CONTRIBUTION OF THE EVACUATION SIMULATOR FOR A TOWN

AREA REMODELING PLAN FOR PROTECTION AGAINST

DISASTERS OF TSUNAMI

HITOSHI GOTOH
Department of Urban and Environmental Engineering, Kyoto University, Kyoto-Daigaku-Katsura,
Nishikyo-ku
Kyoto, 615-8540, Japan

EIJI HARADA
Department of Urban and Environmental Engineering, Kyoto University, Kyoto-Daigaku-Katsura,
Nishikyo-ku
Kyoto, 615-8540, Japan

KEISUKE OHNIWA
School of Urban and Environmental Engineering, Kyoto University, Kyoto-Daigaku-Katsura,
Nishikyo-ku
Kyoto, 615-8540, Japan

Investigation of the evacuation process from tsunami attack is quite important for coastal town
planning against tsunami disaster. Reproduction of imminent situation in a real disaster is impossible
in ordinary evacuation training. For this reason, a numerical simulation is an effective tool to
investigate the evacuation process in tsunami attack. We have been developed the DEM-base human
behavior model to simulate an evacuation behavior. In the present study, a disaster prevention
planning against tsunami in a small model town is conducted from the viewpoint of appropriate
setting of evacuation places.

1 Introduction
Quick evacuation from tsunami attack is a key to survival. A plan of an appropriate
evacuation route and evacuation place is indispensable for a quick evacuation.
Development of supporting tool for evacuation planning is required. A numerical
simulation can deal with various conditions flexibly. To investigate an evacuation process
in detail, several functions are required in the numerical simulation model. For example,
expression of a difference of locomotive power and age of individual are required to
simulate an evacuation process in a human scale. Development of evacuation simulator in
a human scale is also indispensable for simulation of the blockage condition of crowd
evacuation around a bottleneck of evacuation route. The DEM-base evacuation simulator
is effective for dealing with evacuation behavior of individual. We have developed the
DEM-base evacuation simulator, and investigated the crowd evacuation process in
tsunami attack [1] and the crowd evacuation process from flooded underground space[2].
In this study, the disaster prevention in town planning was discussed from the viewpoint
of appropriate selection of the evacuation places by performing the numerical simulations

1
2

in different scenarios. The effectiveness of this kind of evaluation planning method based
on the DEM-base evacuation simulator is clarified.

2 DEM-base Crowd Evacuation Simulator

The DEM-base crowd evacuation simulator proposed by Gotoh et al.[1] is used for the
disaster prevention in town planning. The motion of person is written by translational and
rotational equations as follows:

d u hi
M hi = F hint + F aw (1)
dt

d ~hi d f vh rd hi4
I hi = 2hi ! " f hs , j ; I hi = h 32 (2)
dt j

in which, Fhint=inter-element force between person/person or person/wall; Faw=walking

force of person; uh=moving velocity; Mh=mass of person which is assumed as cylinder
shape; h=specific density of person element(=0.98); eh=coefficient which means the
difference of volume between cylinder and person(=0.34); dhi=diameter of person
element;  h=angular velocity of person element; fhs=external force in the tangential
direction between i- and j-th person elements; and Ihi=moment of inertia of person
element. The force between person/person or person/wall elements is described by
following the standard DEM. These force act in the area as shown in Fig. 1 in
consideration of the view range of person as follows:

r r
Khcf d- E idij E n
3 3
r hij E * d +d (3)
hi hj
] otherwise g
2
2 2
r hij = r j - ri ; r hij = ^ x i - x jh + ^ y i - y jh (4)

u hi $ r hij
cos i dij = (5)
u hi r hij

in which rij=positional vector between i- and j-th person elements; dij=angle between
moving direction of i-th and positional vector j-th person elements; and

hcf=psychological radius between person elements observed by Kiyono et al.[3]. When
Eq. (3) is satisfied, inter-element force is evaluated from spring and dashpot system
which is arranged between elements. Individual moves according to the designated
direction, which is given as location-depending information. The equilibrium velocity of
individual is described in consideration of the number density of people in the view range,
and uniform acceleration is given up to the equilibrium velocity. In addition, this
 3

evacuation simulator shows good agreement with the relation between density and
velocity of crowd, which observed by Koseki and Watanabe[4].

Figure 1. Behavior of individual is controlled according to view range.

3 Town Planning against Tsunami by Crowd Evacuation Simulator

3.1. Initial Condition

A disaster prevention town planning in the model town, which places in tsunami hazard
area, is examined. The model town formed in consideration of the features of a general
coastal town. In this study, the model town is modeled by referring the land use of the
actual town Wakayama prefecture in Japan as shown Fig. 2. A hill as the evacuation
place lies in the calculated domain of about 1km2. Access to a hill is only one route of the
western road. A hill is 30m above sea level. For that reason, the maximum evacuation
velocity is set at 70% of flat area on the route from the town to the hill, which is
assumption peculiar to this simulation. The number of the people and the details (gender
and age) were set by population statistical data of Wakayama prefecture (refer to Table 1).
As shown Table 2, the people are arranged in four blocks at random. Area of the hotel
and the hospital are assumed as non-residential area. Tsunami is supposed to arrive at 900
seconds after the evacuation warning in this simulation. People have to reach the
evacuation place, within 900 seconds. The present evacuation place is only the hill. In the
planning stage, existing building taller than 15m above sea level and tsunami shelter, or
evacuation tower, capacity of which is 60 people, are assumed as additional evacuation
places. In addition, the simulations were carried out in consideration of the capacity of
the buildings.
4

Figure 1. Model coastal town is set by referring to actual town.

