Journalhttp://jfs.sagepub.

com/
of Fire Sciences

Experimental and Numerical Studies on Longitudinal Smoke Temperature

Distribution Upstream and Downstream from the Fire in a Road Tunnel
L. H. Hu, R. Huo, H. B. Wang and R. X. Yang
Journal of Fire Sciences 2007 25: 23
DOI: 10.1177/0734904107062357

The online version of this article can be found at:

http://jfs.sagepub.com/content/25/1/23

Published by:

http://www.sagepublications.com

Additional services and information for Journal of Fire Sciences can be found at:

Email Alerts: http://jfs.sagepub.com/cgi/alerts

Subscriptions: http://jfs.sagepub.com/subscriptions

Reprints: http://www.sagepub.com/journalsReprints.nav

Permissions: http://www.sagepub.com/journalsPermissions.nav

Citations: http://jfs.sagepub.com/content/25/1/23.refs.html

>> Version of Record - Jan 10, 2007

What is This?

Downloaded from jfs.sagepub.com at UMKC Libraries on December 1, 2014

Experimental and Numerical
Studies on Longitudinal Smoke
Temperature Distribution
Upstream and Downstream
from the Fire in a Road Tunnel
L. H. HU,* R. HUO AND H. B. WANG
State Key Laboratory of Fire Science, University of Science and
Technology of China, Hefei, Anhui, China

R. X. YANG
Yunnan General Fire Brigade, Kunming, Yunnan, China
(Revised November, 2005)

ABSTRACT: Four full-scale tests are conducted in a real road tunnel, with
ceiling jet temperature distributions measured 200 m upstream and downstream
from the fire. Two sizes of pool fires, 1.8 and 3.2 MW, with two different fire
surface heights, 0.2 and 1.7 m from the floor level, are considered. Longitudinal
ventilation velocities are also varied. The experimental data obtained in
these four tunnel fire tests are used for validation of FDS 4.0 parallel simulation
on tunnel fires. The ceiling jet temperature distributions upstream and
downstream, and thus the back-layering length are compared. Results show
that the temperatures predicted by FDS 4.0 are near to the measured data.
In near fire regions, for instance, not more than 40–80 m away from the fire,
the temperature predicted is very close to the full-scale data. There is a deviation
between the predicted value and the measured one, shown to be mainly less
than 4–5
C, at positions further away. The back-layering length predicted by
FDS 4.0 also seems to agree fairly well with that deduced from the full-scale
experiments.

KEY WORDS: experimental, numerical, temperature distribution, road tunnel fire.

*Author to whom correspondence should be addressed. E-mail: [email protected]

Figures 1, 2 and 4 appear in color online: http://jfs.sagepub.com

JOURNAL OF FIRE SCIENCES, VOL. 25 – January 2007 23

0734-9041/07/01 0023–21 $10.00/0 DOI: 10.1177/0734904107062357
ß 2007 SAGE Publications

Downloaded from jfs.sagepub.com at UMKC Libraries on December 1, 2014

24 L. H. HU ET AL.

INTRODUCTION

WITH THE FAST development of transportation in China, more and

more tunnels had been constructed. There are about 6000 railway
tunnels with a total length of about 3200 km, and about 1200 road
tunnels with a total length of about 400 km. These tunnels are mainly
located at the western part of China as there are many mountains in the
area. More than half of the railway tunnels are located in the four
provinces of western China: SiChuan Provience, Shangxi Province,
YunNan Provience, and GuiZhou Province. In YunNan Province where
90% of the area is covered by mountains, nearly 30 road tunnels were
constructed or are under construction in the past 10 years [1,2].
Several catastrophic fires have occurred in tunnels in past years.
Examples are the accidents in Mont-Blanc [3] and Tauern in 1999 [4] ,
Kitzsteinhorn funicular in 2000, Gotthard in 2001, Dague in 2003 [5],
and the recent one in Frejus in France [6]. In the past 3 decades, 7 large
railway tunnel fires, six of them due to oilcan train fires and one due to a
gas canister explosion, and 1 large road tunnel fire have also occurred
in China, as listed in Table 1 [1,2]. The latest large tunnel fire, which
occurred in Daegu, Korea on February 18, 2003 caused 198 deaths [5].
Due to the serious results, tunnel fire and its prevention have attracted
much attention in recent years. In order to provide appropriate fire
protection, the physics of fire growth and smoke spreading in a tunnel
should be well understood [7–9].
Many studies have been performed on tunnel fires. Bench-scale
tests have been performed to study the fire characteristics, smoke
movement and control in the case of a tunnel fire [8,9]. Numerical
simulations have also been tried to predict the fire development and to
investigate the efficiency of different smoke control methods in
tunnels [7,10,11]. Functional relationships have been derived from
laboratory-scale tests [8,9]. All the numerical simulation results
and the functional relationships should be validated, especially by
full-scale data.
Some large/full-scale tests have also been carried out in past years
[12,13]: in 1992, tests were performed in a disused Norwegian mine
tunnel, nominally 5.5 m high, 6.5 m wide, and 2.3 km in length under the
EUREKA EU499 project [12]. The main objective of this series of tests
was to determine the heat output of fire in tunnels, during fire fighting
and rescue [12,14,15]. Afterwards, an extensive series of experiments
were conducted at HSL [16], Buxton in a 366 m long, 2.56 m high
tunnel with a cross-section of 5.4 m2, to provide data suitable for the
validation of computational fluid dynamics (CFD) simulation. In 1995,

