Required Travel Distance and Exit Width for Rooms Determined

by Risk-Based Evacuation Safety Design Method

1 2 3 4
YUKA IKEHATA , JUN’ICHI YAMAGUCHI , DAISAKU NII and TAKEYOSHI TANAKA
1
Technology Center, Taisei Corporation
344-1, Nase-cho, Totsuka-ku, Yokohama 245-0051, Japan
2
Technical Research Institute, Obayashi Corporation
4-640, Shimokiyoto, Kiyose-shi, Tokyo 204-8558, Japan
3
National Institute for Land and Infrastructure Management
1, Tachihara, Tsukuba, Ibaraki 305-0802, Japan
4
Professor Emeritus of Kyoto University
485-14 Iwakura Hanazono-cho, Sakyo-ku, Kyoto 606-0024, Japan
ABSTRACT
Introducing fire risk concept in performance-based fire safety design of building is beneficial in many
aspects of evacuation safety. In this study, the risk-based evacuation safety design method developed based
on fire risk concept was applied to simplify the procedure of performance-based evacuation safety
verification of rooms of which the fire risk is relatively low. The critical travel distance to room exit and
required exit width derived from this method are found to be dependent on room area but the critical travel
distance becomes to be relatively insensitive to room area as the area increase. The particular advantage of
this method is that engineers can easily develop the evacuation safety standards for safe room evacuation
and designers can check the evacuation safety performance of a large number of rooms in a building
efficiently using the developed standards. The comparison of the derived standards for travel distance and
exit width with the travel distances and exit widths in actual building rooms revealed that the method in this
paper is reasonable and practicable in actual verification practices of evacuation safety of rooms.

KEYWORDS: performance based design, evacuation risk in fire, simple verification of room evacuation,
standard of evacuation safety design

NOMENCLATURE LISTING
A room area (m2) design-based acceptable evacuation risk
R DA
Ap parent room area (m2) (persons/fire)
B exit width (m) t time (s)
Bcrit required exit width (m) tASE critical evacuation time (s)
Ccas casualty toll per hazardous fire (persons/fire) tRSE required evacuation time (s)
Ci consequence in scenario i (persons) ts smoke filling time (s)
Co initial number of occupants (persons) tstart evacuation starting time (s)
l travel distance to room exit (m) v travel speed of occupants (m/s)
lcrit critical travel distance (m)
N exit flow factor (= 1.5 person/m/s) Greek
q occupant density (persons/m2) α fire growth factor (kW/s2)
pcas probability of casualty occurrence αD fire growth factor for design (kW/s2)
probability of hazardous fire occurrence subscripts
Phf
(fire/m2 year) A acceptable
Pi probability of scenario i to occur crit critical
Qf heat release rate (kW) superscripts
RD design-based evacuation risk (person/fire) D Design based

INTRODUCTION
Existing prescriptive building codes place greater emphasis on fire safety provisions (e.g., smoke exhaust
systems or fire escape stairs) for larger buildings with a greater number of occupants than for smaller
buildings with fewer occupants. This implies that the existing prescriptive codes intend to control the
probability of occurrence of severe fire events according the degree of impact of such an event. In other
words, the consideration for fire risk control is embedded in the existing prescriptive codes, albeit

FIRE SAFETY SCIENCE-PROCEEDINGS OF THE ELEVENTH INTERNATIONAL SYMPOSIUM pp. 919-932 919
COPYRIGHT © 2014 INTERNATIONAL ASSOCIATION FOR FIRE SAFETY SCIENCE/ DOI: 10.3801/IAFSS.FSS.11-919
 implicitly and empirically. In Japan, a performance-based evacuation safety verification method was
introduced by the amendment of the Building Standard Law (BSL) in 2000, aiming to provide greater
flexibility in building design and clarity in building regulatory systems [1]. However, a serious drawback of
the evacuation safety verification method is that it does not consider the fire risk aspect, which is
incorporated into the prescriptive provisions in the BSL. The design fires are prescribed, regardless of the
size of the space and occupant load, i.e., regardless of the potential consequences. This also means that the
stages of evacuation, i.e., (1) room evacuation, (2) floor evacuation, and (3) building evacuation, under the
performance-based fire safety design method in the BSL are treated equally in the evacuation safety
verification method. However, in a high-rise office building, the safety of staircases, which would be used
by all the occupants of the building in the event of a fire, is obviously most important, followed by the
safety of common escape routes such as corridors, which would be used by all the occupants of a floor. On
the other hand, the importance of room evacuation routes is relatively low considering the occupant load
and the probability of a fire occurring in a particular room.
There is tremendous number of rooms in a large building and the rooms are frequently changed with time
due to many reasons, e.g., change of tenants, change of business environment. It is necessary to establish an
efficient method for verifying the evacuation safety of rooms in order to reduce the workload of building
designers and fire safety engineers and thus enable them to devote greater efforts toward floor and the
whole building so as to enhance the safety of building evacuation routes and staircases.
Although fire risk assessment methods have been discussed by many researchers [2], they have hardly been
applied in real designs of the evacuation safety by now, partly because the concept accepting certain
number of casualties as a risk may not fit comfortably in the expressions in a law, and also because the
acceptable level of risk is seldom clarified.
To address the abovementioned problems, we have proposed a performance-based (P-B) fire safety design
(FSD) method based on risk concept named Risk-based Evacuation Safety Design Method (R-B ESDM),
whereby the design fires and scenarios are systematically determined for the verification of evacuation
safety of building spaces by using the value of the evacuation risk based on building use and scale [3-7]. In
the R-B ESDM, the evacuation risk in the event of a fire in the context of P-B FSD is defined as an
expected number of casualties exposed to smoke. If it is verified that no casualty occurs under the design
fire selected under this method, the evacuation risk is automatically controlled below the acceptable level.
In this paper, the design fires selected on the basis of the R-B ESDM are applied to the evacuation safety
verification of rooms and a method is proposed to identify the conditions under which the evacuation safety
of rooms is approved, in terms of required travel distance and exit width. In addition, the conditions for
evacuation safety design derived by the R-B ESDM are compared to those of the rooms in an actual office
and mercantile occupancy building to determine the practicability with regard to the evacuation safety
standards.

