- Email: [email protected]
A study on experiment of human behavior for evacuation simulation
Ocean Engineering 31 (2004) 931–941 www.elsevier.com/locate/oceaneng
A study on experiment of human behavior for evacuation simulation Dongkon Lee , Jin-Hyung Park, Hongtae Kim Korea Research Institute of Ships and Ocean Engineering, KORDI, P.B. Box 23, Yuseong, Daejeon 305 600, South Korea Received 4 August 2003; accepted 5 December 2003
Abstract Walking speed is a very important factor in evacuation analysis for human safety. To provide against emergencies such as abandonment of ship due to accident, the authority body required evacuation analysis for passenger ships. To develop a simulation tool for the evacuation analysis, human behavior data for evacuation situations are fundamentally necessary. In this paper, onboard experiments were carried out twice, once with ship motion and another without ship motion. They were performed assuming the situation where subjects are evacuating under instructions from the crew without panic. Not only individual movement but also group movement with and without motion based on inclination was covered. # 2004 Elsevier Ltd. All rights reserved. Keywords: Onboard experiment; Evacuation; Human safety; Ship design; Human behavior
1. Introduction There have been many accidents of passenger ships at sea and they have caused huge losses of human lives. The interim guidelines for evacuation analyses for new and existing passenger ships were developed by the Maritime Safety Committee (MSC) of the International Maritime Organization (IMO) and circulated as MSC/ Circ. 1033 (IMO, 2002) for the purpose of unified implementation of the requirement of evacuation analysis which is required by regulation. The MSC considered whether the interim guidelines should be made mandatory for large passenger ships
Corresponding author. Tel.: +82-42-868-7222; fax: +82-42-868-7229. E-mail address: [email protected] (D. Lee).
0029-8018/$ - see front matter # 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.oceaneng.2003.12.003
932
D. Lee et al. / Ocean Engineering 31 (2004) 931–941
and, having noted that more data on application of the interim guidelines are needed to validate the methodologies and criteria contained in the above guidelines, agreed that this issue should be further considered by the committee. In 1995, the 41 m long high speed twin hull passenger liner St. Malo (Lockey et al., 1997) was stranded off the shore of France and the passengers evacuated while the ship was listing. The total evacuation time of 308 passengers after the distress signal was recorded as 1 h and 17 min. It is more than nine times the total evacuation time recorded during evacuation drill in stationary conditions, which was 8 min. This is a good example showing that evacuation analyses that does not take account of motion, listing, and the psychological state of the passengers are meaningless in the real world. In this paper, onboard experiments were carried out twice, once with ship motion and another without ship motion. They were performed assuming the situation where subjects are evacuating under instructions from the crew without panic. Not only individual movement but also group movement with and without motion based on inclination was covered. The results showed that a floor with motion decreases the walking speed by 10–20% and an inclined floor without motion also affected walking speed. The results from the experiments will be used to increase the accuracy of the evacuation simulation system, which is being developed. 2. Related works Human behavior study in evacuation situations provides important information to engineers, managers, operators and users. It is necessary to test if an appropriate safety system exists to cope with a possible critical situation in order to allocate resources including trained personnel. In addition, designing alarm systems announcing necessary information to passengers in critical situations without correct understanding of human behavior will decrease the overall efficiency of the evacuation system. Past research reports that walking speed on a flat floor is between 0.98 and 1.39 m/s in case of Asians (Hwang et al., 1991; Fukuchi et al., 1998). Research in Japan (Ando et al., 1988) in busy train stations reports that people of about 20 years of age show the fastest moving speed. On the other hand, research from Europe and Australia reports the walking speed of adult males to be between 1.4 and 1.6 m/s. To model human movement in an evacuation situation accurately, each evacuee is described as an object with a set of parameters including age, gender, and walking speed. The values of these parameters should be acquired from actual experiments for more realistic simulation. The characteristics and experiment results of major related research are summarized. The National Maritime Research Institute of Japan experimented on an anchored ship from 1994 for 3 years (Katuhara et al., 1997, 1998). The subjects were 70–120 students of 20 years age. The subjects moved along a predefined evacuation path and the experiment was recorded using a video camera. The results of the experiment
D. Lee et al. / Ocean Engineering 31 (2004) 931–941
933