Downloaded from jfs.sagepub.com at UMKC Libraries on December 1, 2014

 Table 1. Large railway/road tunnel fires in the past 3 decades in China.

Time Accident tunnel Cause Economic loss Death/injured

1976.3.23, 16 : 36 FengSha Railway 46# tunnel Oilcan lorry fire 400 thousand RMB with traffic 0/0
interruption of 54h and 34min
1976.10.18, 15 : 15 BaoCheng Railway Oilcan lorry fire 1,076.84 thousand RMB with traffic 75/9
BaiShuiJiang140# tunnel interruption of 382h and 15min
1987.8.23, 7 : 44 LongHai Railway Oilcan lorry fire 2400 thousand RMB with traffic 2/0
ShiLiShan 2# tunnel interruption of 201h and 56min
1990.7.3, 14 : 56 XiangYu Railway LiZiYuan tunnel Oilcan lorry fire 5000 thousand RMB with traffic 4/14
interruption of 550h and 56min
1992.9.15, 4 : 58 QingZang Railway 18# tunnel Oilcan lorry fire 1326 thousand RMB with traffic 0/0
interruption of 82h and 19min
1993.6.12, 10 : 53 XiYan Railway Lin Oilcan lorry fire 5614.2 thousand RMB with traffic 8/10
Jiachuan tunnel interruption of 579h and 16min
1998.7.10 GuiZhou Tunnel Gas canisters Not reported >80/not reported
explosion

Downloaded from jfs.sagepub.com at UMKC Libraries on December 1, 2014

2002.1.10, 9 : 50 MaoLiLing Road Tunnel Truck fire Traffic interruption of 18 days 0/0
Longitudinal Smoke Temperature Distribution in a Road Tunnel
25
26 L. H. HU ET AL.

large-scale tests, known as The Memorial Tunnel Fire Ventilation Test

Program (MTFVTP) [17], were carried out in a disused two-lane
highway tunnel, in Virginia (USA), with a length of 850 m and a
longitudinal slope of 3.2%, to evaluate the capacity of longitudinal and
transverse smoke control systems and to obtain the optimal manage-
ment. All of these full-scale tests have provided very useful data, but still
with some limitation: most of them were conducted in spaces with
limited dimensions, while the real scale of a road tunnel should be up to
7 m high and 8 m wide. In EUREKA EU499 tests, unfortunately, the
walls were also rough and the cross-sectional area varied significantly
with location [13]; while in MTFVTP tests, the fire dynamics were not
studied sufficiently.
Some full-scale tests have also been conducted recently, such as
the Shimizu tests [18] and the Benelux tests [19,20] in 2001, and
the Runehamar tests in 2003 [21,22]. The detection and suppression
systems were studied in the Benelux tests, along with the influence of
longitudinal ventilation on fire size and fire development [19]. Large-
scale fire tests were carried out with heavy goods vehicle (HGV) cargos in
the Runehamar tunnel in Norway. Gas temperature development [21]
and heat release rate [22] were investigated.
With the rapid development of computers, CFD or fire-field models
seems to be another way to study fire in tunnels. The software package,
fire dynamics simulator (FDS), based on large eddy simulation (LES)
with a postprocessing visualization tool, SMOKEVIEW, developed
by the National Institute of Standards and Technology (NIST), USA is
now a practical tool for simulating fires. This model had been applied
to study different fire scenarios. The latest version 4.0 of this software
package was released on February 14, 2005 [23]. In order to account
for big jobs with large calculation workloads, a new feature of
parallel processing was introduced into FDS 4.0. Although some
comparisons and verifications of this CFD model have been performed
with some full-scale burning tests on room fire, how it predicts
the smoke spread, especially by parallel processing technology in long
tunnels is still unknown. This should be studied carefully before
applying this model in such scenarios. However, not many studies
have been conducted on this and very few full-scale experiments are
reported in the literature.
For a small fire size such as a burning car, the flame tip does not
touch the ceiling. Buoyancy-induced turbulence will give a smoke
plume and a ceiling jet as shown in Figure 1. The development of