CONCEPT OF RISK-BASED EVACUATION SAFETY DESIGN METHOD

Procedure of Fire Safety Verification

The purpose of the R-B ESDM is to control the design-based evacuation risk for the object building within
the design-based acceptable evacuation risk. The R-B ESDM assumes that evacuation safety verification is
conducted according to the usual P-B FSD procedure, as illustrated in Fig. 1. The R-B ESDM is different
from the current P-B FSD in that the design fires and scenarios to be checked are systematically selected.
The primary components of the R-B ESDM include: (1) acceptable evacuation risk in the context of FSD,
(2) design fire with variable fire growth factor, and (3) scenario event tree, established based on the success
or failure of the fire safety system. In the R-B ESDM, the design fire growth factor, αD, is selected by
considering the probability density function of the fire growth factor, α, and the acceptable evacuation risk
in each scenario. When no casualty occurs under the design fire, Qf, it is verified that the evacuation risk for
scenario is below the acceptable value.

Q f = α Dt 2 (1)

FIRE SAFETY SCIENCE-PROCEEDINGS OF THE ELEVENTH INTERNATIONAL SYMPOSIUM pp. 919-932 920
COPYRIGHT © 2014 INTERNATIONAL ASSOCIATION FOR FIRE SAFETY SCIENCE/ DOI: 10.3801/IAFSS.FSS.11-919
 Start
Area, Occupancy, Fire safety system
Evacuation Safety Acceptable risk
Design
Scenario event tree
Design Fire & Scenario Scenario Probability
generate
Prediction of Scenario acceptable risk
Fire and Evacuation Calculation Methods

Safety Criteria Data  Base

No Fire occurrence
Meet
Criteria? probability ratio

Yes Frequency histogram of

fire growth factor
End P-­‐B  FSD  Procedure Design occupants density

Fig. 1. Procedure of risk-based evacuation safety design method.

Definition of Evacuation Risk

The evacuation risk in the R-B ESDM is defined as the products of the probability of fire occurrence and
the number of casualties exposed to smoke under the fire. According to the concept that the evacuation risk
of an arbitrary space of a building is controlled within the acceptable risk, RA,

Phf (K )A(K )Ccas (K ) ≤ R A ≡ Phf (DH )A(DH )Ccas (DH ) (2)

We have tried to determine RA by using the statistical data. Although there is room for argument regarding
the acceptable evacuation risk, we assume the current risk of general dwelling as in Eq. 2. Using Eq. 2, we
have proposed the acceptable evacuation risk in the context of the R-B ESDM, RAD(K), for the arbitrary
space. The RAD for arbitrary space K is given as [8]:

RAD (K ) = 1.1× Phf* (K )×125 / A(K ) [ = Ccas (DH ){Phf (DH ) / Phf (K )}{A(DH ) / A(K )}] (3)

where Phf* is the ratio of the probability of the occurrence of a hazardous fire per unit area of the
occupancy type of space K relative to that of the dwelling. The 125 on the right-hand side in Eq. 3 is the
average area and 1.1 is the design-based casualty toll per hazardous fire in the dwelling. Although the
casualty toll per hazardous fire obtained from the statistical data is 0.3 persons/hazardous fire, the value is
modified to the design-based value, replacing the average occupant density with that prescribed in the P-B
evacuation safety design [8]. The ratios of hazardous fire occurrence per floor area of typical building types
are obtained by using the national building fire data from the Fire and Disaster Management Agency.
Ratios of the occurrence of hazardous fires are shown in Table 1.
Table 1. Ratios of the occurrence of hazardous fire by unit floor area for different occupancies.
Ratio Retail Apartment
a Theater Restaurant Hotel Hospital School Office Dwelling
P hf Shop building
P hf (DH )
1.2 0.5 7.2 3.1 1.5 9.0 9.7 4.1 1.0
P hf (K )
a
DH : dwelling; K : arbitrary occupancy.
The Ccas of Eq. 2 may be rewritten as the product of the ratio of occupants who fail to evacuate safely, i.e.,
Ccas = pcasC0. Therefore, by using Eq. 2, the screening room area, Asc, for which the evacuation risk is
obviously lower than the acceptable evacuation risk can be obtained as

FIRE SAFETY SCIENCE-PROCEEDINGS OF THE ELEVENTH INTERNATIONAL SYMPOSIUM pp. 919-932 921
COPYRIGHT © 2014 INTERNATIONAL ASSOCIATION FOR FIRE SAFETY SCIENCE/ DOI: 10.3801/IAFSS.FSS.11-919
 pcas (DH ) q(DH ) * ⎡ 0.14 0.06 * ⎤ (4)
Asc (K ) = Phf A(DH ) ⎢= 125 Phf ⎥
pcas (K ) q(K ) ⎣ pcas (K ) q(K ) ⎦

