You are on page 1of 16

Transportation Research Part D 31 (2014) 2136

Contents lists available at ScienceDirect

Transportation Research Part D


journal homepage: www.elsevier.com/locate/trd

A system dynamics modeling for urban air pollution: A case


study of Tehran, Iran
Hamed Vafa-Arani, Salman Jahani, Hossein Dashti, Jafar Heydari , Saeed Moazen
School of Industrial Engineering, College of Engineering, University of Tehran, Iran

a r t i c l e

i n f o

Keywords:
Urban air pollution
System dynamics modeling
Trafc
Public transportation
Energy fuel and automotive industries

a b s t r a c t
Development of strategies to control urban air pollution is a complex process involving a
wide range of sciences. In this study a system dynamics model is proposed in order to estimate the behavior of parameters affecting air pollution in Tehran. The proposed model
includes two subsystems: (1) urban transportation, (2) air polluting industries. In this
paper, several policies are proposed to mitigate air pollution. The proposed model is simulated under several scenarios using historical data of transportation and industrial sectors in
Tehran. Policies are categorized as: (1) road construction, (2) technology improvement in
fuel and automotive industries, (3) trafc control plans, (4) development of public transportation infrastructures. The results show effectiveness of the proposed policies. In this case,
technology improvement in fuel and automotive industries and development of public
transportation infrastructures are more effective policies in order to reduce air pollution.
2014 Elsevier Ltd. All rights reserved.

Introduction
Rapid urbanization may cause different environmental pollutions, such as air pollution, acid rain, water pollution and
land pollution, solid wastes, toxic wastes and deforestation. In the metropolitans of developing countries especially in Asia,
the environmental problems are much greater, because of the overwhelming scale and speed of urbanization (Atash, 2007).
Air pollution has become an undeniable problem of most countries. Reducing urban air pollution has been on the agenda of
policy makers for several decades, especially in United States and Europe. A long list of consecutive initiatives, often of an
international nature, has been taken to reduce emissions of key pollutants (Beckx et al., 2009). But this trend is reverse in
developing countries, especially in Asian countries. Remarkable negative effects of air pollution on public health, environment and economic, can oblige the decision makers to plan for controlling and eliminating the main sources of pollution
in these countries. Since the air pollution can affect several facets of urban human life, it can be affected from several
man-made systems including public and individual systems, industrial plants and household which are the main sources
of urban air pollution. Transportation, industries and houses cannot be considered as some independent systems in urban
areas. These systems have several interactions with each other while all of them are affected by human behavior, so air
pollution is a socio-economic problem in which both facets of it should be analyzed. Thus, all of these systems should be
considered as subsystems of a greater one, which includes all the systems within a city as well as other exogenous systems
inuencing the urban system. In addition to interactions between different subsystems, feedback structure should be taken
into account which implies dynamic loops. System dynamics modeling is one of the best tools to model a socio-economic

Corresponding author. Tel.: +98 21 82084489.


E-mail address: J.heydari@ut.ac.ir (J. Heydari).
http://dx.doi.org/10.1016/j.trd.2014.05.016
1361-9209/ 2014 Elsevier Ltd. All rights reserved.

22

H. Vafa-Arani et al. / Transportation Research Part D 31 (2014) 2136

problem with mentioned characteristics. System dynamics can model feedback structure as well as interactions between
several variables in a system by means of simulation which can demonstrate the future effects of policies in different variables and subsystems. Socio-economic systems- like the one considered in this study-are complex and their behavior cannot
be estimated through analytical methods. In such systems, a simulation approach could be utilized to explore behavior of the
system. Among several simulation methods, systems dynamics approach is an appropriate and effective one for socioeconomic systems.
Tehran Province has a population of around 12.4 million and surpassing 8.3 million in the wider metropolitan area
according to 2012 census (Statistical Centre of Iran, 2012). Ineffective public transportation infrastructures, non-standard
streets and highways, population congestion, non-efcient urban management and planning, and several other reasons
cause high trafc congestion in the metropolitan area of Tehran.
Inversion can lead to pollution such as smog being trapped close to the ground. In recent years, inversion has caused several closures of schools and government centers during winters in Tehran which lead to serious economic losses, environmental damages and mortality increase. Accordingly, in this paper urban transportation subsystem is considered as the
main subsystem which can affect other subsystems and be affected by them. Two major types of trips in urban areas are
public and individual trips. The portion of trips by personal automobiles can show the situation of transportation system
of the city. High level of public trips shows the strength of public transportation infrastructure and public transportation
attractiveness for people. On the other hand, high level of individual trips in a specic city shows high driving attractiveness
which originates from several factors such as low travel time by individual automobiles with respect to public transport,
expensive public transportation fare, cheap automobiles and cheap fuel for automobiles. In order to reduce the trafc congestion which is the most important source of air pollution, several policies can be made. Road construction, trafc congestion control plans, and investment on public transportation system are practical ones in Tehran. The effects of mentioned
policies on the urban air pollution are investigated in the proposed system dynamics model.
Majority of industrial countries are faced with pollutant industries more than the past. Especially, in developing countries
such as China, industrial air pollution is not only a source of urban air pollution, but also is a global problem which causes
global warming and Ozone layer destruction. In the current study, boundary of the model does not include this later issue,
thus it is considered that urban air pollution is caused by suburban factories. Two major groups of polluting industries
around Tehran are automotive and fuel production industries. In addition, these two industries are the most related to
the urban transportation systems. Therefore, industrial pollution resulted from automotive and fuel production industries
form the rst part of industrial subsystem.
Energy is one of the most important resources of the mentioned industries. Fuel and automotive production industries are
two of the most energy consumers in Iran. Meanwhile, required energy for these industries is generated by power plants
around them. In other words, power plants around Tehran should generate required energy for the industries. In urban areas,
household consumers are the major consumers of energy. Thus, two major consumers of energy are industries and households. As the power plants cause major pollution, they are considered as the second part of industrial subsystem.
In this paper a system dynamics model is proposed to cope with complexities of urban air pollution problem which
includes two major subsystems: (1) transportation subsystem, and (2) industrial subsystem. Subsystems and variables
are consistent with case of Tehran. It is rst time which a system dynamics model is developed to investigate the air
pollution problem of Tehran from urban management perspective by considering interactions of two aforementioned subsystems. Remarkable contributions of the model could be classied to two major groups. First, although, usually trafc congestion is considered as the main factor of urban air pollution, pollutant industries should be taken into account as another
source of urban air pollution, especially in industrial cities such as Tehran. In this study, industrial pollution is considered as
one of the main causes of urban air pollution. In addition, fuel and automotive industries as the most effective industries on
the air pollution associated with automobiles. Automobile and fuel quality are variables which are related to industrial and
transportation subsystems. Second contribution of this paper is about several novel applicable policies for mitigating the air
pollution. Road construction, technology development in fuel and automotive industries, applying trafc control plans and
developing public transportation infrastructures are four proposed policies which are investigated.