Table 1. People are arranged according to the statistical data of

Wakayama prefecture.

Category Gender Age Number ratio Velocity

1 Male 10-39 132 15.73 1.45
2 Female 10-39 116 13.83 1.23
3 Male 40-69 157 18.71 1.19
4 Female 40-69 178 21.22 1.04
5 Male Over70 70 8.34 0.99
6 Female Over70 108 12.87 0.89
7 Child 5-9 78 9.30 1.06
Total 839 100.00 % m/s

Table 2. People are arranged in each category.

category 1 2 3 4 5 6 7 Sum.
block L 15 16 21 23 9 9 8 101
block CL 62 57 61 64 31 43 32 350
block CR 27 17 36 47 12 33 21 193
block R 28 26 39 44 18 23 17 195

3.2. Investigation of Present Condition for Evacuation

Figure 3 shows the snapshots and the time series of the accumulated number of the
people who completed evacuation of the simulation No.1, which is performed to
investigate the present condition. It is confirmed that in this condition numerous people
don’t reach the hill from the snapshot at 900s. The evacuation completion time is 1524
seconds in the simulation No.1. The completion rate of evacuation is 40% at the arrival of
tsunami at the time t=900s. From these results, it is found that the present condition isn’t
suitable for a safe evacuation planning. The map, which indicates the initial position of
the people who cannot evacuate within 900 seconds, is shown in Fig. 4. Most of the
people in the south-east part of town are exposed to the dangerous situation of tsunami.
By utilizing the roof of the hotel which is in the south-east part as the evacuation place
 5

will be considered as the effective countermeasure for more effective evacuation. Here,
the simulation No.2 that added the roof of a hotel as the evacuation place was performed.

Figure 3. Snapshots (left) and time series of accumulation people (right) in simulation No1 are shown.

Figure 4. Initial position of people who cannot evacuate within 900s is gathering in the south-east part.

3.3. Simulation for Additional Evacuation Place

The time series of the accumulated number of the people who completed evacuation of
the simulation No.2 and the simulation No.1 is shown in left side of Fig. 5. Whole of the
completion evacuation time is considerably shortened from 1542 seconds of the
simulation No.1 to 999 seconds of the simulation No.2. From this result, the
improvement of the evacuation efficiency by utilizing the hotel as the evacuation place is
understood. However, further improvement of evacuation completion time shorter than
900 seconds is necessary. The initial position, which means the hazardous area in the
simulation No.2, is shown in right side of Fig. 5. The 7% people cannot evacuate. So
construction of middle size evacuation facilities or evacuation tower is effective for
shortening of evacuation time as shown in left side of Fig. 6. So, to complete the
evacuation within 900 seconds, two simulations, namely the simulation No.3 in which the
6

hospital is utilized as the evacuation place; and the simulation No.4 in which the
evacuation towers are constructed as the evacuation place are performed. Right side of
Fig. 6 show the time series of the accumulated number of the people who completed
evacuation of the simulation No.3 and No.4 with the results of the simulation No.1 and
No.2. The time of the completion of evacuation is 795 seconds and 828 seconds
respectively. Both of the simulation No.3 and No.4, all of the people can evacuate within
900 seconds. In an actual disaster prevention town planning, the better plan is chosen in
consideration of the construction costs and landscape.

Figure 5. Time series of accumulated number of people who completed evacuation (left) and hazardous area,
which means initial position of people who cannot evacuate within 900s (right), are shown.

Figure 6. Planning of new facilities for evacuation within 900s (left) and time series of each simulation (right)
are shown.
 7

4 Conclusions
The DEM-base evacuation simulator was applied to the disaster prevention town
planning. Simulations of several plans from the viewpoint of shortening of evacuation
time were performed. And the processes of the disaster prevention town planning were
shown concretely. The DEM-base evacuation simulator is the Lagrangian model. Hence,
motion of person can be tracked and the animation of the simulation results gives the
evacuation situation visually. So, this kind of animation is thought to be an effective tool
for an explanation to inhabitants. We would like to promote this kind of the DEM-base
evacuation simulator as a key tool of town planning. As the example of the CG, the
snapshots of the simulation No.3 and No.4 around the hotel and the evacuation tower are
shown in Fig. 7 respectively. This kind of CG enables us to image the evacuation process
as virtual reality. Camera work is comparatively easy by using the CG technique.
Therefore the various scenes in evacuating processes are created by the CG. We would
like to investigate the possibility of this kind of CG as the virtual training for evacuations
in the near future.

Figure 7. Typical snapshots of simulation No.3 and No.4 by computer graphics, evacuation process is shown
visibly from person behavior.

References

1. H. Gotoh, E. Harada, Y. Kubo and T. Sakai, Annual Jourmnal of Coastal

Engineering, JSCE, Vol.51, 1261-1265 (2004) (in Japanese).
2. E. Harada, H. Gotoh, T. Sakai and Y.Kubo, Annual Journal of Hydraulic
Engineering, JSCE, Vol.50, 589-594 (2006) (in Japanese).
3. J. Kiyono, F. Miura and K. Takimoto, Jour of Structure Mechanics and Earthquake
Engineering, No.537/I-35, 233-244 (1996) (in Japanese).
4. K. Koseki and S. Watanabe, Transaction of the AIJ., 762 (1967) (in Japanese).