2017 4th International Conference on Transportation Information and Safety (ICTIS), August 8-10, 2017, Banff, Canada
Simulation on Occupant Evacuation during Aircraft Emergency Based on Cellular Automata Tianjin, China
[email protected]
DU Hong-bing College of Flight Technique Civil Aviation University of China Tianjin, China
[email protected]
FENG Zhen-yu Tianjin Key Laboratory of Civil Aircraft Airworthiness and Maintenance Civil Aviation University of China Tianjin, China
[email protected]
YU Xiao-fang College of Flight Technique Civil Aviation University of China
Abstract—The characteristics of occupant evacuation during aircraft emergency were analyzed according to emergency evacuation guidance systems, light intensity and evacuees. Evacuees included occupant’s physiological characteristics and their psychological characteristics. The occupant’s physiological characteristics were mainly related to panic, pressure and conformity. The psychological characteristics involved in waist circumference, age, gender and height. The above factors were applied to the aircraft emergency evacuation simulation on cellular automata. Then, the cabin environment model and the occupant model were presented. The occupant model included individual pre-reaction time submodel, cabin channel velocity submodel and escape time submodel of exit. Study shows that the results could provid an auxiliary verification information for aircraft initial certification, and could provide strategic guidance for airline to carry out personnel evacuation and emergency training for employees.
actual situation. In Vacate Air, all particles moving to a same target did not match the actual situation that passengers evacuated to more than one exit. The parameters of ETSIA were modified according to A320-100 validation experimental data and published literature data, and ETSIA was used to verification experiment and the design of the cabin. AAMAS took into account the impact of passenger panic on the escape process during the airfraft emergency incident. EvacuSimulation mainly studied the population density effect on the model, but it did not consider the individual physiological and psychological characteristics effect on it. GUI only created the individual physiological characteristics on occupant evacuation process. CabinEvacu researched the influence of gender, age and degree of panic, but it did not considered the effect of the different cabin area to personnel speed.
Keywords—simulation model; cellular automata; emergency evacuation; civil aircraft
The research determined the impact factors of the civil aircraft emergency evacuation through aircraft, the external environment, and evacuees. These factors were applied to cellular automata theory. Then, the cabin environment model and the occupant model were presented. These models were more consistent with the actual situation. The research results could be used as an auxiliary verification method for initial certification, and they could also provide a theoretical basis for the design of civil aircraft emergency group simulation software.
I.
INTRODUCTION
The 90 seconds occupant evacuation experiment is one of the effective methods to judge the aircraft emergency evacuation capability. At present, this experiment mainly recruited volunteers to complete the real evacuation in the aircraft enviorment meeting airworthiness criterion. However, this approach requires a lot of spending and there is a risk and hazard for the participants. With the maturity of computer simulation method, it reflects the urgency that the 90 seconds emergency evacuation experiment is carried out by the method.
II.
Von Neumann and others put forward the cellular automata theory which core idea was dispersing the time and space. The theory included cellular, time step, neighborhood, and rule, introduced as follows:
And so far, there are some mature evacuation simulation models in the aircraft emergency, such as air EXODUS[1~2], DEM[3], Vacate Air[4~5], ETSIA[6~8], AAMAS[9], [10] EvacuSimulation 、GUI(Graphical User Interfaces)[11~12] 、 CabinEvacu[13~14] and so on. In these models, the individual of air EXODUS was designed to move towards the nearest available exit in the case of no unit guidance or identification. The individual movement in the DEM model was considered to follow the Newtonian kinematics theorem. But many parameters of individual behavior and motion do not match the
716 978-1-5386-0437-3/17/$31.00 ©2017 IEEE
CELLULAR AUTOMATA
Cell distributions in the discrete physical environment and it has two states: "Occupy" and "idle". It is the most basic element of cellular automata.
Time step means the discrete time size where the cell updates the state.
2017 4th International Conference on Transportation Information and Safety (ICTIS), August 8-10, 2017, Banff, Canada
B. Cell Status and Update Rules The cellular state was expanded according to the different cabin areas. Then, a new update rule for the cellular state was set up.
