Sei sulla pagina 1di 14

Risk Analysis, Vol. 28, No. 5, 2008 DOI: 10.1111/j.1539-6924.2008.01104.

Cellular Automata-Based Systematic Risk Analysis


Approach for Emergency Response

Xuewei Ji,1,2 ∗ Wenguo Weng,1 and Weicheng Fan1

Emergency response is directly related to the allocation of emergency rescue resources. Effi-
cient emergency response can reduce loss of life and property, limit damage from the primary
impact, and minimize damage from derivative impacts. An appropriate risk analysis approach
in the event of accidents is one rational way to assist emergency response. In this article, a
cellular automata-based systematic approach for conducting risk analysis in emergency re-
sponse is presented. Three general rules, i.e., diffusive effect, transporting effect, and dissipa-
tive effect, are developed to implement cellular automata transition function. The approach
takes multiple social factors such as population density and population sensitivity into con-
sideration and it also considers risk of domino accidents that are increasing due to increasing
congestion in industrial complexes of a city and increasing density of human population. In
addition, two risk indices, i.e., individual risk and aggregated weighted risk, are proposed to
assist decision making for emergency managers during emergency response. Individual risk
can be useful to plan evacuation strategies, while aggregated weighted risk can help emer-
gency managers to allocate rescue resources rationally according to the degree of danger in
each vulnerable area and optimize emergency response programs.

KEY WORDS: Cellular automata; domino effect; emergency response; risk analysis; social factors

1. INTRODUCTION exposure to disaster impacts. In an effort to minimize


the potential and subsequent impact of disasters on
Disasters occur frequently, taking the form of
life and property, emergency response has received
floods, hurricanes, earthquakes, fires, terrorism, and
more and more attention.(4,5)
nuclear and hazardous material accidents.(1–3) These
Emergency response is “applying science, tech-
emergency situations can result in great loss of life
nology, planning, and management to deal with ex-
and property. Public awareness of hazards, emer-
treme events that can injure or kill large numbers of
gencies, and disasters has increased as the cost of
people, do extensive damage to property, and dis-
disasters has increased dramatically. Rapid popula-
rupt community life.”(3) Emergency response is di-
tion and economy growth in the most hazardous geo-
rectly related to the supplies of manpower and ma-
graphical areas of the country have created increased
terial resources for preventing accidents in an urban
area, as well as the quantum of monetary resources
to be committed for the purpose. Efficient emer-
1 Center for Public Safety Research, Department of Engineering gency response can reduce loss of life and property,
Physics, Tsinghua University, Beijing, P. R. China. limit damage from the primary impact, and minimize
2 School of Aerospace, Tsinghua University, Beijing, P. R. China.
damage from derivative impacts. An appropriate risk
∗ Address correspondence to Xuewei Ji, Center for Public Safety
Research, 4F, Section 1, the West Main Building, Tsinghua Uni-
analysis approach in the event of accidents is one ra-
versity, Beijing, P. R. China; 100084; tel: +86-10-62792402; tional way to conduct efficient emergency response.
jixuewei00@mails.tsinghua.edu.cn. Risk is defined as the likelihood that an event will

1247 0272-4332/08/0100-1247$22.00/1 
C 2008 Society for Risk Analysis
1248 Ji, Weng, and Fan

occur at a given location within a given time pe- analysis approach incorporating the domino effect is
riod and inflict casualties and damage.(6,7) For about very necessary.
10 years, many methodologies have been developed Furthermore, it is important for the emergency
to undertake a quantitative risk assessment, e.g., managers to realize that not only technical aspects
MCAA (maximum credible accident analysis),(8) but also political, psychological, and social processes
FTA (fault tree analysis) study,(9) or any other ex- all play an important role in risk analysis during
ercise in loss prevention and safety implementation emergency response.(4) Without doubt, some social
whose essential inputs come from the probability and factors such as population density and population
the enormity of the likely accidents. However, it has sensitivity can directly affect emergency response.
been found that a risk analysis approach to serve During emergency response, a sparsely populated
emergency response during accidents is still missing area with high individual risk may be not the area in
in literatures. The methodologies mentioned above most need of rescue resources. In contrast, the area
cannot be applied to emergency response. with high individual risk and dense population is the
First, as we know, emergency response is urgent most dangerous area and needs more attention. The
and the emergency managers need to know the risk constitution of the population is of great importance
of an accident scene as soon as possible. Accuracy due to the vulnerability variation of different popu-
and rapidity are two crucial factors for risk analysis lation groups to the hazards, such as children, elderly
that can determine success or failure of the emer- people, and patients. Accordingly, incorporation of
gency response. Accident modeling is the foundation some social factors into risk analysis is necessary.
of risk analysis, so simulating the accident spread- Based on the above considerations, a cellular
ing accurately and rapidly is very important. Em- automata-based systematic approach is presented to
pirical formulas(10,11) or partial difference equations assist emergency response. The approach can fore-
(PDEs)(12–14) are usually used to model accidents in cast the risk of a single accident as well as a chain
risk analysis in previous studies. However, an empir- of accidents (domino effect) and deal with hetero-
ical formula method is simple but not very accurate, geneous environmental conditions. It can also sys-
while the PDE method is accurate but computation tematically consider multiple factors such as popula-
time consuming. One needs to develop an approach tion density and population sensitivity. To meet the
that is both accurate and rapid. requirements of accuracy and rapidity during emer-
Second, the common feature of some method- gency response, a cellular automata (CA) method is
ologies described above is that the accident impact is adopted. After analyzing the spreading mechanism
usually assumed to propagate outward from the ac- of various accidents, three common rules (diffusive
cidental epicenter in a radically symmetrical fashion. effect, transporting effect, and dissipative effect) are
With this basic assumption, the impact areas of these proposed to define cellular automata transition func-
accidents are denoted with circles. The areas corre- tion. Besides, two risk indices are adopted, e.g., in-
sponding to, say, 100%, 50%, and 25% probability of dividual risk and aggregated weighted risk, to assist
death due to an accident are bounded by circles of decision making for emergency managers. Individual
increasing radii, with the accident site serving as the risk can be used as prewarning information to help
center of the circles. But in real situations, the condi- the people who are likely to be affected to select the
tions prevailing in the neighborhood of the accident appropriate escape routes. Aggregated weighted risk
epicenter are rarely homogeneous. The area of im- can help emergency managers to allocate rescue re-
pact of an accident shall have an irregular shape. sources rationally according to the degree of danger
Third, recently, domino accidents are more com- in each vulnerable area and optimize emergency re-
mon on account of increasing congestion in indus- sponse programs.
trial complexes and increasing density of human CA is a class of automata defined on the simula-
population around such complexes.(15) However, the tion space divided into discrete areas called “cells,”
domino effect is often neglected or is badly dealt and was developed by John von Neumann and
with due to lack of related scientific support during Stanislaw Ulam in the United States.(16,17) Though
emergency response. The explosion in Jilin chemical the purpose of developing CA was said to model
plant of China in 2005 leading to a series of emer- a self-reproducing machine or to understand the
gencies, which formed an emergency chain, is the mechanism of a neural system, it is currently being
best example. Therefore, in this complex emergency employed in diverse fields such as architectural de-
response decision-making process, a systematic risk sign, ecology, epidemiology, environmental hazard
Cellular Automata-Based Systematic Risk Analysis 1249

