Scholarly article on topic 'Analysis of Crowd Flow Capacity through a Door Connected to a Crowded Corridor.'

Analysis of Crowd Flow Capacity through a Door Connected to a Crowded Corridor. Academic research paper on "Chemical engineering"

CC BY-NC-ND
0
0
Share paper
Academic journal
Transportation Research Procedia
OECD Field of science
Keywords
{"evacuation experiment" / "flow rate" / "density of corridor" / simulation}

Abstract of research paper on Chemical engineering, author of scientific article — Akihide Jo, Tomonori Sano, Yuka Ikehata, Yoshifumi Ohmiya

Abstract This paper presents the characteristics of pedestrian movement around doors connected to corridors. We focused on the flow rates through a door connected to a corridor to verify the evacuation safety for a performance-based design. In full scale evacuation experiment, when the corridor connected to a door was crowded, the flow rate at the door decreased. Furthermore, the differences between the experimental results and those from multi-agent pedestrian simulator were investigated to examine the validity of the simulation.

Academic research paper on topic "Analysis of Crowd Flow Capacity through a Door Connected to a Crowded Corridor."

Available online at www.sciencedirect.com

ScienceDirect

Transportation Research Procedia 2 (2014) 10-18

Transportation

Procedía

www.elsevier.com/locate/procedia

The Conference on Pedestrian and Evacuation Dynamics 2014 (PED2014)

Analysis of crowd flow capacity through a door connected to a

crowded corridor.

Akihide Joa, Tomonori Sanob, Yuka Ikehatac Yoshifumi Ohmiyad*

aTakenaka Corporation, 1-5-1, Ohtsuka, Inzai-shi, Chiba-ken, 270-1395, Japan, bProfessor, Waseda University, 2-579-15, Mikajima, Tokorozawa-shi, Saitama-ken, 359-1192, Japan cTechnology Center, Taisei Corporation, 344-1, Nase-cho, Totsuka-ku, Yokohama-shi, Kanagawa-ken, 245-0051, Japan dProfessor, Tokyo University of Science, 2641, Yamazaki Noda-shi, Chiba-ken, 278-8510, Japan

Abstract

This paper presents the characteristics of pedestrian movement around doors connected to corridors. We focused on the flow rates through a door connected to a corridor to verify the evacuation safety for a performance-based design. In full scale evacuation experiment, when the corridor connected to a door was crowded, the flow rate at the door decreased. Furthermore, the differences between the experimental results and those from multi-agent pedestrian simulator were investigated to examine the validity of the simulation.

© 2014 Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.Org/licenses/by-nc-nd/3.0/).

Peer-reviewunderresponsibility ofDepartmentof Transport & Planning Faculty of Civil Engineering and Geosciences Delft University of Technology

Keywords: evacuation experiment; flow rate; density of corridor; simulation

1. Introduction

In performance-based fire safety design, to safely evacuate people from a building, the required safe egress time (RSET) needs to be shorter than the available safe egress time (ASET). The RSET is determined by calculating the time from ignition to the time when all the occupants have been evacuated from the building. The RSET includes the following time periods: fire ignition to detection (the detection phase); detection until occupants are notified of a fire emergency (the notification phase); notification until evacuation commences (the pre-evacuation phase); and the

* Corresponding author. Tel.:+81-476-77-1285; fax: +81-0476-47-6460. E-mail address: jou.akihide@takenaka.co.jp

2352-1465 © 2014 Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.Org/licenses/by-nc-nd/3.0/).

Peer-review under responsibility of Department of Transport & Planning Faculty of Civil Engineering and Geosciences Delft University of Technology doi: 10.1016/j.trpro.2014.09.003

start of evacuation until a safe place is reached (the evacuation phase). The evacuation phase is broken down into two parts: the travel time (the time spent moving toward a safe place) and the flow time (the time spent in congestion which is controlled by the flow characteristics). The length of the flow time is influenced by the following: the effective width, population density, speed, flow characteristics, and time taken to pass through a component. The flow time for a room is determined by the time spent passing through an exit door.

The relationship between the density and flow rate has been shown in the SFPE Handbook (Nelson and MacLennan (1995)) (Figure 3-13.8). According to this figure, the flow rate is influenced by the density. When calculating the time spent passing through a door, a constant value for the flow rate is utilized in the conventional method. However, in performance-based fire safety design, a new method should be developed to calculate an evacuation time that incorporates a varying flow rate.

