Physica A 512 (2018) 1140–1151

Contents lists available at ScienceDirect

Physica A
journal homepage: www.elsevier.com/locate/physa

Evacuation experiments in vertical exit passages in an

underwater road shield tunnel
∗ ∗∗
Yu-chun Zhang a,b , , Aixi Zhou c , Yue Xiang d , Chuan He a , , Qizhu Jiao e ,
Bo Wan f , Wei Xie b
a
Key Laboratory of Transportation Tunnel Engineering, Ministry of Education, Southwest Jiaotong University, Chengdu 610031, China
b
Department of Fire Protection Engineering, Southwest Jiaotong University, Chengdu 610031, China
c
Department of Engineering Technology and Construction Management, University of North Carolina at
Charlotte, Charlotte, NC, 28223, USA
d
Chongqing University of Science & Technology, Chongqing 401331, China
e
China Railway Siyuan Survey and Design Group Co. LTD, Wuhan 430063, China
f
Shanghai Tunnel Engineering Co. Ltd, Shanghai 200082, China

highlights

• Evacuation experiments were conducted in an actual underwater road tunnel.

• Concurrent descending/ascending experiments were performed on a staircase.
• Foundational experimental evacuation data were obtained in underwater road tunnels.
• Differences in evacuation behavior were explained in different vertical exits.

article info a b s t r a c t

Article history: Evacuation experiments were conducted to examine the movement speed and evacuation
Received 15 September 2017 capacity in an underwater shield road tunnel where vertical evacuation passages were
Received in revised form 22 March 2018 installed. Behavioral results were collected through several field experiments aimed at
Available online 16 August 2018
investigating vehicle evacuations, the time required to open vertical evacuation passages,
Keywords: slide and stairs descending with and without counter flows. The results show that the
Evacuation experiment individual movement speed in staircase was lower than the speed in slide, but the average
Underwater road tunnel Occupant Flow Rate (OFR) in staircase was higher than the OFR in slide. The probable
Vertical exit passages reasons were discussed and analyzed. Females had worse performance when opening the
Evacuation time exit door than males. The upstairs movement of firefighters was observed not negatively
affect the evacuation performance of occupants down stairs. The replicated experiments
showed that appropriate training or drills can help improve the movement speed of
occupants. The results of this study will be essential to evaluate evacuation capacity and
occupant behavior on stairs and slides of underwater shield tunnels.
© 2018 Elsevier B.V. All rights reserved.

1. Introduction

Over 40 underwater road tunnels have been built in China in the past several decades, due to the extraordinary growth of
economy and urbanization in the country. An underwater road tunnel plays a key role in transportation hub connecting cities

∗ Corresponding author at: Department of Fire Protection Engineering, Southwest Jiaotong University, Chengdu 610031, China.
∗∗ Corresponding author.
E-mail addresses: zycfire@swjtu.cn (Y.-c. Zhang), chuanhe21@163.com (C. He).

https://doi.org/10.1016/j.physa.2018.08.129
0378-4371/© 2018 Elsevier B.V. All rights reserved.
 Y.-c. Zhang et al. / Physica A 512 (2018) 1140–1151 1141

Fig. 1. Illustrations of crossover passages in underwater tunnels.

Fig. 2. Typical vertical passages in underwater shield tunnels.