Literature review
Urban air pollution has become one the most critical problems in several metropolitans over the world. Previous studies
investigated the urban air pollution problem with several perspectives using different approaches and tools. The current
study is not based on technical and chemical aspects of the urban air pollution. But, strategic decision making in urban
air pollution problem is investigate. All of studies in the eld of urban air pollution could be categorized into two major
groups. First, studies that investigate the air pollution through analytical modeling methods such as operations research.
Second group includes studies which utilize system dynamics approach to model the air pollution and not necessarily urban
air pollution. The body of literature in air pollution modeling is reviewed here according to aforementioned classication.
In the rst class, Sivacoumar et al. (2001) utilized a mathematical programming approach to predict the air pollution of
Jamshedpur region of India. They indicated the portion of domestic, industrial and automobiles from NOx pollution which the
results show that the contribution of each part was 53%, 7% and 40%, respectively. Ulke et al. (2001) investigated the

H. Vafa-Arani et al. / Transportation Research Part D 31 (2014) 2136

23

production of urban air pollution that was studied at metropolitan area of Sao Paulo, Brazil. Another estimation of urban air
pollution was performed by Lyons)2003). He illustrated a strong relationship between vehicle kilometers of travel and
urbanized land area to estimate the vehicular emissions. These emission rates were incorporated with simple meteorological
forcing into a box model to form a generalized urban air pollution model for any city all over the world. Jensen et al. (2008)
evaluated different air quality models which estimate the exposure factors and their application in order to assess the external pollution costs of transportation. In this regard, they used an impact pathway methodology. A quantifying approach for
environmental costs of air pollution is developed by Yoo et al. (2008). They attempted to perform a choice experiment in
order to quantify environmental costs of four air pollution effects including mortality, morbidity, soiling damage, and poor
visibility. They used the proposed approach in a case of Seoul. Vlachokostas et al. (2009) used a hybrid mathematical
programming-air quality modeling approach to evaluate the air pollution of Thessaloniki, Greece. They utilized both costeffectiveness and cost-benet analysis to put the problem on the basis of economic efciency from a societal perspective.
An activity-based modeling approach was performed by Beckx et al. (2009) to evaluate the exposure to air pollution. They
developed a new model using micro-simulator of activity-travel behavior for dynamic exposure assessment. Results showed
a remarkable underestimation of static models exposure in comparison to their proposed study. A two-stage joint-probabilistic programming was developed by Lv et al. (2011) for planning a regional air quality management system. They modeled
the system with multiple pollutants and multiple sources via a mathematical programming including two-stage stochastic
programming, probabilistic constraint programming and interval mathematical programming. Dong et al. (2012) proposed a
hybrid inexact optimization approach for supporting energy systems planning and air pollution mitigation under uncertainty. They combined interval-parameter programming, mixed-integer programming and chance-constraint programming
to deal with systematic complexities associated with energy utilization and air pollution mitigation. Strauss et al. (2012)
investigated relation between cyclist volumes and air pollution along bicycles. They focused on intersections and several corridors which bicycles use and measure air pollution level in them. Results showed that intersection and corridors with bicycles have greater air pollution level. Shang et al. (2013) investigated the relationship between short-term exposure to air
pollution and daily mortality. They evaluated several pollutant gasses as the main reasons of daily mortality associated with
exposure to air pollution. Effects of speed limits for vehicles were investigated by Bel and Rosell (2013) in metropolitan area
of Barcelona. They evaluated effects of different speed limit policies including 80 km/h speed limit and variable speed limit
on the urban air pollution via a differences-in-differences method. Results showed that the variable speed limit is a more
effective policy than 80 km/h speed limit for air pollution reduction. Hong and Shen (2013) examines the effect of residential
density on CO2 equivalent from automobile using more specic emission factors based on vehicle and trip characteristics,
and by addressing problems of spatial autocorrelation and self-selection. The results indicate that the effect of residential
density on transportation emissions is inuenced by spatial correlation and self-selection.
Studies which tried to investigate the air pollutions via system dynamics approach are the second body of literature in
this eld. System dynamics allows the researchers to combine different facets of socio-economic problems from several
points of view. Lu and Turco (1997) and Schmidt and Schafer (1998) are the rst studies which utilized system dynamics
approach for air pollution from a technical point of view. Schmidt and Schafer (1998) evaluated the transport induced emission which was approximately 70% of all emissions. They simulated the emissions and air pollutions in urban areas via simulation software. Also a scenario based analysis was performed on urban management policies. ANH (2003) considered
trafc congestion as the main result of urbanization and the main reason for under-standard and overused infrastructures,
environmental pollutions and trafc problems through a system dynamics approach. They proposed public transportation
system development, road network expanding and alternative vehicle enhancing as policies. Another research on trafc congestion modeling via system dynamics was performed by Wang et al. (2008). They incorporated seven subsystems for the
trafc congestion model including population, economic development, number of vehicles, environmental inuence, travel
demand, transport supply, and trafc congestion. They presented a solution to decision makers of Dalian, China which was
the restriction of total number of vehicles. Anand et al. (2006) developed a system dynamics model to evaluate the CO2 emissions from cement industries in India. Three scenarios were incorporated to the system including population growth stabilization, energy conservation and structural management in cement manufacturing processes. Results show that applying the
mentioned scenarios can reduce the air pollution from cement industries by 42%. Another study on air pollution of cement
industries was performed by Ansari and Sei (2013). They modeled energy consumption and CO2 emission in cement industries of Iran. Results showed that after removal of energy subsidy, energy consumption and CO2 emission in Iranian cement
industries will reduce signicantly. A comprehensive study on air pollution was performed by Wong Wing Ying (2007) via a
system dynamics model. He classied the air pollution into street level and regional. He also investigated the health effects
of the air pollution by two health indices. Szarka et al. (2008) developed a system dynamics model for evaluation of effects of
biomass substitution on air pollution of AustrianHungarian cross-border area with economic, social and environmental
subsystems. Meanwhile, they simulated the model for seven scenarios including different types of biomasses and share of
them in fuel market. Shahgholian and Hajihosseini (2009) proposed a system dynamics model to evaluate the effects of
air pollution in Tehran in future decade. They assumed just the population changes as the main source of air pollution; rather
they investigated the effects of urban air pollution on health and populations. Another study which considered trafc congestion dynamics and emissions from it was performed by Armah et al. (2010). They simulated the proposed system dynamics model for a long planning horizon on a case of Accra, Ghana. Motorcycles as one of the main sources of air pollution
should be investigated via a system dynamics approach. Trappey et al. (2012) applied a system dynamics approach to green