Downloaded from jfs.sagepub.com at UMKC Libraries on December 1, 2014

Longitudinal Smoke Temperature Distribution in a Road Tunnel 27

I II III IV

One-dimensional spreading

h u

Internal jump

Side view

I II III IV
Top view

Figure 1. Schematic view of development of smoke layer generated by fire in

tunnels.

a buoyant smoke layer in a tunnel can be summarized into four phases

or regions [24,25].
. Impinging region of rising plume on the ceiling
. Radial spread of smoke under the ceiling after impingement
. Interaction with side walls, and thus the transition region to
one-dimensional spread
. One-dimensional spreading
Before the smoke layer falls into the one-dimensional spreading, an
internal jump [24–26] will happen, resulting in energy loss and
considerable air entrainment, as shown in Figure 1. This can happen
before the smoke layer interacts with the side walls or afterwards,
determined by the width and height of the tunnel according to similar
research by Delichatsios on a corridor smoke layer generated by
fires [25].
For the buoyancy-driven spread of fire smoke along a long tunnel,
the temperature should be a key parameter. The complex heat transfer,

Downloaded from jfs.sagepub.com at UMKC Libraries on December 1, 2014

28 L. H. HU ET AL.

including convective and radiative heat loss to the boundaries,

causes the continuous decay of the smoke temperature along the
tunnel, resulting in buoyancy loss and slowing down of the smoke
flow travels. When a smoke flow travels along the tunnel, there would be
a boundary layer contacting the tunnel ceiling. The smoke flow
temperature would decrease along the tunnel primarily due to the
heat loss to the tunnel ceilings through this boundary layer. In order to
predict the smoke flow temperature distribution or the smoke spread
in a long tunnel, the heat loss to the boundaries should first be treated
carefully. So, the smoke ceiling jet temperature should be one of the
important parameters for validation of fire models in long tunnel
scenarios.
Another feature for fire smoke flow in tunnels is that there should be
two opposite directional flows, upstream and downstream from the fire
source, under limited longitudinal ventilation. The interaction between
smoke flow and longitudinal air wind was of different patterns between
the upstream flow and the downstream flow. The travel direction of the
upstream smoke flow was opposite to that of the longitudinal air wind,
while it was the same for the downstream smoke flow. So, what is the
difference in ceiling jet temperature distribution between upstream and
downstream? This should also be interesting for study.
In this study, four full-scale tests were conducted in a vehicular
tunnel. The temperature distribution of smoke ceiling jet flow, upstream
and downstream of the fire, below the ceiling was measured.
Corresponding numerical simulations were also performed by FDS 4.0
with a parallel processing technology. The predictions by FDS 4.0 were
compared with the data obtained from the full-scale tests for validation
of this model.

FULL-SCALE EXPERIMENTS

Four tests were conducted in a full-scale road tunnel – YuanJiang1#

Road Tunnel. This tunnel is located in the western part of China,
belonging to the highway systems between the cities Yuanjiang and Yuxi
in Yunnan Province, China. The cross-sectional shape and the plane
view of the tunnel are shown in Figure 2. The height and width are
7.2 and 10.8 m respectively. The length of the tunnel is about 1032 m
with a slope of 2.1%. The tunnel has two tubes, with unidirectional two
lanes in each tube. A cross passage was constructed at the middle
of the tunnel, connecting the two tunnel tubes. During the tests,
the cross passage was closed. The smoke temperatures below the tunnel

Downloaded from jfs.sagepub.com at UMKC Libraries on December 1, 2014

Longitudinal Smoke Temperature Distribution in a Road Tunnel 29

(a)

(b)

7.2m

10.8m

(c)

Longitudinal wind direction

Upstream Downstream

Figure 2. Schematic of the tunnel for tests: (a) tunnel entry; (b) schematic diagram of
the cross-section; and (c) plane view of the tunnel.

Downloaded from jfs.sagepub.com at UMKC Libraries on December 1, 2014

30 L. H. HU ET AL.

Table 2. Summary of the full-scale tests.

Effective
Fuel Height of the longitudinal Ambient
Test Pool (liters Heat release pool fire source ventilation temperature
number size used) rates (MW) above the floor (m) velocity (m/s) (
C)