or regions. It should be noted that there are usually large amount of vehicles and passengers in a tunnel. Occupant evacuation
in case of fire or other emergencies is therefore a critical concern for underwater road tunnel design and construction.
Both horizontal and vertical passages can be used as evacuation exits in road tunnels. Horizontal crossover passages
allow quick evacuation in emergencies, and thus have been used as a major evacuation method for dual-bore mountain road
tunnels, as shown in Fig. 1. Shield and boring construction is the main construction method for underwater road tunnels
in China [1]. An underwater shield tunnel is usually divided into two zones: an upper zone served as a ventilation and
driveway zone, and a lower zone served as a safety/service zone or driveway zone. Connections of these two zones are
vertical passages such as slides or staircases. Various styles of evacuation exits are possible in shield underwater road tunnels,
including horizontal crossover passages, vertical passages, and their combinations, as shown in Figs. 1 and 2. For example,
the crossover passages like Fig. 1(a) were used in the Western Scheldt Tunnel (the Netherlands), and slides like Fig. 2(a)
were used in the Trans-Tokyo Bay Highway Tunnel (Japan), the Wuhan Yangtze River Tunnel (China), the Nanjing Yangtze
River Tunnel (China), and the Hangzhou Qingchun Road Tunnel (China). Staircases similar to Fig. 2b have been used in the
Shanghai Shangzhong Road Tunnel (China).
Researches have been performed to examine occupant evacuation issues in road tunnels. Frantzich [2] studied the
passenger’s ability to move inside a tunnel in the event of a fire. Boer and van Zanten [3] examined the self-rescue assumption
and the adequate use of escape provisions in road tunnels. Ronchi et al. [4] analyzed different evacuation models in road
tunnels. Factors influence the evacuation performance were investigated by various studies, including: extinction coefficient,
flashing lights, and the responses of occupants to emergency situations [5–10]. Virtual reality technology was applied to
evacuation experiments in road tunnels [11–13]. However, most studies were conducted in mountain road tunnels without
1142 Y.-c. Zhang et al. / Physica A 512 (2018) 1140–1151

Fig. 3. Cross-section of the Qianjiang Road Tunnel (Unit: mm).

any vertical evacuation passages. The evacuation speed obtained in these studies did not take account of the effect of vertical
evacuation passages on evacuation performance.
A number of previous studies considering the effect of stairs have been performed for buildings [14–31]. On the one
hand, it is not clear whether the data obtained from buildings stairs could be applied in underwater road tunnels, due to
different structure and environment. On the other hand, tunnel evacuation has been researched by experimental study
[2,5,18,9,7,10,13,32–34], theoretical analysis [11,35] as well as numerical models [8,4,33,34]. However, most studies mainly
considered horizontal movement in tunnel, where vertical movement in tunnel slides or stairs was ignored. In the Trans-
Tokyo Bay Highway Tunnel, slides were installed as vertical passages at intervals of 300 m, with the assumption that an
evacuation capacity of 40–60 persons per minute (or 1.0–1.5 s per person) can be achieved in the slide [35]. Unfortunately,
analysis details and experimental data were not provided in the article. The assumed evacuation capacity of 40–60 persons
per minute on the slide in underwater road tunnels needs to be further confirmed by more experimental data.
This paper aims to examine the movement speed and evacuation capacity of the occupants in underwater shield road
tunnels where vertical evacuation passages are installed. To achieve the purpose, evacuation experiments were performed in
a real road shield tunnel with installed vertical passages. Detailed information concerning actions or activities of occupants
were obtained from several field experiments, including vehicle evacuation, passage door opening, slide descending,
staircase descending, and concurrent descending/ascending on a staircase.

2. Method

2.1. Experimental tunnel

The Qianjiang Road Tunnel is an underwater shield road tunnel located in Hangzhou, Zhejiang Province, China. It is a
unidirectional twin-tube tunnel with three lanes in each tube. The cross-section dimension of a tube is shown in Fig. 3. The
cross-section is divided into two zones by a road deck: the upper zone is used for traffic and ventilation, while the lower
zone is used for safety and other purposes. The total length of the tunnel is 4.45 km. The outer and inner diameter of the
tube is 15.0 m and 13.7 m respectively. The total traffic lane width is 11.0 m, with three traffic lanes (3.75 m, 3.75 m, and
3.50 m) plus two shoulders (0.75 m and 0.50 m). The height of the traffic lane is 5.0 m.
Evacuation slides and stairs were installed in the tunnel as vertical passages from the upper traffic zone to the lower safety
zone. Fig. 4 shows a portion sketch layout of the stairs and slides. Evacuation slides were set every 80.0 m along the right
side of the tunnel driveway, with a total of 66 evacuation slides. Rescue stairs were installed every 240.0 m in the tunnel,
with a total of 24 stairs. The height for both slides and stairs is 3.24 m. The width of the slides and the stairs are 0.78 m and
0.80 m respectively. The stairs have 17 steps, with 0.195 m riser-heights for step 1–12 and 0.18 m for step 13–17. A resting
platform (0.9 m × 0.8 m) was set on step 12 on stairs. Fig. 5 shows the structure of stairs and slides in the tunnel.