management, genetics, medical sciences, road traffic ∂φ   ∂ 2φ ∂φ



flow modeling, cryptography, image processing, ur- = i 2 − Ui + Sφ , (4)
∂t i=x,y,z
∂i ∂i
ban dynamics modeling, and others.

where i represents the coordinates in which the pro-


2. THEORETICAL BACKGROUND cesses are considered.
A physical quantity called accident intensity is Accident intensity spreading is dominated by
defined here to express the common characteristic two ingredients: “inner factor” and “outer factor.”
of various accidents, which describes the ability of The dynamics of propagation of the intensity is first
an accident to cause losses to life and property. It controlled by the diffusive effect, which is the inner
expresses different meanings in different accidents, factor. This implies that intensity flux liberated from
such as blast waves and shrapnel in explosions, gas the accident epicenter travels outward to the adjoin-
concentration in toxic gas dispersion, and so on. The ing cells and once each of these incident cells become
risk assessment here is performed by evaluating in- saturated with the intensity, they in turn begin to act
dividual risk for each cell and aggregated weighted as new intensity sources and the intensity flux begins
risk for a cluster of cells where population density to diffuse from these cells into their respective neigh-
and population sensitivity are taken into considera- borhoods. The intensity gradient, which expresses
tion. Population density reflects the population mag- different meanings in different accidents, is the driv-
nitude per unit area, and population sensitivity in- ing force governing this outward propagation of the
dicates the constitution of the general population. intensity flux. It means concentration gradient in
Individual risk (t+1 IRij ) is deemed to be a function of toxic gas dispersion and temperature gradient in heat
accident intensity (t+1 AIij ) and population sensitivity spreading and so on. According to common sense,
(PSij ), while aggregated weighted risk (t+1 AWRij ) de- accident spreading is also influenced by some outer
pends on individual risk (t+1 IRij ), population density factors. For example, wind and terrain slope will af-
(PDij ), and population sensitivity (PSij ). They can be fect the distribution of the toxic gas concentration
expressed as follows: and temperature difference between the ambience
  and the accident hazards will result in convection. We
t+1
IRij = f t+1 AI ij , PSij (1) say that the intensity propagation is also governed by
t+1  the transporting effect. The dissipative effect is also
t+1
AWRij = f IRij , PDij , PSij . (2) an important outer factor, which perhaps changes the
way of the intensity spreading. While spreading, bar-
2.1. Cell Accident Intensity State Analysis riers such as building, river, trees, and sedimentation
may all weaken the accident intensity.
In most accidents, such as fires and gas disper- Based on above analysis, three common rules
sion, the phenomenon of accident intensity spread- can be extracted, i.e., diffusive effect, transporting ef-
ing in an isotropic medium can be expressed by the fect, and dissipative effect, which the accident inten-
following equation: sity spreading obeys. Three generalized coefficients
     
∂ϕ ∂ ∂ϕ ∂ ∂ϕ ∂ ∂ϕ are defined to illustrate the three rules:
= x + y + z
∂t ∂x ∂x ∂y ∂y ∂z ∂z
∂ ∂ ∂ (1) Diffusive coefficient (): the ability of inten-
− (Uϕ) − (Vϕ) − (Wϕ) + Sϕ , (3) sity spreading from high intensity to low in-
∂x ∂y ∂z
tensity. It is the intrinsic attribute of the ac-
where ∂ϕ/∂t is the rate of transfer per unit space, ϕ is cident intensity spreading.
intensity of spreading accident,  is the diffusive co- (2) Transporting coefficient (τ ): the ability that
efficient describing transmissivity or conductance of environment transports the intensity.
the medium, U, V, and W are the convective coeffi- (3) Dissipative coefficient (α): the ability that
cient describing the degree of convection, and Sϕ is outer factors weaken (strengthen) the
the source intensity of accident epicenter. The term intensity.

