The document describes a simulation model for evaluating wide-area evacuation difficulty in densely populated wooden residential areas during disasters. The model simulates property damage, evacuation behavior, and interactions between evacuees. It is used to 1) visualize risk levels in residential areas, 2) quantify risk by evaluating evacuation difficulty before and after improvement projects, and 3) propose effective improvement strategies by testing scenarios like adding evacuation routes. The case study area is a district in Tokyo, and simulations analyze how evacuation difficulty changed after upgrading building fire-resistance from 1991 to 2006.
From idea to production in a day – Leveraging Azure ML and Streamlit to build...
Iscram2016 slide oki
1. Wide-area Evacuation
Difficulty in Densely-built
Wooden Residential Areas
Takuya Oki and Toshihiro Osaragi
Tokyo Institute of Technology
CREST, Japan Science and Technology Agency
1
2. OUTLINE OF PRESENTATION
1. INTRODUCTION
2. OVERVIEW OF SIMULATION MODEL
• Overview of Wide-area Evacuation Simulation Model
• Property Damage Model
• Wide-area Evacuation Behavior Model
• Definition of Wide-area Evacuation Difficulty
3. APPLICATIONS OF SIMULATION TO DISASTER
MITIGATION PLANNING
• Analyzed Area and Assumptions of Simulation
• Visualization
• Evaluation of Wide-area Evacuation Difficulty before/after
Improvement Project
• Effects of Adding New Evacuation Routes between Two
Intersections
4. SUMMARY AND CONCLUSIONS 2
3. OUTLINE OF PRESENTATION
1. INTRODUCTION
2. OVERVIEW OF SIMULATION MODEL
• Overview of Wide-area Evacuation Simulation Model
• Property Damage Model
• Wide-area Evacuation Behavior Model
• Definition of Wide-area Evacuation Difficulty
3. APPLICATIONS OF SIMULATION TO DISASTER
MITIGATION PLANNING
• Analyzed Area and Assumptions of Simulation
• Visualization
• Evaluation of Wide-area Evacuation Difficulty before/after
Improvement Project
• Effects of Adding New Evacuation Routes between Two
Intersections
4. SUMMARY AND CONCLUSIONS 3
4. DENSELY-BUILT WOODEN
RESIDENTIAL AREAS IN JAPAN
• In Japan, there are many
densely-built residential
areas with old wooden
houses and complex road
networks, consisting of
narrow streets.
• As these areas are high risk
in the case of a large
earthquake, it has been
suggested to improve the
spatial characteristics as
soon as possible.
4
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
5. PROBLEMS IN DENSELY-BUILT
WOODEN RESIDENTIAL AREAS
• However, improvements in densely-
built wooden residential areas have
not been smoothly performed due
to consensus-building with local
residents and the costs of projects.
• Therefore, these residential areas,
where it is difficult to prevent big
fires during a large earthquake and
to smoothly evacuate to evacuation
areas, have still remained.
5
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
6. WHY ARE THE IMPROVEMENTS
SO DIFFICULT?
(1) It is not easy to understand how dangerous the
residential area in case of a large earthquake is by
using hazard maps based on disaster mitigation
planning.
(2) Stake holders (such as local residents, residential
developers, the government and municipalities)
cannot easily realize the effects of the improvement
projects and how to improve the areas.
6
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
7. OUR APPROACH
• We have attempted to build methods to:
(1) visualize and quantify risk in densely-built wooden
residential areas
(2) propose effective and efficient ways to improve the
areas.
• We have constructed a simulation model, which describes
people’s evacuation behavior under the assumption of
various hazards caused by a large earthquake.
• Using this simulation model, we also demonstrate the
evaluation of risk in densely-built wooden residential
areas and the effects of improvement projects from the
viewpoint of evacuation difficulty.