Test 1 1 m2 Gasoline/20 1.8 1.7 1.1 21.5

Test 2 1 m2 Gasoline/15 1.8 0.2 0.6 22
Test 3 2 m2 Gasoline/30 3.2 0.2 0.6 25
Test 4 2 m2 Gasoline/30 3.2 0.2 0.9 24

ceiling, upstream and downstream of the fire, were measured under

different fire sizes, fire surface heights, and longitudinal ventilation
velocities.
Pool fires were set as fire sources. Gasolin-90# was used as fuel. The
transient mass of the fuel was measured by a weighting system with
6 strain gages. The system can measure mass up to 60 kg. The pool size
and height from the fire surface to the tunnel ceiling were varied. The
set up of the pool fires is summarized in Table 2. The pool fire was placed
about 500 m into the tunnel.
The longitudinal ventilation velocities in the tunnel were controlled by
jet fans with different rotating speeds through a frequency inverter, or
just by natural longitudinal winds in the tunnel with jet fans turned off.
The representative longitudinal ventilation velocities were measured
by a hot-wire anemometer. Typical longitudinal ventilation velocities
in all tests are also shown in Table 2.
The horizontal distribution of smoke temperature under the ceiling
was acquired by thermal resistors. These thermal resistors were
positioned at 0.1 m below the tunnel ceiling. They were all located
at horizontal intervals of 20 m. The thermal resistors were assigned
both upstream and downstream from 200 m to 20 m and from þ20 m
to þ200 m, taking the horizontal coordinate position of the fire as zero.
The directions of ‘upstream’ and ‘downstream’ were defined in reference
to the direction of longitudinal ventilation. The accuracy was 0.5
C
for these thermal resistors.

Numerical Simulation

Numerical simulations were performed by FDS. It was developed by

the National Institute of Standards and Technology, USA. The newest
version is FDS 4.0 which was released on September 15, 2005.

Downloaded from jfs.sagepub.com at UMKC Libraries on December 1, 2014

Longitudinal Smoke Temperature Distribution in a Road Tunnel 31

The FDS solves the basic conservation of mass, momentum, and

energy equations for a thermally expandable, multicomponent mixture
of ideal gases. The basic sets of equations are [23]:
Conservation of mass:

@

þ r 
u ¼ 0:
@t

Conservation of species:

@
_ 000
ð
Yl Þ þ r 
Yl u ¼ r 
Dl rYl þ m l :
@t

Conservation of momentum:


@u

þ ðu  rÞu þ rp ¼
g þ f þ r  :
@t

Conservation of energy:

@ Dp X
ð
hÞ þ r 
hu ¼  r  qr þ r  krT þ r  hl
Dl rYl :
@t Dt L

Pool fires with the same areas as those in the tests were set as fire
sources. The heat outputs of these pool fires were set according to the
heat release rate per unit area (HRRPUA) in the FDS. The heat release
rate history of the pool fires was input by ‘RAMP’ command in FDS,
according to the mass loss rates measured with combustion efficiency
of 0.75. The ‘REACTION’ type in FDS simulation was set to be ‘CRUDE
OIL’ with ‘SOOT_YIELD ¼ 0.1’, in the FDS reaction database [23].
Concerning the computing capacity, the size of the simulation domain
was only 420 m long, with a width of 10.8 m and height of 7.8 m, being
the same size as the data sampling domain in the full-scale tests.
In order to get more efficiency by multiprocessing, the job assigned
to each processor should be similar. The simulation domain was
thus divided into four axi-symmetric similar sub-domains, each with a
corresponding mesh, as shown in Figure 3. A grid system, with smaller
grids of 0.13 m assigned near to the fire and bigger grids of 0.4 m in other
spaces, was reported [27] to give good prediction on tunnel fires
simulation by the FDS. Based on this, the size of the sub-domain was
211 m long, 5.9 m wide and 7.8 m high, divided into 630  18  24

Downloaded from jfs.sagepub.com at UMKC Libraries on December 1, 2014

32 L. H. HU ET AL.

(a) 6
5
4
3
2 Sub-domain1 Sub-domain3
1
0 Sub-domain5
−1
−2
−3
−4 Sub-domain2 Sub-domain4
−5
−6
−20 −15 −10 −5 0 5 10 15 20

(b) 8
6

4 Sub-domain1 & 2 Sub-domain5 Sub-domain3 & 4

2
02
−20 −15 −10 −5 0 5 10 15 20

(c) 8

7 Blocked

4
Sub-domain Sub-domain5 Sub-domain
3 1&2 3&4

−6 −5 −4 −3 −2 −1 0 1 2 3 4 5 6
Figure 3. Design of the simulation domain and the grid system: (a) x–y plane view;
(b) x–y longitudinal sectional view; and (c) y–z cross-sectional view.

meshes. For the communication of these four sub-domains, they were all
partly overlapped between their edges. The width of the overlapped area
for these four sub-domains was 2 and 1 m in the x- and y-direction,
respectively. And a small sub-domain 5 was also designed at the same