where 0.14 is used for pcas(DH) in dwellings, which was obtained from a past study [8]. There is rarely a
case in which all occupants of a fire room die in actual fires, i.e., pcas(K) = 1.0. Nevertheless, the extreme
example assumes that all occupants fail to evacuate for other occupancies; thus, using the hazardous fire
occurrence ratio Phf* in Table 2, we can calculate Asc as shown in the first row of Table 2. The number of
occupants is estimated by the occupant density, q, set forth in the BSL for the evacuation safety verification
method, which is also shown in Table 2. It may appear that the areas that can be screened out are too small
for most occupancies. This is attributable to the extreme assumption that pcas = 1.0. Usually, the dwelling is
the occupancy whose pcas is the highest. The screening areas calculated by using pcas(K) = 0.14 of the
dwelling are also shown in Table 2. In such a room, because the evacuation risk is evidently lower than the
acceptable evacuation risk, there is no need for any particular verification of the evacuation safety; it is
only necessary to check the room area. Therefore, a room that is larger than Asc needs to be assessed for
evacuation safety.
Table 2. Room areas for which the evacuation risk is below the acceptable evacuation risk, Asca.
Occupancy Retail Apartment
Theater Restaurant Hotel Hospital School Office Dwelling
Shop building
p cas (K ) (q =1.5) (q =0.7) (q =0.5) (q =0.16) (q =0.06) (q =0.125) (q =0.7) (q =0.125) (q =0.06)
1.0 10 10 43 50 57 97 43 65 47
0.14 28 26 116 134 152 260 114 175 125
a
Units: Asc (m2), q (persons/m2).

Fire Scenario Based on Success or Failure of Fire Protection System

The fire spread for evacuation varies, depending on whether or not the fire protection systems function.
However, only the worst cases are considered in P-B FSD. Multiple scenarios considering the success or
failure of the equipment for fire protection are evaluated in this method as shown in Fig. 2.

Sprinkler system Smoke control system Scenario Scenario Number of

Start of Fire suppresses the fire is activated ID Probability Casualties
Success
psm 1 P1 C1
Success
psp Failure
2 P2 C2
Hazardous 1-psm
Fire Success
psm 3 P3 C3
Failure
1-psp Failure
4 P4 C4
1-psm

Fig. 2. Scenario event tree constructed based on success or failure of sprinkler system and smoke control
system.
Moreover, the design-based evacuation risk, RD, is defined as the sum of Pi and Ci in the arbitrary fire
scenario, i, obtained for the event trees developed based on the success or failure of the equipment for fire
protection, as shown in Eq. 5:

R D = ∑ Pi Ci (5)

FIRE SAFETY SCIENCE-PROCEEDINGS OF THE ELEVENTH INTERNATIONAL SYMPOSIUM pp. 919-932 922
COPYRIGHT © 2014 INTERNATIONAL ASSOCIATION FOR FIRE SAFETY SCIENCE/ DOI: 10.3801/IAFSS.FSS.11-919
 In the R-B ESDM, safety criteria to verify the evacuation safety in buildings are set at a strict level, such as
“not exposed” or “only slightly exposed” to smoke. The purpose of the R-B ESDM is to control RD for the
object building within the design-based acceptable evacuation risk, RAD, as shown in Eq. 6:

R D ≤ R AD (6)

Design Fire Growth Factor

The fire growth factor, α [9], is regarded as the probability variable in this paper, because in a real fire, its
distribution depends on various factors, e.g., the location, material property, and quantity of the fire source
and the manner of ignitions. The design fire growth factor, αD, can be obtained by solving Eq. 7,
considering the number of occupants in the room, C0 (= qA), and RAD obtained by Eq. 3:

∞ R AD ⎛ R AD ⎞
(7)
∫α f (α )dα = ⎜⎜ = ⎟⎟
D C0 ⎝ qA ⎠

where f(α) is the probability density function of α. The αD is dependent upon room area, A, because RAD/C0
is a function of A. In Eq. 7, because the fire equipment is not considered, the probability of the scenario is
equal to 1. The concepts of Eq. 7 is explored in some references ([3–5, 9]) and [ANNEX-A]. The f(α) is
modeled as a lognormal distribution, defined as in Eqs. 8–10 based on the distribution of α, estimated from
the statistical data from the fire report of National Fire Defense Agency for fire statistics, as shown in Fig. 3
[9]:

1 ⎡ − {log(α ) − λ}2 ⎤
f (α ) = exp⎢ ⎥ (8)
2π ζα ⎣ 2ς 2 ⎦

1
λ = ln µ − ς 2 (9)
2

⎛ σ 2 ⎞
ς 2 = ln⎜⎜1 + 2 ⎟⎟ (10)
⎝ µ ⎠

where µ is mean, σ is standard deviation, λ is mean of logarithmic normal distribution, and ζ is standard
deviation of logarithmic normal distribution, of which the values are shown for the occupancies of a retail
shop and an office.
0.20 100% 0.20 100%
Cumulative  Probability(%)
Cumulative  Probability(%)

0.16 μ=0.0236 80% 0.16 μ=0.0236 80%

Relative  Frequency
Relative  Frequency

σ=0.030 σ=0.034
0.12 60% 0.12 60%

0.08 40% 0.08 40%

Statistics Statistics
Theoretical  Curve Theoretical  Curve
0.04 20% 0.04 20%

0.00 0% 0.00 0%
0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.10
0.11
0.12

0.00
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.10
0.11
0.12

Fire  Growth  Factor α (kW/s2) Fire  Growth  Factor  α  (kW/s2)

(a) Retail shop (b) Office
Fig. 3. Distribution of fire growth factor [9].

FIRE SAFETY SCIENCE-PROCEEDINGS OF THE ELEVENTH INTERNATIONAL SYMPOSIUM pp. 919-932 923
COPYRIGHT © 2014 INTERNATIONAL ASSOCIATION FOR FIRE SAFETY SCIENCE/ DOI: 10.3801/IAFSS.FSS.11-919
 Determining f(α) by inserting µ and σ in Fig. 3 into Eqs. 9 and 10, Eq. 7 can be solved for αD. The αD for
the retail shop and the office building is shown as a function of room area in Fig. 4. As Fig. 4 indicates, αD
becomes large as the area of room increases, because it is assumed that the hazardous fire occurrence
probability increases in proportion to the area of the room, and accordingly, the evacuation risk increases.

The starting point of αD does not correspond with 0 m2 in Fig. 4 because the evacuation risk does not
exceed the acceptable evacuation risk in small rooms, even if all occupants fail to evacuate.