∂x
(x ∂ϕ
∂x
) + ∂∂y ( y ∂ϕ
∂y

) + ∂z (z ∂ϕ
∂z
) expresses diffusive
effect and the term ∂ x (Uϕ) + ∂∂y (Vϕ) + ∂z
∂ ∂
(Wϕ) ex- The three general coefficients are critical for sim-
presses transporting effect. The above equation may ulation results, and they may be functions of other
be generalized as: physical quantities in some complex scenarios. They
1250 Ji, Weng, and Fan

 t+1 
express different meanings in different accidents. For t+1
AI ij = t AI ij + τ neighbor AI neighbor − t+1 AI ij
example, the diffusive coefficient expresses gas dif- neighbor
fusive coefficient in toxic gas dispersion, while it → →
means thermal energy diffusive coefficient in heat  r · Uij t+1 
= AI ij +
t
n AI neighbor − t+1 AI ij ,
spreading. rneighbor
neighbor
Under two-dimensional situations, “Moore (7)
neighborhood,” which comprises of eight immediate →
neighbors to each cell, is used. The amount of where rneighbor and r are respectively scalar and vec-
the accident intensity in a cell ij(t+1 AIij ) at time tor indicating the distance between neighborhood
step t + 1 would depend on the magnitude of the cells and target cell, n is the distance impact factor,

intensity present in the cell at the time t (t AIij ), and Uij is a vector indicating the direction of accident
the magnitude of the intensity present in neigh- spreading.
borhood cells at the time t + 1 (t+1 AINeighbour ),
(3) Dissipative effect rule.
and attributes of the cell and its neighborhood
cells’ ( i j , τ i j , α i j ). The attributes of each cell We regard the dissipative effect as a “barrier,”
include accident intensity state, natural factors, and which is considered to be a property of the cell. The
hazardous factors. Natural factors refer to wind effect of barriers is likely to be anisotropic, e.g., the
velocity, wind direction, barriers, and so on. Haz- effect of a wall as a barrier to toxic dispersions de-
ardous factor expresses whether there are potential pends on wind direction. The anisotropy of barriers
hazardous sources in the cell or not, which may is not discussed here, and an isotropic effect of bar-
cause derivative accidents. The transition function riers is assumed. That is to say, if there is a “barrier”
to compute accident intensity is established as in the cell ij, the accident intensity increases or de-
follows: creases certain amount of quantity linearly. The rule
  based on the above assumption is expressed as:
t+1
AI ij = f t AI ij , t+1 AI Neighbor , ij , τij , αij . (5)    
t+1
AI ij = 1 − t+1 αij f t AI ij , t+1 AI Neighbor , ij , τij ,
(1) Diffusive effect rule. (8)
where αij is positive if barriers weaken the intensity,
Two types of neighbors depending upon their
and negative otherwise.
orientation around the central cell—the adjacent
Incorporating above three rules, the cell transi-
neighbors and nonadjacent neighbors—are consid-
tion function can be set up:
ered.(18) Thus:  
   t+1
AI ij = t AI ij + 1 − t+1 αij
t+1
AI ij = t AI ij + ij ωa t+1
AI adjacent − t+1 AI ij ⎡
adjacent   
 t+1  × ⎣ij ωa t+1
AI adjacent − t+1 AI ij
+ ij ωb AI nonadjacent − t+1 AI ij , adjacent
nonadjacent  t+1 
(6) + ij ωb AI nonadjacent − t+1 AI ij
nonadjacent
where ωa is the adjacent neighbor’s weight to the tar- ⎤
→ →
get cell and ωb is the nonadjacent neighbor’s weight.  r · Uij t+1 
The two coefficients obey the following relationships: + AI neighbor − t+1 AI ij ⎦ .
rn
ωa > ωb, ωa + ωb = 1. neighbor neighbor

(9)
(2) Transporting effect rule.
2.2. Domino Effect Analysis
In this approach, the transporting effect is sup-
posed to be only affected by the cell, the line between To study the domino effect, the cells are sorted
whose center and the target cell’s is not orthogonal into two categories, general cells (cellg ) and haz-
with the transporting force direction. The distance ardous cells (cellh ). General cells are the cells that
between neighborhood cells and target cell is also an do not contain the potential hazardous source, and
impact factor. Thus: the hazardous cells are the cells that contain the
Cellular Automata-Based Systematic Risk Analysis 1251