24

H. Vafa-Arani et al. / Transportation Research Part D 31 (2014) 2136

transportation policy evaluation in Taiwan in terms of air pollution. Gasoline motorcycles and electric scooters were the
competitors in their model. CO2 emission and energy consumption were the most effective factors in competition between
gasoline motorcycles and electric scooters. Finally, behavior of CO2 emission was simulated for a long horizon and effects of
several scenarios based on two types of policies were investigated. Yu and Wei (2012) proposed a hybrid system dynamicsgenetic algorithm for evaluation of coal demand, production and environmental pollution in China. They simulated the proposed system dynamics model for several scenarios for pollution mitigation, but the results showed pollution increasing,
despite the coal intensity decreasing. In their study, the genetic algorithm was utilized for parameter calibration. Feng
et al. (2013) accomplished a system dynamics modeling for urban energy consumption and CO2 emission in case of Beijing,
China. They proposed residential, transport, agricultural, industrial and service energy consumption as the main subsystems
of the urban energy consumption system. Results showed that the service sector will be one of the main energy consumers in
the future.
The body of literature indicates shortage of strong works on the urban air pollution from a systematic point of view. As
the literature shows, air pollution of Tehran as one of the metropolitan areas in Asia has been rarely investigated via a macro
view. Previous studies did not consider the interactions between transportation and industrial subsystems while the origin
of vehicular air pollution is due to industries, especially automotive and fuel industries. Major contribution of the current
study is proposing a new approach towards urban air pollution with interaction between subsystems including transportation and industrial plants. Thus, remarkable novelties of this paper are summarized as follows: rst, interaction between
industrial subsystem and transportation subsystem are considered which has not been considered in previous studies.
Second, the system dynamics as a powerful modeling approach is used to cover more aspects of the urban air pollution problem in comparison with previous studies. Besides, in this study, four novel air pollution mitigation policies are taken into
account: (1) road construction, (2) technology improvement in fuel and automotive industries, (3) trafc control plans,
(4) development of public transportation infrastructures. The mentioned policies are simulated over a planning horizon
and the most effective policy is indicated.
Methodology
The system dynamics was introduced by Forrester from MIT in 1950s. The approach is based on the feedback control theory, equipped with computer simulation technology, and used in quantitative researches of complicated socioeconomics
eld. The methods of system dynamics are realized by involving feedback loops, variables, and equations. The feedback loop
is dened as a closed chain of causes and effects. The variables include (i) stock (level) variable, the one that accumulates a
ow over continuous time periods; (ii) rate variable, the one that represents a ow during a time period; (iii) auxiliary variable, the one that identies rate variables. The three kinds of variables are linked by equations taking the form of integral,
differential, or other types. Because urban transportation system is much complex and covers a wide range, it is not suitable
to use the traditional approaches to simulate and analyze. Therefore, the system dynamics approach has been applied in
complex system analysis (Wang et al., 2008).
In this study a system dynamics based computer simulation model for estimation of air pollution in Tehran metropolitan
is developed. The boundary of the model is the total administrative area of Tehran city and the considered time period for
dynamic modeling is from 2012 to 2030. In light of the research purpose and data availability, the model is comprised of
three subsystems including industrial, transportation and household pollution. In order to limit our model boundary, household pollution subsystem is considered as an exogenous variable since its effects are uncontrollable. Hence, two main subsystems are industrial and transportation pollution. Air pollution as our main stock variable in this study is affected directly
by variables from these subsystems.
It is worth mentioning that describing the dynamics of a socio-economic system with a single empirical equation is
impossible. Accordingly, a set of equations is dened and the relationships between the input and output variables are
extracted. Considering these equations simultaneously helps us perceive the whole dynamics of the system.
Fig. 1 illustrates the subsystems diagram which addresses the main variables of each subsystem and relations between
subsystems. Industrial pollution and transportation subsystem are considered to be two effective subsystems for the Tehran
air pollution model. The subsystem diagram gives a general sense to the decision makers and stock-ow diagram gives more
details about the interrelations between the main variables within each subsystem. Air pollution associated with energy production is considered as a primary part of the industrial pollution subsystem. It addresses the air pollution which is caused
by power plants in suburban area of Tehran. In addition to the energy consumption by industries, household energy consumption has a remarkable share on energy demand in urban areas. Thus, the household energy consumption is taken into
account as an exogenous variable because the dynamics of household energy consumption is not controllable from this study
perspective. Tehran air pollution is assumed to be the sum of direct industrial pollution and energy production pollution as
industrial pollution subsystem and air pollution caused by transportation subsystem.
Industrial pollution subsystem
Like other metropolitan areas around the world, Tehran is located among its major industries. Recent data shows that
energy consumed by the industrial sector located in the suburban area of Tehran is increasing dramatically over time.

H. Vafa-Arani et al. / Transportation Research Part D 31 (2014) 2136

Industrial subsystem

Transportation subsystem

Fuel Production Capacity

Driving Attractiveness

Automobile Production
Capacity

Road Capapcity

Automobile Industries
Technology Improvement

Traffic Congestion

Fuel Industries Technology


Improvement

25

Traffic Congestion Control


Plans

National Income
Travel Time
Fuel Price

Household
Energy
Consumption

Automobile Price

Investment on Public
Transportation Infrastructures

Fuel Quality

Public Transit Fare

Automobile Quality

Air Pollution
of Energy
Production

Industrial
Pollution

Automobiles Pollution

Air Pollution

Fig. 1. Subsystems diagram.