0.40
Design fire growth
factor αD[kW/s2]
0.30
Shop
0.20
0.10 Office

0.00
0 500 1000 1500
Room area, A[m2]
Fig. 4. Design fire growth factor versus floor area.

SIMPLE FORMULAS FOR VERIFYING ROOM EVACUATION SAFETY (CRITICAL TRAVEL

DISTANCE AND REQUIRED EXIT WIDTH)

Derivation of Basic Formula

Because virtually no provision for evacuation safety is imposed in usual dwellings, the safety verification
of room evacuation should be considerably simplified for small rooms in a building, because the evacuation
risk attributable to the occurrence of fire in a small room is often extremely low. For this purpose, we
propose a calculation method that explicitly estimates the critical travel distance and the required exit width,
which can allow occupants to exit a room without being exposed to smoke under the determined design fire.
The room evacuation safety is verified based on the concept of RSET and ASET, because it is the most
familiar among most fire safety engineers. We verify that the evacuation time, tRSE, is below the critical
evacuation time, tASE, as shown in Eq. 11:

t RSE ≤ t ASE (11)

The tRSE is determined either by the travel time of the occupant from the remotest location to exits or the
time of queuing at exits. Therefore tRSE is given as follows:

⎛ l qA ⎞ (12)
t RSE = max⎜ , ⎟
⎝ v NB ⎠

Zstart tstart
ts
Zcrit tASE Queuing time
pA
Travel time NB
l tRSE
v

Fig. 5. Description of tASE and tRSE.

FIRE SAFETY SCIENCE-PROCEEDINGS OF THE ELEVENTH INTERNATIONAL SYMPOSIUM pp. 919-932 924
COPYRIGHT © 2014 INTERNATIONAL ASSOCIATION FOR FIRE SAFETY SCIENCE/ DOI: 10.3801/IAFSS.FSS.11-919
 The tASE is the difference between the critical evacuation time, ts, and the evacuation starting time, tstart, as
follows:

t ASE = t s − t start (13)

Equation 14 is obtained by substituting Eqs. 12 and 13 into Eq. 11.

max{l / v, qA /(NB)}≤ t s − t start (14)

Here, ts and tstart are calculated based on the smoke height. The tstart assumes that occupants start to egress
when the smoke layer descends to a certain height (Zstart [= 90 % of the ceiling height m]) [10]. The ts is the
time when the smoke layer descends to a height that is hazardous to occupants (Zcrit [=1.8 m]). Both tstart
and ts are calculated by using a generally used smoke filling formula for a time-squared fire. Applying the
fire given by Eq. 1, we obtain [5, 11, 12]:

3/ 5
⎧⎪ 5 x s ρ s A p ⎛ 1 1 ⎞⎟⎫⎪
t start = ⎨ ⋅ ⎜ − (15)
1/ 3 2/3 ⎬
⎪⎩ 2 C mα D ⎜ Z start
⎝ H 2 / 3 ⎟⎠⎪
⎭
3/ 5
⎧⎪ 5 x s ρ s Ap ⎛ 1 1 ⎞⎟⎫⎪
t s = ⎨ ⋅ ⎜ − ⎬ (16)
1/ 3 2/3
⎪⎩ 2 C mα D ⎜ Z crit
⎝ H 2 / 3 ⎟⎠⎪⎭

where H is room ceiling height (m), ρs is smoke layer density (= 1.0 kg/m3), xs is the adjustment coefficient
of smoke layer density, and Cm is the coefficient of the plume flow rate equation (kg/kJ1/3/m5/3/s2/3).
Moreover, in actual office rooms, many rooms are often arranged within a tenant area for various purposes,
e.g., meeting rooms and executive office rooms. In the event of fire, the occupants in such a room have to
evacuate through the room that connects to a corridor with an exit. Here, let us call the room connecting to
the corridor the “parent room” (Fig. 6). In this case, the conditions of a tenant area for evacuation safety are
calculated by substituting the area of a parent room (m2), Ap, in Eqs. 15 and 16, when a fire occurs in a
parent room.
When simple calculation formulas are used to estimate ts, the density of the smoke layer is often assumed
as constant (1.0 kg/m3) and the distance from the virtual point heat source is neglected. However, in a room
with a large floor area, smoke filling takes considerable time, during which the temperature of the smoke
layer may rise considerably because of the increase of the heat release rate with time so that the change of
smoke density becomes no longer negligible. Therefore, we introduce the adjustment coefficient of smoke
layer density, xs, in Eqs. 15 and 16, which is calculated as shown in ANNEX-B.

Ac1 Ac2
Ap (=A m2)
Parent room Ap
Fire (Ap+Ac1+Ac2=A m2)
Fire

Exit door Exit door Exit door Exit door

Corridor Corridor

Fig. 6. Definition of smoke filling an area.

FIRE SAFETY SCIENCE-PROCEEDINGS OF THE ELEVENTH INTERNATIONAL SYMPOSIUM pp. 919-932 925
COPYRIGHT © 2014 INTERNATIONAL ASSOCIATION FOR FIRE SAFETY SCIENCE/ DOI: 10.3801/IAFSS.FSS.11-919
 From Eq. 14, two critical conditions for safe evacuation can be derived: (1) critical travel distance to the
room exit and (2) required exit width, as described in the following.
(1) Critical travel distance, lcrit
Invoking the travel time from the remotest position to the exit, l v ≤ t s − t start ,

the critical travel distance, lcrit, is expressed as

lcrit ≡ (t s − t start )v (17)