potential hazardous source and serve as potential human vulnerability.(26,27) The probit function for fa-
seed cells. The domino effect is performed through tal injuries can be expressed as:
two levels.(15)  
At the first level, a screening of all the hazardous
t+1
Pij = (5 − m/σ ) + (1/σ ) ln t+1 AI nij × t , (11)
cells is done in order to identify hazardous cells and
where t+1 Pij is the probit value, m is the mean value
general cells. Then, we need to determine whether
of the intensity fatal to population, and σ is the stan-
certain hazardous cell can cause accidents accord-
dard deviation, both of which are from experiments
ing to the accident intensity the cell has perceived.
on animals or the historical data.
For this purpose, three methods have been proposed
Based on the assumption that the strength of the
in the literatures: (1) vulnerability threshold mod-
human body to accident intensity is normally dis-
els (the physical effect on the secondary target is
tributed with mean value of 5 and standard deviation
higher than a threshold value for damage and then
of 1,(25) the probability of loss of life in cell ij on re-
domino effect happens);(19–21) (2) propagation func-
ceiving certain accident intensity AI ij is given by:
tions based on empirical decay relations for physical
effects;(22) (3) propagation functions based on spe- t+1 Pij −5  2
1 u
cific probabilistic models.(23,24)
t+1
IRij = √ exp − du. (12)
2π −∞ 2
At the second level, a more detailed analysis is
conducted to compute the intensity states of all cells The approach proposed here to treat a sensi-
under domino effect. After obtaining the intensity tive population is based on the assumption that the
states of all cells resulting from the primary accident mean value of intensity fatal to a sensitive population
at a time t, we further judge which cell can be served is lower than the mean value of intensity fatal to a
as the seed cell based on the method referred at the normal population, and also that the standard devia-
first level. In the next time step, simulation is per- tion σ of the distribution for a sensitive population is
formed based on the propagation occurring from all lower than the standard deviation for a normal popu-
seeds. With time going by, we will analyze the tertiary lation. Suitable probit function can be generated for
accident, the quartus accident, and so on in the same the estimation of individual risk to a sensitive popu-
way. In the approach, the vulnerability threshold lation through adjusting σ and m.(28)
model is used. The domino effect can be expressed The overall individual risk due to contempo-
as: rary exposition to different types of physical ef-
fects (e.g., a toxic release and a fire, and so on)
If t AI ij > AI threshold
ij , t+1 AI ij = accident epicenter is also considered. It is calculated as a combina-
  tion of each related individual risk. The combination
otherwise t+1 AI ij = f t AI ij , t+1 AI Neighbor , ij , Uij , αij .
can be performed by different strategies. Individual
(10) risk is actually probability value; thus probabilistic
rules are required for the combination. Four methods
have been reported elsewhere.(29) In the approach,
2.3. Individual Risk and Aggregated Weighted
one can choose one or several methods to compute
Risk Analysis
the overall individual risk referring to detailed dis-
Accident intensity provides a basis for systematic cussions by Cozzani et al. (2005).
comparison of different exposure. However, in case Aggregated weighted risk (AWR) is cited to de-
of emergency response, the emergency managers are scribe the risk of a cluster of cells. It expresses “the
more concerned about human lives rather than ac- relationship between individual risk and the number
cident intensity itself. Furthermore, accident inten- of people suffering from a specified level of harm in
sity from different phenomena, e.g., thermal heat a given population from the realization of specified
flux and toxic concentration is not directly compa- hazards.”(30) The factor of population density varies
rable. On the other hand, the type of health dam- spatially and temporally, that is to say, different cells
age that an exposure might cause provides a common have different population density states at different
basis that allows direct comparison.(25) In the pro- times. Aggregated weighted risk is calculated by mul-
posed approach, we will only consider death proba- tiplying population density inside a certain area (A)
bility of an exposed individual. The probability can with its IR ij level in cell ij. The effect of protection
be calculated from the physical effects and the es- due to buildings is not accounted for here, so the re-
timated time of exposure using probit models for sults are conservative.
1252 Ji, Weng, and Fan


are selected and accident scenarios are developed.
t+1
AWR A = t+1
IRij × PDij (t + 1)d A (13)
A The algorithm evaluates the intensity state of each
cell in the matrix with respect to the neighborhood
cells, taking into account diffusive effect, transport-
2.4. Algorithm of the Systematic Approach
ing effect, and dissipative effect. After obtaining the
An algorithm of the systematic approach has intensity state of each cell, we judge whether some
been developed (Fig. 1), where simulation starts at cell contains a potential hazardous source; if yes,
t = 0 and T is the maximum time, user defined. The we judge whether the intensity of the cell exceeds
study area is divided into a two-dimension lattice of the threshold; if yes, we take the cell as a seed cell in
cells, and the attributes (hazardous factors, natural the next time step. Otherwise, we directly do the next
factors, and social factors) of each cell are identi- simulation. During each simulation step, we will take
fied. Seed cells that can cause the primary accidents population density state and population sensitivity