2.2. Participants

The participants were composed of volunteers and employees from a tunnel construction company in Hangzhou, China.
A total of 48 participants (37 females and 11 males) were included in this experiment. The average age was 20.8 ± 2.2
years, ranging from 16 to 30 years. The average height, weight and shoulder breadth was 163.8 ± 6.7 cm, 53.1 ± 9.1 kg and
 Y.-c. Zhang et al. / Physica A 512 (2018) 1140–1151 1143

Fig. 4. Layout of the stairs and slides in the tunnel.

Fig. 5. Structure of exist stairs and slides.

39.4 ± 2.8 cm respectively. The average body mass index was 19.7 ± 2.3 kg/m2 . All participants were able bodied. Before
each experiment, the participants were clearly informed the experimental purpose, the experimental procedure, and the
safety issues that need attention. Participants were free to participate or quit during the experimental process.

2.3. Experimental procedure

2.3.1. Vehicle evacuation

Four replications were performed for the vehicle evacuation experiment, with a 46-passenger motor coach was used. Prior
to each experiment, 44 participants were randomly assigned to be seated. Participants started to evacuate from the vehicle
after hearing a command given by a staff, and then the exit door was opened. Fig. 6a shows a snapshot of the evacuation
process. Time for evacuation from the vehicle represented the time difference between the starting command and the landing
of last passenger. Exit time interval represented the landing time difference between two neighboring participants.

2.3.2. Vertical passage door opening

A crossover passage door can be opened manually by simply unlocking and pushing (or pulling) the door. It would be more
difficult for opening a vertical passage door, which requires more effort and longer time. Participants need to manually lift
and turn a door handle to activate the hydraulic device. The door of slide is 0.78 m × 1.92 m, while the door of stair is 0.8
m × 1.0 m or 1.09 m × 1.8 m. A sign was deployed on the wall indicating the procedure of opening the passage door. A
tunnel staff also explained and demonstrated the door opening process before each experiment. Participants who did not
feel comfortable during experiment were able to choose to quit. As a result, the number of participants for slide door opening
experiment and staircase door opening experiment were 20 and 21 respectively. Snapshots of door opening experiments
were illustrated in Fig. 6b and c.
1144 Y.-c. Zhang et al. / Physica A 512 (2018) 1140–1151

Fig. 6. Snapshots of experimental procedures.

2.3.3. Slide and staircase descending

The number of participants for staircase descending experiments and slide descending experiments were 48 and 43
respectively. At the beginning of each experiment, the participants were informed with the experimental procedures and
safety issues. A starting command was given from a staff, and then participants began to move through the slides or stairs. The
time that the first participant arrived at the door was determined as the starting time. It was considered as the completion
when the last participant passed through the slide or staircase, which time was recorded as the ending time. A total of five
replications were performed for the slide and staircase descending experiments. Fig. 6d and e show the snapshots of the
descending experiments.

2.3.4. Concurrent staircase descending/ascending

The concurrent staircase descending/ascending experiment was conducted for further understanding of occupants
behavior and movement speed down stairs in emergencies while emergency responders climbing upward at the same time.
There were 43 participants walking downward stairs and 5 firefighters walking upward simultaneously. All participants
were briefed with the experimental procedures and safety issues prior to the experiment. Occupants started to move down
stairs as soon as the start signal was given. Fire fighters began to climb up the stairs when the participants reached a steady
descending speed, as shown in Fig. 6f.
 Y.-c. Zhang et al. / Physica A 512 (2018) 1140–1151 1145

Fig. 7. The layout of experimental setup.

Fig. 8. Exit time intervals during motor coach evacuation.

