

0379-7112/84/$3.00 © Elsevier Sequoia/Printed in The Netherlands

108

20% lower than the supply fans. Nonetheless, building in the winter; and (2) reduce the
in a building with nonoperable windows, flow of building air to the outside during the
the flush mode may be the most expeditious cooling season.
m e t h o d for the fire department to purge The closing of the elevator hoistway doors
the smoke after the fire is extinguished. is the only one which would n o t apply during
Atrium smoke removal, based on the guid- normal building operations. The Safety Code
ance in the Life Safety Code [1], can be for Elevators and Escalators (ANSI/ASME
considered a flush system. The major differ- A17.1-1981) [2] requires that, upon automat-
ence between it and the air handler flush ic elevator recall to the designated floor, the
mode is that the supply air is designed to elevator doors shall open and remain open.
have a capacity of 75% of the exhaust. The National Fire Protection Associations
Technical Committee on Fire Doors and
SMOKE CONTROL WITH BUILDING AIR- Windows (NFPA-80) proposed a tentative
HANDLING SYSTEMS interim a m e n d m e n t (TIA) to incorporate this
elevator code requirement. After considering
If building air-handling systems are to be the negative comments received, the proposed
successfully utilized in smoke control sys- TIA was sent back to the Committee by the
tems, limited fire size and reduced stack Standards Council*. Unfortunately, ANSI/
action are required to offset the limited fan ASME A17.1-1981 makes no exceptions, even
capacities. if the hoistways are mechanically pressurized.
The force of the expanding fire gases must
be minimal. With a properly designed sprin- TESTING METHODS
kler system, it is unlikely that the fire will
exceed an area of 10ft X 10ft (3.05m X Two c o m m o n methods are used for smoke
3.05m) in a typical office environment. Thus control testing -- measurement of tracer gas
the m a x i m u m pressure differential from the concentrations and pressure differentials.
expanding gases is not expected to exceed The tracer gas m e t h o d involves bulky equip-
0.05 - 0.06 in water gauge (wg) (12 - 15 Pa). ment, is much more time consuming, and
Also, because of the assumed operation of the requires the building to be purged prior to
sprinkler system, the design of the building proceeding with another test. With the pres-
air-handling systems can be limited to control- sure differential method, a magnehelic gauge
ling "cold smoke". or differential manometer is all that is needed.
Detrimental pressure differentials created However, in this m e t h o d smoke candles may
by stack action must be reduced to levels be required in exhaust plenums to visually
that can be sufficiently overcome within the observe if exhaust air is being recirculated
fan capacities. Stack action can be reduced into the building.
by: The use of smoke candles alone should not
(1) closing elevator hoistway vents; be the sole basis for smoke control testing.
(2) closing elevator hoistway doors after The basic data should be available to explain
automatic elevator recall; the results. However, for those who must "see
(3) enclosing vestibules for loading docks; to believe" this may be the only way to re-
(4) providing revolving doors and vesti- ceive acceptance from the building code
bules at building entrances; officials and to establish confidence in the
(5) providing air locks at mechanical rooms system by fire department personnel.
and elevator pits;
(6) maintaining slight positive floor pres- Testing procedure pressure-differential
sures relative to the outside. method
Positive pressures to the outside can be The following steps are recommended:
accomplished, within the fan and duct capac-
ities, by automatically adjusting the static *At the annual meeting in May 1983, the NFPA
pressures across the supply fans to referenced approved the following revision to NFPA 80:
7-2.4 When required by the authority having
outside pressures. jurisdiction, doors serving elevators required to con-
All of the above are intended to: (1) re- form with rule 211.3 (see 7-2.3) shall be automati-
duce the unwanted outside air entering the cally closed after a predetermined time interval.
 109