Fig. 1. Algorithm of the cellular


automata-based systematic risk analysis
for emergency response.
Cellular Automata-Based Systematic Risk Analysis 1253

state into consideration. And then, we can get indi- are river and virescence areas to separate the indus-
vidual risk state in each cell and aggregated weighted trial and residential areas. Therefore, the model can
risk in a cluster of cells at time t + 1. represent a complex city with a cluster of potential
The algorithm is run synchronously for every cell hazardous sources and heavy population density. The
present in the lattice. The emerging scenario includ- city model is shown in Fig. 2.
ing domino effect in the subsequent time step is au- The fictitious city is divided into uniform grids
tomated. The algorithm is run repeatedly in order to of 35 × 35 m. As illustrated in the article, the pur-
generate the scenarios for all subsequent time steps. pose of the approach proposed here is for risk anal-
This is pursued until the number of time steps be- ysis during emergency response. According to Abra-
comes larger than the user-defined constant. Conse- hamsson,(31) the process of risk analysis has a lot of
quently, the algorithm is able to generate real-time uncertainty and the more detailed spreading simula-
individual risk and aggregated weighted risk at the tion might not play a greater role for the results of
end of every time step. risk.(31) Therefore, the size of cell needs not be too
small. Besides, in our city model, the attribute of each
3. APPROACH APPLICATION: AN cell is homogeneous when the size is 35 × 35 m. If the
ILLUSTRATIVE EXAMPLE size is larger, the attribute of some cells may not be
homogeneous. This will bring great trouble for the
3.1. Hypothetical Accident Scenarios
computations and may affect the precision of the re-
To illustrate the application of the systematic ap- sults. Taking the above analysis into consideration,
proach, we made simulations on a fictitious city based we think that the size (35 × 35 m) is appropriate in
on THU-CPSR city model, which was designed to our example.
simulate a real city with the scale of 5,250 × 4,410 m. L, M, and N (see Fig. 2) are potential hazardous
There are two commercial buildings, 10 residential sources that may cause pool fires. The vulnerable ar-
buildings, two workshops, three LNG gas contain- eas are A to J (see Fig. 2), which represent residential
ers, and a petrochemical industry in the city. There areas with heavy population. A hypothetical scenario

Fig. 2. THU-CPSR city model.


1254 Ji, Weng, and Fan

is used to illustrate the application of the approach. for nonadjacent neighbors in this example. The exact
A gasoline leak in one of the gasholders formed a liq- values need to be calibrated based on experiments.
uid pool at M and flashing of the liquid gasoline re- If a strong wind oriented from northeast to
sults in a pool fire. If the thermal intensity at L and N southwest were assumed, the amount of intensity
exceeds the threshold value (10 KW/m2 is used here), transported into the cell ij from the northeast neigh-
derivative pool fires may happen. bors would be more than that from its southwest
During a fire incident, the dominant form of en- neighbors compared to the conditions in the absence
ergy that causes maximum damage is the thermal of any wind currents. Transporting effect is affected
intensity, which propagates via conduction, convec- by wind in our case study. Here, we do not consider
tion, and radiation. Therefore, accident intensity ex- the effect of wind on pool fire. Instead, we consider
presses thermal intensity in this situation. For pool the effect of wind on thermal energy spreading. The
fires, there is an initial unsteady phase in which the effect of this factor depends on the distance between
burning rate accelerates, a steady state where equi- neighbor cell and “seed cell” and the direction of
librium is achieved, and finally a dying down phase wind. Thus, the magnitude of the transporting coeffi-
in which the accident intensity dissipates to zero. The cient may be expressed as:
amount of thermal intensity is influenced by factors
such as pool diameter, flame lip effects, the amount 1
of fuels, and so on. In most pool fires, it passes τij = uij , (16)
rn
very quickly from unsteady state to steady state.(32,33)
Therefore, only the steady state is considered here, where r is distance from neighbor cell to the “seed
and thermal intensity from pool fire is considered to cell,” n is an exponent, and u ij is the directional
be constant, i.e., 100 KW/m2 . constant associated with the specific direction. The
A constant heat value distribution in space numerical values assigned to these directional con-
would be expected from a pool fire having a constant stants are selected according to the importance and
thermal intensity, but the constant value cannot be degree of the directionality. In this example, we
achieved as soon as the fire reaches the stable state. It have assigned to the exponent value 1 for simplic-
needs time for each cell to achieve the constant state ity and we assume the direction of wind is northeast.
in the steady state. This time is vital for emergency So we have une > un = ue > 0, unw = use = 0, usw <
response. If efficient emergency rescue is performed uw = us < 0.
during this time, the loss of life and property may be Now, we turn to dissipative effect. Unlike in
reduced and a domino effect may also be avoided. the case of an absolute vacuum, the participating
In our approach application, the process will be dis- media may attenuate the released thermal inten-
cussed. sity due to absorption and scattering. Referring to
Section 2, an isotropic effect of barriers is assumed.
The dissipative coefficients used here obey the prin-
3.2. Methods
ciple that different objects have different dissipative
When heat flux passes from the source to the ability. The exact values need to be obtained from
receptor, significant attenuation of intensity occurs. experiments.
The atmospheric conditions are the main factor to According to F. P. Lees,(32) the mean value of
affect diffusivity. Therefore, the following expression the intensity fatal to normal population m is 7.775,
for diffusive parameter has been proposed:(34) the standard deviation σ is 0.3903. The individual
risk is computed with Equations (11) and (12). The
ij = 2.02 × (Pij × r )−0.09 , (14) aggregated weighted risk in the residential areas A
where: to J were mainly analyzed. The aggregated weighted
  risk can be calculated using Equation (13). Table I
5328
Pij = 1013.255 × (RHij ) × exp 14.4144 − presents the information about the population den-
Tij
sity, and the total number of the population of each
(15)
residential area. We regard population density as
and r is the length scale of the cells, T ij the ambient constant in the example for simplicity. The combina-
temperature in K at cell ij, and RH ij is the relative- tion of site monitoring and a GIS database can be
humidity in cell ij. The weight term for the adjacent used to obtain the real-time data of population den-
neighbors has been taken as ω a = 0.88 and ω b = 0.12 sity in practical application.
Cellular Automata-Based Systematic Risk Analysis 1255