Therefore, lcrit is obtained by substituting Eqs. 15 and 16 into Eq. 17 as

3/5 3/ 5 3/ 5
⎧⎛ 1 1 ⎞⎟ ⎫⎪⎧⎪ (x s ρ s Ap ) ⎫⎪
3/5
⎛ 5 ⎞ ⎪⎜ 1 ⎞⎟ ⎛ 1
⎜
lcrit = ⎜⎜ ⎟⎟ ⎨⎜ 2 / 3 − 2 / 3 ⎟ − ⎜ 2 / 3 − 2 / 3 ⎟ ⎬⎨ 1/ 5 ⎬v
⎝ 2C m ⎠ ⎪⎩⎝ Z crit H ⎠ ⎝ Z start H ⎠ ⎪⎭⎪⎩ α D ⎪⎭
(18)
3/5 3/5
⎧⎛ 1 1 ⎞ ⎛ 1 1 ⎞ ⎫⎧ (x ρ A ) 3/5
⎫⎪
⎪ ⎪⎪ s s p 3/ 5
= 8.1⎨⎜ 2 / 3 − 2 / 3 ⎟ − ⎜ 2 / 3 − 2 / 3 ⎟ ⎬⎨ 1/ 5 ⎬v = MK l Ap
⎜ Z crit H ⎟ ⎜ Z start H ⎟
⎪⎩⎝ ⎠ ⎝ ⎠ ⎪⎭⎪⎩ α D ⎪⎭

where M and Kl are given by Eqs. 19 and 20, respectively, as follow:

3/ 5 3/ 5
⎧⎛ ⎛ 1 ⎞ ⎫⎪
⎪⎜ 1 1 ⎞⎟ 1 (19)
M = 8.1⎨ 2 / 3 − 2 / 3 − ⎜ 2 / 3 − 2 / 3 ⎟ ⎬
⎜
⎪⎩⎝ Z crit H ⎟
⎠
⎜ Z start H ⎟ ⎪
⎝ ⎠ ⎭

Kl = v
(x s ρ s )3 / 5 (20)
α 1D/ 5

(2) Required exit width, Bcrit

Invoking queuing time at the exit, Bcrit is also calculated as follows:

qA (21)
Bcrit ≡
N (t s − t start )

qA α 1D/ 5 K A (22)
Bcrit = = B 3/ 5
N M (xs ρ s Ap )3/ 5
M Ap

where KB is given by Eq. 23 and M is given by Eq. 19:

qα 1D/ 5 (23)
KB = 3/ 5
N (x s ρ s )

when a room is not divided, Eqs. 18 and 22 can be transformed as follows:

lcrit = MK l A
3/ 5
, Bcrit = K B A 2 / 5 (24)
M

FIRE SAFETY SCIENCE-PROCEEDINGS OF THE ELEVENTH INTERNATIONAL SYMPOSIUM pp. 919-932 926
COPYRIGHT © 2014 INTERNATIONAL ASSOCIATION FOR FIRE SAFETY SCIENCE/ DOI: 10.3801/IAFSS.FSS.11-919
 CALCULATION RESULTS OF CRITICAL TRAVEL DISTANCE AND REQUIRED EXIT
WIDTH
The critical travel distance and the required exit width in the room obtained by this verification method are
compared with those from the provision of the BSL.
In the BSL, there is no standard for restricting lcrit and Bcrit in a room, although there is a limitation of
maximum travel distance from the exit to the outdoors or to the evacuation stair. However, the limit value
is constant, indifferent to room area. For example, the travel distance to the stairs in a retail shop is
restricted to 40 m when a principal structure involves semi-fireproof construction or the interior finish
material is inflammable and to 50 m when the interior finishing material is nonflammable. On the other
hand, the exit width to the stairs must be 0.27 m or more per 100 m2 of area on each floor above ground
level in the case of retail shops.
Figure 7 shows the relation of lcrit and A of a retail shop and an office for the cases in which the ceiling
heights are 2.4 m and 3.0 m; the calculations were made by Eq. 24 using v = 1.0 m/s for a retail shop and v
= 1.3 m/s for an office room. Moreover, the travel distance limits to the escape stairs (40 and 50 m)
prescribed in the BSL provision are shown in Fig. 7.
The limits of the travel distances in the BSL vary with the fire rating of the structure and lining, but are
constant, indifferent to the area of the room. Fig. 7 shows that the calculation results of critical travel
distance with this method tend to be approximately constant for the area at 500 m2 or more. In the retail
shop, the lcrit for H = 2.4 m of the BSL does not meet the lcrit derived from this method, as shown in Fig.
7(a). In the office building, the lcrit for room ceiling height H = 2.4 m is comparable to the lcrit of BSL, 50 m
for A = 500 m2 or more.
100 100
Critical travel distance,

Critical travel distance,

80 80 This method (H=3.0m)

This method (H=3.0m)
60 60 This method (H=2.4m)
lcrit [m]

lcrit [m]

BSL BSL
40 40
This method (H=2.4m)
20 20

0 0
0 500 1000 1500 0 500 1000 1500
Room area, A [m2] Room area, A [m2]
(a) Retail shop (b) Office
Fig. 7. Comparison of critical travel distance between this method and the BSL.
Figure 8 shows the relation of exit width, Bcrit, calculated by Eq. 24 as a function of the room area, A, in
which N = 1.5 persons/m/s is used. Figure 8(a) shows the required exit widths for the rooms of the retail
shop, with ceiling height H = 2.4 and 3.0 m. The widths of exits to the evacuation stairs prescribed in the
BSL, 0.27 m/100 m2, are also shown in Fig. 8(a). The Bcrit required in the provisions of the BSL is the
smallest among the required exit widths in Fig. 8(a), so the demand of the BSL is much looser than the Bcrit
derived by this method. If we convert the Bcrit in Fig. 8(a) to the required exit width per occupant by using
the occupant density given in the evacuation safety verification method of the BSL, 0.5 person/m2, the
values become approximately 1.9, 1.0, and 0.54 cm per person for H = 2.4 m, H = 3.0 m, and the BSL
provision, respectively. Because similar codes in many countries require approximately 0.8–1.0 cm/person,
the calculated required exit width for H=3.0 m, 1.0 cm/person, is approximately equivalent to the
international average, and the width prescribed by the BSL may be too small, if the occupant density, 0.5
person/m2, is not an overestimation. Figure 8(b) shows the required exit width for office rooms with ceiling
height H = 2.4 and 3.0 m. If the values are again converted to the widths per person, the values are
approximately 0.8 and 1.6 cm per person for the ceiling heights H = 2.4 and 3.0 m. The 0.8 cm/person for
H = 2.4 is equivalent to the international average. From the previously shown comparisons of the critical
travel distances and required exit widths, it may be suggested that the maximum travel distance in the BSL
should be less restrictive and that the required exit width in the BSL should be more demanding.