7
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
8. ADVANTAGES OF OUR METHOD
1. Visual and quantitative grasp of the risk
2. Applicability to various situations only by
changing input datasets
3. Evaluation of the risk at a building or street level
4. Consideration of multiple kinds of property
damage
5. Consideration of the interaction between
property damage and people’s evacuation
behavior
8
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
9. OUTLINE OF PRESENTATION
1. INTRODUCTION
2. OVERVIEW OF SIMULATION MODEL
• Overview of Wide-area Evacuation Simulation Model
• Property Damage Model
• Wide-area Evacuation Behavior Model
• Definition of Wide-area Evacuation Difficulty
3. APPLICATIONS OF SIMULATION TO DISASTER
MITIGATION PLANNING
• Analyzed Area and Assumptions of Simulation
• Visualization
• Evaluation of Wide-area Evacuation Difficulty before/after
Improvement Project
• Effects of Adding New Evacuation Routes between Two
Intersections
4. SUMMARY AND CONCLUSIONS 9
10. OUTLINE OF PRESENTATION
1. INTRODUCTION
2. OVERVIEW OF SIMULATION MODEL
• Overview of Wide-area Evacuation Simulation Model
• Property Damage Model
• Wide-area Evacuation Behavior Model
• Definition of Wide-area Evacuation Difficulty
3. APPLICATIONS OF SIMULATION TO DISASTER
MITIGATION PLANNING
• Analyzed Area and Assumptions of Simulation
• Visualization
• Evaluation of Wide-area Evacuation Difficulty before/after
Improvement Project
• Effects of Adding New Evacuation Routes between Two
Intersections
4. SUMMARY AND CONCLUSIONS 10
11. WIDE-AREA EVACUATION SIMULATION
MODEL
11
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
The concept of Multi-Agent Simulation (MAS) enables us to
describe the interaction between agents (such as evacuees, the
spread of fire, congestion of streets, etc.) and the process by
which the conditions of the city vary. * Step interval = 30 sec
12. WIDE-AREA EVACUATION SIMULATION
MODEL
12
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
The concept of Multi-Agent Simulation (MAS) enables us to
describe the interaction between agents (such as evacuees, the
spread of fire, congestion of streets, etc.) and the process by
which the conditions of the city vary. * Step interval = 30 sec
13. WIDE-AREA EVACUATION SIMULATION
MODEL
13
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
The concept of Multi-Agent Simulation (MAS) enables us to
describe the interaction between agents (such as evacuees, the
spread of fire, congestion of streets, etc.) and the process by
which the conditions of the city vary. * Step interval = 30 sec
14. WIDE-AREA EVACUATION SIMULATION
MODEL
14
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
The concept of Multi-Agent Simulation (MAS) enables us to
describe the interaction between agents (such as evacuees, the
spread of fire, congestion of streets, etc.) and the process by
which the conditions of the city vary. * Step interval = 30 sec
20. EVACUATION BEHAVIOR MODEL
20
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
Temporary refuge
Small park
School
Temple/Shrine
etc.
Evacuation area
University campus
Riverbed
etc.
Estimated from the
Person-Trip Survey
(2008)
22. WIDE-AREA EVACUATION DIFFICULTY
Calculated based on the results of the wide-area evacuation
simulation, and considered as the risk in the residential
areas from the viewpoint of wide-area evacuation.
[Definition 1: for each building]
Ndi : Number of the damage patterns in which someone in
building i cannot evacuate to any temporal refuges and
evacuation areas
N : Number of simulation trials
[Definition 2: for each area]
Npjk : Number of people with
difficulty in wide-area evacuation in area j for trial k
Npj : Number of people inside buildings and pedestrians in
area j
22
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
Bi diD N N
1
1
( )
N
Aj pjk pj
k
D N N
N
23. WIDE-AREA EVACUATION DIFFICULTY
Calculated based on the results of the wide-area evacuation
simulation, and considered as the risk in the residential
areas from the viewpoint of wide-area evacuation.
[Definition 1: for each building]
Ndi : Number of the damage patterns in which someone in
building i cannot evacuate to any temporal refuges and
evacuation areas
N : Number of simulation trials
[Definition 2: for each area]
Npjk : Number of people with
difficulty in wide-area evacuation in area j for trial k
Npj : Number of people inside buildings and pedestrians in
area j
23
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
Bi diD N N
1
1
( )
N
Aj pjk pj
k
D N N
N
24. WIDE-AREA EVACUATION DIFFICULTY
Calculated based on the results of the wide-area evacuation
simulation, and considered as the risk in the residential
areas from the viewpoint of wide-area evacuation.