Neighborhood and rule mean the state of the cell is updated with its own state and other cells within its neighborhood. Update rule as follows: Cit+1=f {Cit, Ci+1t, Ci+2t,…Ci+nt}
(1)
Cit+1=f {Cit, Ci+1t,…Ci+nt,}+φ{Pit,Pi+1t…Pi+mt}
t
Where Ci+n is the state of the n-th cell aside I cell at time t; Cit+1 is the state of the ith cell at time t+1.
Where Ci+nt is the state of the n-th cell aside I cell at time t; Cit+1 is the state of i-th cell at time t+1; Pi+mt is the property of the m-th cell aside I cell at time t; f{} is the effect of n cell states aside i-th cell to the i-th cell state at time t; φ{} is the effect of m individuals aside i-th cell to the i-th cell state at time t.
Cellular automata method could simulate complex behavior with a simple way, and it could set different rules to reflect the interaction between individuals and the characteristics of the movement. Its grid is divided into uniform and large size, the general size is 0.5m * 0.5m, or 0.5m * 0.4m. The size reflects the accuracy of the simulation.
V. AIRCRAFT EMERGENCY EVACUATION SIMULATION MODELS
III. AIRCRAFT EMERGENCY EVACUATION FACTORS
According to the aircraft emergency evacuation factors and the emergency evacuation experimental data, taking the Boeing 737-800 aircraft as an example, cabin environment model and occupant model were established.
A. Aircraft According to the experimental condition of CCAR-25 Airworthiness Standard for Transport Aircraft, the aircraft environmental characteristic was analyzed from the aircraft structure and the emergency evacuation guidance systems.
A. Cabin Environment Model 1) Classification of the cabin area
B. Aircraft External Environmental The occupant’s evacuation behaviors are affected directly by the fire growth rate, smoke concentration, toxicity, temperature, light intensity and so on. When an aircraft accident happens, it is accompanied by fire, lighting system paralysis, etc. Light intensity was the only considered factor in the 90s certification trials of aircraft because of the limitation of experimental condition.
According to the characteristics of personnel flow, the cabin was divided into four areas: the obstacle, the front and rear passenger gate, the main channel, the inter-seat channel. The obstacle area is the places where the occupant could not cross, including seat, partition, kitchen, etc. The front and rear passenger gate areas refer to the places between the front boarding gate and the front service door, and between the rear boarding door and the rear service door. These areas are more spacious than the cabin channel, so they could accommodate more people. At the same time, they are also crowded easily. The main channel area is the aisle that is used to connect the front passenger gate and rear passenger gate. The inter-seat channel is between the two rows of seats in the same row. That was divided into three regions due to the different location, which includes economy class seat channel, first class seat channel, and the distance between the seats at the wing exit.
C. Evacuees Combined with the emergency evacuation process of civil aircraft and individual cognitive process, the ones of occupant’s physical and psychical characteristics were analyzed. The physical factors included age, sex, waist circumference, height, etc. The psychical factors included panic, pressure, conformity, etc. IV. CONSTRUCTION RULES OF AIRCRAFT EMERGENCY EVACUATION SIMULATION MODEL
2)
A. Determination of Physical Region Representation According to the characteristics of cabin environment and individual distance, the fine grid was chosen to divide the cabin. In order to reduce the computational pressure of computer, and to describe aircraft internal environment and the differences between individual accurately, the area where the individual distance was large and the movement between occupants was small was set by setting the grid cell. Fig.1 (a) is the schematic of the large size grid cell, Fig.1 (b) is the schematic of the improved grid cell.
(a)
(b)
Fig.1. Different grid cellular schematic
717 978-1-5386-0437-3/17/$31.00 ©2017 IEEE
(2)
Division of the cabin grid
In order to improve the simulation accuracy, the 0.15m * 0.15m was used as the basic grid size with reference to the size of Boeing 737-800 aircraft. Then, according to Chinese adult human body size and the cabin actual size, the number of the basic grid that was occupied by personnel and the different cabin area was determined, see TABLE I. TABLE I.