FIRE SAFETY SCIENCE-PROCEEDINGS OF THE ELEVENTH INTERNATIONAL SYMPOSIUM pp. 919-932 927
COPYRIGHT © 2014 INTERNATIONAL ASSOCIATION FOR FIRE SAFETY SCIENCE/ DOI: 10.3801/IAFSS.FSS.11-919
 16 16

Required exit width, Bcrit [m]

Required exit width, Bcrit [m]

14 14
12 12
10 10
8 This method (H=2.4m) This method 8 This method (H=3.0m)
(H=3.0m)
6 6
This method (H=2.4m)
4 4
2 BSL 2
0 0
0 500 1000 1500 0 500 1000 1500
Room area, A [m2] Room area, A [m2]
(a) Retail shop (b) Office
Fig. 8. Relation between required exit width and room area.

Comparison of Conditions between This Method and Actual Buildings

Figures 9 and 10 show comparisons of the conditions for this method and those in actual buildings where
the fire safety designs have been evaluated by the current evacuation safety verification method of the BSL.
The number of rooms investigated in actual buildings includes 273 retail shops and 81 offices for the area,
the travel distance, the exit width, and the ratio of the parent room, Ap, to A. The percentages that the ratio
of the parent room is less than 0.5 are 1.1 % in retail shops and 30.9 % in office rooms; thus, office rooms
are used by dividing the parent room to make many annexed rooms. In this comparison, the rooms with
typical ceiling height are considered, i.e., 2.8 and 3.5 m for retail shops and 2.6 and 3.0 m for office rooms.
However, lcrit and Bcrit are not considered in the ratio of the parent room.
100 100 lcrit(H=3.0)
Travel Distance [m]

80
Travel Distance [m]

80
lcrit(H=3.5) lcrit(H=2.6)
60 lcrit(H=2.8) 60

40 40

20 20

0 0
0 500 1000 1500 0 500 1000 1500
Room Area [m2] Room Area [m2]
(a) Retail shop (b) Office
Fig. 9. Comparison between lcrit of this method and travel distance of actual buildings.
12 12
10 10
Exit Width [m]
Exit Width [m]

8 Bcrit(H=2.8) 8
6 6
4 4 Bcrit(H=2.6)

2 BSL 2
Bcrit(H=3.5) Bcrit(H=3.0)
0 0
0 300 600 900 1200 1500 0 300 600 900 1200 1500
Room Area [m2] Room Area [m2]
(a) Retail shop (b) Office
Fig. 10. Comparison between Bcrit of this method and exit width of actual buildings.

FIRE SAFETY SCIENCE-PROCEEDINGS OF THE ELEVENTH INTERNATIONAL SYMPOSIUM pp. 919-932 928
COPYRIGHT © 2014 INTERNATIONAL ASSOCIATION FOR FIRE SAFETY SCIENCE/ DOI: 10.3801/IAFSS.FSS.11-919
 In both occupancies, the travel distances of actual buildings tend to increase for relatively small room areas
and to be constant from approximately 500 m2, just like the lcrit of this method. However, the lcrit calculated
in this study is larger than the actual travel distances. This is why the actual travel distances are decided by
the restriction of travel distance to the stair, based on the provision in BSL. Moreover, lcrit is not considered
about the ratio of the parent room.
In actual retail shops, the trend that the exit width is proportional to A is shown in Fig. 10(a). The exit
widths of actual retail shops tend to be close to the Bcrit, except for some small rooms, whereas there are
only small numbers of rooms that do not satisfy Bcrit. The lower limit of Bcrit is regarded as the BLT. On the
other hand, the exit widths of actual office rooms are much wider than the required exit width, Bcrit, and
scatter widely, as shown in Fig. 10(b).
As shown in Figs. 9 and 10, the overall tendencies of travel distances and exit widths in rooms of actual
building are that they can reasonably clear the proposed levels based on the method in this paper. This
implies that the proposed critical travel distance and required exit width can be accepted without pains to
the owners, users, or designers. Considering that the function of any standards for the fire safety of
buildings is to reject extraordinarily unsafe building features, the proposed method for critical travel
distance and required exit is assessed to be reasonable.

CONCLUSION
We proposed a method for calculating the critical travel distance and required exit width to evacuate
without exposing occupants to smoke in a room based on the R-B ESDM. The effects of the difference
between building occupancy and the ceiling height of the room were assessed. The standards for evacuation
safety design using this method were compared with the prescriptive provisions of the BSL. In addition, the
travel distance to exits and exit widths of the rooms in the actual buildings designed by using the
evacuation safety verification method were compared with the proposed standards. Accordingly, the
following results were obtained.
• Because the design fire growth factors are determined according to the evacuation risk reflecting room
area, reasonable standards can be derived for critical travel distances to exits and required exit widths.
• The critical travel distance according to this method depends on room area, but becomes relatively
insensitive when the room area is 500m2 or larger. On the other hand, the required exit width is
proportional to room area, as in the case of the BSL provision.
• Compared with the BSL, the critical travel distance in this method tends to be easier, whereas the
required exit width is more severe.
• The ceiling height has a significant effect on the critical travel distance and the required exit width in
this method.
• It is possible to develop simple verification standards for the evacuation safety of rooms by using this
method so that verification of the evacuation safety of rooms can be efficiently conducted, including
when the layout of a room is changed from the original.
Finally, this method can be applied when a sprinkler system is taken into account for evacuation safety,
although this paper only dealt with non-sprinkler situations.