[Definition 1: for each building]
Ndi : Number of the damage patterns in which someone in
building i cannot evacuate to any temporal refuges and
evacuation areas
N : Number of simulation trials
[Definition 2: for each area]
Npjk : Number of people with
difficulty in wide-area evacuation in area j for trial k
Npj : Number of people inside buildings and pedestrians in
area j
24
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
Bi diD N N
1
1
( )
N
Aj pjk pj
k
D N N
N
25. OUTLINE OF PRESENTATION
1. INTRODUCTION
2. OVERVIEW OF SIMULATION MODEL
• Overview of Wide-area Evacuation Simulation Model
• Property Damage Model
• Wide-area Evacuation Behavior Model
• Definition of Wide-area Evacuation Difficulty
3. APPLICATIONS OF SIMULATION TO DISASTER
MITIGATION PLANNING
• Analyzed Area and Assumptions of Simulation
• Visualization
• Evaluation of Wide-area Evacuation Difficulty before/after
Improvement Project
• Effects of Adding New Evacuation Routes between Two
Intersections
4. SUMMARY AND CONCLUSIONS 25
26. OUTLINE OF PRESENTATION
1. INTRODUCTION
2. OVERVIEW OF SIMULATION MODEL
• Overview of Wide-area Evacuation Simulation Model
• Property Damage Model
• Wide-area Evacuation Behavior Model
• Definition of Wide-area Evacuation Difficulty
3. APPLICATIONS OF SIMULATION TO DISASTER
MITIGATION PLANNING
• Analyzed Area and Assumptions of Simulation
• Visualization
• Evaluation of Wide-area Evacuation Difficulty before/after
Improvement Project
• Effects of Adding New Evacuation Routes between Two
Intersections
4. SUMMARY AND CONCLUSIONS 26
27. STUDY AREA
27
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
Wakabayashi
3 chome & 4 chome,
Setagaya Ward,
Tokyo
Temporary refuge
Railway crossing
28. ASSUMPTIONS OF SIMULATION
28
Scenario Earthquake North Tokyo Bay Earthquake (M 7.3)
(Seismic velocity and acceleration were determined by
each building based on set instrumental seismic intensity.)
Weather Condition Sunny with 8 m/s north wind
Number of Fire-
outbreak Buildings
93 buildings in whole Setagaya Ward
(This is the average number of 100 times fire-outbreak
simulations.)
Earthquake Occurrence
Time
6:00 pm on a weekday in winter
Number of Simulation
Trials
100 times
(We prepared 100 patterns of property damage estimated
by the property damage model, and carried out one trial
for each pattern.)
Reference Damage estimation published by Tokyo Metropolitan
Government (2012)
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
29. SIMULATION RESULTS
1. Visualization (= Visualize risk in densely-built
wooden residential areas)
2. Evaluation of wide-area evacuation difficulty
before/after improvement project (= Quantify
risk in densely-built wooden residential areas)
3. Effects of adding new evacuation routes between
two intersections (= Propose effective and
efficient ways to improve the areas)
29
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
30. SIMULATION RESULTS
1. Visualization (= Visualize risk in densely-built
wooden residential areas)
2. Evaluation of wide-area evacuation difficulty
before/after improvement project (= Quantify
risk in densely-built wooden residential areas)
3. Effects of adding new evacuation routes between
two intersections (= Propose effective and
efficient ways to improve the areas)
30
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
31. RESULT 1: VISUALIZATION
31
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
■ Evacuation area
■ Collapsed building
■ Burned building
― Blocked street
● Refugees
● People who have difficulty
in wide-area evacuation
32. SIMULATION RESULTS
1. Visualization (= Visualize risk in densely-built
wooden residential areas)
2. Evaluation of wide-area evacuation difficulty
before/after improvement project (= Quantify
risk in densely-built wooden residential areas)
3. Effects of adding new evacuation routes between
two intersections (= Propose effective and
efficient ways to improve the areas)
32
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
33. SIMULATION RESULTS
1. Visualization (= Visualize risk in densely-built
wooden residential areas)
2. Evaluation of wide-area evacuation difficulty
before/after improvement project (= Quantify
risk in densely-built wooden residential areas)
3. Effects of adding new evacuation routes between
two intersections (= Propose effective and
efficient ways to improve the areas)
33
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
36. 36
Composition ratio of each structure and comparison
of the rate of fireproof area
1000 20 40 60 80
Composition ratio [%]
0 20 40 60
Rate of fireproof area [%]
2006
2006
1991
Naked-
wooden
Fire-
proofing
Semi-fire-
resistant
Fire-
resistant
(a) Composition ratio of each structure (b) Rate of fireproof area
1991
3-chome4-chome
Wakabayashi
2006
2006
1991
1991
3-chome4-chome
Wakabayashi
(Rate of fireproof area) = SA x 0.8SB / S
where SA, SB, and S are total sum of area of fire-resistant buildings,
semi-fire-resistant buildings, and all buildings, respectively.
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
RESULT2:EVALUATIONOFWIDE-AREAEVACUATION
DIFFICULTYBEFORE/AFTERIMPROVEMENTPROJECT
44. 44
Advantage
1. The cost of an improvement project can be reduced
because the number of buildings involved with adding
new evacuation routes (i.e., the number of people
required consensus-buildings) is fewer than the case
of widening streets.
2. Connecting two streets with an evacuation route
simultaneously shortens the length of both streets.
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
RESULT3:EFFECTSOFADDINGNEWEVACUATION
ROUTESBETWEENTWOINTERSECTIONS
45. 45
Simulation procedure
1. Execute the wide-area evacuation simulation 100
times for the different property damage patterns,
based on the data for 2006.