Population Population
Residential Density Total Number Residential Density Total Number
Area No. (no./km2 ) of Population Area No. (no./km2 ) of Population
Table I. Population Density and the
Total Number of Population in Each A 31,020 4,484 F 17,143 1,323
Residential Area B 29,388 4,132 G 4,082 1,323
C 31,020 3,648 H 14,694 1,188
D 31,020 4,750 I 14,694 1,512
E 33,469 5,248 J 14,694 1,584

3.3. Results and Discussions be obtained at N, and the two values are 8.84 and
6.61 KW/m2 . Due to the impact from the intercepting
The results under homogeneous conditions and
objects, time to cause derivative accidents is also dif-
heterogeneous conditions were computed to illus-
ferent and the existence of intercepting objects can
trate the CA method’s advantage in treating hetero-
delay the time. Under homogeneous conditions, L
geneous conditions. Here, homogeneous conditions
and N can cause derivative accidents at t = 1005s and
mean that there exist no intercepting objects in all
t = 697s and the results are 1425 and 975s under het-
cells (α i j = 0). Heterogeneous conditions arise when
erogeneous conditions.
there are intercepting objects in some cells and the
The contours of thermal intensity were also com-
conditions are different in different cells. Domino ef-
puted (Fig. 4). The results under homogeneous con-
fect under the two conditions was also analyzed.
ditions are regular curves, while the impact of hetero-
Fig. 3 gives the thermal energy varying with time
geneous conditions on the results can be seen, which
at potential hazardous sources L and N. A significant
makes the curves not smooth and is closer to real-
difference of thermal intensity value can be observed
ity. One can also conclude that designing isolated
under the two conditions. The thermal intensity un-
areas to separate an industry area and a residential
der homogeneous conditions is higher than that un-
area, and different industry areas is advantageous to
der heterogeneous conditions at the same time. For
reduce accident impact and prevent domino effect.
example, at t = 600s, the thermal intensity under ho-
Comparing the two groups of figures, one can see that
mogeneous conditions is 5.17 KW/m2 and its coun-
the existence of river and virescence areas greatly re-
terpart is 3.32 KW/m2 at L. A similar conclusion can
duces the thermal intensity. For example, in the area
G thermal intensity has been more than 10 KW/m2
without intercepting objects, but under heteroge-
neous conditions the value is less than 10 KW/m2 at
t = 800s (Figs. 4(C) and 4(D)). At t = 800s, thermal
intensity at N is over 10 KW/m2 and a secondary ac-
cident has happened under homogeneous conditions
(Fig. 4(D)). Due to the absorption of buildings, the
intensity is less than 10 KW/m2 at L and no secondary
accident happened at t = 800s (Fig. 4(C)). The po-
tential hazardous source L received nearly 10 KW/m2
thermal intensity at 1000s (Fig. 4(F)), and the tertiary
accident will occur soon, while the counterpart inten-
sity is only about 5 KW/m2 (Fig. 4(E)), which can-
not cause any derivative accident. Besides, according
to the results, one can determine emergency rescue
time. At t = 975s, the secondary accident occurs at N.
So if emergency rescue power is sent to N during this
time, a secondary accident may be prevented. If not,
at least one can delay the time of the secondary hap-
Fig. 3. Domino effect analysis. Derivative accident occurs when
pening to obtain more time for people evacuation.
the thermal intensity at potential hazardous sources (L and N) is Individual risk is analyzed and the simulation re-
not less than 10 KW/m2 . sults of individual risk contours are discussed (Fig. 5).
1256 Ji, Weng, and Fan

Fig. 4. Contours of thermal intensity


under two conditions at different times.
At t = 0, the accident happened at M.
Figs. A, C, E, and G represents the
results under heterogeneous conditions
and Figs. B, D, F, and H represents the
results under homogeneous conditions.

From contours of individual risk, one can obtain the ample, at t = 800s (Fig. 5(B)), although the thermal
real-time information about the distribution of in- intensity in most of the residential areas is more than
dividual risk. As expected, for an emergency man- 1 KW/m2 , the individual risk is much less than 1 ×
ager, individual risk is much more practical than ac- 10−6 . Therefore, the manager can easily judge that at
cident intensity during emergency response. For ex- this time most of the residential areas are safe (we
Cellular Automata-Based Systematic Risk Analysis 1257

Fig. 5. Contours of individual risk under


heterogeneous conditions at different
times. At t = 0, the accident happened at
M.