2.4. Data extraction

The experimental zone was 120.0 m long with a slide and a stair. 18 video cameras were deployed to record the
experimental process of all participants, as shown in Fig. 7. Key parameters were obtained, including the starting and ending
time, the door opening time, the travel time, and the total movement time. During experiments, the lighting system in the
tunnel worked in normal conditions. The average luminance of tunnel driveway was measured as 4.0 cd/m2 by referencing
the procedures in JTJ026.1-1999 [36].

3. Results

3.1. Vehicle evacuation

Time for evacuation from the vehicle in the four replicated experiments was 47.6 s, 51.0 s, 50.5 s, and 52.1 s respectively.
The average evacuation time was 50.3 ± 1.66 s.
Fig. 8 shows the exit time interval during evacuation from the vehicle. It could be found that in the first experiment, the
waiting time was the longest and had the widest variations. In the following three replications, the average waiting time
became shorter and less scattered, as the participants got familiar with the evacuation process after the first experiment.
The tunnel occupants may not be familiar with the evacuation procedures in real emergencies, thus the average waiting time
in the first experiment could be more applicable. It also indicated that appropriate training or drills are likely to reduce the
waiting time and speed up the evacuation process in emergency.

3.2. Vertical passage door opening

The door opening time was determined from the time of touching the door handle, to the time the exit door was fully
opened. The comparison of slides and stairs door opening time between male and female is shown in Table 1. More details
are illustrated in Fig. 9. The average time to open a staircase door was 12.9 s, ranging from 5.2 s to 43.3 s. Comparatively,
the average time to open a slide door was 13.4 s, with minimum and maximum time of 6.9 s and 35.7 s respectively. Some
1146 Y.-c. Zhang et al. / Physica A 512 (2018) 1140–1151

Table 1
Vertical passage door opening time (in seconds)
Door opening time Average Median Shortest Longest
Slide — Female 13.4 10.0 6.9 33.7
Slide — Male 13.5 10.8 6.9 35.7
Slide — All 13.4 10.6 6.9 35.7
Stairs — Female 16.0 11.2 6.8 43.3
Stairs — Male 9.4 6.9 5.2 21.6
Stairs — All 12.9 10.7 5.2 43.3

Fig. 9. Comparison of door opening time: slide and staircase.

Fig. 10. Comparison of door opening time: female and male.

participants had difficulty in operating the door and had to ask for assistance from another participant, contributing to longer
door opening time. This condition is likely to happen in a real emergency, thus more concerns are required to guide the door
opening.
Fig. 10 compares the door opening time between female and male occupants. The average time for females to open the
staircase door was 16.0 s, which was 70% higher than male (9.4 s). However, no significant difference was found in slide door
opening time between male (13.5 s) and female (13.4 s). Female performed worse in stair door opening test, which led to
the longest opening time (43.3 s). During our experiments, several female participants were not able to open the door, and
had to seek help from other participants.

3.3. Slide and staircase descending

Individual travel time for descending the slides and the stairs are illustrated in Table 2. A comparison of individual travel
time is shown in Fig. 11. The error bars indicate the standard deviation from each experiment. The average travel time for
 Y.-c. Zhang et al. / Physica A 512 (2018) 1140–1151 1147

Table 2
Descending individual travel time in vertical passages (in seconds)
Descending time Average Median Shortest Longest
Slide — Female 7.3 7.0 4.6 14.7
Slide — Male 6.3 6.3 4.5 8.7
Slide — All 7.0 6.5 4.5 14.7
Staircase — Female 8.3 8.4 3.6 10.3
Staircase — Male 8.3 8.6 4.23 10.5
Staircase — All 8.3 8.4 3.6 10.3

Fig. 11. Comparison of descending individual travel time in vertical passages.

Fig. 12. Comparison of average speed in building stair and tunnel vertical passage (Building stair data from [37], Junho 2014).

the slide was 7.0 ± 0.7 s, which was slightly lower than that of the staircase (8.3 ± 0.3 s). The maximum travel time was
observed in slide.