THE NUMBER OF BASIC GRID AND DISTRIBUTION OF INDIVIDUAL
Basic Grid Number
The Number of Individual
Occupant
2×2
/
Economy class seat channel
6×2
3
Channel between the first row of seats in economy class
6×2
3
The seats at the wing exit
6×3
3
First class seat channel
6×4
2
Zone Name
2017 4th International Conference on Transportation Information and Safety (ICTIS), August 8-10, 2017, Banff, Canada The main channel
2×128
64
Male
0.93
The front and rear passenger gate area
14×6
21
Female
1.07
Economy class seat
2×2
/
1447-1625
1.09
1625-1676
First class seat
3×2
/
0.99
Bulkhead between first class and economy class
6×2
/
1676-1727
0.95
1727-1803
0.97
Kitchen at the front of the cabin
6×2
/
1803-2006
0.99
584-787
0.85
787-864
0.90
864-965
0.95
965-1041
1.08
The cloakroom at the front of the cabin
6×2
/
The toilet at the rear right of the cabin
6×2
/
Two bathrooms at the rear left of the cabin
6×6
/ a.
3)
Gender
Height (mm)
Waist (mm)
1041-1575 Continue the previous column
Definition of the grid attribute
ti1=[(α+β)×(KiA+KiG+KiH+KiW)]/5(i=1,2,3…)
2) Cabin channel velocity submodel Cranfield University had conducted an eight-day emergency evacuation experiment under straight channel and dark scene, but there was an accident on the fourth day in the experiment. The individual moving speed was determined by analyzing these experimental data except the fourth-day's[16]. And the distance from the individual to the exit was divided into three regions that were (2,5] m, (5,8] m, (8,11] m. According to the each regional experiment data, the velocity characteristics and distribution of each region were obtained, see TABLE III and Fig.2. TABLE III.
1) Pre-reaction time submodel In initial reaction process, the pre-reaction time submodel with unlocking the seat belt and go to channel from the seat as the core was presented. The model considered the effects of age, sex, height and waist on emergency evacuation. And the impact proportions of these attributes were determined by referring to The Aircraft Accident Statistics and Knowledge (AASK) database V4.0 [15], see TABLE II.
Age
Impact Proportion
18-22
0.83
23-32
0.89
33-42
0.97
43-52
1.06
53-65
1.25
718 978-1-5386-0437-3/17/$31.00 ©2017 IEEE
THE INDIVIDUAL SPEED OF MOVEMENT AT DIFFERENT DISTANCES FROM EXIT
Name
THE IMPACT PROPORTIONS OF OCCUPANT’S PHYSICAL CHARACTERISTIC TO THE PRE-REACTION TIME
Attribute Ordering
(3)
Where i is the different passengers; ti1 is the pre-reaction time of i; α is the average time for passengers to unlock the seat belt; β is the average time for passenger leave from the seat, KiA is the impact factor of i’s age, KiG is the impact factor of i’s gender, KiH is the impact factor of i’s height, KiW is the impact factor of i’s waist.
B. Occupant Model According to the aircraft emergency evacuation process, the occupant escape models were constructed in stages.
Base Attribute
Continue the previous column
The pre-reaction time submodel was presented as follows.
There were four main types of the cabin grid attributes, which included grid type, grid location and scope, grid structure, grid competition rules. The grid type indicates whether the grid is in a state of traffic. "0" means the accessible area grid, "2" means the non-accessible area grid, and “4" means other obstacle grids in the non-accessible area. The grid location and scope were the location basis for the virtual personnel's movement. The grid structure indicates that the individual mobility constraints in different regions. It was divided into three attributes: the grid structure of inter-seat channel, the main channel, the front and rear passenger gate. When there is a competitive condition in the evacuation process, the occupants’ movements are following rules: men have priority over women when they are in the same position; when they have the same gender, under 50 years old is better than the over-50s; when they have the similar age, the small is better than large in waist size; when there is no difference in the nature of the personnel, the system chooses the individual randomly to occupy the position.