showed that the movement speeds were 1.4 m/s in corridors and 0.7 m/s for stairs. The maximum density was 3 persons/m2. The Research Institute of Marine Engineering of Japan used a total of 20 male and female adults as subjects and experimented inside a corridor (Murayama et al., v v 2000). The experiment conditions include list angles between +20 and 20 . In addition, six subjects were experimented with while the corridor was in motion v with a motion angle of 10 and a cycle time of 5 and 10 s. In case of trim, the walking speed was between 0.82 and 1.38 m/s and the speed decreased as the v v upward trim increased. However, heeling angles between 0 and 20 had insignificant effects on the walking speed. In the ship motion experiment, the walking speed was between 0.71 and 0.77 m/s, which is about 70% less than in stationary conditions (0.90 m/s). The Australian Maritime Engineering-Cooperative Research Centre (AMECRC) experimented on the walking speed in corridors and on ladders as a part of the Australian Maritime Safety Authority Program (Koss et al., 1997). The subjects were a total of 67 male and female adults 18–25 years of age. The walking speed increased as the downward trim increased. In case of upward trim and heel, the walking speed was similar to the case where there was no list. However, there was decrease in speed in case of heel when two persons walked together along a corridor of 1.2 m width. In another research in 2000 with 985 subjects, the results showed that significant decrease in speed was noticed when the subject was more than 65 years of age, and males were approximately 18% faster than females. In case of handicapped persons, the decrease in speed was significant. TNO Human Factors in Netherlands experimented on the effects of ship motion and list in corridors and on stairs on human walking speed (Bles et al., 2001). The subjects were a total of 150 adults 18–83 years of age. A ship motion simulator, where a 4:0 m 2:4 m 2:3 m sized cabin was placed on a hydraulic system, was used for the experiment. The experiments with positive trim angle showed substantial decrease in the walking speed. On the whole, ship trim resulted in 35% decrease in speed. For the ship motion experiments, walking speed decreased up to 15% as the cycle time and angle increased. 3. Onboard experiment The corridor model (10:0 mm 1:2 mm 1:9 mm) was used in the experiment. v v v The model can have trim angle between 20 and 20 , and heel angle between 0 v and 20 . A handrail is present and stairs are located at the end of the corridor as shown in Fig. 1. The following are the measuring and recording devices used in the experiment. – Eight-Channel Image Grabber Board equipped systems (digital video network transmission S/W, digital video recording S/W). – Eight CCD high resolution cameras. – Ship motion measuring device (Seatex Motion Reference Unit, MRU-5).
934
D. Lee et al. / Ocean Engineering 31 (2004) 931–941
Fig. 1. A scene of onboard experiment at sea and corridor model.
All of the experiments were carried out twice, once with ship motion and another without ship motion, except the doorway passing experiments. The experiments were carried out assuming the situation where subjects are evacuating under the instructions of the crew without panic. The following experiments were carried out. 1. Walking along a corridor without counter-flow a. Measuring the time required for a single person to cover the length of a corridor of 10 m length. b. Measuring the total time required for a group of 21 persons in single file to walk the length of a corridor of 10 m length. 2. Walking along a corridor with counter-flow a. Measuring the total time required for two groups of 10 persons each in single file to walk the length of a 10 m long corridor in opposite directions starting from each end. b. Measuring the total time required for groups of 10 persons and a single person to walk single file along the 10 m long corridor in opposite directions starting from each end. 3. Moving through a doorway a. Measuring the total time required for a group of 21 persons in single file to walk through a doorway of width 0.9 m. To replicate ship motion, the corridor model was constructed on the aft deck of a training ship of the Korea Maritime University. The experiments were carried out while the ship was sailing from Pusan harbor to Yesu harbor, 1 night and 2 days. The amount of roll and pitch was measured using the ship’s motion sensor.
D. Lee et al. / Ocean Engineering 31 (2004) 931–941
935
Fig. 2. Data capture system with Eight-Channel Image Grabber Board and CCD high resolution camera.
The subjects in the experiment were students from the Korea Maritime University. They were a total of 21, including 18 male students and three female students. The subjects were wearing safety helmets and life vests and serial numbers were painted at each side of the helmet. The experiments are recorded by eight video cameras installed in the corridor model and the recordings were stored in the computer through the Eight-Channel Image Grabber Board as shown in Fig. 2. 4. Experimental results The experiments can be classified into the following four categories. – – – –
Individual movement experiment with ship trim and heel. Group movement experiment with ship trim and heel. Individual movement experiment with ship motion. Doorway passing experiment.
Below are the summarizations of the experimental results for each category (see Table 1). 4.1. Individual movement experiment under trim and heel It is difficult to compare the results of the experiment in this research with other experiments since the experiment conditions and subject characteristics are different. Fig. 3 shows the results of this experiment with the data from AME-CRC (Koss et al., 1997) and the Research Institute of Marine Engineering in Japan (Murayama et al., 2000).