A large number of walking speed values have been obtained from different types of conditions (such as fire drills, normal
use , real fires, complied works and experiment studies) on building staircases [37]. Bryan and James [37] summarized the
effect of some variables on evacuation processes, including stair width, stair conditions, exit selection, weather, groups,
occupant spacing, gender, carrying, and body size, et al. To calculate walking speeds, there are two main tread methods
regarding travel distance: the horizontal speed and the slope speed, and many studies used the later one [37]. Based on the
slope distance calculation, the slowest average walking speed (the first experiment) for the staircase was 0.75 m/s, and the
fastest average walking speed was 0.83 m/s. These values were very close to the data obtained by Junho 2014 (0.74 m/s∼0.83
m/s), Kratchman 2007 (0.7∼0.8 m/s), and Pauls 1971 (0.61∼0.81 m/s) [31,37]. One the other hand, the slowest and fastest
movement speed on the slide was 0.85 m/s and 1.14 m/s respectively, which were higher than the value of most stairs (see
Fig. 12).

Occupant flow rate (OFR) of an exit passage was determined as the number of participants passing through the slide or
staircase divided by the corresponding movement time. In this study, the unit for OFR is persons/minute (or simplified as
P/min). Fig. 13 compares the OFRs for the slide and the staircase. The average OFR for the slide was 23.8 ± 3.3 P/min, with
the lowest value of 19.5 P/min and the highest value of 28.9 P/min. The average OFR for the staircase was 36.4 ± 1.2 P/min,
while the lowest value of 34.3 P/min and the highest value of 38.0 P/min.
1148 Y.-c. Zhang et al. / Physica A 512 (2018) 1140–1151

Fig. 13. Comparison of OFRs in descending experiments.

Fig. 14. Time interval of slide and staircase.

The results indicate that the OFRs of staircase were higher than the OFRs of slide. The average OFR for staircase (36.4
P/min) was about 53% higher than the average OFR for slide (23.8 P/min). Two possible reasons were provided: Firstly, the
slide and the staircase have different structural details, as illustrated in Fig. 5. Escape process for the slide and the staircase
were different. Movement behavior of participants almost remained the same on stairs, due to familiarity to the structure
of stairs. Participants tended to follow the person in front of them. However, most participants were normally not familiar
with the structure of a curved slide. He or she must squat down and prepare for sliding first. Thus, the time interval on the
slide was longer than that of the staircase. The second and the main reason is that only one or two occupants were allowed
on a slide simultaneously during evacuation, while the staircase was able to accommodate more than two occupants at the
same time, leading to a higher OFR value. The above mentioned two reasons led to the longer entrance/ exit time interval
on the slide than those on the stair, as shown in Fig. 14. The longer time intervals on the slide then negatively influence the
OFR.
It should be noted that the obtained OFRs for the slide (23.8 ± 3.27 P/min) in this study are much lower than the assumed
value of 40–60 P/min on the slide in the Trans-Tokyo Bay Highway Tunnel [35]. The possible reasons could be: Firstly, the
time to open the vertical passage door were included in this study but not considered in Yamada’s research. Secondly, the
much higher OFRs in Yamada’s study may be due to better training participants while the participants in this experiment
were not trained before experiment.

3.4. Concurrent staircase descending/ascending

Three replicated experiments were conducted for concurrent descending and ascending a staircase in the tunnel.
Fig. 15 compares the OFR values between concurrent experiments and descending-only experiments. The average OFR
for concurrent descending and ascending was 37.3 ± 0.6 P/min, which was slightly higher (about 2.5%) than the average
OFR for descending experiments (36.4 ± 1.2 P/min). This was not expected before the experiments, as one could estimate
that the OFR of concurrent descending and ascending would be lower than that of the descending OFR. The experimental
 Y.-c. Zhang et al. / Physica A 512 (2018) 1140–1151 1149

Fig. 15. Comparison of OFRs on stairs: descending only vs. concurrent descending/ascending.