(1) Building pressure differential profile (5) Verification of programmed events

Determine the building pressure differen- Physically check that supply, return and
tial profile (BPDP) with HVAC systems exhaust dampers are properly positioned and
operating in the normal mode. The BPDP has all other functions in the fire management
proven useful in discovering problems which system actually do occur.
should be corrected prior to testing in the
smoke control mode. Using a differential (6) Pressure differential readings
manometer or magnehelic gauge, pressure Determine pressure differentials between
differentials are measured between the ver- the smoke control zone and building openings
tical openings and the floors. These data within and surrounding the smoke control
determine the stack actions which are the zone. Supplemental pressure differential
result of the interaction of stack effect and readings should also be taken at shaft open-
the air handling systems. ings in the pressurized zones.
The results may indicate no pressure
differentials between a shaft and two or (7) Analyze data to determine effectiveness
more floors located several levels apart. of smoke control system
Some shafts may be positive to a particular Compare the operation of the system rel-
floor while others m a y be negative. Sudden ative to predetermined objectives.
variations in the pressure differentials be-
tween adjacent floors and a c o m m o n shaft (8) Investigate for adverse effects
should be investigated. An air balancing Is excessive force required to open stair-
damper may have closed, or the floor may well doors? Will the doors relatch after being
be designed to be negative relative to its opened? Will hoistway/elevator cab doors
surrounding, such as a cafeteria floor or open and close?
laboratory space.
(a) Opening o f stairwell d o o r s - Some
(2) Choose smoke control zone consensus codes require that the unlatching
In order to expedite the smoke control force on panic hardware does not exceed 15
testing, an engineering judgement should be lb (66.7 N). The Life Safety Code [1] limits
made as to which smoke control zone pro- the force to fully open any door in the means
vides the greatest challenge*. The BPDP, of egress to 50 lb (222.4 N) applied to the
type of air handling systems, fan capacities, latch stile. The increased friction on a stair-
supplemental pressurization fans, air han- well door latch results from greater pressure
dling zones, programmed sequence of events, differentials between the floor and the shaft.
preplanned relocation of occupants, and Since the stairwell doors open in the direction
existing weather conditions must be consid- of egress, the required force to open the doors
ered in the decision. increases with increased positive shaft pres-
sures to the floors.
(3) No tifica tion The force to open an egress door can be
Inform building occupants and fire depart- easily calculated by multiplying one-half the
m e n t or central station prior to and after area of the door face by the pressure differ-
completion of smoke control testing. ential. Reasonably accurate results can also
be obtained with the use of a spring scale.
(4) Initiation o f smoke control mode
With the spring scale attached to the handle
Activate a sensor programmed to initiate
and the door unlatched, the m a x i m u m scale
the smoke control system in the desired
reading to pull the door open should approxi-
zone, e.g. sprinkler waterflow at the inspec-
mate the calculated force. If corrective action
tors test pipe, smoke detector, heat detector,
is required, possible solutions include instal-
etc.
lation or adjustment of barometric vents,
static pressure controllers and damper link-
* L i m i t e d resources and available t i m e m a y p r e c l u d e
ages, hardware alignment, etc.
detailed testing o f every s m o k e c o n t r o l zone. H o w -
ever, all s m o k e c o n t r o l z o n e s at least s h o u l d be
activated t o ensure t h a t all p r o g r a m m e d f u n c t i o n s (b) Relatching of stairwell d o o r s - If a
d o occur. stairwell door does not relatch after opening
110

with floor being positive to the stairwell,

an adjustment or the replacement of a worn
door closer may be necessary.

(c) E l e v a t o r s d o o r s - - The Safety Code for

Elevators and Escalators [2] requires that
the force necessary to prevent closing of
hoistway doors or cab doors from rest shall
not exceed 30 lb (133.4 N). Furthermore,
the kinetic energy of the hoistway door and
all parts rigidly connected thereto, computed
for the average closing speed, shall not exceed
7 ft pdl (0.29 J). These limited forces may
prevent the elevator doors from opening or
closing when the building smoke management
systems, such as hoistway pressurization,
pressurized elevator lobbies or building smoke
control, are activated. This can be roughly
simulated by applying pressure with an open
hand on the face of the hoistway or cab door.
Co*
The friction forces created by hand pressure
may be sufficient to overcome the force of
the power-operated door mechanism. This
can occur when the hoistway is at a relative
negative or positive pressure to the elevator
lobby. After automatic elevator recall, the
elevator doors should be checked for proper J
operation on those floors with the greatest
pressure differentials.