936
D. Lee et al. / Ocean Engineering 31 (2004) 931–941
Table 1 Experimental conditions Experiment under static condition
Experiment under dynamic condition
Corridor
Corridor dimension Trim Heel Combination of trim and heel Corridor dimension Trim Heel Rolling angle Roll period Coaming height of door Breadth of door
10:0 m 1:2 m 1:9 m v v v v v +20 , +10 , 0 , –10 , –20 v v v 20 , 10 , 0 Combination 10:0 m 1:2 m 1:9 m v v v +10 , 0 , 10 v v v 20 , 10 , 0 v 3–4 5–12 s 0.23 m 0.9 m
Most of the results are similar to the data reported in other experiments, but in case of trim, the results show difference. For downward trim, the speed was expected to increase, but the results showed decrease in speed. This reflects that psychological state affects human behavior. In case of upward trim, the speed did not decrease as much as in the other experiments. This is because the floor of the corridor model used in this experiment was carpeted, and thus less slippery than in case of the corridor model used in other experiments. 4.2. Group movement experiments with ship trim and heel Group movement experiments with ship trim and heel were carried out using three different methods. In the first one, 21 subjects moved along the corridor without delay to simulate the evacuation situation. Fig. 4 shows the results of this experiment compared to the individual movement experiment where the subjects moved at intervals of 3 m. It shows that the movement speed was 20% slower in group movement than in individual movement. Fig. 5 shows the results when ship trim and heel are both present. The results showed substantial decrease in speed for the case where either trim or heel was present. In the second method, two groups of 10 people each walked in opposite directions starting from each end of the corridor. This can be one of the typical situa-
Fig. 3. Individual movements with various trim and heel angles.
D. Lee et al. / Ocean Engineering 31 (2004) 931–941
937
Fig. 4. Individual and group movements with various trim and heel angles.
Fig. 5. Individual and group movements with both trim and heel angles.
tions during evacuation. Fig. 6 shows the results of these experiments. The people at the tail of each group showed slower walking speed than the people at the head of the group. This pattern was noticeable when both trim and heel were present.
Fig. 6. Group movements in opposite directions with various trim and heel angles.
938
D. Lee et al. / Ocean Engineering 31 (2004) 931–941
Fig. 7. Individual and group movements in opposite directions with various trim and heel angles.
In the third method, a group of 10 people and a single person walked in opposite directions starting from each end of the corridor. Fig. 7 shows the results, and it is noticeable that the movement speed of the single person is slower than the movement speed of the group. 4.3. Individual movement experiment with ship motion Ship motion was simulated by natural ship motion while it was anchored out at sea. The corridor model moved with the ship since it was fixed on the deck. Since no motion simulator was used, only limited experiments were possible. A Seatex Motion Reference Unit (MRU-5) was used to measure the amount of ship motion. Fig. 8 shows an example of period and amplitude of the rolling motion. The amplitude of roll is small because the weather condition was good. v The experiment with 20 trim was not carried out due to safety concerns.
Fig. 8. Example of roll amplitude and period.
D. Lee et al. / Ocean Engineering 31 (2004) 931–941
939
Fig. 9. Individual movements with roll motion and trim. v
v
Fig. 9 shows the walking speed of the subjects in 10 and 10 trim situations with ship motion and without the motion. The results showed that ship motion decreases the speed by 10–20%. It means that the dynamic effect is a very important factor in walking speed prediction. 4.4. Doorway passing experiment The time required by 21 people in single file to pass through a doorway (coaming height 0.23 m, width 0.9 m) located at the center of the corridor of 1.2 m width was studied. As it was not possible to change the width of the doorway and the corridor, the experiment was carried out for only one condition. As Fig. 10 shows, the specific flow of persons (Fs) was between 1.05 persons/ms and 1.24 persons/ms with a mean of 1.13 persons/ms. The specific flow of persons here is defined as the number of people passing the doorway in 1 s. In general evacuation simulation, a value of 1.5 persons/ms (in case of corridors or horizontal passages) or 1.3 persons/ms (in case of stairs and vertical passages) is used. On the other hand, MSC/Circ. 909 (IMO, 1999) uses 1.1 persons/ms (descending stairs), 0.88 persons/ms (ascending stairs), and 1.3 persons/ms (in corridors and doorways).
Fig. 10. Specific flow of persons at door.