Fig. 16. Phases in the first concurrent staircase descending/ascending experiment.

results showed a different pattern than prediction. Possible reason could be that the participants got familiar with
the stairs after the descending experiments, thus movement speed increased in the concurrent descending/ascending
experiments. Additionally, it could be found that the movement of participants appeared in phases in the first concurrent
descending/ascending experiment as shown in Fig. 16, which may help explain the results.
As shown in Fig. 16, the evacuation process could be divided into four phases. In Phase I, the beginning phase of the
evacuation, fast movement speed was observed since few participants just arrived at the entrance. Then movement speed
gradually decreased with increasing number of participants and tended to be stable in Phase II. In Phase III, firefighters moved
upward, contributing to reverse conflict with participants who were moving downward. The walking speed of participants
decreased slowly because they stopped and moved to the side of the stairs in order to give way to the firefighters. In Phase IV,
the movement speed of participants began to rise again since all firefighters reached the top exit. During the whole process of
evacuation, movement speed of participants kept a high level firstly, then followed by a sharp decrease due to movement of
firefighters, and finally increased again and kept stable at the end of the evacuation process. The phased movement could be
helpful to explain the slight higher average OFR of the concurrent descending/ascending experiments than the descending
experiments.

4. Limitations

Due to limited resources concerning occupant behavior and evacuation speed on stairs and slides of underwater shield
tunnels, evacuation experiment was conducted to provide more data. The described experimental conditions would bring
some limitations of this study, shown as follows:
(1) The experiments were not conducted under the condition of actual emergency. Thus any dangerous behavior during
evacuation process was prohibited. The results provide important data to help recognize and evaluate occupant behavior as
well as the basic evacuation capacity of vertical exit passage in underwater road shield tunnels.
(2) Obesity and age have an effect on the walking speed in buildings [38]. The participants of our studies were mostly
young adults with relatively low BMI. The results could be used as a baseline or start point for design purpose.
1150 Y.-c. Zhang et al. / Physica A 512 (2018) 1140–1151

(3) It must be recognized that the firefighters carry special equipment with a certain weight while ascending stairs in
tunnel emergencies. In this experiment, firefighters did not carry any equipment. Therefore, the obtained OFR would be
slightly higher than the real case but would not seriously influence the walking speed and evacuation capacity.

5. Conclusions

Evacuation experiments were conducted to examine the movement speed and evacuation capacity of the occupants in
underwater shield road tunnels where vertical evacuation passages are installed. Experimental data concerning movement
speed, occupant flow rate (OFR), and evacuation time were obtained. Based on the present results, the following conclusions
can be reached.
The average vehicle evacuation time for a 46-passenger motor coach was 50.3 ± 1.7 s. The average time to open a staircase
door was 12.9 s, which was slightly lower than the time to open a slide door (13.4 s). Females spent more time opening the
door than the males. Some females could only open the door with the help of others. The appropriate training or drills are
suggested to perform to help improve the evacuation performance of occupants.
The individual travel time for slide (7.0 ± 0.7 s) was lower than the time for staircase (8.3 ± 0.3 s), but the Occupant
Flow Rate of staircase (36.4 ± 1.2 P/min) was higher than the OFR of slide (23.8 ± 3.3 P/min). Possible reason could be the
different structure of slides and staircases. On the one hand, participants were more familiar with the movement on stairs. On
the other hand, the stairs were able to evacuate more participants simultaneously while only one or two participants were
allowed on slide. The average walking speed in tunnel staircase was roughly equal to that in normal buildings as preliminary
literatures introduced. The obtained OFRs for the slide (23.8 ± 3.3 P/min) were much lower than the assumed value of 40–60
P/min on the slide in the Trans-Tokyo Bay Highway Tunnel [35].
The rescue movement of fire fighters upward stairs in emergencies must be taken into consideration while occupants
evacuate down stairs in an opposite direction. The average OFR for concurrent/descending experiment and descending-
only experiment was very close, indicating the movement of fire fighters did not negatively affect the escape process in
emergencies.
The results of this study will be essential to recognize and evaluate evacuation capacity and behavior of vertical exit
passages in underwater shield tunnels. We also hope more work will be expanded in the future studies to include more
variables, such as environmental condition, age, gender, carrying, and body size.