Downloaded from jfs.sagepub.com at UMKC Libraries on December 1, 2014

Longitudinal Smoke Temperature Distribution in a Road Tunnel 33

time for near the fire region, overlapping all the other four sub-domains.
The sub-domain 5 was 6 m long, 4 m wide and 7.8 m high, divided into
60  40  80 meshes. Finer grids were assigned in this sub-domain 5,
near the fire, as complex combustion and heat transfer process takes
place there. The longitudinal range of the sub-domain was 2 m upstream
and 4 m downstream from the fire, on account of the flame tilting
under longitudinal ventilation. The commands for setting up the
above sub-domains and the corresponding meshes in FDS4.0 are listed
as follows:
&GRID IBAR ¼ 60, JBAR ¼ 40, KBAR ¼ 80/sub-mesh5
& P D I M X B A R 0 ¼ 2 , X B A R ¼ 4 , Y B A R 0 ¼ 2 , Y B A R ¼ 2 ,
ZBAR ¼ 7.8/sub-domain 5
&GRID IBAR ¼ 630, JBAR ¼ 18, KBAR ¼ 24/sub-mesh1
&PDIM XBAR0 ¼ 210, XBAR ¼ 1, YBAR0 ¼ 0.5, YBAR ¼ 5.4,
ZBAR ¼ 7.8/sub-domain1
&GRID IBAR ¼ 630, JBAR ¼ 18, KBAR ¼ 24/sub-mesh2
&PDIM XBAR0 ¼ 210, XBAR ¼ 1, YBAR0 ¼ 5.4, YBAR ¼ 0.5,
ZBAR ¼ 7.8/sub-domain2
&GRID IBAR ¼ 630, JBAR ¼ 18, KBAR ¼ 24/sub-mesh3
&PDIM XBAR0 ¼ 1, XBAR ¼ 210, YBAR0 ¼ 0.5, YBAR ¼ 5.4,
ZBAR ¼ 7.8/sub-domain3
&GRID IBAR ¼ 630, JBAR ¼ 18, KBAR ¼ 24/sub-mesh4
&PDIM XBAR0 ¼ 1, XBAR ¼ 210, YBAR0 ¼ 5.4, YBAR ¼ 0.5,
ZBAR ¼ 7.8/sub-domain4.

The simulation domain was partly filled by ‘CONCRETE’ blockages,

as shown in Figure 3(c), to have its inner cross-section dimensions
similar to those in the full-scale burning tests. The longitudinal slopes
of the tunnels were considered by specifying the gravity vectors along
both the x- and z-direction in the FDS simulations. The internal
boundary material of the simulated tunnel was set to be same as that in
the tests. The ceiling, side walls, and floor were also set to be concrete.
The property of the ‘CONCRETE’ was just set according to the database
of FDS 4.0:
&SURF ID ¼ ‘CONCRETE’
FYI ¼ ‘Quintiere, Fire Behavior’
RGB ¼ 0.66, 0.66, 0.66
C_P ¼ 0.88
DENSITY ¼ 2100.
KS ¼ 1.0
DELTA ¼ 0.1/.

Downloaded from jfs.sagepub.com at UMKC Libraries on December 1, 2014

34 L. H. HU ET AL.

As the computing domain was not as long as the full-scale tunnel, the
longitudinal ventilation velocity u in the tunnel was initially set through
the domain by command of
&MISC U0 ¼ ‘u’/,
rather than being blown into the domain at the opening end.
All numerical simulations were conducted in a personal computer
with 2 CPUs of 3.0 GHz, total 4 CPU threads under hyper-threading
technology, and 4 GB RAM. All cases were run for a simulation time
of 750 s.

RESULTS AND DISCUSSION

The heat release rate of the pool fire was estimated according to the
mass loss rate measured:
 
Q ¼   m 
H
 
where Q is heat release rate,  is combustion efficiency, m is mass
loss rate, and
H is heat of combustion. In these full-scale tests,
Gasoline-90# was taken as the fuel with the heat of combustion
45,000 kJ kg1.
The combustion efficiency of these pool fires was justified to be about
0.75 by oxygen consumption calorimetry on measuring both the heat
release rate and mass loss rate in the laboratory. The square gasoline
pool fires with sizes of 0.5  0.5 m, 0.6  0.6 m, 0.7  0.7 m were tested.
In these tests, the combustion products were collected by an exhaust
hood above the fire and further analyzed, together with the mass loss
rate measured by the weighting system placed below the pool at the
same time, as shown in Figure 4. The combustion efficiency of these
pool fires was finally estimated to be about 0.75. The heat release
rates measured by the oxygen consumption method and that deduced
from the mass loss rate method for these pool fires are shown in
Figure 5(a)–(c). It can be seen that good agreement was shown between
the two methods. Taking the combustion efficiency to be 0.75, a typical
mass loss curve and corresponding heat release rate measured for a
1  1 m square gasoline pool fire in the tunnel are also shown in Figure
5(d). The peak heat release rate of this fire was shown to be about
1.8 MW. The peak heat release rates of the gasoline pool fires with areas
of 1 and 2 m2 were then determined to be about 1.8 and 3.2 MW,
respectively.