In this study, full-scale evacuation experiments were conducted to confirm that a decrease in the flow rate would occur when passing through a door. This would be caused by the flow from the room merging with that of the evacuation route. The study also revealed the congestion point of the doorway and the characteristics of the flow from the other rooms that merge with that of the corridor.

A flow rate of 1.5 persons/m/s is assumed as the convention for the Building Standard Law of Japan. However, when a corridor connected to a door is crowded, the flow rate at the door decreases. In effect, when evacuees pass through an evacuation route, the flow rate can be said to be 1.5 persons/m/s until a density threshold is reached; after that, queuing begins and the door flow declines. Therefore, the effective flow rate, which changes in relation to the effective available area of the corridor, was used in Japanese performance-based codes to accurately evaluate the evacuation times of buildings during a fire.

Furthermore, the differences between the results of this experiment and "SimTread" (a multi-agent pedestrian simulator) were investigated to examine the validity of the simulation.

2. Review

In this study, we focused on the flow rate where a door connects to a corridor to verify the evacuation safety, based on a performance-based design. Much research has been conducted on the crowd flow and evacuation behavior in the event of a fire. This includes studies of the following: the basic behavior of pedestrian movement, human behavior during fires using the results from fire evacuation drills, and evacuation modeling.

There has been considerable experimental research on basic pedestrian movement (Tian et al. (2012), Boyce and Shields (2009)). The time required for evacuating of corridors (Tian et al. (2012)) and stairs (Boyce and Shields (2009)) has been investigated. The velocity, density, flow rate, and interrelationships among these factors have been obtained from experiments. A survey collecting the accounts of the survivors of real emergency situations such as those in high-rise buildings (Shields et al. (2009), Proulx (1995)) and football stadiums (Oberije et al. (2009)) has also been conducted. The modeling and calculation methods for evacuations have been studied(Nelson and MacLennan (1995), Hoskins and Milke (2012)). An overview of 30 current egress models was conducted, which included a checklist to apply when building a new model. The relative performances of the static calculation method and an agent-based dynamic pedestrian model have been compared (Waterson et al. (2010)). Finally, a method for measuring occupant speed on stairs has been discussed (Sano et al. (2010)).

A large amount of effort has gone into researching pedestrian movement at bottlenecks. However, these studies failed to take into account the decrease in the flow rate through an opening where the flow from a room merges with that of a crowded evacuation route.

3. Experiment Method

3.1. Experiment Facility

An evacuee experiment was conducted in a full-scale compartment, in which evacuees passed through a door to a corridor. Figure 1 shows the plan of the compartment. The width of the room was set at 8,100 mm, with a depth of 5,400 mm. The experimental compartment could be divided into sub-compartments A, B, and C. Each room had both a door that exited into a corridor and a door that exited directly to the outside of the experiment facility. Both

were sliding doors. The door width was set at either 1,600 or 800 mm. The doors were closed when not in use. The experiment facility was located indoors. It had no roof installed, which allowed video cameras to record the experiments.

r Dom r 3om 3 r oomv \

S 10 S 00 S 10

or 1600 or 1600 or 1600

do 3rd do jrßl roc rA 1

corr ¡dor

M " I " I " I " I " I " I " M

[_8100_I 1800 |

a) Experimental facility

Fig. 1. Experimental compartment [unit : mm]

case of 800mm 1600

J—L_

case of 1600mm b) Pattern of door opening

3.2. Evacuees

The evacuee group was comprised of 34 healthy men and 22 healthy women (56 people in total). All the evacuees were between 21 and 39 years of age (mean 30.8; standard deviation 5.5). Since the experiments were conducted during winter, the evacuees wore coats.

3.3. Measurement method

The movements of the evacuees were observed from above using six video cameras. They wore surgical caps with different colors to allow them to be easily counted. The evacuees started the evacuation simultaneously at the signal of a whistle.