940
D. Lee et al. / Ocean Engineering 31 (2004) 931–941
5. Observations on experimental results In this research, human behavior under ship list and motion conditions was experimented with, taking account of the other experimental results related to ship evacuation. The quality of the experiments is not very different from that of other experiments carried out in other research. But the experiment for two groups, and a group and one person in opposite directions (counter-flow effect) was a new approach. Counter-flow happens in the evacuation process unavoidably and therefore it will be became a very important factor in real situations. It was confirmed that the walking speed reduction by counter-flow is near 30–60% in this study. Furthermore, also a new approach in this research, the results when ship trim and heel are both present were confirmed to be important. The results of this study will be used to improve the accuracy of the evacuation analysis and the simulation tool that is currently being developed. Most of the hindrances to evacuees, such as ship list, motion or smoke, can be quantified through experiments. However, group movement patterns and psychological factors in real evacuation situations are difficult to reflect in experiments or simulations. First, decrease in speed of group movement by confusion can be partially analyzed as shown in the above experiments, but confusion in extreme hazard situations cannot be reproduced in experimental conditions. Therefore, factors such as decrease in walking speed with crowd density need to be further reflected. In addition, psychological factors such as initial recognition of the accident, perception of the seriousness, report and propagation of the accident, safety zone search, beginning evacuation and evacuation time can be reflected in the evacuation simulation based on the interviews and surveys with experienced people. In addition, factors such as alarm systems, training and knowledge of accidents can be reflected. However, reflections of these psychological factors pose difficulties in gathering data and verifying them. In future, to improve the accuracy of evacuation simulation, the following experiments need to be carried out. – Additional experiments on the effect of ship list for female, aged and handicapped persons. – Additional experiments on the effect of ship motion using a ship motion simulator. – Additional experiments on doorways and stairs of varying width and height. – Experiments on the effect of smoke on evacuees in case of fire.
6. Concluding remarks Human behavior and walking speed under list and dynamic conditions of ships are very important factors in evacuation analysis for Ro/Ro passenger and large passenger ships. To get realistic data for evacuation analysis and simulation, an onboard experiment was performed in this research. The walking speed reduction
D. Lee et al. / Ocean Engineering 31 (2004) 931–941
941
on inclined floors with and without motion is quantified. Especially, it was confirmed that counter-flow was a very important factor for walking speed. Additional onboard experiments are being planned to confirm the effect of severe ship motion within the near future. Acknowledgements The contents of this paper are the results of the Inherent Research Project of KRISO/KORDI, ‘‘Development of Base Technology for Integrated Maritime Risk Management System’’. References Ando, K., et al., 1988. Forecasting the flow of people. Railway Research Review 45 (8), 8–14. Bles, W., Nooy, S., Boer, L.C., 2001. Influence of ship listing and ship motion on walking speed. In: Proceedings of First Conference on Pedestrian and Evacuation Dynamics, Duisburg, Germany, pp. 437–452. Fukuchi, N., Shinoda, T., Imamura, T., 1998. Establishing the methodology for safe evacuation in the event of a marine fire. Journal of the Society of Naval Architects of Japan 184, 579–590. Hwang, K., Chung, D., Lee, D., 1991. An analysis of gait characteristic parameters for the Korean normal adults. Journal of the Human Engineering Society of Korea 10 (2), 15–22. IMO, 1999. Interim guidelines for a simplified evacuation analysis on Ro–Ro passenger ships. MSC/ Circ. 909. IMO, 2002. Interim guidelines for evacuation analyses for new and existing passenger ships. MSC/Circ. 1033. Katuhara, M., et al., 1997. Simulation of human escape on board-I. Journal of Japan Institute of Navigation 96, 283–293. Katuhara, M., et al., 1998. Simulation of human escape on board-II. Journal of Japan Institute of Navigation 98, 141–150. Koss, L., Moore, A., Porteous, B., 1997. Human mobility data for movement on ships. In: Proceedings of International Conference on Fire at Sea, pp. 1–11. Lockey, D., et al., 1997. Injuries sustained during major evacuation of the high-speed catamaran St Malo off Jersey. Injury 28 (3), 187–190. Murayama, M., Itagaki, T., Yoshida, K., 2000. Study on evaluation of escape route by evacuation simulation. Journal of the Society of Naval Architects of Japan 188, 441–448. Dongkon Lee is a Principal Researcher at the Korea Research Institute of Ships and Ocean Engineering/ KORDI. He received his Ph. D. in Naval Architecture from Busan National University, Korea, in 1995. His research interests are in ship design systems and design for safety of ships including damage survivability, human evacuation, fire safety and so on. Hongtae Kim is a Senior Researcher at the Korea Research Institute of Ships and Ocean Engineering/ KORDI. He received his Ph. D. in I.E from Korea University in 2003. His primary research interests are modeling and simulation of manufacturing and logistics systems for maritime industry. Jin-hyoung Park is a Researcher at the Korea Research Institute of Ships and Ocean Engineering/ KORDI. He received his BS and MS in Computer Science from KyungPook National University, Daegu, South Korea, in 1994 and 1996, respectively. His research interests are in safety engineering and human ergonomics for maritime applications with computer model.
