See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/251575575

Elevator shaft pressurization for smoke control in tall buildings: The Seattle
approach

Article  in  Fuel and Energy Abstracts · November 2011

DOI: 10.1016/j.buildenv.2011.05.007

CITATIONS READS

16 1,738

1 author:

Richard Steven Miller

Clemson University
70 PUBLICATIONS   2,208 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Thermal management View project

All content following this page was uploaded by Richard Steven Miller on 20 January 2020.

The user has requested enhancement of the downloaded file.

This article appeared in a journal published by Elsevier. The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
 Author's personal copy

Building and Environment 46 (2011) 2247e2254

Contents lists available at ScienceDirect

Building and Environment

journal homepage: www.elsevier.com/locate/buildenv

Elevator shaft pressurization for smoke control in tall buildings: The Seattle
approach
Richard S. Miller*
Department of Mechanical Engineering, Clemson University, Clemson, SC 29634-0921, USA

a r t i c l e i n f o a b s t r a c t

Article history: CONTAM simulations of both commercial and residential tall building models are conducted in order to
Received 18 February 2011 study recently adopted Seattle code requirements for elevator shaft pressurization systems. In contrast to
Received in revised form the International Building Code (IBC) requirements, the Seattle approach specifies across elevator door
27 April 2011
pressure minimums and maximums on only four “fire floors” (including one above, and two below, the
Accepted 8 May 2011
fire floor). This is accomplished using a minimal pressurization of the entire elevator shaft in conjunction
with venting of the four fire floors. The present study adresses the feasibility of calibrating such a system
Keywords:
to meet the design objectives in tall buildings (system performance during an actual fire event is not
Smoke control
Pressurization
considered). The two building models correspond to 37 story buildings with dual elevator and dual
Elevator stairwell shafts extending the entire height of the building. Each model is calibrated to experimental
Hoistway data. Simulations are conducted for a variety of ambient temperatures and exterior building door
Seattle positions. Coupled pressurization of the stairwells is also considered. The system requirements are found
International building code to be achievable for both elevator only and coupled elevator and stairwell pressurization systems.
However, the observed pressure differences do change with changes in the ambient temperature as well
as changes in the ground floor exterior door position. It is therefore recommended that such systems
should be calibrated for pressure differences intermediate to the prescribed minimum and maximum
values to compensate for changes to the system performance. Providing a relief vent to ambient on any
recall floor may also be advisable.
Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction is not required, provided that interior exit stairways are pressurized
to a minimum of þ25 Pa and a maximum of þ87 Pa in the shaft
Stairwells and elevator shafts can be dangerous conduits of relative to the building measured with all stairway doors closed
smoke migration throughout buildings during fire situations. under maximum anticipated stack effect pressures.” Stairwell
Smoke penetrating the shaft can spread through both the buoyancy pressurization has been approved and adopted for a relatively long
of hot gases as well as by the stack effect which occurs when there time. In contrast, the use of elevator shaft pressurization has only
is a temperature difference between the air in the shaft and the air recently received approval by the IBC and relatively little research
in the outside environment [1]. One method for controlling smoke has been done in this area. The pertinent section of the code rele-
flow is by using shaft pressurization. The intent of pressurization vant to elevator shaft pressurization (Section 708.14.2) states in
systems is to use outside air to pressurize a shaft such that only part: “Elevator hoistways shall be pressurized to maintain
positive across door pressures are achieved on all floors. This a minimum positive pressure of þ25 Pa and a maximum positive
requires the specification of a suitable minimum pressure needed pressure of þ62.5 Pa with respect to adjacent occupied space on all
to prevent smoke from entering the shaft, as well as the specifi- floors. This pressure shall be measured with all elevator cars at the
cation of a suitable maximum pressure difference in order to ensure floor of recall and all hoistway doors on the floor of recall open and
proper door functioning. For example, the pertinent sections of the all other doors closed.”
International Building Code (IBC) 2009 relevant to stairwell pres- The author has sought to fill this gap in the literature and has
surization systems states in part (Section 909.20.5): “the vestibule been studying elevator pressurization using numerical simulations
[2e5]. The results of these studies have shown that elevator pres-
surization is much more complex than stairwell pressurization.
* Tel.: þ1 864 656 6248; fax: þ1 864 656 4435. Stairwells are characterized by relatively well sealed doorways
E-mail address: [email protected]. which are in the closed position during pressurization operation. As

0360-1323/$ e see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.buildenv.2011.05.007
 Author's personal copy