Acknowledgments

We would like to express our gratitude to all participants who participated in this study. The study was supported by the
China National Natural Science Foundation (Grant # 51578464), the National Key and Development Program of China (Fund
# 2016YFC0802201), and the Science and Technology Project of the Ministry of Transport of China (Grant # 2009353333340).
YZ and AZ would like to thank the China Scholarship Council, China for financial support to make this collaboration possible.
Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not
necessarily reflect the views of the China National Natural Science Foundation, the National Key and Development Program
of China, the Ministry of Transport of China, and the China Scholarship Council.

References

[1] C. He, Underwater Tunnel (in Chinese), Southwest Jiaotong University Press, Chengdu, China, 2011.
[2] H. Frantzich, Occupant behaviour and response time-results from evacuation experiment, in: Proceedings of the Second International Symposium in
Human Behaviour in Fire, Boston USA, Vol. 28, 2001, pp. 159–165.
[3] L.C. Boer, D.W.V. van Zanten, Behaviour on tunnel fire, in: N. Waldau, et al. (Eds.), Predestrian and Evacuation Dynamics, Springer, New York, 2005,
pp. 91–98.
[4] E. Ronchi, P. Colonna, J. Capote, et al., The evaluation of different evacuation modes for assessing road tunnel safety analysis, Tunnelling Underground
Space Technol. 30 (2012) 74–84.
[5] H. Frantzich, D. Nilsson, Evacuation experiments in a smoke filled tunnel. The third international symposium on human behaviour in fire, Belfast UK,
2004.
[6] D. Nilsson, H. Frantzich, W. Saunders, Coloured flashing lights to mark emergency exits - experiences from evacuation experiments, Fire Saf. Sci. 8
(2005) 569–579.
[7] D. Nilsson, M. Johansson, H. Frantzich, Evacuation experiment in road tunnel: A study of human behaviour and technical installations, Fire Saf. J. 44
(2009) 458–468.
[8] J.A. Capote, D. Alvear, O. Abreu, et al., A real-time stochastic evacuation model for road tunnels, Saf. Sci. 52 (2013) 73–80.
[9] K. Fridolf, E. Ronchi, D. Nilsson, et al., Movement speed and exit choice in smoke-filled rail tunnels, Fire Saf. J. 59 (2013) 8–21.
[10] M. Seike, N. Kawabata, M. Hasegawa, Experiments of evacuation speed in smoke-filled tunnel, Tunnelling Underground Space Technol. 53 (2016)
61–67.
[11] M. Kinateder, M. Müller, M. Jost, et al., Social influence in a virtual tunnel fire -Influence of conflicting information on evacuation behavior, Appl.
Ergon. 45 (2014) 1649–1659.
[12] E. Ronchi, M. Kinateder, M. Müller, et al., Evacuation travel paths in virtual reality experiments for tunnel safety analysis, Fire Saf. J. 71 (2015) 257–267.
[13] E. Ronchi, D. Nilsson, S. Kojic, et al., A Virtual reality experiment on flashing lights at emergency exit portals for road tunnel evacuation, Fire Technol.
52 (2016) 623–647, Amerind Publishing, New Delhi.
 Y.-c. Zhang et al. / Physica A 512 (2018) 1140–1151 1151