A study on experiment of human behavior for evacuation simulation Dongkon Lee , Jin-Hyung Park, Hongtae Kim Korea Research Institute of Ships and Ocean Engineering, KORDI, P.B. Box 23, Yuseong, Daejeon 305 600, South Korea Received 4 August 2003; accepted 5 December 2003
Abstract Walking speed is a very important factor in evacuation analysis for human safety. To provide against emergencies such as abandonment of ship due to accident, the authority body required evacuation analysis for passenger ships. To develop a simulation tool for the evacuation analysis, human behavior data for evacuation situations are fundamentally necessary. In this paper, onboard experiments were carried out twice, once with ship motion and another without ship motion. They were performed assuming the situation where subjects are evacuating under instructions from the crew without panic. Not only individual movement but also group movement with and without motion based on inclination was covered. # 2004 Elsevier Ltd. All rights reserved. Keywords: Onboard experiment; Evacuation; Human safety; Ship design; Human behavior
1. Introduction There have been many accidents of passenger ships at sea and they have caused huge losses of human lives. The interim guidelines for evacuation analyses for new and existing passenger ships were developed by the Maritime Safety Committee (MSC) of the International Maritime Organization (IMO) and circulated as MSC/ Circ. 1033 (IMO, 2002) for the purpose of unified implementation of the requirement of evacuation analysis which is required by regulation. The MSC considered whether the interim guidelines should be made mandatory for large passenger ships
Corresponding author. Tel.: +82-42-868-7222; fax: +82-42-868-7229. E-mail address: [email protected] (D. Lee).
0029-8018/$ - see front matter # 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.oceaneng.2003.12.003
932
D. Lee et al. / Ocean Engineering 31 (2004) 931–941
and, having noted that more data on application of the interim guidelines are needed to validate the methodologies and criteria contained in the above guidelines, agreed that this issue should be further considered by the committee. In 1995, the 41 m long high speed twin hull passenger liner St. Malo (Lockey et al., 1997) was stranded off the shore of France and the passengers evacuated while the ship was listing. The total evacuation time of 308 passengers after the distress signal was recorded as 1 h and 17 min. It is more than nine times the total evacuation time recorded during evacuation drill in stationary conditions, which was 8 min. This is a good example showing that evacuation analyses that does not take account of motion, listing, and the psychological state of the passengers are meaningless in the real world. In this paper, onboard experiments were carried out twice, once with ship motion and another without ship motion. They were performed assuming the situation where subjects are evacuating under instructions from the crew without panic. Not only individual movement but also group movement with and without motion based on inclination was covered. The results showed that a floor with motion decreases the walking speed by 10–20% and an inclined floor without motion also affected walking speed. The results from the experiments will be used to increase the accuracy of the evacuation simulation system, which is being developed. 2. Related works Human behavior study in evacuation situations provides important information to engineers, managers, operators and users. It is necessary to test if an appropriate safety system exists to cope with a possible critical situation in order to allocate resources including trained personnel. In addition, designing alarm systems announcing necessary information to passengers in critical situations without correct understanding of human behavior will decrease the overall efficiency of the evacuation system. Past research reports that walking speed on a flat floor is between 0.98 and 1.39 m/s in case of Asians (Hwang et al., 1991; Fukuchi et al., 1998). Research in Japan (Ando et al., 1988) in busy train stations reports that people of about 20 years of age show the fastest moving speed. On the other hand, research from Europe and Australia reports the walking speed of adult males to be between 1.4 and 1.6 m/s. To model human movement in an evacuation situation accurately, each evacuee is described as an object with a set of parameters including age, gender, and walking speed. The values of these parameters should be acquired from actual experiments for more realistic simulation. The characteristics and experiment results of major related research are summarized. The National Maritime Research Institute of Japan experimented on an anchored ship from 1994 for 3 years (Katuhara et al., 1997, 1998). The subjects were 70–120 students of 20 years age. The subjects moved along a predefined evacuation path and the experiment was recorded using a video camera. The results of the experiment
D. Lee et al. / Ocean Engineering 31 (2004) 931–941
933