TABLE II.
1.23 b.
Distance Partition (2,5]m
(5,8]m
(8,11]m
Average speed
1.14
0.53
0.36
Maximum speed
2.54
1.09
0.53
Minimum speed
0.50
0.23
Standard deviation
0.54
0.30 0.16
0.07
2017 4th International Conference on Transportation Information and Safety (ICTIS), August 8-10, 2017, Banff, Canada TABLE IV.
THE IMPACT PROPORTIONS OF THE AGE AND SEX TO THE MOVEMENT SPEED (F)
Frequency
Age Group 35 30 25 20 15 10 5 0
18-22
23-32
33-42
43-49
50-52
53-62
Male
1.1
1.1
1.1
1.1
0.88
0.88
Female
0.97
0.97
0.97
0.97
0.78
0.78
Gender
The cabin channel velocity model was presented as follows. Vi2=f×Vi0(i=1,2,3
)
(4)
Where Vi2 is I’s movement speed in the cabin channel, f is the impact factor of i’s physical characteristics, Vi0 is the reference speed.
Moving Speed (m/s)
3) Escape time submodel of wing exit According to the report Study on the Influencing Factors of Type-Ⅲ Wing Exit Emergency Evacuation in Cabins[18], the individual with different age, sex, waist, and height had effects on the time that through the Type- III wing exit and the time were determined. According to each attribute classification corresponding to the time variance, the impact proportions of each physiological characteristic to the escape time were determined, see TABLE V.
(a) (2, 5] m Frequency 60 50 40 30 20 10 0
TABLE V.
THE AVERAGE WING EXIT ESCAPE TIME OF INDIVIDUAL WITH DIFFERENT PHYSIOLOGICAL CHARACTERISTICS
Base Attribute Moving Speed (m/s)
Age (b) (5, 8] m Frequency 35 30 25 20 15 10 5 0
Gender
Height (mm)
Moving Speed (m/s)
Waist (mm) (c) (8, 11] m Fig.2. The regional velocity profile
Escape Time (s)
23-32
1.44
33-42
1.57
43-52
1.71
53-65
2.01
Male
1.49
Female
1.70
1447-1625
1.74
1625-1676
1.59
1676-1727
1.52
1727-1803
1.55
1803-2006
1.58
584-787
1.35
787-864
1.43
864-965
1.51
965-1041
1.72
1041-1575
1.96
Impact Proportion
1.34
0.263
0.299
0.024
0.414
The escape time model of wing exit was presented as follows.
The results showed that the moving space and individual speed located in the long range were directly influenced by forwarding individual, and needed longer waiting time to escape. The individual barrier-free moving speed couldn’t be reflected well by these people’s moving speed data. For these reasons, the average velocity (1.14 m/s) from the nearest exit area ((2,5] m) was selected as the reference velocity (Vi0). This model considered these attributes, including age and sex. And their impact proportions were determined by reference to McLean GA’s study [17], see TABLE IV.
ti3=aMiA+bMiG+cMiH+dMiW(i=1,2,3
)
(5)
Where ti3 is i spends time through the wing exit, a is the factor’s proportion of age to individual through the wing exit, b is the factor’s proportion of sex to individual through the wing exit, c is the factor’s proportion of height to individual through the wing exit, d is the factor’s proportion of waist to individual through the wing exit, MiA is the age of individual I
719 978-1-5386-0437-3/17/$31.00 ©2017 IEEE
Attribute Ordering 18-22
2017 4th International Conference on Transportation Information and Safety (ICTIS), August 8-10, 2017, Banff, Canada
corresponds to the time through the wing exit, MiG is the sex of individual I corresponds to the time through the wing exit, MiH is the height of individual I corresponds to the time through the wing exit, MiW the waist of individual I corresponds to the time through the wing exit. 4) Escape time submodel of non-wing exit In order to obtain the average non-wing exit escape time of individuals with different gender and degree of panic, a jumping slide experiment was established. The impact proportions of an individual with sex and panic to escape time was determined by analyzing the experimental results, and the impact proportons is shown in TABLE VI. At the same time, the study summarized that the average decline time of the first body in each group was 1.5s. TABLE VI. Base Attribute Gender
Level of panic
Fig.3. the simulation result
B. Analyzing the Simulation Experimental Result Fig.4(a) shows the frequency distribution curve of the evacuation time by analyzing the evacuation time of the 1000 simulation experiments. As can be seen from the diagram, the frequency distribution of the evacuation time is similar to the normal distribution. And the results were compared with previous studies, it found that the result was similar to E.R. Galea’s research results, see Fig.4 (b). The result proves the validity of the experiment.