2248 R.S. Miller / Building and Environment 46 (2011) 2247e2254

such, stairwell pressurization is relatively straight forward and pressurization based on the Seattle approach. All results were
requires only relatively modest fan flow rates to achieve estab- obtained via computer simulations using the CONTAM software
lished pressure difference minimums. In contrast, elevator shafts developed by the Indoor Air Quality and Ventilation Group at the
are characterized by doors with relatively large leakages, and often National Institute of Standards and Technologies. The software is
multiple doors per floor. In addition, the IBC 2009 recognizes that a zonal model in which a building geometry is composed of
minimum and maximum pressure differentials be achieved during a number of zones (rooms, shafts, floors, etc.). Each zone is treated
Phase 1 “with all elevator cars at the floor of recall and all hoistway as a lumped parameter with only hydrostatic pressure variations
doors on the floor of recall open and all other hoistway doors within the zone (dynamic pressure variations being five or more
closed.” Therefore, substantially larger fan flow rates are required to orders of magnitude smaller for the present application). Only the
pressurize an elevator shaft; in some cases more than ten times the “long time” equilibrium pressure distributions are predicted.
flow rates required to pressurize a stairwell. Pressurized elevator Details of the software can be found at http://www.bfrl.nist.gov/
air flows exit the hoistway and enter the building corridors on all IAQanalysis/index.html. Note too that only the ability of the
floors, with especially large flow rates at the recall floor. Such large system to meet its design minimum and maximum pressure
amounts of air flooding the building can result in complex pressure differences as prescribed by the SBC are considered in this paper.
profiles as well as strong interactions with stairwell pressurization Neither the wisdom of only considering the four fire floors nor the
systems. In particular, if the recall floor does not have an open performance of the systems under an actual fire event are
pathway to the outside world for excess elevator shaft air flows to considered.
exit the building, then it is impossible to achieve pressure profiles
within the both the stairwells and elevator shafts that do not 2.1. Building models
violate either the pressure difference minimums and maximums
specified by the IBC 2009 code. Such a venting is not currently Results are presented for two different building models (see
required by the IBC 2009. Substantial coupling effects between the Figs. 1 and 2). Model 1 is a simple open floor plan commercial
elevator and stairwell pressurization systems were also observed. building model with no internal pressure barriers between the
The presence of an open garage level can also impede the ability to hoistway and the exterior walls of the building. Model 2 is a resi-
achieve proper system performance. For buildings with other dential building which includes both an underground garage level
interior doors, such as those to residences, very large pressures may and interior residences. For the sake of simplicity, all building
inadvertently be created by the elevator pressurization system models are the same height and have the same floor area. Pres-
across these doors [5]. The reader is referred to the above citations surization of elevator shafts is achieved by roof mounted fans.
for more details concerning the IBC approach to pressurization. Pressurization of stairwells is achieved by fans located in the
The objective of the current study is to extend the investigation basement level (using ambient air). The SBC additionally requires
of pressurization systems beyond the methodology adopted by the venting of the tops of the stairwells which is modeled by two out
IBC. In particular, a new approach recently adopted by the Seattle flowing fans mounted on the roof and having fixed flow rates of
jurisdiction is studied. The “Seattle approach” to elevator shaft 1.18 m3/s (2500 cfm). Schematics of each building model’s floor
pressurization is based on a modification of the IBC code language plans are provided in Figs. 1 and 2. Additional data for each of the
which results in the allowance of a substantially different approach. buildings is provided in Tables 1 and 2. Interior leakage values were
The Seattle Building Code (SBC) (Section 708.14.2.1) allows for obtained from established published values [1]. Exterior leakages
a “four-floor” approach aimed at targeting the primary fire floor, were calibrated to experimental measurements from Ref. [2]
the floor directly above, and two floors immediately below the fire (commercial) and Ref. [6] (residential). The calibration procedure
floor (the approach was first proposed by Dr. John Klote, co-author is discussed in detail in Ref. [2]. In brief, the outer building leakages
of the Principles of Smoke Management [1]). The across elevator and residential door leakages are calibrated to match the relative
door pressure differences on these four floors are treated identi- proportion of the stack effect pressure differential to experimental
cally to the IBC language and must remain within a pressure range data.
of 25e62.5 Pa. However, the code allows for the following excep-
tion: “The pressure differential is permitted to be measured relative
to outdoor atmosphere on floors other than the.” four fire floors.
In practice, for floors other than the four fire floors the pressure is
raised via pressurization until a minimum 25 Pa to atmosphere
pressure drop from the hoistway is achieved. This total pressure
drop will be comprised of the sum of pressure drops across the
elevator doors, any interior pressure drops, plus the pressure drop
across the outer building wall. As such, the actual across elevator
door pressure differences will be (considerably) less than 25 Pa on
these floors. The four fire floors are then vented to develop the
desired pressure differentials across the doors by relieving the
over-pressures on these floors. Both exhaust shafts and the HVAC
system have been proposed as sources of the venting. In addition,
the SBC now allows the same procedure to be used for the stair-
wells. The code change was adopted during the summer of 2010
and the author is not aware of any publications addressing such
a design.