showed that the movement speeds were 1.4 m/s in corridors and 0.7 m/s for stairs. The maximum density was 3 persons/m2. The Research Institute of Marine Engineering of Japan used a total of 20 male and female adults as subjects and experimented inside a corridor (Murayama et al., v v 2000). The experiment conditions include list angles between +20 and 20 . In addition, six subjects were experimented with while the corridor was in motion v with a motion angle of 10 and a cycle time of 5 and 10 s. In case of trim, the walking speed was between 0.82 and 1.38 m/s and the speed decreased as the v v upward trim increased. However, heeling angles between 0 and 20 had insignificant effects on the walking speed. In the ship motion experiment, the walking speed was between 0.71 and 0.77 m/s, which is about 70% less than in stationary conditions (0.90 m/s). The Australian Maritime Engineering-Cooperative Research Centre (AMECRC) experimented on the walking speed in corridors and on ladders as a part of the Australian Maritime Safety Authority Program (Koss et al., 1997). The subjects were a total of 67 male and female adults 18–25 years of age. The walking speed increased as the downward trim increased. In case of upward trim and heel, the walking speed was similar to the case where there was no list. However, there was decrease in speed in case of heel when two persons walked together along a corridor of 1.2 m width. In another research in 2000 with 985 subjects, the results showed that significant decrease in speed was noticed when the subject was more than 65 years of age, and males were approximately 18% faster than females. In case of handicapped persons, the decrease in speed was significant. TNO Human Factors in Netherlands experimented on the effects of ship motion and list in corridors and on stairs on human walking speed (Bles et al., 2001). The subjects were a total of 150 adults 18–83 years of age. A ship motion simulator, where a 4:0 m 2:4 m 2:3 m sized cabin was placed on a hydraulic system, was used for the experiment. The experiments with positive trim angle showed substantial decrease in the walking speed. On the whole, ship trim resulted in 35% decrease in speed. For the ship motion experiments, walking speed decreased up to 15% as the cycle time and angle increased. 3. Onboard experiment The corridor model (10:0 mm 1:2 mm 1:9 mm) was used in the experiment. v v v The model can have trim angle between 20 and 20 , and heel angle between 0 v and 20 . A handrail is present and stairs are located at the end of the corridor as shown in Fig. 1. The following are the measuring and recording devices used in the experiment. – Eight-Channel Image Grabber Board equipped systems (digital video network transmission S/W, digital video recording S/W). – Eight CCD high resolution cameras. – Ship motion measuring device (Seatex Motion Reference Unit, MRU-5).
934
D. Lee et al. / Ocean Engineering 31 (2004) 931–941
Fig. 1. A scene of onboard experiment at sea and corridor model.
All of the experiments were carried out twice, once with ship motion and another without ship motion, except the doorway passing experiments. The experiments were carried out assuming the situation where subjects are evacuating under the instructions of the crew without panic. The following experiments were carried out. 1. Walking along a corridor without counter-flow a. Measuring the time required for a single person to cover the length of a corridor of 10 m length. b. Measuring the total time required for a group of 21 persons in single file to walk the length of a corridor of 10 m length. 2. Walking along a corridor with counter-flow a. Measuring the total time required for two groups of 10 persons each in single file to walk the length of a 10 m long corridor in opposite directions starting from each end. b. Measuring the total time required for groups of 10 persons and a single person to walk single file along the 10 m long corridor in opposite directions starting from each end. 3. Moving through a doorway a. Measuring the total time required for a group of 21 persons in single file to walk through a doorway of width 0.9 m. To replicate ship motion, the corridor model was constructed on the aft deck of a training ship of the Korea Maritime University. The experiments were carried out while the ship was sailing from Pusan harbor to Yesu harbor, 1 night and 2 days. The amount of roll and pitch was measured using the ship’s motion sensor.
D. Lee et al. / Ocean Engineering 31 (2004) 931–941
935
Fig. 2. Data capture system with Eight-Channel Image Grabber Board and CCD high resolution camera.
The subjects in the experiment were students from the Korea Maritime University. They were a total of 21, including 18 male students and three female students. The subjects were wearing safety helmets and life vests and serial numbers were painted at each side of the helmet. The experiments are recorded by eight video cameras installed in the corridor model and the recordings were stored in the computer through the Eight-Channel Image Grabber Board as shown in Fig. 2. 4. Experimental results The experiments can be classified into the following four categories. – – – –
Individual movement experiment with ship trim and heel. Group movement experiment with ship trim and heel. Individual movement experiment with ship motion. Doorway passing experiment.
Below are the summarizations of the experimental results for each category (see Table 1). 4.1. Individual movement experiment under trim and heel It is difficult to compare the results of the experiment in this research with other experiments since the experiment conditions and subject characteristics are different. Fig. 3 shows the results of this experiment with the data from AME-CRC (Koss et al., 1997) and the Research Institute of Marine Engineering in Japan (Murayama et al., 2000).