Table II. Aggregated Weighted Risk in the 10 Residential Areas at Different Times

600s 800s 1200s 1600s 2000s 2600s

A 0.00000000 0.00000000 0.00000000 0.00000000 0.00000000 0.05801328


B 0.00000000 0.00000000 0.00000000 0.00000000 0.03297678 70.69564293
C 0.00000000 0.00000000 0.00000000 0.00000000 0.00007334 0.45263844
D 0.00000000 0.00000000 0.00000102 0.00110798 0.20017541 34.11987993
E 0.00000238 0.00072004 0.18783947 4.60474650 98.56040847 1126.08295703
F 0.00000000 0.00000000 0.00008104 0.01595472 0.70950394 27.93128237
G 0.00462710 0.07861287 1.46759059 9.16648734 35.34787203 115.56288804
H 0.00000000 0.00000000 0.00000000 0.00007405 0.01880954 3.57464209
I 0.00000000 0.00000000 0.00005910 0.02229424 1.22450960 46.84007340
J 0.00000000 0.00001079 0.01515049 1.31135140 24.86859071 316.95609140
1258 Ji, Weng, and Fan

think that individual risk less than 1×10−6 is safe). proposed approach for risk analysis and emergency
At t = 1200s (Fig. 5(C)), the individual risk around response has been illustrated with an example. The
building G is more than 1 × 10−6 , and the area is not results demonstrate that the CA method can easily
safe anymore. With time going by, the unsafe area treat heterogeneous conditions, and give an irregular
expands gradually. At t = 2600s (Fig. 5(D)), almost shape of the area of impact of an accident, which is
all residential buildings are in danger except building much more realistic assessment for risk analysis. Be-
A. Besides, according to the contours of individual sides, according to the results, we conclude that indi-
risk, emergency managers can issue prewarning in- vidual risk is useful to plan evacuation strategies and
formation to guide the people to evacuate. If one ne- aggregated weighted risk can help emergency man-
glects tertiary accidents, it is a good strategy to guide agers to allocate rescue resources rationally and op-
people to evacuate to the northwest area. However, a timize emergency response programs.
high risk source L exists in the north and it will cause In reality, before the approach is brought to bear
an accident at t = 1425s. Comprehensively consider- on any complex problems, there are three issues mer-
ing these factors, a west side evacuation is strongly iting special attention. First and foremost, it must
recommended (Figs. 5(E) and 5(F)). be emphasized that cellular automaton is a compu-
Aggregated weighted risks in the 10 residential tational tool, whose effectiveness depends upon its
areas were also computed at different time. The re- accurate calibration on the basis of data from exper-
sults are given in Table II. Some significant conclu- imental results as well as those from past accident
sions can be obtained to help emergency managers histories. Research in these areas has been continu-
allocate emergency rescue resources. For example, ing across the world but needs greater impetus. Be-
the AWR is very low in all the 10 areas at t = 600s and sides, discussion of the size of the cells is also needed
t = 800s. At t = 1200s and t = 1600s, the AWR at area according to real problems in the future research.
E, G, and J is much higher than that at other areas, Secondly, the process of emergency response is very
especially area G where the AWR is the highest. At complicated, and some other social factors may also
t = 2000s, the area E is the most dangerous area. Af- affect the results of risk assessment, such as impor-
ter t = 2000s, nearly all areas have high AWR. Based tant facility density, transportation system density,
on above analysis, the following resources allocation rescue power distribution, and so on. The detailed
strategy is advised: (1) at the first half an hour of discussion of those factors should be conducted in
emergency response, more rescue resources should future research. Last but not least, the merit of the
be sent to E, G, and J; (2) after that, D, F, H, and I approach should be judged in real-life scenarios, al-
gradually need more resources. though it is hard. It is a good way to validate and
improve the approach in future emergency response
drills. The data gathered from real cases are also
meaningful for this purpose.
4. CONCLUSIONS
Emergency response is applying science, tech-
nology, planning, and management to deal with ex- REFERENCES
treme events. Developing a meaningful model ca-
1. Ford, J. K., & Schmidt, A. M. (2000). Emergency response
pable of conducting emergency response is an ar- training: Strategies for enhancing real-world performance.
duous task that requires the contributions of many Journal of Hazardous Materials, 75, 195–215.
individuals from several disciplines. The cellular 2. Haimes, Y. Y. (1999). The role of the society for risk analysis in
the emerging threats to critical infrastructures. Risk Analysis,
automata-based systematic risk analysis approach for 19(2), 153–157.
emergency response presented here is a preliminary 3. Wulf, W. A., Haimes, Y. Y., & Longastaff, T. A. (2003). Strate-
attempt to contribute to the gigantic efforts. gic alternative responses to risks of terrorism. Risk Analysis,
23(3), 429–444.
In this article, three cellular automata transition 4. Lindell, M. K., Prater, C., & Perry, R. W. (2007). Introduction
rules, which work through diffusion, transport, and to Emergency Management. New York, Wiley Press.
dissipation mechanism, are proposed. The approach 5. Dombroski, M., Fischhoff, B., & Fischbeck, P. (2006). Predict-
ing emergency evacuation and sheltering behavior: A struc-
can take multiple factors into account such as physi- tured analytical approach. Risk Analysis, 26(6), 1675–1688.
cal and social characteristics of the site and can also 6. Kaplan, S., & Garrick, J. (1981). On the quantitative definition
handle domino accidents. As an outcome, two risk of risk. Risk Analysis, 1(1), 11–27.
7. Haimes, Y. Y. (2006). On the definition of vulnerabilities in
indices are proposed to assist emergency managers measuring risks to infrastructures. Risk Analysis, 26(2), 293–
to make decisions. The potential application of the 296.
Cellular Automata-Based Systematic Risk Analysis 1259