CASE S T U D Y ¢~1

Seattle Federal Building

This totally sprinklered building, with
mechanical equipment rooms in the base- Fig. 1. Air handling zones in the Seattle Federal
ment, sixth, 15th and 36th floors, is 487 ft Building.
(148.5 m) above grade (First Avenue) and
has a gross area of 830 114 ft 2 (77 201 m2). containing the fire floor are turned off. The
Air distribution (Figs. 1 and 2) is divided companion return air fans go to 100% exhaust
into seven vertical air handling zones serving and the adjacent air handling zones go to
the following floors: basement through 100% fresh air supply with their return fans
fifth, sixth and seventh, eighth through 13th, turned off. The number of air handling zones
14th (this is the computer area), small por- which go to 100% supply varies depending
tions of the 15th and 16th through 23rd, upon the size and location of the air handling
24th through 31st, 32rid through 35th with zones. Fans are used to recirculate building
a small portion of the 36th. All air handlers return air through the elevator shafts. During
are designed for constant volume air flow. the smoke control mode, these two-speed
There are no fire or smoke dampers where the fans pressurize the elevator shafts with 100%
ducts penetrate the air shafts. Furthermore, if outside air.
this building were to be constructed today, The building pressure differential profile
the return air duct smoke detectors would be (Fig. 3) indicates the most adverse pressure
eliminated. differential occurred at the east stairwell on
In the smoke control mode, the supply the 18th floor (the stairwell was at a negative
fans for the vertical air handling zone 0.065 in water gauge (wg) (16.3 Pa) relative
 111

" -, ,- II -i
Is lINE
4'1 l'b I! -i
w

.dh. !1 .ji
"4F" '"
W
•'IF" J"
m •'IF" J"
D "IF" '"
W •4 F - ,-

° *5F* ~"
-4F- ,-

, *'IF" -!-
. "4F" -~"

o -41,- ~-
. .....
*.IF- i
~-

Fig. 2. Air handling system during normal operation.

112

E / ~ - V',~ T @ ~ / - / O / s - 7-/~tF y'S"

|
/do.v.h z, $ a/o~ ~,~.=;
~./,F,..¢'/" .~-,'~,',~~,~E~ ~ .~:"=~'v.,~,,'-o,.'z' ~"l-~'g.~-f'¢~¢ 2"~/P-5"/~/''
~ $-,'~"~r .o.~/:4rnf~',¢.----~l b F'~'/'/I-;'- JL, 8~I'k',¢"' / f " ~ L, 'g'~4"~'~B" ~' L.,,C'~/'/r~" ,..J
P ~I I TM q I TM qD qP q

0 o 0 o o

~2_

i . J , , , . . , , i , , i | | . t . s . . | , , = . i • , . , o • .

- .0 8 -.O# *,Og *.08 "fit f i g +.@~/r'.08 *./2. ",@q ,*.Oq ,*.og +;,/2. -,Og *.Oq ".08 ~.IZ " • ~Oq

Fig. 3. B u i l d i n g pressure differential profiles for the Seattle Federal B u i l d i n g obtained at n o r m a l and smoke
c o n t r o l modes o f operation. --×--× the n o r m a l mode, implies t h a t all air handling zones are in n o r m a l operation.
the smoke c o n t r o l mode, in o p e r a t i o n in the smoke c o n t r o l zone on floors 16 t h r o u g h 23. Pressure dif-
ferentials are expressed in inches w a t e r gauge. Outside air temperature ~ 67 °F: return air t e m p e r a t u r e - 77 ~'F.
A l l readings were taken on August 2, 1979.

to the floor). Furthermore, all other shaft on the 18th floor. The results were very
openings to this floor were significantly more successful. Although the elevator hoist-
negative than on the other floors tested. For ways were pressurized by dedicated fans,
this reason, the first smoke control test was substantially the same results were obtained
conducted in the zone containing the 18th in the stairwells. This is a good example of
floor. (A subsequent investigation indi- creating "localized stairwell pressurization
cated that a return air-balancing damper was where it is needed without stairwell pressur-
closed.) ization fans".
If the outside temperature was 10 °F
(--12.2 °C) rather than 67 °F (19.5 °C) when
1st test
this test was performed, successful smoke
control might have been dependent upon
Smoke control zone --16th through 23rd the manual pressurization of an additional
floor
air handling zone, which can be accomplished
Pressurized zones - - 2 4 t h through 31st
at the control center. However, the adverse
floor
effects of the temperature extremes diminish
-- 8th through 14th with decreasing temperature differentials in
floor the pressurized zones (100% outside air).
All other floors - - normal operation
It would be interesting ,to determine the
results if the stairwell doors: were propped
For comparison, Fig. 3 shows the pres- open on one pressurized f l o o r above and
sure differential profile in the normal mode one below the smoke control zone with the
and in the smoke control mode after the stairwell doors remaining open on the fire
return air-balancing damper was adjusted floor.
 113