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
RESULT3:EFFECTSOFADDINGNEWEVACUATION
ROUTESBETWEENTWOINTERSECTIONS
2006
■ 0%
■ 0% - 10%
■ 10% - 20%
■ 20% - 30%
■ 30% - 40%
■ 40% - 50%
■ 50% -
100 m
N
2. List all streets in order of
the estimated number of
people who cannot
evacuate to any
evacuation areas.
46. 46
Simulation procedure
3. Add new evacuation routes in the 1st, 2nd and
3rd stages so as that the total length of added
routes is equal to about 1%, 2% and 3%
respectively of the total length of pre-existing
streets in the study area.
4. Execute the simulation 100 times under the
condition of adding new evacuation routes from
1% to 3% step by step in each damage pattern.
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
RESULT3:EFFECTSOFADDINGNEWEVACUATION
ROUTESBETWEENTWOINTERSECTIONS
47. 47
Effects of reducing Wide-area Evacuation Difficulty
(Wbi) by adding new evacuation routes
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
RESULT3:EFFECTSOFADDINGNEWEVACUATION
ROUTESBETWEENTWOINTERSECTIONS
48. 48
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
RESULT3:EFFECTSOFADDINGNEWEVACUATION
ROUTESBETWEENTWOINTERSECTIONS
Effects of reducing Wide-area Evacuation Difficulty
(Wbi) by adding new evacuation routes
49. 49
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
RESULT3:EFFECTSOFADDINGNEWEVACUATION
ROUTESBETWEENTWOINTERSECTIONS
Effects of reducing Wide-area Evacuation Difficulty
(WAj) by adding new evacuation routes
50. 50
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
RESULT3:EFFECTSOFADDINGNEWEVACUATION
ROUTESBETWEENTWOINTERSECTIONS
Decrease of the number of people who cannot
evacuate to any evacuation areas for each target street
51. OUTLINE OF PRESENTATION
1. INTRODUCTION
2. OVERVIEW OF SIMULATION MODEL
• Overview of Wide-area Evacuation Simulation Model
• Property Damage Model
• Wide-area Evacuation Behavior Model
• Definition of Wide-area Evacuation Difficulty
3. APPLICATIONS OF SIMULATION TO DISASTER
MITIGATION PLANNING
• Analyzed Area and Assumptions of Simulation
• Visualization
• Evaluation of Wide-area Evacuation Difficulty before/after
Improvement Project
• Effects of Adding New Evacuation Routes between Two
Intersections
4. SUMMARY AND CONCLUSIONS 51
52. OUTLINE OF PRESENTATION
1. INTRODUCTION
2. OVERVIEW OF SIMULATION MODEL
• Overview of Wide-area Evacuation Simulation Model
• Property Damage Model
• Wide-area Evacuation Behavior Model
• Definition of Wide-area Evacuation Difficulty
3. APPLICATIONS OF SIMULATION TO DISASTER
MITIGATION PLANNING
• Analyzed Area and Assumptions of Simulation
• Visualization
• Evaluation of Wide-area Evacuation Difficulty before/after
Improvement Project
• Effects of Adding New Evacuation Routes between Two
Intersections
4. SUMMARY AND CONCLUSIONS 52
53. SUMMARY AND CONCLUSIONS
• We constructed a wide-area evacuation simulation
model, which integrated the property damage
model and the evacuation behavior model.
• Wide-area Evacuation Difficulty was defined to
quantify the risk in densely-built wooden
residential areas in terms of wide-area evacuation.
• Using some examples, we demonstrated the
usefulness of our simulation model as a tool for
visualizing/quantifying the risk, evaluating projects
and discussing how to effectively/efficiently reduce
wide-area evacuation difficulty.
53
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
54. SOME IDEAS OF FURTHER IMPROVEMENT &
APPLICATION OF THE SIMULATION MODEL
• Incorporating other kinds of people’s actions
immediately after a large earthquake
Returning home on foot
Rescue / firefighting activities by local residents
• Improving the programming codes for high-speed
and large-scale simulation
• Applying to other cases (e.g. evacuation from
tsunami, etc.)
54
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS
55. SOME IDEAS OF FURTHER IMPROVEMENT &
APPLICATION OF THE SIMULATION MODEL
• Incorporating other kinds of people’s actions
immediately after a large earthquake
Returning home on foot
Rescue / firefighting activities by local residents
• Improving the programming codes for high-speed
and large-scale simulation
• Applying to other cases (e.g. evacuation from
tsunami, etc.)
55
1. INTRODUCTION 2.SIMULATIONMODEL 3. APPLICATION 4. CONCLUSIONS