2. Modeling approach

The following presents results obtained by computer simula- Fig. 1. Schematic of the commercial building Model 1 (not to scale): (a) all upper
tions of model buildings utilizing both elevator and stairwell floors, (b) ground floor level, and (c) roof level.
 Author's personal copy

R.S. Miller / Building and Environment 46 (2011) 2247e2254 2249

a b

c d

Fig. 2. Schematic of the residential building Model 2 (not to scale): (a) all upper floors, (b) ground floor level, (c) roof level, and (d) basement garage level.

2.2. Thermal modeling


P 0 hRTO x
Pressurization uses outside air which is in general at a different TS ðxÞ ¼ TB  ðTB  TO Þexp  : (1)
CP QPatm
temperature than the building interior. A model for the bulk aver-
aged air temperature within the shaft relevant to the CONTAM code The shaft position is x (from the intake; top or bottom), TB is the
has been developed in Ref. [2] which is further improved upon as building temperature (assumed to be equal to the shaft wall
follows. Briefly, the model is based on heat transfer within temperature), TO is the outside ambient air temperature, the
a constant wall temperature shaft based on the hydraulic diameter, convective heat transfer coefficient is h, the volumetric flow rate of
Dh ¼ 4A/P, where A is the shaft cross sectional area and P is the the ambient intake air is Q, P0 is the effective wetted perimeter
wetted perimeter. The axially varying average shaft air temperature
is then [7]:
Table 2
Base case building model parameters for building Model 2.
Table 1
Building model parameters for building Model 1 calibrated to commercial building Number of floors above ground 37
data. Number of floors below ground 1
Floor height 3.00 m
Number of floors above ground 37 Building height 111 m
Number of floors below ground 0 Floor area 1210 m2
Floor height 3.00 m Number of stairwells 2
Building height 111 m Stairwell floor area 13.9 m2
Floor area 1210 m2 Stairwell perimeter 15.2 m
Number of stairwells 2 Stairwell hydraulic diameter 3.66 m
Stairwell floor area 13.9 m2 Closed stairwell door leakage 103 cm2
Stairwell perimeter 15.2 m Open stairwell door leakage 1.95 m2
Stairwell hydraulic diameter 3.66 m Number of elevator shafts 2
Closed stairwell door leakage 103 cm2 Cars/doors per elevator shaft 2
Open stairwell door leakage 1.95 m2 Elevator shaft floor area 13.9 m2
Number of elevator shafts 2 Elevator shaft perimeter 15.2 m
Cars/doors per elevator shaft 4 Elevator hydraulic diameter 3.66 m
Elevator shaft floor area 28.8 m2 Closed elevator door leakage 484 cm2
Elevator shaft perimeter 26.8 m Open elevator door leakage 5580 cm2
Elevator hydraulic diameter 4.30 m Number of residential units per floor 12
Closed elevator door leakage 484 cm2 Residential door leakage 116 cm2
Open elevator door leakage 5580 cm2 External building leakages:
Number of residential units per floor 0 Exterior leakage area to wall surface area ratio 3.7  104
External building leakages: Ground floor closed exterior door leakage 1355 cm2
Exterior leakage area to wall surface area ratio 3.4  104 Ground floor open exterior door leakage 3.72 m2
Ground floor closed exterior door leakage 645 cm2 Number of garage level lobby doors 2
Ground floor open exterior door leakage 3.72 m2 Garage level lobby door leakage 323 cm2
Building temperature 21  C Building temperature 21  C
 Author's personal copy

2250 R.S. Miller / Building and Environment 46 (2011) 2247e2254

(discussed below), and the heat capacity of the air is Cp. The bulk achieved between the elevator shaft and the ambient for floors
averaged shaft air temperature is therefore: other than floors 17e20. Since additional pressure drops remain
between the elevator doors and the ambient, the across elevator
ZH door pressure differences are necessarily smaller than 25 Pa.
1
TS ¼ TS ðxÞdx; (2) Four separate, but identical, fans are then placed on floors 17e20
H
0 which draw air directly out of the corridor and to the ambient. The
purpose of these fans is to drop the pressure on these floors until
where H is the height of the shaft. Substituting for TS(x) from Eq. (1) the across elevator door pressures are within the range
yields: 25 / 62.5 Pa. For the purposes of this study, these fan speeds are