Industrial pollution subsystem demonstrates the effects of suburban industries on air pollution of Tehran. In this study,
suburban industries consist of fuel industries and automotive industries which seem to have more tangible role in Tehran air
pollution. One output of this subsystem is the industrial energy consumption and consequently pollution resulting from consumption of energy which the latter variable directly affects the air quality of Tehran. In fact, suburban industries require
energy to build added value. In this study, this required energy is assumed to be electricity. Another effective variable of this
subsystem is industrial pollution itself. Each suburban industry consumes energy which in turn leads to air pollution. The
variables and their interrelations are displayed in Fig. 2.
As output of this subsystem, industrial pollution is affected by the capacity of automotive industries and the capacity of
fuel industries. These capacities have been considered as the only stock variables in the industrial pollution subsystem. We
considered an auxiliary variable called percent of investment in industrial capacity (PII), which indicates the percentage of
investment in industries to develop their capacity. Investment in automotive industries (IAI) and investment in fuel industries (IFI) are two variables which are affected directly by the national income (NI). Using historical data, it is estimated that
IAI and IFI are two constant fractions of the national income. PII together with IAI indicate the growth rate of automotive
industries capacity. Similarly, PII and IFI determine the growth rate of fuel industries capacity. A fraction of investment
in automotive industries is called PIT which indicates the fraction of IAI to develop technology in automotive industries.
PIT along with IAI determine the investment in technology development of automotive industries (AITI). Similarly IFI and
PIT determine the investment in fuel industries (FITI). An increase in AITI leads to corresponding improvement in the quality
of automobiles (AQ). The result follows for FITI and fuel quality (FQ), similarly. As the investment in technology development
of industries increases, industrial pollution intensity decreases. To show this dynamics we take advantage of two variables
namely Beta and Alpha. Beta indicates the amount of pollution in ton per manufactured automobile and Alpha indicates the
amount of pollution in ton per liter of produced fuel. Regression analysis is performed on data from 2000 to 2012 for industrial pollution intensity to obtain these variables. The results of regression analysis are as follows:

p
Beta k  exp  AITI  q

p
0
Alpha k  exp  FITI  q0

where k, k , q and q are constants. The main variable of this subsystem, industrial pollution (IP) is obtained by:

IP AIC  Beta FIC  Alpha

26

H. Vafa-Arani et al. / Transportation Research Part D 31 (2014) 2136


PEP

PIEP

HEC
IEC
<AIC>

PII
FIC
FIGR

IP
Beta
Alpha

IFI

AIC
AIGR

AITI

<NI>

IAI

FQ
FITI

AQ
PIT
<NI>

<PII>

Fig. 2. Industrial pollution subsystem.

Energy consumed in automotive and fuel industries determines the total amount of industrial energy consumption (IEC).
The energy consumption intensities were estimated according to historical trends. As mentioned above, household energy
consumption is considered as an exogenous variable. Another exogenous variable named pollution intensity of energy
production (PIEP) along with HEC and IEC indicate pollution caused by energy production (PEP) as follows:

PEP PPEP  HEC IEC

Denition of rate variables as the main equations of the model is as follows for fuel industries capacity and automotive
industries capacity:

FIGR PII  IFI

AIGR PII  IAI

where IFI and IAI are total investment in fuel and automotive industries, respectively. PII is a fraction of investments that is
spent for development of production capacity in fuel and automotive industries which is affected by air pollution and air
pollution reduction policies.
Transportation subsystem
Transportation is a major source of air pollution in urban areas. The rapid growth of population in Tehran resulted in
severe trafc congestion and air pollution problems. Besides, highest percentage of air pollution comes from private automobile emissions.
In this study, transportation subsystem attempts to demonstrate the effects of private automobiles on Tehran air pollution.
Hence, main output of this subsystem is private automobile emission. This subsystem is divided into two sectors including
public transportation and individual trafc. Driving attractiveness variable (DA) plays an important role in this subsystem.
Because of weak public transportation in Tehran, personal automobiles are preferred by people even with high trafc congestion. Therefore, driving attractiveness plays a critical role in urban air pollution in Tehran. In fact, driving attractiveness shows
people tendency to use private automobiles. It means that the portion of urban trips which is performed via personal automobiles is characterized as driving attractiveness while it is assumed to be affected by several variables illustrated in Fig. 3.
Therefore, driving attractiveness is a fraction of trips which is dimensionless. Another critical variable in this subsystem which
is used frequently over the paper is driving attractiveness. Mathematical denition of it is expressed in Eq. (7):

DA a1 f1 FP a2 f2 PTF a3 f3 AuPr a4 f4 TCCP a5 f5 TT a6 f6 IPT

27

H. Vafa-Arani et al. / Transportation Research Part D 31 (2014) 2136

TCERC

RCF

<NPuT>
<FP>

<AIC>
PTI

AuPr

<IPT>

PTF
IPTEDA

<TC>

RC

RCR

TC
TCCR

PTFEDA

AuPrEDA

PTA

<RC>

NPuT
Pop0
TTEDA

TT

DA
<AQ>

TD

Pop

<Time>

<FQ>

FPEDA

TCCPEDA

ANP

NPeT
ATDT

TCCP

PPK
FP
TCRTP
<NPeT>

TATDA

<TC>
CRTP

AuP
<FIC>

ATD

<NPuT>

<RC>

Fig. 3. Transportation subsystem.

where fi(x) denotes an increasing or decreasing functions of x. Driving attractiveness is considered to be positively inuenced
by fuel price (FP), automobile price (AuPr), travel time (TT), trafc congestion control plans (TCCP), and negatively inuenced
by public transportation fare (PTF), and investment on public transportation (IPT). As shown in Fig. 3, all of mentioned variables will affect driving attractiveness with a delay.
Transportation subsystem is illustrated in Fig. 3. According to Fig. 3 transportation subsystem consists of two stock
variables: trafc congestion (TC) and road capacity (RC). Travel demand (TD) is a variable indicating citizens demand for
transportation which is affected by population.
As the investment in public transportation (IPT) increases, public transportation safety and convenience increases and
this results in reduction of relative driving attractiveness. On the other hand, increasing in public transportation fare
(PTF) makes individual trafc to increase. Meanwhile, an increase in public transportation fare leads to corresponding
increase in the public transportation income (PTI) which results in more investment in public transportation, consequently.
Increasing investment in public transportation makes driving attractiveness (DA) to decrease. Public transportation attractiveness (PTA) as the counterpart of DA shows citizen tendency to use public transportation. Hence, increasing rate of driving
attractiveness results in the reduction of public transportation attractiveness. An increase in public transportation attractiveness leads to corresponding increase in the number of trips by public transportation (NPuT) which in turn increases the public transportation income. These dynamics are illustrated in Fig. 4.
In individual trafc sector, driving attractiveness is affected by travel time and trafc control plans. Trafc congestion as a
stock variable is affected by three factors: (1) road capacity, (2) number of individual trips, (3) number of public

PTA
NPuT

TD
DA
Pop
<Time>

Pop0
PTFEDA
PTF

PTI

IPT
IPTEDA

Fig. 4. Public transportation dynamics.