Downloaded from jfs.sagepub.com at UMKC Libraries on December 1, 2014

Longitudinal Smoke Temperature Distribution in a Road Tunnel 35

(a)

Hood of ISO 9705

1m

3m
3m

Pool fire
1.6m

Strain gage

Weighting system

(b)

Hood of ISO 9705

Figure 4. Combustion efficiency deducing experiments: (a) schematic drawing and

(b) photo.

Pool fires with the same areas as those in the tests were set as fire
sources. The heat outputs of these pool fires were set according to the
HRRPUA in FDS. The heat release rate histories of the pool fires were
set by ‘RAMP’ command in the FDS, according to that measured, as
shown in Figure 6.

Downloaded from jfs.sagepub.com at UMKC Libraries on December 1, 2014

36 L. H. HU ET AL.

(a) 500
0.5 × 0.5 m, by oxygen consumption
450
0.5 × 0.5 m, by mass loss rate
Heat release rate (kW)
400
350
300
250
200
150
100
50
0
0 100 200 300 400 500
Time (s)

(b) 700
0.6 × 0.6 m, by oxygen consumption
600 0.6 × 0.6 m, by mass loss rate
Heat release rate (kW)

500

400

300

200

100

0
0 100 200 300 400 500 600 700
Time (s)
Figure 5. Heat release rate estimation of pool fire in the full-scale tests: (a) for
0.5  0.5 m pool fire; (b) for 0.6  0.6 pool fire; (c) for 0.7  0.7 m pool fire;
and (d) mass loss curve measured and the heat release rate deduced in the
full-scale tests.

Typical temperature history curves measured upstream and down-

stream during one test are shown in Figure 7. It can be seen that the
ceiling jet temperature reduced significantly when traveling away from
the fire along the tunnel. Temperatures near to the fire source increased
much faster than those at positions far away from the fire. The
beginning of the rise of the temperature was recorded earlier for
positions nearer to the fire. In this test, only the first 6 thermal resistors

Downloaded from jfs.sagepub.com at UMKC Libraries on December 1, 2014

Longitudinal Smoke Temperature Distribution in a Road Tunnel 37

(c) 1000
900 0.7 × 0.7 m, by oxygen consumption
0.7 × 0.7 m, by mass loss rate
Heat release rate (kW) 800
700
600
500
400
300
200
100
0
0 100 200 300 400
Time (s)

(d) 54 2000

52 1800

Heat release rate (kW)

1600
Transient mass (kg)

Mass
50
HRR 1400
48 1200
46 1000

44 800
600
42
400
40 200
38 0
0 100 200 300 400 500 600 700
Time (s)
Figure 5. Continued.

had temperature rises. This indicated that the upstream smoke flow
only traveled about 120 m from the fire, and did not reach the position
140 m upstream.
The maximum smoke temperature measured at each point of
the thermal resistor was compared with that predicted by FDS 4.0 in
Figure 8. To start with, it can be seen that the smoke temperature
distributions predicted along the tunnel were with the same decay
pattern as the measured value. However, the predicted temperatures
seemed to be generally somewhat higher than the measured value
for the upstream flow, and lower than that for the downstream
flow. As a whole, the smoke temperature predicted was close to
that measured for positions near the fire, for instance, that not

Downloaded from jfs.sagepub.com at UMKC Libraries on December 1, 2014

38 L. H. HU ET AL.

2000

1750
FDS
1500 Measured

1250
HRR (kW)

1000

750

500

250

0
0 100 200 300 400 500 600 700 800
Time (s)
Figure 6. Tracking of heat release rate history in FDS simulation.

further than 40–80 m away from the fire. While in positions further
away, the difference between the predicted and the measured value
seemed to be larger. However, the deviations seemed mainly within
4–5
C.
Another feature of the tunnel fire smoke spread was the upstream
travel distance of the smoke flow, the so-called ‘back-layering length’.
In this study, it was indicated by the distance of the thermal resistor,
being the furthest one with temperature rise records. It can be seen that
the back-layering length was very sensitive to the longitudinal
ventilation velocity. As the longitudinal ventilation velocity increased,
the back-layering length decreased. It can also be seen that the back-
layering length should increase with the increase of fire size, according
to the comparison between tests 2 and 3. However, the back-layering
length predicted was shown to generally agree with the value deduced
from the measured temperature, though being a bit shorter or longer.