936
D. Lee et al. / Ocean Engineering 31 (2004) 931–941
Table 1 Experimental conditions Experiment under static condition
Experiment under dynamic condition
Corridor
Corridor dimension Trim Heel Combination of trim and heel Corridor dimension Trim Heel Rolling angle Roll period Coaming height of door Breadth of door
10:0 m 1:2 m 1:9 m v v v v v +20 , +10 , 0 , –10 , –20 v v v 20 , 10 , 0 Combination 10:0 m 1:2 m 1:9 m v v v +10 , 0 , 10 v v v 20 , 10 , 0 v 3–4 5–12 s 0.23 m 0.9 m
Most of the results are similar to the data reported in other experiments, but in case of trim, the results show difference. For downward trim, the speed was expected to increase, but the results showed decrease in speed. This reflects that psychological state affects human behavior. In case of upward trim, the speed did not decrease as much as in the other experiments. This is because the floor of the corridor model used in this experiment was carpeted, and thus less slippery than in case of the corridor model used in other experiments. 4.2. Group movement experiments with ship trim and heel Group movement experiments with ship trim and heel were carried out using three different methods. In the first one, 21 subjects moved along the corridor without delay to simulate the evacuation situation. Fig. 4 shows the results of this experiment compared to the individual movement experiment where the subjects moved at intervals of 3 m. It shows that the movement speed was 20% slower in group movement than in individual movement. Fig. 5 shows the results when ship trim and heel are both present. The results showed substantial decrease in speed for the case where either trim or heel was present. In the second method, two groups of 10 people each walked in opposite directions starting from each end of the corridor. This can be one of the typical situa-
Fig. 3. Individual movements with various trim and heel angles.
D. Lee et al. / Ocean Engineering 31 (2004) 931–941
937
Fig. 4. Individual and group movements with various trim and heel angles.
Fig. 5. Individual and group movements with both trim and heel angles.
tions during evacuation. Fig. 6 shows the results of these experiments. The people at the tail of each group showed slower walking speed than the people at the head of the group. This pattern was noticeable when both trim and heel were present.
Fig. 6. Group movements in opposite directions with various trim and heel angles.
938
D. Lee et al. / Ocean Engineering 31 (2004) 931–941
Fig. 7. Individual and group movements in opposite directions with various trim and heel angles.
In the third method, a group of 10 people and a single person walked in opposite directions starting from each end of the corridor. Fig. 7 shows the results, and it is noticeable that the movement speed of the single person is slower than the movement speed of the group. 4.3. Individual movement experiment with ship motion Ship motion was simulated by natural ship motion while it was anchored out at sea. The corridor model moved with the ship since it was fixed on the deck. Since no motion simulator was used, only limited experiments were possible. A Seatex Motion Reference Unit (MRU-5) was used to measure the amount of ship motion. Fig. 8 shows an example of period and amplitude of the rolling motion. The amplitude of roll is small because the weather condition was good. v The experiment with 20 trim was not carried out due to safety concerns.
Fig. 8. Example of roll amplitude and period.
D. Lee et al. / Ocean Engineering 31 (2004) 931–941
939
Fig. 9. Individual movements with roll motion and trim. v
v
Fig. 9 shows the walking speed of the subjects in 10 and 10 trim situations with ship motion and without the motion. The results showed that ship motion decreases the speed by 10–20%. It means that the dynamic effect is a very important factor in walking speed prediction. 4.4. Doorway passing experiment The time required by 21 people in single file to pass through a doorway (coaming height 0.23 m, width 0.9 m) located at the center of the corridor of 1.2 m width was studied. As it was not possible to change the width of the doorway and the corridor, the experiment was carried out for only one condition. As Fig. 10 shows, the specific flow of persons (Fs) was between 1.05 persons/ms and 1.24 persons/ms with a mean of 1.13 persons/ms. The specific flow of persons here is defined as the number of people passing the doorway in 1 s. In general evacuation simulation, a value of 1.5 persons/ms (in case of corridors or horizontal passages) or 1.3 persons/ms (in case of stairs and vertical passages) is used. On the other hand, MSC/Circ. 909 (IMO, 1999) uses 1.1 persons/ms (descending stairs), 0.88 persons/ms (ascending stairs), and 1.3 persons/ms (in corridors and doorways).
Fig. 10. Specific flow of persons at door.