28

H. Vafa-Arani et al. / Transportation Research Part D 31 (2014) 2136

transportation trips. In this study, trafc congestion of streets with trafc plan (TCRTP) (trip/kilometer) is dened as the total
number of trips divided by the capacity of streets with trafc plan (CRTP) and the trafc congestion control plans (TCCP)
variable indicates the number of streets with trafc plan. The dynamics of this sector are illustrated in Fig. 5.
An increase in trafc congestion, leads to corresponding increases in number of streets with trafc plan which in turn
reduces the attractiveness of driving with individual cars. Decrease of driving attractiveness results in a decrease of the
number of individual trips and an increase in the road capacity, which affects the travel time and can increase the driving
attractiveness. Also the increase in the road capacity causes the CRTP to increase as well. Another effective variable in trafc
congestion is the number of trips by the public transportation.
As the Driving attractiveness increases people tendency to use private cars increases. As a result, number of private trips
increases. Increasing number of private trips indicates more private vehicle usage and leads to more private automobile
emissions. To show this dynamics, a few auxiliary variables are dened. First, it is been assumed that the average travel distance per trip (ATDT) (kilometer/trip) is unity so number of personal trips (NPeT) determines the average traveled distance
(ATD) of each individual (kilometer). Second, through historical data and Figures it is estimated that the average number of
passenger (ANP) in each private vehicle is 3.5. Hence, total average of traveled distance (TATDA) per vehicle (in kilometers) is
given by:

TATDA

ATD
ANP

We consider an auxiliary variable namely pollution per kilometer (PPK) to show the pollution intensity of vehicles per
kilometer. PPK links transportation subsystem with industrial pollution subsystem through automobile quality (AQ) and fuel
quality (FQ). By using regression analysis on data from 2000 to 2009, this pollution intensity indicator is formulated as
follows:

PPK k  exp AQ  FQ

where, k is a constant. The main output of this subsystem, automobile pollution (AuP), is determined by PPK and TATDA as
follows:

AuP PPK  TATDA

10

The full name and type of each variable in the system is presented in Appendix A. The main important stock variable in transportation subsystem is road capacity (RC) which is expressed in Eq. (11):

RCR f  RCF  RC

11

where RCF is road construction factor which is affected by trafc congestion. Also, road construction rate (RCR) is considered
to be a function of road capacity, and f is fraction of road capacity which is planned to be added to the road capacity (RC). In
addition trafc congestion (TC) is dened as number of total trips per kilometer available road capacity.

<TC>

TT
TTEDA

<RC>

DA

TCCPEDA
TCCP

TCRTP

ANP

CRTP

<AQ>

<FQ>

ATDT
<TC>

NPeT

<RC>

PPK
ATD
TD

TATDA

NPuT
AuP
Pop
<Time>

Pop0
Fig. 5. Individual transportation dynamics.

29

H. Vafa-Arani et al. / Transportation Research Part D 31 (2014) 2136

Interactions between two subsystems


Considering interactions between industrial and transportation subsystems is a remarkable contribution of current study.
Several direct connections between two main subsystems are taken into accounts which are explained here. Automobile
quality (AQ) and fuel quality (FQ) are two variables which affect pollution associated with automobiles (PPK) in transportation subsystem which can be seen in Figs. 2 and 5. On the other hand, automobile and fuel quality are indicated in industrial
subsystem while investment of fuel and automotive industries on technology improvement indicates quality level of fuel
and automobiles. In addition, three variables which are used to calculate driving attractiveness (DA) in transportation
subsystem are indicated in industrial pollution including fuel price (FP), public transportation fare (PTF) (which is affected
by fuel price), and automobile price (AuPr). Fig. 3 illustrates effective factors on driving attractiveness. Another interaction is
related to public transportation development. On the one hand, investment on public transportation (IPT) is provided
by national income (NI). On the other hand, earnings of automotive and fuel industries and other sections form national
income. It means that public transportation is developed using earnings of fuel and automotive industries, directly. In addition to interactions listed above, there another indirect interaction between two subsystems which is neglected here to
explain.
Air pollutions associated with automobiles, industries, and energy production forms total air pollution as the main stock
variable of the system. It is clear that the rate of air pollution is a weighted sum of mentioned sources of pollutions. Eq. (12)
shows denition of air pollution rate:

APIR w1 AuP w2 IP w3 PEP

12

where AuP, IP, and PEP are automobile pollution, industrial pollution, and pollutions associated with energy production,
respectively. Also, wi is assumed to be weight of pollution sources which shows effect of each one on total urban air
pollution.
Numerical experiments
In order to show real applicability of the proposed system dynamics model, a case study of Tehran as one of the most
polluted metropolitans all over the world in which air pollution problem has become a crisis for people is been considered
in this study. There are a lot of sources for air pollution in Tehran. As mentioned before, the proposed model considered
industrial and transportation sources of air pollution. There are so many several industries, power plants and reneries in
suburban area of Tehran. Besides, high population of Tehran makes transportation an issue. The proposed system dynamics
model has taken all of above mentioned sources of air pollution into account, simultaneously.
System dynamics models need a lot of data for simulation of behavior of systems. Since the air pollution is the most critical problem for the decision makers in Tehran, the related organizations report the main data of air pollution, annually. We
used the real data from the annual report of Tehran transportation from 2007 to 2012 and Iran energy balance report from
2007 to 2011(Ministry of Energy, 2011; Municipality of Tehran, 2011). We attempted to simulate the behavior of system for
the next decade under various scenarios. Times in diagrams are assumed to be quarterly. So, the simulation is performed to
predict the air pollution of Tehran for 40 future seasons. The simulation is performed via Vensim software.
Recent data from Tehran municipality and ministry of health shows that mortality from urban air pollution is increased,
tangibly (Hosseinpoor et al., 2005). Table 1 shows several pollutants in Iran and a portion of various sections in Tehran air
pollution in 2011. As illustrated in this table, air pollution in Tehran includes NOx, SO2, SO3, CO, SPM, CO2, CH4, and N2O. Air
pollution (AP) indicator used in this study is a combination of these pollutants. The main point of this table is major portion
of transportation, industries and power plants in air pollution which conrms taking transportation and industries into
account as the main subsystems in this study. In addition, air pollution associated with power plants is considered as a part
of industrial subsystem which includes energy consumption by households and industries. According to Table 1, since the
main pollutant section in urban area is transportation section, pollutant CO in air pollution indicator has the greatest weight,
while agriculture section has the lowest weight. Industrial pollution (IP) is another main variable of the model which