CONCLUSIONS

Experimental data obtained from four full-scale tunnel fire tests

were used for validation of FDS 4.0 parallel simulation on tunnel fires.
The ceiling jet temperature distributions upstream and downstream
were taken into account, with a consideration of back-layering length.
Results showed that the temperatures predicted by FDS 4.0 were near

Downloaded from jfs.sagepub.com at UMKC Libraries on December 1, 2014

Longitudinal Smoke Temperature Distribution in a Road Tunnel 39

(a) 110
20m 40m
100 60m 80m
90 100m 120m
140m 160m
Temperature (°C)
80 180m 200m
70
60
50
40
30
20
0 100 200 300 400 500 600 700 800
Time (s)

(b) 95
90
85 20m 40m
80 60m 80m
75 100m 120m
Temperature (°C)

70 140m 160m
65
180m 200m
60
55
50
45
40
35
30
25
20
0 200 400 600 800 1000
Time (s)
Figure 7. Typical temperature history curves measured in the full-scale tests: (a) for
upstream smoke flow and (b) for downstream flow.

to the measured data. The temperature predicted would be very close

to the measured value in regions near the fire, for instance, not more
than 40–80 m from the fire, while there was a deviation between
the predicted value and the measured one, shown to be mainly less
than 4–5
C, at positions further away. The back-layering length
predicted by FDS 4.0 also seemed to agree fairly well with that deduced
from the experiments.

Downloaded from jfs.sagepub.com at UMKC Libraries on December 1, 2014

40 L. H. HU ET AL.

(a) 80
75 Upstream, by FDS4.0
70 Upstream, measured
Temperature (°C)
65 Downstream, by FDS4.0
Downstream, measured
60
55
50
45
40
35
30
25
20
0 20 40 60 80 100 120 140 160 180 200 220
Distance from the fire (m)

(b) 75
70 Upstream, by FDS4.0
Upstream, measured
65
Downstream, by FDS4.0
60 Downstream, measured
Temperature (°C)

55
50
45
40
35
30
25
20
0 20 40 60 80 100 120 140 160 180 200 220
Distance from the fire (m)
Figure 8. Smoke temperature distribution upstream and downstream the fire,
measured and predicted by FDS 4.0: (a) test 1; (b) test 2; (c) test 3; and (d) test 4.

Note that the above results hold only for relatively small tunnel fires
such as a burning car. For a medium fire such as a bus fire, the flame
will reach the ceiling and will be directed horizontally under the ceiling,
in addition to forming a smoke ceiling jet. And for a large fire such as
a burning heavy good vehicle (HGV), a large flame-filled domain is
formed. The integral configurations of flame, plume, and ceiling
jet for these two types of fires are different from that of a small fire.

Downloaded from jfs.sagepub.com at UMKC Libraries on December 1, 2014

Longitudinal Smoke Temperature Distribution in a Road Tunnel 41

(c) 110

100
Upstream, by FDS4.0
90 Upstream, measured
Temperature (°C) Downstream, by FDS4.0
80
Downstream, measured
70

60

50

40

30

20
0 20 40 60 80 100 120 140 160 180 200 220
Distance from the fire (m)

(d) 110

100 Upstream, by FDS4.0

Upstream, measured
90
Downstream, by FDS4.0
Temperature (°C)

80 Downstream, measured

70

60

50

40

30

20
0 20 40 60 80 100 120 140 160 180 200 220
Distance from the fire (m)
Figure 8. Continued.

These two cases are not taken into account in this article and will be
reported in the future.

ACKNOWLEDGMENTS

This work was jointly supported by the National Natural Science

Foundation of China under Grant No. 50376061 and Specialized

Downloaded from jfs.sagepub.com at UMKC Libraries on December 1, 2014

42 L. H. HU ET AL.

Research Fund for the Doctoral Program of Higher Education (SRFDP)

underground Grant No. 20030358051.