   adjusted until the maximum pressure difference of 62.5 Pa is ach-
ðTB  TO ÞCp QPatm P 0 hRTO H
TS ¼ TB þ exp   1 : (3) ieved across any of the elevator doors on these four floors. Smaller
P 0 hRTO H Cp QPatm values within this range could have been chosen; however, it is of
The primary issue remaining is properly modeling the convec- interest to determine how changes to the building geometry,
tive heat transfer coefficient. For this, the Dittus-Boelter Nusselt outside air temperature, etc. could result in violation of prescribed
number correlation for fully developed flow in a constant wall pressure differences. The choice of the maximum pressure is dis-
temperature duct is employed [7]: cussed further below.
For cases which include stairwell pressurization the same
Nu ¼ 0:023Re0:8 n
D Pr ; (4) procedure is used for the stairwells (in conjunction with the roof
mounted venting fans described above). The minimum pressure on
where the Nusselt number is Nu ¼ hDh/k and k is the thermal any of the non-fire floors is set to 25 Pa with respect to outside
conductivity of the air. In the above the pipe flow Reynolds number ambient. The pressures on the four fire floors cannot be set inde-
is: ReD ¼ UavgDh/n where Uavg ¼ Q/A and n is the kinematic viscosity pendently of the elevator doors. Any attempt to change the vent
of the air. Also, the Prandtl number is Pr ¼ n/a, where a is the flow rate to control the across stairwell door pressure differences
thermal diffusivity of the air. The exponent, n, takes the values would cause a simultaneous change to the across elevator door
n ¼ 0.3 for cooling and n ¼ 0.4 for heating. All properties are taken pressures. Therefore, for the purposes of this study the maximum of
at the average temperature: (TB þ TO)/2. Implications of the fully either the across elevator or stairwell door pressure is set equal to
developed flow assumption are discussed in Ref. [2] and in the 62.5 Pa. This could be for either the stairwell or the elevator doors,
Appendix below. depending upon the building conditions. Again, there is some
The above thermal model is most applicable to the empty ambiguity in how to best set the pressures. However, one purpose
elevator hoistways which are effectively simply large ducts upon of this study is to determine how much the system behavior
which the above model is based. In contrast, the stairwells have changes with changes in building conditions, and if violations of
interior features (the stairs) which provide a significantly larger either the minimum or maximum pressure differences could
surface area for heat transfer. For this reason the effective wetted potentially occur. Suggestions for proper calibration of the system
perimeter has been defined above. For elevator shafts P0 ¼ P. For are provided in the Conclusions.
stairwells, we model P0 ¼ bP, where b is a geometry correction Finally, once all fan flow rates are determined they are input into
factor. Absent any experimental data the geometry correction is the thermal model to obtain new shaft air temperatures. Changes to
taken to be b ¼ 2 for stairwells hereinafter. However, CONTAM is the shaft air temperatures result in changes to the pressures.
not sensitive to small changes in the shaft air temperatures. Therefore, the entire process is iterated until a final solution is
The approach described above predicts the average shaft air converged. Final fan flow rates and average shaft air temperatures
temperature assuming no leakages throughout the shaft. In reality are provided in Table 3 for all simulations conducted for this study.
air exits the shafts at each floor through either stairwell or elevator
doors. These effects can be taken into account in the thermal
3.1. Elevator pressurization only
modeling. A derivation of such a thermal model is presented in the
Appendix. However, the detailed thermal model is exceedingly
As a first step in the analysis simulations for elevator-only
difficult to use in conjunction with CONTAM. It requires the floor by
pressurization are conducted for both the commercial and the
floor flow leakages to be provided from CONTAM, and the process
residential building models. Cases 1 and 2 (see Table 3) correspond
would have to be iterated as changes to the predicted shaft air
to the above description: systems calibrated for the commercial
temperature change the flow leakages. Nevertheless, two cases
and residential building models with the exterior building doors in
were treated in this manner in order to justify the simpler thermal
the open position and with the exterior doors remaining in this
model. The results of the Appendix show that the above simple
open position. Across door pressure differences as a function of the
approach is in good agreement with the more detailed model;
within z 2  C for the average shaft air temperature for an
ambient air temperature of 12  C. CONTAM is insensitive to such Table 3
small changes in temperature. Therefore, the simpler form pre- Simulation parameters: Case name, building model, ambient air temperature, cali-
sented above is used in what follows. brated elevator, stairwell, and fire floor vent fan flow rates, and the average elevator
and stairwell air temperatures. All cases are calibrated with the exterior building
door in the open position, except for Case 5 which is calibrated with the exterior
3. Results door closed. Case 6 corresponds to calibration on a cold day (Case 4) but with the
system operating on a hot day.
Elevator pressurization of the building models is achieved as Cases Building TO [ C] E. fan [m3/s] S. fan [m3/s] V. fan [m3/s] TE [ C] TS [ C]
follows. All elevator cars are located on the ground floor with all 1 Model 1 12 18.6 N/A 3.24 8 21
doors in the open position. The exterior building door is set to open 2 Model 2 12 11.9 N/A 1.59 8 21
and the ambient air temperature is 12  C unless otherwise stated. A 3 Model 1 12 17.0 2.45 2.64 8 2
fire alarm on the 19th floor is simulated. Two identical roof 4 Model 2 12 10.9 2.50 1.49 8 2
5 Model 2 12 7.67 1.84 1.79 8 2
mounted fans blow air into each of the elevator shafts. The fan flow 6 Model 2 32 10.9 2.50 1.49 31 28
rate is adjusted until a minimum pressure difference of 25 Pa is
 Author's personal copy