Table 1
Portion of various sections in emission of pollutants in Tehran (%).
Section/pollutant

NOx

SO2

SO3

CO

SPM

CO2

CH4

N2O

Energy end users


Household, commercial and public
Industries
Transportation
Agriculture

6.3
8.8
47.2
3.3

3.7
13.4
28.8
4.4

4.3
20.8
33.9
2.8

0.6
0.3
97.1
0.2

2.9
4.3
78.3
6.8

25.0
16.7
23.1
2.3

7.1
3.8
79.7
1.2

4.3
2.4
50.4
36.7

Energy production
Reneries
Power plants

34.4

49.8

38.1

2.7
30.2

0.6
7.7

0.3
5.9

1.8

7.8

30

H. Vafa-Arani et al. / Transportation Research Part D 31 (2014) 2136

represents air pollutions associated with fuel and automotive industries because of the double effects of these industries in
industrial and transportation sections. So, a fraction of air pollution associated with all industries is considered as an industrial pollution variable in this model.
Pollution reduction policies
There are a lot of policies which the government could apply to reduce the urban air pollution. We propose a number of
policies for air pollution reduction in Tehran. Four applicable policies in Tehran are proposed to reduce urban air pollution.
The considered policies are as follows: (1) road construction, (2) Technology improvement in fuel and automotive industries,
(3) Applying trafc control plans, and (4) developing of public transportation infrastructures. It is assumed that all of parameters of the model are in a common level and only the related parameters for each policy will be changed. It is worth mentioning that the proposed policies are the most practicable policies according to political and economic conditions of Tehran.
Road construction is a common policy around the world. In terms of technology improvement, the municipality of Tehran
and Iranian government can force the industries around Tehran to shut down, therefore, industrial rms have to improve
their technologies to reduce air pollution and improve quality of automobiles and fuels. Trafc control plans is a trafc rule
in Tehran and trafc police can expand them to more roads. Also, 12 lines of subway and several lines of bus rapid transits
(BRT) are constructing in the metropolitan area of Tehran while nancing for them might be a suitable air pollution reduction policy.
One of the most regular policies for trafc reduction is road construction. Because of population growth in Tehran, the
government should construct more roads to facilitate the trafc in Tehran. But, the road construction rate must be greater
than growth of the number of automobiles to ease the trafc in an urban area; otherwise, it does not affect the trafc and air
pollution. On the other hand, road construction motivates the people to buy automobiles. Therefore, road construction not
only does not help for the air pollution reduction but also increases the air pollution because of the abundance of automobiles which use the roads. The main output variable of system dynamics model is air pollution. In addition, driving attractiveness, as the most important variable in transportation subsystem, has a critical role in air pollution. The results of
simulation for these two variables are presented in Figs. 6 and 7.
The applied policies affect the road construction rate in transportation subsystem. The rst, second and third policies are
high, medium and low level of road construction rate, respectively. It means that in the rst, second and third scenarios, the
length of roads, which were constructed during a period, will be high, medium and low, respectively. According to Figs. 6 and
7, the road construction growth does not have signicant effect on the reduction of air pollution over the planning horizon.
Air pollution has an overshoot behavior which is because of other air pollution reduction policies which will be introduced in
the next sections.
Second policy is technology improvement in fuel and automotive industries which is applied to the urban air pollution
system. In the proposed policy the government forces industries to improve their technology in order to reduce the air pollution associated with them. If the air pollution increases, the industries should invest on the technology improvement. The
industries divide their budget between technology improvement and production rate increase annually. The proposed policy
makes the share of technology improvement greater in annual budget of the industries. Technology of industries affects two
main variables in the system including direct industrial pollutions and pollution caused by fuel consumption. Firstly, technology causes less air pollution associated with industrial operations. Secondly, technology improvement in fuel and
automotive industries causes production of less pollutant fuels and automobiles. Therefore, technology improvement affects

Fig. 6. Air pollution under three policies in road construction.

H. Vafa-Arani et al. / Transportation Research Part D 31 (2014) 2136

31

Fig. 7. Driving attractiveness under three policies in road construction.

both industrial and transportation subsystems. Figs. 8 and 9 address the air pollution and industrial pollution simulation
results, respectively.
Three applied policies in regard of technology development affect fraction of budget that should be invested for technology improvement in automotive and fuel industries which is affected by air pollution level, directly. High, medium, and low
levels of sensitivity to air pollution are three policies which are forced by policy makers.
Applying trafc control plans is third policy which can reduce the urban air pollution. In Tehran, trafc control plans are
being used such as one-way streets in order to reduce the trafc congestion and urban air pollution, consequently. As these
plans can lengthen the distances for people who use private automobiles, that increases travel time and reduce the driving
attractiveness, consequently. The proposed model considers these popular policies in Tehran in order to reduce air pollution.
Length of roads with trafc control plans is a function of amount of air pollution. It means that the trafc control plans
increases, if the air pollution exceeds a threshold. An exponential function is employed because of high sensitivity of city
managers in higher amount of air pollution. We proposed three different scenarios for the trafc control plans: the better,
the normal, and the worse situations as rst, second and third policy curves in Figs. 10 and 11. Better, normal, and worse
situations illustrate a high, medium, and low level of decision makers sensitivity to air pollution for applying trafc control
plans. According to the gures, rst, second and third policies show low, medium, and high level of sensitivity to the air pollution. Figs. 10 and 11 illustrate the air pollution and driving attractiveness over the planning horizon, respectively. More
sensitivity of the managers to the air pollution and consequent increase in the trafc control plans causes less urban air pollution and driving attractiveness. A comparison between Figs. 8 and 10 shows less effect of trafc control plans on the air
pollution. It means that Tehran municipality should not use more trafc control plans and they should focus on technology
improvement in automotive and fuel industries.

Fig. 8. Air pollution under three policies on technology improvement.

32

H. Vafa-Arani et al. / Transportation Research Part D 31 (2014) 2136

Fig. 9. Industrial pollution under three policies on technology improvement.