[14] J.L. Pauls, B.K. Jones, Building evacuation: research methods and case studies, in: D. Cantor (Ed.), Fires and Human Behavior, Wiley, New York, 1980,
pp. 227–249.
[15] P.N. Daly, F. McGrath, T.J. Annesley, Pedestrian speed/ flow relationships for underground stations, Traffic Eng. Control 32 (1991) 75–78.
[16] G. Proulx, Evacuation time and movement in apartment buildings, Fire Saf. J. 24 (1995) 229–246.
[17] G. Proulx, A. Kaufman, J. Pineau, Evacuation times and movement in of fire buildings. National Research Council of Canada, internal report 711, 1996.
[18] H. Frantzich, Study of movement on stairs during evacuation using video analysis techniques. Department of Fire Safety Engineering, Lund Institute
of Technology, Lund University, 1996.
[19] T.J. Shields, K.E. Boyce, G.W.H. Silcock, et al., The impact of a wheelchair bound evacuee on the speed and flow of evacuees in a stairway during an
uncontrolled unannounced evacuation, J. Appl. Fire Sci. 7 (1997) 29–39.
[20] K.E. Boyce, T.J. Shields, G.W.H. Silcock, Toward the characterization of building occupancies for fire safety engineering: capabilities of disabled people
moving horizontally and on an incline, Fire Technol. 35 (1999) 51–67.
[21] E.R. Galea, S.J. Blake, Collection and analysis of data relating to the evacuation of the World Trade Centre buildings on 11 September 2001. Report
produced for the UK ODPM. Fire Research Technical Report 6/2005, London, 2004.
[22] T. Fujiyama, N. Tyler, An explicit study on walk speeds of pedestrians on stairs, in: 10th international conference on mobility and transport for elderly
and disabled people, Mamamatsu, Japan, 2004, pp. 643-652.
[23] S.M.V. Gwynne, E.R. Rosenbaum, Employing the hydraulic model in assessing emergency movement, 2008.
[24] J. Ye, X. Chen, C. Yang, J. Wu, Walking behavior and pedestrian flow characteristics for different types of walking facilities, Transp. Res. Rec. 2048
(2008) 43–51.
[25] E.R. Galea, G. Sharp, P.J. Lawrence, et al., Approximating the evacuation of the World Trade Center North Tower using computer simulation, J. Fire
Prot. Eng. 18 (2008) 85–115.
[26] T.J. Shields, K.E. Boyce, N. McConnell, The behavior and evacuation experiences of WTC 9/11 evacuees with self-designated mobility impairments,
Fire Saf. J. 44 (2009) 881–893.
[27] A.J. Blair, The Effect of Stair Width on Occupant Speed and Flow of High Rise Buildings (MS thesis), University of Maryland, College Park, 2010.
[28] R.D. Peacock, J.D. Averill, E.D. Kuligowski, Stairwell evacuations from buildings: what we know we don’t know. National Institute of Standards and
Technology, NIST technical note, 1624, 2009.
[29] B.L. Hoskins, J.A.A. Milke, Study of Movement Speeds Down Stairs, Springer, New York, 2013.
[30] E. Ronchi, D. Nilsson, Assessment of Total Evacuation Systems for Tall Building, Springer, New York, 2014.
[31] C. Junho, H. Wonhwa, Individual stair ascent and descent walk speeds measured in a Korean highrise building, Fire Technol. 50 (2014) 267–295.
[32] Cuesta Arturo, Abreu Orlando, Balboa Adriana, Alvear Daniel, An experimental data-set on merging flows in rail tunnel evacuation, Tunnelling
Underground Space Technol. 70 (2017) 155–165.
[33] Miho Seike, Nobuyoshi Kawabata, Masato Hasegawa, Evacuation speed in full-scale darkened tunnel filled with smoke, Fire Saf. J. 91 (2017) 901–907.
[34] Miho Seike, Nobuyoshi Kawabata, Masato Hasegawa, Quantitative assessment method for road tunnel fire safety: Development of an evacuation
simulation method using CFD-derived smoke behavior, Saf. Sci. 94 (2017) 116–127.
[35] N. Yamada, Y. Ota, Safety systems for the Trans-Tokyo Bay Highway Tunnel project, Tunnelling Underground Space Technol. 14 (1999) 3–12.
[36] MOT. Specifications for Design of Ventilation and Lighting of Highway Tunnel (JTJ 026.1-1999), Ministry of Transport (MOT), Beijing, China. Online
available at: http://www.zsda.gov.cn/uploads/soft/091203/4-0912031A050.pdf (last access 9/14/16), 2000.
[37] L.H. Bryan, A.M. James, Study of Movement Speeds Down Stairs, Springer, New York Heidelberg Dordrech London, 2014.
[38] S. Michael, A. Hamish, The effect of an ageing and less fit population on the ability of people to egress buildings, Saf. Sci. 50 (2012) 1675–1684.