REFERENCES

1. Chen, Y.J., Zhang, R.K. and Liu, F.W. (1998). Case Studies and Prevention
of Railway Tunnel Fires [M], The China Railway Press, Beijing.
2. Yun’Nan (2004). General Fire Brigade of China, Code for Design,
Construction and Management of Fire Protection System in Road Tunnels.
3. Vuilleumier, F., Weatherill, A. and Crausaz, B. (2002). Safety Aspects of
Railway and Road Tunnel: Example of the Lotschberg Railway Tunnel and
Mont-Blanc Road Tunnel, Tunnelling and Underground Space Technology,
17(2): 153–158.
4. Leitner, A. (2001). The Fire Catastrophe in the Tauern Tunnel: Experience
and Conclusions for the Austrain Guidelines, Tunnelling and Underground
Space Technology, 16(3): 217–223.
5. Hong, Won, hwa (2004). The Progress and Controlling Situation of Daegu
Subway Fire Disaster, In: 6th Asia-Oceania Symposium on Fire Science and
Technology [C], Daegu, Korea, March 17–20, pp. 28–46.
6. SCMP (2005). Fire Engulfs Alpine Tunnel, Killing Two. South China
Morning Post (SCMP), International, p. A15, 6 June, 2005.
7. Chow, W.K. (1996). Simulation of Tunnel Fires Using a Zone Model,
Tunnelling and Underground Space Technology, 11(2): 221–236.
8. Kurioka, Hitoshi, Oka, Yasushi, Satoh, Hiroomi and Sugawa, Osami (2003).
Fire Properties in Near Field of Square Fire Source with Longitudinal
Ventilation in Tunnels, Fire Safety Journal, 38(4): 319–340.
9. Wu, Y. and Bakar, M.Z.A. (2000). Control of Smoke Flow in Tunnel Fires
Using Longitudinal Ventilation Systems – A Study of the Critical Velocity,
Fire Safety Journal, 35(4): 363–390.
10. Jojo, S.M. and Chow, W.K. (2003). Numerical Studies on Performance
Evaluation of Tunnel Ventilation Safety Systems, Tunnelling and
Underground Space Technology, 18(5): 435–452.
11. Modic, Jurij (2003). Fire Simulation in Road Tunnels, Tunnelling and
Underground Space Technology, 18(5): 525–530.
12. Grant, George, B. and Drysdale, Dougal (1997). Estimating Heat Release
Rates from Large-scale Tunnel Fires, In: Fire Safety Science-Proceedings of
the Fifth International Symposium, Melbourne, Australia, pp. 1213–1224.
13. Grant, G.B., Jagger, S.F. and Lea, C.J. (1998). Fires in Tunnels, Phil. Trans.
R. Soc. Theme Issue on Fire Dynamics, 356(1748): 2873–2906.
14. Richter, E. (1995). Smoke and Temperature Development in Tunnels –
Experimental Results of Full Scale Fire Tests, In: 2nd Int. Conf. Safety in
Road and Rail Tunnels, Granada, Spain, pp. 295–302.
15. Malhotra, H.L. (1995). Goods Vehicle Fire Test in a Tunnel, In: 2nd Int.
Conf. on Safety in Road and Rail Tunnels, Granada, Spain, pp. 237–244.

Downloaded from jfs.sagepub.com at UMKC Libraries on December 1, 2014

Longitudinal Smoke Temperature Distribution in a Road Tunnel 43

16. Bettis, R.J., Jagger, S.F. and Moodie, K. (1994). Reduced Scale Simulations
of Fires in Partially Blocked Tunnels, In: Proceeding of Int. Conf. Fires in
Tunnels, Borås, October 10–11, pp. 162–186.
17. Bechtel/parsons Brinckerhoff, Memorial Tunnel Fire Ventilation Test
Program, Comprehensive Test Report, Massachusetts Highway
Department, November, 2005.
18. Kunikane, Y., Kawabata, N., Ishikawa, T., Takekuni, K. and Shimoda, A.
(2002). Thermal Fumes and Smoke induced by Bus Fire Accident in Large
Cross Sectional Tunnel, In: Proc. 5th JSME-KSME Fluids Engineering
Conference, Nov. 17–21, Nagoya, Japan.
19. Huijben, Lezing, van Ir. J.W. (2002). Tests on Fire Detection Systems and
Sprinklers in a Tunnel, Available from www.rws.nl.
20. ‘‘Project ‘Safety Test’ – Report on Fire Tests’’ Directorate-General for
Public Works and Water Management, Civil Engineering Division, Utrecht,
The Netherlands, August 2002, Available from www.minvenw.nl.
21. Lonnermark, Anders, and Ingason, Haukur (2005). Gas Temperature in
Heavy Goods Vehicle Fires in Tunnels, Fire Safety Journal, 40(6): 506–527.
22. Ingason, Haukur and Lonnermark, Anders (2005). Heat Release Rates
from Heavy Goods Vehicle Trailer Fires in Tunnels, Fire Safety Journal,
40(7): 646–668.
23. McGrattan, Kevin and Forney, Glenn (2005). Fire Dynamics Simulator
(Version 4) User’s Guide, National Institute of Standards and Technology.
24. Kunsch, J.P. (1999). Critical Velocity and Range of a Fire-gas Plume in a
Ventilated Tunnel, Atmospheric Environment, 33(1): 13–24.
25. Delichatsios, M.A. (1981). The Flow of Fire Gases under a Beamed Ceiling,
Combustion and Flame, 43(1): 1–10.
26. Chan, W.R., Zukowski, E.E. and Kubota, T. (1993). Experimental and
Numerical Studies on Two-dimensional Gravity Currents in a Horizontal
Channel, NIST-GCR-93-630, National Institute of Standards and
Technology.
27. McGrattan Kevin, B. and Hamins, Anthony (2002). Numerical Simulation of
the Howard Street Tunnel Fire, Baltimore, Maryland, July 2001, NISTIR
6902, National Institute of Standards and Technology.

Downloaded from jfs.sagepub.com at UMKC Libraries on December 1, 2014