Fig. 10. Air pollution simulation under the three policies on trafc control plans.

Fig. 11. Driving attentiveness simulation under the three policies on trafc control plans.

H. Vafa-Arani et al. / Transportation Research Part D 31 (2014) 2136

33

The last policy is development of public transportation infrastructures. As it is mentioned before, two groups of policies
could be applied to the system in order to reduce the air pollution. The rst group includes short-term policies which might
take effect in a short period and becomes ineffective in a short period, as well. Trafc control plans are one of such policies.
The second group of policies includes long-term ones. Although they take more time to be applied in comparison to the rst
group, but the effects of these policies are more stable than the rst group. Technology improvement and development of
public transportation infrastructures are the most important parts of the second group. Development of public transportation needs a remarkable investment. Financial resources of investment in public transportation are public transportation
prot and governmental investment. Therefore, a portion of the mentioned resources should be indicated. The best scenario
considers that all of the prot of public transportation and a percentage of national income are to be invested in development
of infrastructures. These amounts decrease in normal and worst scenarios. It means that rst, second and third policies in
following Figures show high, medium, and low rate of investment on public transportation infrastructures which are characterized as better, normal, and worse scenarios. Figs. 12 and 13 show the air pollution and driving attractiveness under the
proposed scenarios over the planning horizon. The more investment in the public transportation infrastructures causes less
air pollution and driving attractiveness. Figures show that this policy takes ve years to take effect on air pollution due to the
length of development projects. Also, Fig. 13 shows that this policy causes the most reduction in driving attractiveness comparing to previous policies.
A comparison between Figs. 6, 8, 10 and 12 provides a managerial insight for municipality of Tehran to make the most
effective decisions in order to reduce air pollution. Accordingly, technology improvement and public transportation development are more effective than trafc control plans and road construction, obviously. But, in the current situation of Tehran
from transportation point of view, trafc control plans can be more effective than road construction.

Fig. 12. Air pollution simulation under three scenarios on public transportation development.

Fig. 13. Driving attractiveness simulation under three scenarios on public transportation development.

34

H. Vafa-Arani et al. / Transportation Research Part D 31 (2014) 2136

Fig. 14. Model validation via a comparison between real and estimated data.

Validation
In order to show accuracy of the proposed model, a comparison between real data from 2007 to 2011 and estimation of
the model is implemented. Fig. 14 shows real and estimated data in a diagram. According to this gure, the system dynamics
model can be used as a good estimator for air pollution forecast. It is worth mentioning that the proposed system dynamics
model has been developed for several air pollution reduction policies while effective policies such as public transportation
development and technology improvement were not applied in reality during 20072011 in Tehran. On the other hand,
development of automotive industries, automobile import, and using low quality fuels in Tehran causes a huge growth in
air pollution which is shown in results of previous section.
Conclusion
Urban air pollution has become one of the most critical problems in several metropolitans over the world. In this
paper, Tehran, as a city that suffers from severe air pollution, was considered as a case study. The air pollution problem
comprises complex and nonlinear interactions between variables, difcult-to-explore variables and several delays. Therefore, a system dynamics approach was proposed to model the air pollution of Tehran and nd some efcient policies to
reduce the air pollution as short and long-term plans. Two subsystems were proposed in the proposed system dynamics
model including transportation and industrial subsystem. Transportation subsystem included private and public trips in
Tehran and related air pollution. Industrial subsystem investigated industrial pollution caused by plants around Tehran
and air pollutions associated with energy production/consumption. Fuel and automotive industries were considered in
the industrial subsystem because they affect the transportation subsystem directly via fuel and automobile quality and
price. Four policies were proposed in order to reduce air pollution in Tehran. First policy was road construction in order
to have less trafc congestion. Second policy was about forcing factories to improve their technologies for two purposes:
(1) reduction of direct air pollution caused by plants, (2) improvement of produced fuel and automobile quality. Third
policy was about applying trafc control plans which was referred as a short-term policy. Fourth policy was developing
of public transportation infrastructures. This policy was a long-term one and contained several delays, but its effects were
more stable. According to the simulation results, technology improvement in fuel and automotive industries in addition
to development in public transportation infrastructures were the better policies which had more stable and long-term
effect on the air pollution of Tehran. Road construction did not have remarkable effect on the trafc congestion
and air pollution. In other words, trafc control plans are temporary policies which could be applied in emergency
conditions.
The proposed policies in the current study might be extended by considering green fuels such as biodiesel instead of fossil
fuels, development of green parks and also applying macro-state policies such as decentralization in order to reduction of
population density in Tehran.
Acknowledgements
Authors would like to appreciate two anonymous reviewers for their valuable comments and suggestions which helped
to improve the paper.

35

H. Vafa-Arani et al. / Transportation Research Part D 31 (2014) 2136

Appendix A.
Complete list of variable in casual loop diagrams.
Variable
type

Variable

Description

Unit

Stock

AIC
AP
FIC
RC
TC

Capacity of automotive industries


Air pollution
Capacity of fuel industries
Road capacity
Trafc congestion

Number/month
Ton
Litter/month
Kilometer
Trip/kilometer

Rate

AIGR
APDR
APIR
FIGR
RCR
TCGR

The
The
The
The
The
The

Auxiliary

AITI

The investment in technology development of automotive


industries
Pollution intensity of fuel industries

Alpha
ANP
AQ
ATD
ATDT
AuP
AuPr
AuPrEDA
Beta
CRTP
DA
FITI
FP
FPEDA
FQ
HEC
IAI
IEC
IFI
IP
IPT
IPTEDA
NI
NPeT
NPuT
PEP
PII
PIT
POP
PPK
PTA
PTF
PTFEDA
PTI
TATDA

growth rate of automotive industries capacity


decomposition rate of air pollution
growth rate of air pollution
growth rate of fuel industries capacity
growth rate of road capacity
growth rate of trafc congestion

The average number of passengers in a private vehicle


Automobile quality
Average traveled distance of an individual
Average travel distance per individual trip
Automobile pollution
Automobile price
Effect of automobile price on driving attractiveness
Pollution intensity of automotive industries
The capacity of streets with trafc plan
Driving attractiveness by private automobiles
Investment in technology development of fuel industries
Fuel price
Effect of fuel price on driving attractiveness
Fuel quality
Household energy consumption
Investment in automotive industries
Industrial energy consumption
Investment in fuel industries
Industrial pollution
Investment in public transportation
Effect of investment in public transportation on driving
attractiveness
National income
Number of personal trip
Number of trips by public transportation
Pollution caused by energy production
Percent of investment in industrial capacity
Percent of investment for developing technology in automotive
industries
Population
Pollution per kilometer
Public transportation attractiveness
Public transportation fare
Effect of Public transportation fare on driving attractiveness
Public transportation income
Total average of traveled distance