THE TIME INTERVAL OF INDIVIDUALS WITH DIFFERENT CHARACTERISTICS TO JUMP SLIDE
Attribute Ordering
Average Time Interval (s)
Male
0.732
Female Class A
0.545 0.550
Class B
0.859
Class C
1.165
Impact Proportion 0.271
Frequen 800 cy 700 600 500 400 300 200 100 0 60
0.729
The escape time submodel of non-wing exit was presented as follows. ti4=jNiG+kNiP+1.5(i=1, 2, 3
)
(6)
Where ti4 is I spends time from the non-wing exit to ground, j is the factor’s proportion of sex to individual through the non-wing exit, k is the factor’s proportion of panic to individual through the non-wing exit, NiG is the sex of individual I corresponds to the average separation time through the non-wing exit, NiP is the panic of individual I corresponds to the average separation time through the non-wing exit.
65
70
75
80 85 90 Evacuation time (s)
(a)
VI. SIMULATION EXAMPLES AND RESULTS A. Simulation Example With Visual Studio 2010 development platform, the aircraft emergency evacuation simulation software (EES-air) was developed using Visual C++ according to the cabin and occupant model. In the example of Boeing 737-800, the evacuation plan had conducted 1000 simulations using the EES-air software, the simulation test plan is shown in TABLE VII, and the result is shown in Fig.3. (b) TABLE VII.
EXPORT DISTRIBUTION PLAN
Export Part Front (the boarding doors and the service doors) Middle(wing exits) Rear(the boarding doors and the service doors)
Fig.4. Evacuation time-frequency distribution of 1000 simulation experiments
Distribution Plan
VII. CONCLUSION
The 1st, 2nd, 11th, 12th, 13th, 14th, 15th, 16th, 17th, 18th row The 19th, 20th, 21th, 22th, 23th, 24th, 25th, 26th row The 27th, 28th, 29th, 30th, 31th, 32th, 33th, 34th, 35th, 36th, 37th row
Based on the above research and analysis, the following conclusions are drawn: 1) The aircraft environmental characteristics were the objective condition for emergency evacuation and effected the evacuation efficiency. The characteristics of occupant’s
720 978-1-5386-0437-3/17/$31.00 ©2017 IEEE
2017 4th International Conference on Transportation Information and Safety (ICTIS), August 8-10, 2017, Banff, Canada
physical and psychical played an important role in the evacuation.
[7] Hedo, J. M., “Modelización Computacional del Ensayo de Evacuación
2) According to the experimental data and the influence factors of occupant evacuation in aircraft emergency, cabin environment model and occupant model were presented. These models were more consistent with the actual evacuation situation.
[8]
3) The experiment was a local experiment, so the validity of the data combination was reduced. In further studies, a complete evacuation experiment should be considered.
[10]
[9]
[11]
ACKNOWLEDGMENT This work was partially supported by the Civil Aviation University of China.