940
D. Lee et al. / Ocean Engineering 31 (2004) 931–941
5. Observations on experimental results In this research, human behavior under ship list and motion conditions was experimented with, taking account of the other experimental results related to ship evacuation. The quality of the experiments is not very different from that of other experiments carried out in other research. But the experiment for two groups, and a group and one person in opposite directions (counter-flow effect) was a new approach. Counter-flow happens in the evacuation process unavoidably and therefore it will be became a very important factor in real situations. It was confirmed that the walking speed reduction by counter-flow is near 30–60% in this study. Furthermore, also a new approach in this research, the results when ship trim and heel are both present were confirmed to be important. The results of this study will be used to improve the accuracy of the evacuation analysis and the simulation tool that is currently being developed. Most of the hindrances to evacuees, such as ship list, motion or smoke, can be quantified through experiments. However, group movement patterns and psychological factors in real evacuation situations are difficult to reflect in experiments or simulations. First, decrease in speed of group movement by confusion can be partially analyzed as shown in the above experiments, but confusion in extreme hazard situations cannot be reproduced in experimental conditions. Therefore, factors such as decrease in walking speed with crowd density need to be further reflected. In addition, psychological factors such as initial recognition of the accident, perception of the seriousness, report and propagation of the accident, safety zone search, beginning evacuation and evacuation time can be reflected in the evacuation simulation based on the interviews and surveys with experienced people. In addition, factors such as alarm systems, training and knowledge of accidents can be reflected. However, reflections of these psychological factors pose difficulties in gathering data and verifying them. In future, to improve the accuracy of evacuation simulation, the following experiments need to be carried out. – Additional experiments on the effect of ship list for female, aged and handicapped persons. – Additional experiments on the effect of ship motion using a ship motion simulator. – Additional experiments on doorways and stairs of varying width and height. – Experiments on the effect of smoke on evacuees in case of fire.
6. Concluding remarks Human behavior and walking speed under list and dynamic conditions of ships are very important factors in evacuation analysis for Ro/Ro passenger and large passenger ships. To get realistic data for evacuation analysis and simulation, an onboard experiment was performed in this research. The walking speed reduction
D. Lee et al. / Ocean Engineering 31 (2004) 931–941
941
on inclined floors with and without motion is quantified. Especially, it was confirmed that counter-flow was a very important factor for walking speed. Additional onboard experiments are being planned to confirm the effect of severe ship motion within the near future. Acknowledgements The contents of this paper are the results of the Inherent Research Project of KRISO/KORDI, ‘‘Development of Base Technology for Integrated Maritime Risk Management System’’. References Ando, K., et al., 1988. Forecasting the flow of people. Railway Research Review 45 (8), 8–14. Bles, W., Nooy, S., Boer, L.C., 2001. Influence of ship listing and ship motion on walking speed. In: Proceedings of First Conference on Pedestrian and Evacuation Dynamics, Duisburg, Germany, pp. 437–452. Fukuchi, N., Shinoda, T., Imamura, T., 1998. Establishing the methodology for safe evacuation in the event of a marine fire. Journal of the Society of Naval Architects of Japan 184, 579–590. Hwang, K., Chung, D., Lee, D., 1991. An analysis of gait characteristic parameters for the Korean normal adults. Journal of the Human Engineering Society of Korea 10 (2), 15–22. IMO, 1999. Interim guidelines for a simplified evacuation analysis on Ro–Ro passenger ships. MSC/ Circ. 909. IMO, 2002. Interim guidelines for evacuation analyses for new and existing passenger ships. MSC/Circ. 1033. Katuhara, M., et al., 1997. Simulation of human escape on board-I. Journal of Japan Institute of Navigation 96, 283–293. Katuhara, M., et al., 1998. Simulation of human escape on board-II. Journal of Japan Institute of Navigation 98, 141–150. Koss, L., Moore, A., Porteous, B., 1997. Human mobility data for movement on ships. In: Proceedings of International Conference on Fire at Sea, pp. 1–11. Lockey, D., et al., 1997. Injuries sustained during major evacuation of the high-speed catamaran St Malo off Jersey. Injury 28 (3), 187–190. Murayama, M., Itagaki, T., Yoshida, K., 2000. Study on evaluation of escape route by evacuation simulation. Journal of the Society of Naval Architects of Japan 188, 441–448. Dongkon Lee is a Principal Researcher at the Korea Research Institute of Ships and Ocean Engineering/ KORDI. He received his Ph. D. in Naval Architecture from Busan National University, Korea, in 1995. His research interests are in ship design systems and design for safety of ships including damage survivability, human evacuation, fire safety and so on. Hongtae Kim is a Senior Researcher at the Korea Research Institute of Ships and Ocean Engineering/ KORDI. He received his Ph. D. in I.E from Korea University in 2003. His primary research interests are modeling and simulation of manufacturing and logistics systems for maritime industry. Jin-hyoung Park is a Researcher at the Korea Research Institute of Ships and Ocean Engineering/ KORDI. He received his BS and MS in Computer Science from KyungPook National University, Daegu, South Korea, in 1994 and 1996, respectively. His research interests are in safety engineering and human ergonomics for maritime applications with computer model.