Ton/liter of produced
fuel

Kilometer
Kilometer/trip

Ton/automobile

(continued on next page)

36

H. Vafa-Arani et al. / Transportation Research Part D 31 (2014) 2136

Appendix A. (continued)
Variable
type

Constant

Variable

Description

Unit

TCCP
TCCPEDA
TCERC
TCRTP
TD
TT
TTEDA

Trafc congestion control plans


Effect of trafc congestion control plans on driving attractiveness
Effect of trafc congestion on road capacity
Trafc congestion of streets with trafc plan
Travel demand
Travel time
Effect of travel time on road capacity

Trip/kilometer

DF
POP0
PIEP

Decomposition factor of air pollution


Initial number of population
Pollution intensity of energy production

References
Anand, S., Vrat, P., Dahiya, R.P., 2006. Application of a system dynamics approach for assessment and mitigation of CO2 emissions from the cement industry.
J. Environ. Manage. 79 (4), 383398.
Anh, T.T., 2003. System dynamic applied to study the urban trafc. Proc. East. Asia Soc. Transp. Stud. 4, 16931697.
Ansari, N., Sei, A., 2013. A system dynamics model for analyzing energy consumption and CO2 emission in Iranian cement industry under various
production and export scenarios. Energy Policy 58, 7589.
Armah, F.A., Yawson, D.O., Pappoe, A.A.N.M., 2010. A systems dynamics approach to explore trafc congestion and air pollution link in the city of Accra,
Ghana. Sustainability 2 (1), 252265.
Atash, F., 2007. The deterioration of urban environments in developing countries: mitigating the air pollution crisis in Tehran, Iran. Cities 24 (6), 399409.
Beckx, C. et al, 2009. A dynamic activity-based population modelling approach to evaluate exposure to air pollution: methods and application to a Dutch
urban area. Environ. Impact Assess. Rev. 29 (3), 179185.
Bel, G., Rosell, J., 2013. Effects of the 80 km/h and variable speed limits on air pollution in the metropolitan area of Barcelona. Transp. Res. Part D Transp.
Environ. 23, 9097.
Dong, C. et al, 2012. An inexact optimization modeling approach for supporting energy systems planning and air pollution mitigation in Beijing city. Energy
37 (1), 673688.
Feng, Y.Y., Chen, S.Q., Zhang, L.X., 2013. System dynamics modeling for urban energy consumption and CO2 emissions: a case study of Beijing, China. Ecol.
Model. 252, 4452.
Hong, J., Shen, Q., 2013. Residential density and transportation emissions: examining the connection by addressing spatial autocorrelation and selfselection. Transp. Res. Part D Transp. Environ. 22, 7579.
Hosseinpoor, A.R. et al, 2005. Air pollution and hospitalization due to angina pectoris in Tehran, Iran: a time-series study. Environ. Res. 99 (1), 126131.
Jensen, S.S. et al, 2008. Evaluation of exposure factors applied in marginal external cost analysis of transportation related air pollution. Transp. Res. Part D
Transp. Environ. 13 (4), 255273.
Lu, R., Turco, P., 1997. An integrated air pollution modeling system for urban and regional scales: 1. Structure and performance. J. Geophys. Res. 102 (D5),
60636079.
Lv, Y. et al, 2011. A two-stage inexact joint-probabilistic programming method for air quality management under uncertainty. J. Environ. Manage. 92 (3),
813826.
Lyons, T.J., 2003. An international urban air pollution model for the transportation sector. Transp. Res. Part D 8, 159167.
Ministry of Energy, 2011. Iran Enery Balance Sheet, Tehran.
Municipality of Tehran, 2011. Annual Report of Tehran Transportation Sector, Tehran.
Schmidt, M., Schafer, R., 1998. An integrated simulation system for trafc induced air pollution. Environ. Model. Softw. 13, 295303.
Shahgholian, K., Hajihosseini, H., 2009. A dynamic model of air pollution, health, and population growth using system dynamics: a study on Tehran-Iran
(with computer simulation by the software Vensim). World Acad. Sci. Eng. Technol. 35, 245252.
Shang, Y. et al, 2013. Systematic review of Chinese studies of short-term exposure to air pollution and daily mortality. Environ. Int. 54, 100111.
Sivacoumar, R. et al, 2001. Air pollution modeling for an industrial complex and model performance evaluation. Environ. Pollut., 111.
Statistical Centre of Iran, 2012. Iran Statistical Year Book. Available at: <www.amar.org>.
Strauss, J. et al, 2012. Investigating the link between cyclist volumes and air pollution along bicycle facilities in a dense urban core. Transp. Res. Part D
Transp. Environ. 17 (8), 619625.
Szarka, N. et al, 2008. Atmospheric emissions modeling of energetic biomass alternatives using system dynamics approach. Atmos. Environ. 42 (3), 403
414.
Trappey, A.J.C. et al, 2012. An evaluation model for low carbon island policy: the case of Taiwans green transportation policy. Energy Policy 45, 510515.
Ulke, A.G., Fa, M., Andrade, H., 2001. Modeling urban air pollution in Sao Paulo, Brazil: sensitivity of model predicted concentrations to different turbulence
parameterizations. Atmos. Environ. 35, 17471763.
Vlachokostas, C. et al, 2009. Decision support system for the evaluation of urban air pollution control options: application for particulate pollution in
Thessaloniki, Greece. Sci. Total Environ. 407 (23), 59375948.
Wang, J., Lu, H., Peng, H., 2008. System dynamics model of urban transportation system and its application. J. Transp. Syst. Eng. Inf. Technol. 8 (3), 8389.
Wong Wing Ying, C., 2007. A Study of Regional Characteristic of Air Pollution Problem in Hong Kong Using System Dynamics. City University of Hong Kong.
Yoo, S., Kwak, S., Lee, J., 2008. Using a choice experiment to measure the environmental costs of air pollution impacts in Seoul. J. Environ. Manage. 86 (1),
308318.
Yu, S., Wei, Y., 2012. Prediction of Chinas coal production-environmental pollution based on a hybrid genetic algorithm-system dynamics model. Energy
Policy 42, 521529.

You might also like