R.S. Miller / Building and Environment 46 (2011) 2247e2254 2251

floor number for Cases 1 and 2 are presented in Fig. 3. The pressure
differences labeled “elevator” correspond to the elevator shaft 35
minus the corridor pressure (across elevator door). Pressure 30
differences labeled “stairwell” correspond to the stairwell minus

Floor Number
25
the corridor pressure (across stairwell door). Finally, the label
“residential” for Model 2 depicts the pressure of the corridor minus 20
that of the residence (across the residential doors). The residential 15
door pressures are included because standard IBC type elevator
10
pressurization has been shown to create very large pressures on
Model 1
these doors in some situations [5]. Therefore, positive values 5
Model 2
correspond to flow out of either the elevator or stairwell shafts, and 0
into the residences from the corridor for residential doors. Fig. 4 0 25 50 75 100
depicts the corresponding pressure differences between the ΔP across door [Pa]
elevator hoistway and ambient used for the non-fire floor calibra-
tions for both models. Note that these pressure profiles must be Fig. 4. Pressure differences measured from the elevator shaft to the ambient for Cases
linear as purely hydrostatic pressure distributions exist within both 1 and 2 corresponding to Fig. 5(a) and (b), respectively.

the shaft and the ambient.

Several trends are observed in Fig. 3. First, a strong stack effect is
maximum across stairwell door pressures remain less than the
observed for the (unpressurized) stairwells as illustrated by the
maximum recommended value of 87 Pa for nearly all floors
near linear variation from negative to positive values for the stair-
(although the values on the roof and garage level are slightly
well doors. In contrast, the elevator hoistways are characterized by
exceeded due to strong shaft effect and direct connection to the
near vertical profiles with negligible stack effect influence. This is
outside pressures on these floors). In addition, even though
consistent with the author’s previous findings for IBC systems.
the pressure within the stairwells varies completely linearly within
Elevator pressurization requires large volumes of (cold) outside air
the shaft, a large increase in the across door pressure differences is
flooding the shaft. This reduces the shaft air temperature to near
found on the four fire floors. This is entirely due to the reduced
ambient (see Table 3), thereby negating the stack effect. Also,
pressures on these floors due to venting (same for the elevators).
Furthermore, even though the minimum hoistway to ambient
pressure is 25 Pa (Fig. 4), the across elevator door pressures on the
a non-fire floors are much smaller than 25 Pa and relatively negli-
35 gible. These pressures would not be sufficient to prevent smoke
migration if it were to spread to any of these floors.
30
Floor Number

25 3.2. Coupled stairwell and elevator pressurization

20 Coupled stairwell and elevator shaft pressurization systems are

addressed next. Cases 3 and 4 are discussed in this section. These
15
correspond exactly to Cases 1 and 2 (Fig. 3) except that the stair-
10 wells are also pressurized. Results for Cases 3 and 4 are presented in
Fig. 5. One primary difference from the elevator-only pressurization
5 Elevator
Stairwell cases is the location of the maximum pressure difference across
doors on the four fire floors is now on the stairwell doors. For
0
example, in Fig. 5 the stairwell door pressure difference on the 20th
-100 -50 0 50 100 150 floor has been calibrated to 62.5 Pa pressure difference. However,
ΔP across door [Pa] the maximum obtainable pressure difference for the elevator doors
on this floor is approximately 50 Pa. This is still within the allow-
b able range of pressures; however, it is apparent that caution should
35
be taken in ensuring that both the elevator and stairwell doors can
30 be calibrated within the allowable ranges of pressure differences
based on the specific building geometry being considered. The
Floor Number

25 operation of the two systems is strongly coupled. Another obser-

20 vation is that the stack effect slope of the pressure profiles is greatly
diminished for the stairwell shafts. Under pressurization the
15 stairwell air temperatures are relatively low; again, negating the
stack effect (see Table 3).
10