ACKNOWLEDGEMENT
This study is conducted as the work of the committee of the P-B design of fire the Japan Association for
Fire Science and Engineering. We would like to express our appreciation to all the committee members for
their helpful cooperation.

ANNEX-A:
The conceptual distribution of the fire growth factor, α, and between the number of casualties and α, is also
illustrated on the left-hand side of Fig. A1. Here, α is described as a stochastic variable. To determine the
shape of distribution for α, an analysis was performed, Deguchi [9], which was modeled as logarithmic
normal distribution. Although no occupant will be injured during a small α, casualties will begin to occur at
the instant of a certain level, αD. Although it is thought that casualties actually increase gradually under αD
or more, the shape of the distribution for C (α) is unknown. It is assumed that all occupants (C = C0) cannot

FIRE SAFETY SCIENCE-PROCEEDINGS OF THE ELEVENTH INTERNATIONAL SYMPOSIUM pp. 919-932 929
COPYRIGHT © 2014 INTERNATIONAL ASSOCIATION FOR FIRE SAFETY SCIENCE/ DOI: 10.3801/IAFSS.FSS.11-919
 evacuate safely at αD or more, because the maximum number of casualties is C0. Hence, the evacuation risk
can be represented as a function of the fire growth factor. The threshold value of αD that satisfies Eq. 6 can
be obtained by solving Eq. 7.

Evacuation failure, C(α)

Probability density, f(α)
f(α)
C0

Evacuation risk
C(α) C0 ∫α∞ f (α )
Actual (unknown) D

αD Fire growth factor, α αD Fire growth factor, α

Fig. A1. Conceptual relation among f(α), C(α), and α.

ANNEX-B: CALCULATION METHOD FOR ADJUSTMENT COEFFICIENT OF SMOKE

LAYER DENSITY
The simple prediction equations for smoke filling time for t2 fire, Eqs. 15 and 16, are provided as practical
fire safety engineering tools [10, 11]. In this study, the smoke layer density of Eqs. 15 and 16 are slightly
modified to take into account of the change of smoke layer densities and the distance of virtual point heat
source due to the increase of HRR during smoke filling in terms of the adjustment coefficient of smoke
layer density, xs, determined by the calculation method shown below. The calculation procedure is as
follows [14]:
(Step 1) Calculate heat release rate of design fire and representative length of fire source

Q f (t ) = α D t 2 (b1)

D(t ) = Q f (t ) qʹ′ʹ′ (b2)

where Qf is heat release rate of design fire (kW) , t is the time (s) (dt = 1), αD is fire growth coefficient of
design fire (kW/s2), D is representative length of fire source (m) and q" is fire release rate per unit floor
area (kW/m2)
(Step 2) Calculate virtual point of heat source distance and flow rate of fire plume
2/5
z 0 (t ) = 1.02 D(t ) − 0.083Q f (t ) (b3)

5/3
m p (t ) = C m Q f (t )1 / 3 {Z (t − 1) + z 0 (t )} (b4)

where z0 is virtual point of heat source distance (m) , mp is flow rate of fire plume (kg/s), Cm is plume
coefficient (=0.076 kg/kJ1/3/m5/3/s2/3) and Z is smoke layer height (m)
(Step 3) Calculate wall area which is exposed to smoke layer
Aw (t ) = A + Lw{H − Z (t − 1)} (b5)

where Aw is wall surface area which is exposed to smoke layer (m2), A is floor area (m2), Lw is perimeter of
wall (＝4 A ) (m) and H is ceiling height (m)

FIRE SAFETY SCIENCE-PROCEEDINGS OF THE ELEVENTH INTERNATIONAL SYMPOSIUM pp. 919-932 930
COPYRIGHT © 2014 INTERNATIONAL ASSOCIATION FOR FIRE SAFETY SCIENCE/ DOI: 10.3801/IAFSS.FSS.11-919
 (Step 4) Calculate smoke layer temperature, smoke layer density and smoke layer height
Q f (t ) − c p m p (t )(Ts (t − 1) − T0 ) − hk Aw (t )(Ts (t − 1) − T0 )
Ts (t ) = Ts (t − 1) + Δt (b6)
c p ρ s (t − 1) A(H − Z (t − 1) )

ρ s (t ) = 353 Ts (t ) (b7)

1 ⎡ Q f (t ) − hk Aw (t )(Ts (t ) − T0 ) m p (t ) ⎤ (b8)
Z (t ) = Z (t − 1) − ⎢ + ⎥ Δt
A ⎣⎢ c p ρ 0T0 ρ 0 ⎦⎥

where Ts is smoke layer temperature (K), T0 is ambient temperature (=293 K), hk is effective heat transfer
(=0.015 kJ/K.m2) , ρs is smoke layer density (kg/m3) and ρ0 is ambient density (kg/m3)
(Step 5) Calculate available safe evacuation time by computer model, from Step1 to Step 4
t ASE (com) = t (Z = Z ) − t (Z = Z )
cr it start
(b9)

where tASE(com) is available safe evacuation time calculated by computer model, t(Z=Hlim) is smoke filling time
to the critical smoke layer height, Zcrit (=1.8m), and t(Z=Zstart) is smoke filling time to the smoke layer height
at which occupants start to evacuate, Zstart(=0.9H)
(Step 6) Calculate adjustment coefficient of smoke layer density
The adjustment coefficient of smoke layer density xs is obtained by equating the ASET, tASE by simple
calculation formula and the precise computation above and solving the equation for xs, i.e.:

5/3
1 / 3 ⎡ ⎤
2 C mα D ⎢ t ASE (com ) ⎥ (b10)
xs = ⋅ ⋅ 3 /5 3 / 5 ⎥
5 ρ s A ⎢ 1 / Z crit
⎣
2/3
− 1/ H 2 / 3 ( 2/3
− 1 / Z start − 1/ H 2 / 3 ) ( ) ⎦

The results are shown in below Fig. B1 for retail shop and office occupancy with different ceiling heights.
As seen in Fig. B1, the adjustment coefficient of smoke layer density xs, decreases as room area increases
but xs is not affected by the difference in ceiling height.
This is an unexpected but very convenient feature for use in evacuation safety design practice. This
interesting result is suspected to be attributed to the balance of the mass and heat that are accumulated in
the smoke layer until the layer descend to the critical height ( =1.8m) but we need further investigation for
clearer understanding.

1.2 1.2
CH2.5 CH3.0 CH2.5 CH3.0
1 CH3.5 CH4.0 1 CH3.5 CH4.0
0.8 CH8.0 CH12.0 CH8.0 CH12.0
0.8
Approximated curve Approximated curve
xs [-]
xs [-]

0.6 0.6
0.4 0.4
0.2 0.2
y = -0.17Ln(x) + 1.40 y = -0.20Ln(x) + 1.61
0 0
0 500 1000 1500 0 500 1000 1500
floor area A [m2] floor area A [m2 ]
(a) Retail shop (b) Office
Fig. B1. Adjustment coefficient of smoke layer density.

FIRE SAFETY SCIENCE-PROCEEDINGS OF THE ELEVENTH INTERNATIONAL SYMPOSIUM pp. 919-932 931
COPYRIGHT © 2014 INTERNATIONAL ASSOCIATION FOR FIRE SAFETY SCIENCE/ DOI: 10.3801/IAFSS.FSS.11-919
 REFERENCES
[1] “Japanese Verification Methods for Determining safe Evacuation of a Floor and Building”, The
Building Center of Japan, 2005.
[2] For example, Shida, K., Tsujimoto, M. and Yanagisawa, M., “Evaluation Method of Life Risk in
Case of Building Fires”, Architectural Institute of Japan, No.368, 1986, pp.69-78 (in Japanese).
[3] Tanaka, T., “Risk-based Selection of Design Fires to ensure an Acceptable Level of Evacuation
Safety”, IAFSS Symposium, 2008, pp.49-61.
[4] Tanaka, T., Nii, D., Yamaguchi, J., Notake, H., Ikehata, Y., “The Risk-based Evacuation
Safety Design Method and Its Practicability in the Safety Verification of Realistic Buildings”,
Fire Sci. Tech. -- Proc. 8th Asia-Oceania Symposium, 2010.
[5] Tanaka, T., “Integration of Fire Risk Concept into Performance-Based Evacuation Safety
Design of Buildings”, Fire Safety Science -- Proc. of 10th Int’l Symposium, Maryland, USA,
2011.6 (Howard Emmons Invited Plenary Lecture).
[6] Nii, D., Yamaguchi, J., Notake, H., Ikehata, Y. and Tanaka, T., “Risk-based Selection of design
Fire Scenarios in Performance based Evacuation Safety Designs of Buildings”, 8th SFPE
Symposium, Lund, Sweden, 2010.
[7] Ikehata, Y., Notake, H., Yamaguchi, J., Tanaka, T., “Analysis of Fire Statistics for Establishing
Benchmark Fire Risk for Evacuation Safety Designs of Buildings”, 8th SFPE Symposium, Lund,
Sweden, 2010.
[8] Notake, H., Ikehata, Y., Yamaguchi, J. and Tanaka, T., “Estimation of Design-based Benchmark
Fire Risk Derived from Statistical Analysis by Type of Buildings- Study on Risk-based Evacuation
th
Safety Design Method -”, Fire Safety Science -- Proc. of 10 Int’l Symposium, IAFSS, 2011,
pp. 1073-1086.
[9] Deguchi, Y., Notake, H., Yamaguchi, J. and Tanaka, T., “Statistically Estimations of Distribution
of Fire Growth Factor - Study on Risk-based Evacuation Safety Design Method -”, Fire Safety
Science -- Proc. of 10th Int’l Symposium, IAFSS, 2011, pp. 1087-1100.
[10] Yamaguchi, J., Nagaoka, T., Notake, H., Kuwana, H., Ikehata, Y., Takahashi, W., Sano, T. and
Hagiwara, I., “Development of Verification Methods for Evacuation Safety In Building Fire –Part
1 Estimation method of the pre-movement time of evacuation-“, Proceedings of annual meeting
AIJ, A-2, 2010, pp.287-290 (in Japanese).
[11] Yamaguchi, J. and Tanaka, T., “Simple Equations for Predicting Smoke Filling Time in Fire
Rooms with Irregular Ceilings”, Fire Science and Technology, Vol. 24, No. 4.
[12] NFPA 92B: Smoke management System in Malls, Atria, and Large Areas, 1991 Edition, NFPA,
1991.
[13] Tanaka, T. and Yamada S., “BRI2002 Two Layer Smoke Transport Model”, Fire Science and
Technology, Vol. 23, No. 1(special issue), 2004.
[14] Kujime, M. and Tanaka, T., “An Evaluation Method for Evacuation Safety from Fire Room Based
on Required Smoke Extraction Rates”, Transaction of AIJ, No. 586, 2004, pp.1-8 (in Japanese).

FIRE SAFETY SCIENCE-PROCEEDINGS OF THE ELEVENTH INTERNATIONAL SYMPOSIUM pp. 919-932 932
COPYRIGHT © 2014 INTERNATIONAL ASSOCIATION FOR FIRE SAFETY SCIENCE/ DOI: 10.3801/IAFSS.FSS.11-919