8. Khan, F. I., & Abbasi, S. A. (2002). A criterion for developing 21. Delvosalle, C. (1998). A Methodology for the Identification and
credible accident scenarios for risk assessment. Journal of Loss Evaluation of Domino Effects. Report CRC/MT/003, Brussels,
Prevention in the Process Industry, 15, 467–475. Belgium, Belgian Ministry of Employment and Labor.
9. Khan, F. I., & Abbasi, S. A. (2000). Analytical simulation and 22. Bagster, D. F., & Pitblado, R. M. (1991). The estimation of
PROFAT II: A new methodology and computer automated domino incident frequencies: An approach. Process Safety En-
tool for fault tree analysis in chemical process industries. Jour- vironment, 69, 195–199.
nal of Hazardous Materials, A75, 1–27. 23. Khan, F. I., & Abbasi, S. A. (1998). Models for domino ef-
10. Arystanbekova, N. K. (2004). Application of Gaussian plume fect analysis in chemical process industries. Process Safety
models for air pollution simulation at instantaneous emissions. Progress, 17(2), 107–123.
Mathematics and Computers in Simulation, 67, 451–458 24. Cozzani, V., & Salzano, E. (2004). The quantitative assessment
11. Gong, Y. W., Lin, W. S., Gu, A. Z., & Lu, X. S. (2004). A sim- of domino effects caused by overpressure. Part I. Probit mod-
plified model to predict the thermal response of PLG and its els. Journal of Hazardous Materials, 107(3), 67–80.
influence on BLEVE. Journal of Hazardous Materials, A108, 25. Papazoglou, I. A., & Aneziris, O. N. (1999). Uncertainty quan-
21–26. tification in the health consequences of the boiling liquid ex-
12. Hankin, R. K. S., & Britter, R. E. (1999). TWODEE: The panding vapor explosion phenomenon. Journal of Hazardous
Health and Safety Laboratory’s shallow layer model for dense Materials, A67, 217–235.
gas dispersion. Part 1. Mathematical basis and physical as- 26. Committee for the Prevention of Disasters. (1989). Methods
sumptions. Journal of Hazardous Materials, 66(3), 211–226. for the Calculation of Possible Damage (Green Book). The
13. Chen, H.- J., & Lin, M.- H. (1999). Modeling a boiling-liquid, Hague, The Netherlands: Committee for the Prevention of
expanding-vapor explosion phenomenon with application to Disasters.
relief device design for liquefied ammonia storage. Industrial 27. Uijt de Haag, P. A. M., & Ale, B. J. M. (1999). Guide-
& Engineering Chemistry Research, 38, 479–487. lines for Quantitative Risk Assessment (Purple Book). The
14. Hankin, R. K. S. (2003). Shallow layer simulation of heavy Hague, The Netherlands: Committee for the Prevention of
gas released on a slope in calm ambient. Part I. Contin- Disasters.
uous releases. Journal of Hazardous Materials, A103, 205– 28. Christou, M. D., & Mattarelli, M. (2000). Land-use planning in
215. the vicinity of chemical sites: Risk-informed decision making
15. Khan, F. I., & Abbasi, S. A. (2001). An assessment of the likeli- at a local community level. Journal of Hazardous Materials,
hood of occurrence, and the damage potential of domino effect 78, 191–222.
(chain of accidents) in a typical cluster of industries. Journal of 29. Cozzani, G., Gubinelli, G., Antonioni, G., Spadoni, G., &
Loss Prevention in the Process Industries, 14, 283–306. Zanell, S. (2005). The assessment of risk caused by domino
16. Von Neumann, J. (1966). In A. Burks (Ed.), Theory of Self- effect in quantitative area risk analysis. Journal of Hazardous
Reproducing Automata. Urbana, IL: University of Illinois Materials, A127, 14–30.
Press. 30. Jonkmana, S. N., van Gelder, P. H. A. J. M., & Vrijling, J. K.
17. Ulam, S. (1952). Random processes and transformations. Pro- (2003). An overview of quantitative risk measures for loss of
ceedings of the International Congress on Mathematics, 2, 264– life and economic damage. Journal of Hazardous Materials,
275. A99, 1–30.
18. Sarkar, C., & Abbasi, S. A. (2006). Cellular automata-based 31. Abrahamsson, M. (2002). Uncertainty in quantitative risk
forecasting of the impact of accidental fire and toxic dispersion analysis: Characterization and methods of treatment. Re-
in process industries. Journal of Hazardous Materials, A137, search report 1024, Lund.
8–30. 32. Lees, F. P. (1996). Loss Prevention in the Process Industries,
19. Gledhill, J., & Lines, I. (1998). Development of Methods to 2nd ed. Oxford, UK: Butterwort-Heinemann.
Assess the Significance of Domino Effects from Major Haz- 33. Centre for Chemical Process Safety (CCPS) of the American
ard Sites. CR Report 183. London, UK: Health and Safety Institute of Chemical Engineers. (2000). Guidelines for Chem-
Executive. ical Process Quantitative Risk Analysis, 2nd ed. New York:
20. Contini, S., Boy, S., Atkinson, M., Atkinson, M., Labath, M., AiChE.
Banca, M., et al. (1996). Domino effect evaluation of major 34. Pietersen, C. M., & Huerta, S. C. (1985). Analysis of the LPG
industrial installations: A computer aided methodological ap- Incident in San Juan Ixhuatepec, Mexico City, 19 Nov 1984.
proach. In Proceedings of the European Seminar on Domino TNO84-0222. Apeldoorn, The Netherlands: Netherlands Or-
Effects. Leuven, Belgium. ganization for Applied Scientific Research.

Potrebbero piacerti anche