5 Elevator 3.3. Effects of the exterior door position

Stairwell
0 Residential
The effect of the exterior building door position is illustrated in
-100 -50 0 50 100 150 Figs. 6 and 7. The author’s previous results have shown that this has
ΔP across door [Pa] a strong impact on the ability of the traditional IBC type elevator
pressurization system to function properly. As mentioned above, if
Fig. 3. Pressure differences for elevator shaft pressurization only calibrated for cold
day (12  C) conditions with a fire simulated on the 19th floor with the exterior
there is no outlet for the large elevator air flow rates to exit the
building doors open. The exterior building doors are in the open position: (a) Case 1, building at the ground floor, strict adherence pressurization is
Model 1 (commercial), and (b) Case 2, Model 2 (residential). impossible to achieve. For the purposes of this study only the
 Author's personal copy

2252 R.S. Miller / Building and Environment 46 (2011) 2247e2254

a 35
35

30 30

Floor Number
25
Floor Number

25

20 20

15 15

10 10

5 Elevator
5 Elevator
Stairwell Stairwell
0 Residential
0
-100 -50 0 50 100 150 -100 -50 0 50 100 150
ΔP across door [Pa]
ΔP across door [Pa]
Fig. 7. Pressure differences for Model 2 (residential) with coupled stairwell and
b elevator shaft pressurization calibrated for cold day (12  C) conditions with a fire
35 simulated on the 19th floor with the exterior building doors closed. The results are for
Case 5.
30
Floor Number

25
Therefore, it is important to explore both how the system responds
20 to changes in conditions like the exterior door position, as well as if
the system can be calibrated for different conditions.
15
Fig. 6 addresses the first point. It presents the same building
10 simulation as Case 4 and Fig. 5(b). That is, a coupled pressurization
Elevator system calibrated with the exterior doors open. However, in this
5
Stairwell case now the exterior doors have been closed. Fig. 6 provides the
0 Residential resulting pressure profiles. Upon closing the doors the air flow is
-100 -50 0 50 100 150 restricted from exiting the building. This results in a substantial
ΔP across door [Pa] increase in the internal building pressure across all floors. The
primary effect of this pressure increase on the fire floors is a nearly
Fig. 5. Pressure differences for coupled stairwell and elevator shaft pressurization 25 Pa increase in the across door pressure differences for both
calibrated for cold day (12  C) conditions with a fire simulated on the 19th floor with stairwells and elevators on the four fire floors. The commercial
the exterior building doors open. The exterior building doors are in the open position:
building model is more tightly sealed and experiences a nearly
(a) Case 3, Model 1 (commercial), and (b) Case 4, Model 2 (residential).
75 Pa increase. Also note that the across stairwell door pressure on
residential building Model 2 is shown hereinafter. Results for the the roof level is now significantly larger than 87 Pa. If the building is
commercial building model are similar. originally calibrated near the maximum across elevator door
Pressurization systems are typically calibrated with the building pressure differences on the fire floors (as is the current situation),
doors propped open (although Seattle is an exception). Once cali- operation without a flow pathway to the ambient on the ground
brated the fan flow rates are typically fixed constant and operate floor could result in code violations if the system is operated with
identically irregardless to changes in the building configuration, the exterior doors closed. However, an automatic venting system
ambient air temperature, wind, etc. There is no dynamic for the recall floor could negate this over pressurization effect.
compensation to ensure proper pressure differences are achieved As mentioned above, the standard IBC type elevator pressuri-
for other system configurations (or for actual fire situations). zation system is essentially impossible to calibrate to strict adher-
ence if the exterior building doors are closed in the model [5] (due
to large air flow rates pressurizing the building). Fig. 7 provides
pressure profiles for this situation for the Seattle approach. The data
35 corresponds to Case 5 which is identical to Case 4 except that the
system is calibrated with the exterior door closed. These results
30 show that the Seattle approach actually has an opposite effect of
the exterior door position than observed in the author’s previous
Floor Number

25
studies. Under the standard code language the pressure differen-
20 tials are measured across the elevator doors on all floors. If the
15 exterior building door is closed then the air flow exiting the ground
floor elevator doors acts to pressurize the ground floor itself. This
10 acts to reduce the “across door” pressure difference. More air must
5 Elevator be forced into the elevator shaft to finally raise the across door
Stairwell pressure difference; resulting in a highly pressurized building. The
0 Residential standard approach would, therefore, require tremendous fan flow
-100 -50 0 50 100 150 rates if it were to meet strict adherence with the exterior doors
ΔP across door [Pa] closed.
In contrast, the Seattle approach only considers the pressure
Fig. 6. Pressure differences for Model 2 (residential) with coupled stairwell and
elevator shaft pressurization calibrated for cold day (12  C) conditions with a fire
difference from the hoistway to the ambient for the non-fire floors.
simulated on the 19th floor with the exterior building doors open. The results corre- With the exterior doors open sufficient air must be supplied to
spond to Case 4 but operating with the exterior building door in the closed position. simply raise the pressure within the shaft to 25 Pa above the
 Author's personal copy