3.4. Experimental conditions

Figure 2 lists the different permutations of the experimental conditions. The experiments were conducted to determine how the flow rate at the door connected to the corridor changed according to different evacuation conditions. The following conditions were varied:

1. The different size of the effective area of the corridor (case A).

2. The flow from the other rooms that merged with that of the corridor (case B).

3. The impact of a crowded corridor (case C)

Case Area [m2] Width of door [m] Evacuee [persons] Effective area of corridor [m2] Width of corridor [m2]

room A room B room C door A door B door C room room A B room C corridor total

A-1 19.44 - 1.6 - - 56 - 0 56 2.88 16

A-2 19.44 - - 1.6 - 56 - 0 56 7.20

A-3 - 19.44 - - 1.6 56 - 0 56 11.52

C-1 9.72 - 0.8 - - 28 - 28 56 10.00 16

C-2 9.72 - 1.6 - - 28 - 28 56 10.00

Fig 2. Experimental conditions

In case A, in order to confirm the change in the flow rate from the room in relation to a change in the crowd density in the corridor, the effective area of the corridor was changed. In case B, door A and door B were used to confirm the influence of people merging from the other rooms. In addition, an experiment involved changing the width of the door. In case C, it took time for the evacuees to enter the corridor when nobody was there already. Cases were also considered where the corridor was crowded from the outset. Figure 3 shows the effective area of the corridor. This was the area that was actually used for walking from the door of the room to the exit of corridor. This paper focuses on cases A-1-A-3 and case C-1-C-2, in order to confirm the influences of the effective area of the corridor and corridor congestion.

effective area of corridor (a) Case A-1

roomC roomB 56 person

®3><a «ai_______

doorC I 1600 I

effective area of corridor (b) Case A-3

effective area of corridor

(c) Case C-1

effective area of corridor-^ (d) Case C-2

Fig. 3. Experimental conditions

4. Experiment Results

4.1. Influence on flow rate of variation in density of effective area of corridor

Figure 4 shows the flow rates over time for cases A-1 to A-3, from the room to the corridor. Figure 4 shows the density of the evacuees in the effective area of the corridor.

When the density of the corridor is low (i.e., cases A-1 and A-3), the flow rate increases to over 3.0 persons/s. In case A-1, the maximum density is approximately 3.6 persons/m for 11~20s. The time taken to pass through the door of the room became longer as the effective area of the corridor became smaller. This was because the flow from the room was restricted, in response to congestion when the corridor area was smaller. The flow rate started to decrease as the density of the corridor increased (approximately 3.0 persons/m2) and became static at 1.5 persons/m during the periods of density in the corridor. This static state was seen for several seconds in the second half of case A-2, but was not seen at all in case A-3 because of the larger capacity of the corridor. The flow rate during the static state was halved compared with the rate when the density was not high. This depended on a corridor width of 800 mm.

- / / ''' \ \

—CaseA-1 —CaseA-2 - - CaseA-3 \ \

Time [s]

Fig. 4. a) Flow rate at door area connected to corridor, b) Density of effective area of corridor

4.2. Influence on flow rate of variation in density of effective area of corridor

The flow rates from door A in case A-1 (without initial congestion in the corridor) and case C-2 (with initial congestion in the corridor) are shown in Figure 4. The flow rate in case C-2 was half of the rate observed in case A-1, because the evacuees in the corridor merged in front of door A. The flow rate at door A when the corridor was crowded was proportional to the door width at the inflow from a room and the outflow from the corridor.

0 10 20 30 40

Time [s]

Fig. 5. Flow rate at door connected to corridor

—CaseA-1 —CaseC-2 _

A key map is shown in Figure 6, which assumes that the flow rate at room A is Rj and that the flow rate at the corridor is R2. In the static state, the total of their flow rates equals the flow rate at the exit (Rneck), which is shown in equation (1). Bj is the width of a door in a room, B2 is the width of the corridor.

R2 ^ Bnect>

Fig. 6. Flow rate at each door

R1 + R2 = Rneck

At the area in the effective corridor where the inflow from the room merged with the flow in the corridor, it was assumed that the densities and walking speeds of the evacuees passing through both these sections were equal, shown by equation (2), because R1 and R2 were proportional to the width.

Rj: R2 = Bj: B2

B1 and B2 were assumed to the widths of the door and corridor, respectively. The flow rates for these can be calculated using equation (3) and equation (4), respectively, based on from equation (1) and equation (2).

Rj = Bj/ (Bj + B2 )x R

R2 - B2/{BX + B2 )x R

4.3. Comparison of door width ratios for static state of corridor

Figure 6a shows the changes in the flow rate at the door for case C-1 and case C-2. The average flow rates in the static state for these cases (case C-1 was for 10-18s, and case C-2 was for 11-18s) were 0.53 persons/s and 0.73 persons/s respectively. The theoretical flow rates at door A for these cases, a calculated using equation (3), are 0.58 persons/s and 0.87 persons/s, respectively. The tendency for the flow rate of case C-1 to be lower than that for case C-2 was similar to that of the experiment results.