[12]
REFERENCES [13]
[1] Galea, E. R., Blake. “The Use of Evacuation Modelling Techniques in
[2] [3] [4]
[5]
[6]
the Design of Very Large Transport Aircraft and BlendedWing Body Aircraft,” The Aeronautical Journal, Vol. 107, No. 1070, 2003, pp. 207– 218. Galea, E. R. “Proposed Methodology for the Use of Computer Simulation to Enhance Aircraft Evacuation Certification,” Journal of Aircraft, Vol. 43, No. 5, 2006, pp. 1405–1413. doi:10.2514/1.20937 Robbins, C. R., and McKee, S., “Simulating the Evacuation of a Commercial Airliner,” The Aeronautical Journal, Vol. 105, No. 1048, June 2001, pp. 323–328. Xue, Z., and Bloebaum, C. L., “A Particle Swarm Optimization-Based Aircraft Evacuation Simulation Model: VacateAir,” 46th AIAA Aerospace Sciences Meeting, Reno, NV, AIAA Paper 2008-0180, Jan. 2008. Xue, Z., and Bloebaum, C. L., “Experimental Design-Based Aircraft Egress Analysis Using VacateAir: An Aircraft Evacuation Simulation Model,” 12th AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference, Victoria, B.C., Canada, AIAA Paper 2008-6061, 2008. Hedo J M, Martinez-Val R. Assessment of narrow-body transport airplane evacuation by numerical simulation[J]. Journal of Aircraft, 2011, 48(5): 1785-1794.
[14]
[15] [16]
[17]
[18]
721 978-1-5386-0437-3/17/$31.00 ©2017 IEEE
de Emergencia de Aviones de Transporte,” Ph.D. Thesis, Univ. Politécnica de Madrid, Madrid, May 2009 (in Spanish). Hedo, J. M., and Martinez-Val, R., “Computer Model for Numerical Simulation of Emergency Evacuation of Transport Aeroplanes,” The Aeronautical Journal, Vol. 114, No. 1161, 2010, pp. 737–746. Miyoshi T, Nakayasu H, Ueno Y, et al. An emergency aircraft evacuation simulation considering passenger emotions[J]. Computers & Industrial Engineering, 2012, 62(3): 746-754. Du Hong-bing, Zhang Qing-qing, Chen Chen. Occupant Evacuation Simulation Model during Civil Aircraft Emergency[J]. Journal of Southwest Jiaotong University, 2016,(01):161-167. Wang Wei-jie. Emergency Evacuation Simulation for Large Civil Aircraft and Evacuation Guiding System Optimization[D]. University of Electronic Science and Technology of China, 2013. Yu Liu, Weijie Wang, Hong-Zhong Huang. A new simulation model for assessing aircraft emergency evacuation considering passenger physical characteristics[J]. Reliability Engineering & System Safety, 2014, 121: 187-197. Chen Chen. Research on Behavior of Civil Aircraft Cabin Passengers during Emergency Evacuation[D]. Tianjin: Civil Aviation University of China, 2014. Du Hong-bing, Chen Chen, Wang Yan-qing. Research on Crowds Behavior Characteristics of Civil Aircraft Cabin Passengers during Emergency Evacuation [J]. Fire Science and Technology, 2014, 33(7): 818-821. Galea E R, Finney K M, Dixon A J, et al. Aircraft Accident Statistics and Knowledge Database: Analyzing Passenger Behavior in Aviation Accidents[J]. Journal of Aircraft, 2006, 43(5):1272-1281. Helen Muir, Ann Cobbett MSc. Aircraft Evacuation Tests - An Initial Assessment of the Influence of Various Aisle Configurations and Lighting Conditions under Different Evacuation Scenarios[R]. Transport Canada Publication No TP 12832E,1996. McLean GA, Corbett CL,Larcher K. Access-to-Egress I: Interactive Effects of Factors that Control the Emergency Evacuation Of Naïve Passengers through the Transport Airplane Type-III Overwing Exit[R]. Office of Aerospace Medicine Report, 2002,8. Hedo J M, Martinez-Val R. Assessment of Narrow-body Transport Airplane Evacuation by Numerical Simulation[J]. Journal of Aircraft, 2011,48(5):1785-1794.