R.S. Miller / Building and Environment 46 (2011) 2247e2254 2253

air temperature is also found to have a significant impact on the

35 system performance. For a system calibrated on a cold day significant
reductions of all across door pressure differences on the fire floors
30
occurs when the system operates on a hot day. Pressure variations of
Floor Number

25 up to 25 Pa are observed when a system calibrated with the exterior

doors open operates with the exterior door closed. This could be
20
alleviated by adding an automatic venting system for the recall floor.
15 Finally, it is recommended that elevator hoistways pressurized using
10 the Seattle approach be calibrated to across door pressure differences
intermediate to the allowed minimum and maximum in order to
5 Elevator
allow for seasonal and other changes.
Stairwell
0 Residential
-100 -50 0 50 100 150 Appendix. Thermal model verification
ΔP across door [Pa]
The thermal model developed above neglects several effects
Fig. 8. Pressure differences for Model 2 (residential) with coupled stairwell and
elevator shaft pressurization calibrated for cold day (12  C) conditions with a fire
that may be significant in real buildings. Eq. (1) employs a constant
simulated on the 19th floor with the exterior building doors open. The results are for heat transfer coefficient, h, corresponding to the fully developed
Case 6 which corresponds to Case 4 but operating on a hot day (32  C). flow coefficient and therefore neglects entrance region effects. In
reality the heat transfer coefficient does not settle to its fully
ambient. If the exterior doors are closed this is actually easier to developed value until approximately x/Dh z 25 [7]. Near the
accomplish as the building retains the pressure with the exterior entrance the heat transfer coefficient can be up to 25% larger than
doors closed. Table 3 shows that smaller fan flow rates are needed its fully developed value. In addition, Eq. (1) assumes that the mass
when the system is calibrated with the exterior door closed. Pres- flow rate from the pressurization fans are entirely conserved,
surizing the building is actually a “good” thing (from the perspec- whereas in reality much of the flow is “lost” along the route
tive of the Seattle code requirements) because the hoistway is through the door leakages on each floor. Variations in the air
always at a larger pressure than the building and only the pressure properties are treated in a relatively simplified manner, and friction
difference to the ambient is considered. effects are neglected.
Therefore, in order to test the validity of Eq. (1) a detailed
thermal model has been developed in Excel. Each floor of the shaft
3.4. Effects of the ambient temperature
is modeled using Eq. (1) individually. Each floor uses an inlet air
temperature corresponding to the outlet temperature of the
Effects of the ambient air temperature are addressed next. As
previous floor. Properties are taken at the local floor averaged air
mentioned above, pressurization systems are generally calibrated
temperatures. Flow rate losses are accounted for along each floor as
on a particular day of the year and the fan flow rates are fixed
follows. The total mass flow rate into and exiting each shaft are
thereafter. Therefore, the system must be sufficiently robust in
equal under steady conditions:
order to properly function on all days of the year and under all
weather conditions. Fig. 8 presents the operation a system cali- X
rO Qin ¼ ri Qout;i ; (5)
brated on a cold day (Case 4), but operating on a hot day. The same
fan flow rates are retained but the shaft air temperatures change where r is the air density and the summation is over all floors/
(see Case 6, Table 3). The results reveal a decrease in all across door outflows. Invoking the ideal gas law, P ¼ rRT, and noting that the
pressure differences on the four fire floors of approximately 10 Pa. total pressure is approximately constant across the height of the
Pressures measured to ambient for the remaining floors are also shaft (hydrostatic pressure variations are w1% of atmospheric
reduced; in some cases below the code minimum of 25 Pa (not pressure):
shown).
X TO
Qin ¼ Qi ; (6)
4. Conclusions TI
where Ti is the average shaft air temperature of floor i. CONTAM
The Seattle approach is substantially more robust and effective
models flow rates using the “orifice equation” in which Q w ALDP1/2,
than the “standard” IBC elevator pressurization in which all hoistway
where AL is the leakage area, and DP is the pressure drop across the
doors require a minimum 25 Pa across door pressure difference.
leakage. Comparing with Eq. (6), the relative flow loss for each
These standard systems require excessively large air flows into the 1=2
leakage is wAL;i DPi =Ti . The excel spread sheet makes use of this
building which can create many problems (including excessive
relation in calculating the flow loss from floor to floor. For elevators
pressures across both elevator and stairwell doors). By essentially
the entire flow is lost at the point of the last elevator door. For
excluding all but the four fire floors the Seattle approach alleviates
stairwells 1.18 m3/s (2500 cfm) remains after the roof level stairwell
many of these issues. Floors other than the four fire floors display
door leakage due to the vent (approximately 50% for Cases 3 and 4).
relatively low to negligible pressure differences. However, prescribed
An improved Nusselt number correlation is also used indepen-
pressure limits are attainable on the four fire floors under a large
dently for each floor:
range of building and weather conditions. Nevertheless, all across
door pressure differences (elevator, stairwell, and residence) are Nux ¼ NuN aE aF ; (7)
shown to change significantly when either the exterior building
doors open or close as well as when the ambient air temperature where the Nusselt number for fully developed flow is now modeled
changes. Coupled stairwell and elevator shaft pressurization systems using the SiedereTate Equation [7]:
are shown to be more difficult to calibrate to the Seattle code  0:14
requirements as the across stairwell and elevator door pressures mm
Nu ¼ 0:023Re0:8
D Pr
1=3
: (8)
cannot be varied independently on the four fire floors. The ambient mw
 Author's personal copy