4.4. Relative density of corridor and flow rate

The density of the corridor and the flow rate in caseA-1 for the static state of the corridor are shown in Figure 7b. As a result of a one-way analysis of variance for the average value of each level, the density of the corridor was confirmed to have a significance of 5% for 2.4-2.6 persons/ m2, and for 2.6-2.8 persons/ m2. Thus, the threshold value at which the flow rate decreased in corridor could be assumed. Moreover, the decreasing flow rate coefficient for the door of a room is a value provided for the flow rate by the bottleneck at the exit of the corridor. As the result, it is evident that the flow rate at the door of a room depends on the density of the connected corridor when numerous of evacuees remain in front of the exit. It is seen that the flow rate has a low value when the density of the corridor in the figure has a low value. As a result, it is thought that the lack of congestion in front of a door occurred immediately after the experiment began and near the end of the experiment.

—CaseC-1 —CaseC-2 .

1 ^ I

20 Time [s]

0.2 0.6 1.0 1.4 1.8 2.2 2.6 3.0 3.6 4.0 Density of corridor [parsons/m2]

Fig. 7. a) Flow rate at door connected to corridor, b) Comparison of flow rate at door to density of corridor

5. Simulation Method

5.1. Pedestrian Simulator "SimTread"

In this section, we will examine the evacuation conditions using an evacuation simulator called "SimTread." (Sano et al. (2010)). SimTread is a type of multi-agent model, where the crowd flow characteristics are determined by the movement of individual agents, to whom the same behavioral rules are applied. The elements of the SimTread model include the following:

• An agent, or walker, whose variables include their position, direction, and maximum speed.

• A space, or building plan, whose variables include obstacles such as walls and furniture.

• A destination, whose variables include arbitrary sets, plural placements, and a target that agents walk toward.

• Agents have the a numbered list of destinations and move toward a destination on this list

The operation starts with drawing a plan, which is followed by arranging the agents and destinations. These are prepared using computer aided design (CAD) drawing software.

5.2. Comparison offlow rates from experiments and simulations for case A-1 and case A-3

The flow rates from at the door connected to the corridor of the experiment and the simulation of cases A-1 and A-3 are shown in figures 8a and 8b, respectively. A comparison in cases A-1 and A-3 shows that when the corridor started to become crowded the flow rate decreased more in the simulation than in the experiment. In the simulation, the flow rate at the door connected to the corridor became about 0.75 persons/m/s when the corridor was crowded. Moreover, it was confirmed that the flow rate at the door connected to the corridor became about 1.5 persons/m/s when the corridor was not crowded. These values are lower than that in the experiment. Moreover, it is clear that the simulation model was able to reproduce the same crowd-flow dynamics as the experiment.

Fig. 8. a) Flow rate and density of corridor for case A-1, b) Flow rate and density of corridor for case A-3

Figure 9 shows pictures taken during the experiment, along with the images of the simulation; all were taken 15s after the start of the experiment/simulation. The gray agents are moving slowly, and those in black are stationary. This demonstrates that similar walking behaviors occurred in the experiment and simulation.

5.3. Reproducing results using large number of evacuees

The static state did not continue for a long period in the experiment because the number of evacuees was limited. It was difficult to observe the static state for a long period because the evacuation finished quickly in each case. It was confirmed that the simulation and experiment were reproducible by making a comparative study of 5.2. In this chapter, the static state condition was simulated using SimTread. In cases A-1-A-3, the number of evacuees was increased and the situation in which a constantly crowded corridor existed was reproduced. Figure 10 shows the flow rate results. In all three cases, the maximum value of the flow rate was reached about 3s after the start. Then, the corridor started becoming crowded after 5 s and the flow rate became constant. This flow rate was about 1.3 persons/s. The results seemed to indicate that the flow rate was influenced by the ratio of the room's door width and the exit width of the corridor.