2 n d test to be marginally successful in preventing

smoke from entering the elevator shafts
Smoke c o n t r o l zone -- 32nd through 35t h (+0.005 to +0.035 in wg) (1.25 - 8.4 Pa).
floor E xcept for the basement, smoke would enter
Pressurized zone -- 24th t hr ough 31st the stairwells (--0.01 to - - 0 . 0 2 5 in wg)
floor (2.49 - 6.23 Pa).
All o t h e r floors -- normal operation Single floor pressurization is only planned
on the sixth and seventh floors. This results
This test was primarily c o n d u c t e d t o deter- in two pressurized floors versus six "fire
mine the effect of open stairwell doors. floors". With the additional pressurization
When the 33rd floor is the fire floor, the of the eight through 13th floors, the neces-
greatest n u m b e r o f floors are involved in the sary positive pressures should be developed
planned relocation o f occupants. The occu- to prevent smoke migration into the stair-
pants on floors 32 t hr ough 35 relocate t o wells.
the 30th and 31st floors. In all probability, Although the elevator hoistways are pres-
the stairwell doors on all six floors could be surized by dedicated fans, their effectiveness
open at th e same time. T o determine the on the lowest floors is minimal since the fans
capability o f the smoke cont r ol system to are located near the t op o f the hoistways.
prevent smoke from entering the two stair- Their effectiveness is f u r t h e r diminished
wells u n d e r these severe circumstances, all since the elevator doors remain open after
stairwell doors f r om the 30th through the elevator recall. These doors should be pro-
35th floors were blocked open with the air grammed to close after a short time interval
handling zone in the smoke c ont r ol mode. to reduce the pressure losses through the
Each pressure differential reading was taken hoistway openings at the designated level
with the d o o r closed. All the stairwells in (nonfire floor). The additional pressurized
the smoke c o n t r o l zone remained positive floors will also enhance the hoistway pres-
with a range o f 0.03 - 0.04 in wg (7.5 - 10 Pa). sures.
With all the doors closed the range was 0.18 - Some smoke cont rol systems are designed
0.26 in wg (44.9 - 64.8 Pa). to pressurize all floors within a vertical air
A secondary purpose f or this test was to handling zone except the fire floor. In such
determine the m a x i m u m force required to cases, smoke dampers are required where the
open the stairwell doors. Since the stairwells floor supply and return air ducts penet rat e
were at a positive pressure relative to the the air shafts. The supply smoke dampers
corridors and the office areas were negative are closed on the fire floor and the return
to the corridors, the m a x i m u m force required air smoke dampers are closed on the nonfire
to open the stairwell doors on each f l oor floors. Ret urn air is exhausted and 100%
was d e p e n d e n t on t he s u m m a t i o n o f these outside air is supplied to the pressurized
two pressure differentials. This would occur floors. Since the return air fan only serves
when the corridor doors were opened. On the one floor in the smoke cont rol mode, its
35th floor the greatest pressure differential effective fl oor capacity m ay be increased
was developed between the west stairwell several fold, t hereby creating localized shaft
and the office space ( 0 . 2 6 - 0.15 in wg) pressurizations on the fire floor. This could
(64.8 - 37.4 Pa). Excluding the friction forces, eliminate the need for dedicated pressuriza-
the calculated force to open the west stair- tions fans in hoistways and stairwells.
well d o o r was 26 lb {115.6 N). Constant volume, variable volume, and
single fan systems can all be adapted to
3rd test provide smoke control.

Smoke c o n t r o l zone - Basement t hr ough

-

5th floor
Pressurized zo n e -- 6th and 7th f l oor REFERENCES
All o t h e r floors - Normal operation
-

1 Life Safety Code, NFPA 101, National Fire

Protection Association, Quincy, MA, 1981.
With the basement t hr ough fifth floors in 2 Safety Code for Elevators and Escalators, ANSI/
the smoke c o n t r o l mode, the pressure differ- ASME A17.1 -- 1981, American Society of
entials indicated the smoke cont r ol system Mechanical Engineers, New York.