2254 R.S. Miller / Building and Environment 46 (2011) 2247e2254

The above described thermal model predicts the axially

35 varying air temperature within the shaft as a piece-wise function
of average air temperatures for each floor. The average over all
30
floors is also calculated in the spread sheet. This is the single
25 average shaft air temperature that would be used for a CONTAM
Floor

simulation. In practice, it would be extremely cumbersome to use

20
in conjunction with the CONTAM simulations because the rela-
15 tive leakages at each floor are implicit functions of the pressure
Case 3: Elevator
10 Case 3: Stairs drops across the doorways. However, a set of post processed
Case 4: Elevator validations is possible. For this purpose we have taken Cases 3
Case 4: Stairs
5 and 4 from Table 3. The pressure drops from each of these
0 simulations were imported into the spread sheet and the fan
-20 -10 0 10 20 speeds listed in Table 3 are used directly. The air temperature
Temperature [0C] profiles predicted by the detailed thermal model are provided in
Fig. 9. The figure caption also contains the averaged shaft air
Fig. 9. Detailed thermal model predictions of the shaft air temperatures for Cases 3
temperatures which can be compared directly to those calculated
and 4. The integrated average shaft air temperatures for the four cases in the legend
are 8  C, 1  C, 7  C, and 0  C, respectively. via Eq. (3) from Table 3. The detailed model is shown to be in
excellent agreement with the simplified model employed for the
CONTAM simulations. Average shaft air temperatures are within
The absolute air viscosities are calculated at the average of the 2  C for the two models. Such small changes are insignificant
inlet and wall temperatures for each floor, mm, and at the wall for the CONTAM simulations. This verifies the use of the simpler
temperature, mw. This form of the correlation is considered to be thermal model.
more accurate than the Dittus-Boelter expression when significant
variations in species properties are present. It is also valid over
a larger range of Reynolds numbers [7]. Entrance length effects are References
accounted for using:
[1] Klote JH, Milke JA. Principles of smoke management. Atlanta, Georgia:

American Society of Heating, Refrigerating and Air-Conditioning Engineers
x
aE ¼ 1 þ 0:25 exp 0:125 : (9) (ASHRAE), Inc.; 2002.
Dh [2] Miller RS, Beasley D. On stairwell and elevator shaft pressurzation for smoke
control in tall buildings. Building and Environment 2009;44:1306e17.
Standard entrance length corrections often are not valid “too [3] Miller RS, Beasley D. Smoke control by pressurisation in stairwells and eleva-
near” the entrance as they exhibit singularities (1/x). Therefore, the tors. The Singapore Engineer; February 2009:6e11.
[4] Bowers D, Ellison J, Beasley DE, Miller RS. Numerical study of elevator and
above expression is simply an extrapolated fit to the Pr ¼ 0.7 data of stairwell shaft pressurization systems using detailed building models. In:
Fig. 11-5 in Ref. [7]. Friction factor effects are included using: Proceedings of the eighth international conference on performance-based
codes and fire safety design methods. Society of Fire Protection Engineers;
 1=2 2010. p. 397e408.
f
aF ¼ ; (10) [5] Miller RS, Beasley D. On elevator shaft pressurzation system standards and
fs codes for smokes control in tall buildings. Journal of Fire Protection Engineering
2011;50:30e40.
taken from pg. 495 of Ref. [7]. In the above, f is the rough wall [6] Jo J, Lim J, Song S, Yeo M, Kim K. Characteristics of pressure distribution and
friction factor and fS is the smooth wall friction factor. A ratio of solution to the problems caused by stack effect in high-rise residential build-
ings. Building and Environment 2007;42:262e77.
f/fs ¼ 1.1 is used for current purposes; corresponding to concrete [7] Burmeister LC. Convective heat transfer. 2nd ed. New York: John Wiley and
walls and a Reynolds number of 104 from the Moody diagram. Sons, Inc.; 1993.

View publication stats