—CaseA-1 (Const) —CaseA-2 (Const) --CaseA-3 (Const) ••••CaseA-1 (Const Exit) -

; i V VA-7 (a7V / \ ,-' \ « i \

/\ A / V \\* ••"7V9

20 Time [s]

Fig. 10. Flow rate for large number of evacuees

6. Discussion

Figure 11 is a conceptual diagram of the flow rate through a door connected to a corridor. The flow rate through a door connected to a corridor varies according to the corridor density. When a corridor is vacant, an evacuee is able to smoothly outflow to the corridor. However when the density of the corridor becomes p1 (=2.4 persons/s) the flow rate through a door in a room reaches a maximum value (Rmax). As the corridor becomes more crowded and the density of the corridor in front of a door in a room increases, the flow rate gradually decreases and reaches a constant value. When the effective area of corridor is large, the time needed to reach in the maximum flow rate is great because the time to reach p1 is great. It depends on the flow rate at the exit because the time to reach pmax (= 2.8 persons/s) depends on the density of the corridor and when the flow rate through a door connected to the corridor became steady state. When the density of the corridor reaches the maximum value, the inflow in the corridor depends on the flow rate, which is found by dividing the flow rate of the bottleneck in the corridor proportionally according to the ratio of the flow through a door to the connected a corridor and its neighborhood as shown in equation (5). When all the occupants have exited a room, the flow rate becomes 0 person/m/s.

Pmax <P

Effective Area of Corridor :Small Effective Area of Corridor :Middle Effective Area of Corridor :Large

t1 t2 t3 t4 t5 t6 t7 t8 t9 t [s]

t1 t3 t5

t2 t4 t6

Fig. 11. Conceptual diagram of flow rate and density through door connected to corridor

Summary

In this paper, the change in the flow rate from a room according to a change in the density of the corridor connected to the room by a door was clarified using the full-scale evacuation experiments. The correspondence between the full-scale experiment and simulation was confirmed by reproducing the situation of the full-scale experiment with a multi-agent model simulator. The conclusions of this paper can be summarized as follows:

• The time to reach the limiting density of a corridor depends on the effective area of the corridor.

• The flow rate from a door in a room when the merging in the corridor reaches the limiting density depends on

the flow rate at the exit of the corridor.

• The threshold to which the flow attenuates by the merging in the corridor is 2.8 persons/m2.

• The reproducibility of the multi agent simulator was verified.

• The flow rates from the room according to the change in the merging density of the corridor over time were the

same in the simulation and experiment.

Acknowledgements

This experiment was supported by the Japanese Ministry of Land, Infrastructure, Transport, and Tourism's "Building Standard Law of Japan Improvement Project."

References

Boyce K. E., Shields J., 2009. Experimental studies to investigate merging behaviour in a staircase, Proceedings of the 4th International

Symposiumu on Human Benaviour in Fire 36(4-5), pp. 383-398. Hoskins, B. L., Milke, J. A., 2012. Differences in measurement methods for travel distance and area for estimates of occupant speed on stairs, Fire Safety Journal, 48, pp. 49-57.

Nelson, H. E., MacLennan, H. A. 1995. Emergency movement. In P. J. DiNenno (Ed.), Proulx, G., Evacuation time and movement in apartment

buildings, Fire Safety Journal, 24(3), pp. 229-246. Oberije, N., Kobes, M., Weges, J., Post, J., 2009. "Fire in Euroborg Football Stadium; Analysis of human behavior The Netherlands",

Proceedings of the 4th International Symposiumu on Human Benaviour in Fire Sano, T., Yoshida, Y., Takeichi, N., Kimura, T., Minegishi, Y., 2011. "Experimental study of crowd flow passing through a simple-shaped room and validation for an evacuation simulator", in: Conference on Pedestrian and Evacuation Dynamics, In Peacock, R. D., Kuligowski, E., Averill, J.D. (Eds.), pp. 587-599.

Shields, T.J., Boyce K. E., McConnell, N., The behaviour and evacuation experiences of WTC 9/11 evacuees with self-designated mobility

impairments, Fire Safety Journal, 44(6), August 2009, pp. 881-893 Tian, W., Song, W., Ma, J., Fang, Z., Seyfried, A., Liddle, J., Experimental study of pedestrian behaviors in a corridor based on digital image

processing, Fire Safety Journal 47, pp.8-15 Waterson, N.P., Bail, S. L., Casle, C. J. E., 2010. "A Comparison of Evacuation predictions made using Agent-based Simulation and Cade-based Approaches", 8th International Conference on Performance -Based Codes and Safety Design Methods, pp. 311-322