Exploration of the Growing Trend of Electric Vehicles in
Beijing with System Dynamics method and Vensim model
Chaoning Zhang', Caiyan Qin”
chaoningzhang1990@ gmail.com
> caiyanhit@ gmail.com
12 Delft University of Technology, Netherlands & Harbin Institute of Technology,
China
*Both authors are following a joint Master study between Delft and Harbin
Abstract
This research is conducted to explore the growing trend of private vehicles in Beijing,
China, in the coming 25 years using the system dynamics (SD) method. The vensim
software is used to build the SD model and do simulations. First, the paper
introduces the background of the private vehicles in Beijing and analyzes the issues
of traffic congestion and air pollution caused by them. Second, the method system
dynamics is justified to analyze the concerned issues of Beijing vehicle and the
attendant traffic congestion and air pollution. Third, qualitative analysis is done to
explore what possible policies the Beijing government can adopt to increase the
percentage of electrical vehicles (EVs) and control the number of total private
vehicles at the same time. A causal loop diagram is drawn to facilitate the way of
thinking. Afterwards, a quantitative SD model about the system is built based on the
causal loop diagram and qualitative analysis is made. Finally, the SD model is
simulated to explore the plausible future trend about private vehicles and study the
implications of different policies.
Key words: system dynamics; vensim; Beijing; electric vehicle; air pollution; traffic
congestion
1. Introduction
1.1 Research background
Beijing has been facing severe air pollution since the end of 2012. According to “The
Guardian” news(V aughan, 2013), all over 2013, only 13 days in Beijing is considered
"good" based on the US index, while 70 days moderate air pollution, 64 days
unhealthy for sensitive groups, 148 days unhealthy, 45 days very unhealthy, 14 days
hazardous and one day "beyond index", ie., off the scale. The primary cause is the
huge amount of pollutant emission, among which vehicle emission ranks second after
industrial emission.
Another significant problem caused by private vehicles in Beijing is traffic
congestion. In the past decade, the number of private vehicles in Beijing has
increased from 1.5 million in 2000 to 4.1 million in 2013, which made a great
contribution to the economy increase, but in the meantime, the traffic in Beijing is
badly affected with such a huge number of vehicles (Wikipedia, 2012).
The Beijing government is putting great effort in alleviating the pollution and traffic
congestion. To control the increase of vehicles, the ‘license plate lottery’ policy (Lim,
2011) has been implemented since 2011. Since the implementation of this policy,
citizens who want to get a new license plate to buy a vehicle must register online and
join the lottery. Every month a certain number of license plates are issued to these
registers randomly. License plate lottery policy is chosen as the first base policy in
this paper. To encourage citizens to buy EV, citizens can join a lottery special for EV
and the chance to get an EV license plate is higher than for conventional vehicles
(CVs). Besides, the Beijing government provides about 40% subsidy to those who
buy EV.
One interesting policy about private vehicles in Beijing is that some private vehicles
are not allowed to be on road on weekdays. For instance, those vehicles whose license
plate number ends with 0 or 6 are not allowed to be on road on Monday, and number
ending with 1 or 7 on Tuesday, etc. This policy is chosen as the second base policy in
this paper.
Currently, two hot debates are undergoing. One is whether the restriction should be
only on CV. In other words, some citizens propose that those who want to buy EV
have no need to join the lottery to get the license whenever they want. The other is
whether the Beijing government should adopt the license plate policy as in Shanghai
(Yue, 2013). In Shanghai, there is also restriction on new license plates; however, it
takes place in form of auction. Citizens who want to get a license plate should
participate in an auction rather than join the ‘license plate lottery’ process in Beijing.
Shanghai citizens have no need to wait for years like in Beijing to get a new license
plate but they need to pay about 70000RMB (Yue, 2013) for a license plate. The
paper will use quantitative system dynamics model to explore the advantages and
disadvantages of the two policies.
1.2 Research method
A lot of studies on transportation and traffic congestions applying system dynamics
have been conducted by researchers, among which are (Springael, 2002) and (Barlas,
2002). In terms of air pollution, many researchers have also been done with system
dynamics method, a good example is (Hamed, 2014). In our research, system
dynamics is also used to explore the relationship of private vehicles, traffic congestion
and air pollution.
System dynamics is a discipline emerged in late 1950s, which attempts to explore
complex long-term policies in both public and private domains (Sterman, 2000). It is
a tool to help address complex issues involving delays, feedbacks and nonlinearities.
As the real world is a multi-loop, multi-state, non-linear feedback system (Forrester,
1969), system dynamics provides a scientific way to facilitate our thinking of the
complex world.
Causal loop diagram is a tool to understand the dynamic, interrelated nature of the
real world. By linking together the key variables, the causal relationships between
them are indicated. By stringing together several loops, a coherent story about a
specific issue can be depicted (Kim, 1999).
Vensim is a visual system dynamics s simulation modelling tool, with which users
can conceptualize, document, simulate, analyze, and optimize models of dynamic
systems (Ventana, 2010). Based on causal loop diagram, a simulation model can be
built in the vensim environment, in a presenting form of stock and flow diagram.
Relations among system variables are built and recorded by connecting words with
arrows. When completed, the model can be simulated, analyzed and its behaviours
can also be explored thoroughly by users (Akhtar, 2011).
In this paper, system dynamics thinking is applied both qualitatively and
quantitatively. First, qualitative analysis is done to explore what possible policies the
Beijing government can adopt to increase the percentage of EV and control the total
number of private vehicles simultaneously; Second, causal loop diagram of the private
vehicle market in Beijing is drawn to facilitate the way of thinking; Finally, SD model
using vensim software is built to explore the plausible future trend about private
vehicles and implications of different policies.
1.3 Research questions
More private vehicles will lead to more traffic congestion, and more CVs will cause
more air pollution. Hence, to solve the two problems, on the one hand, the total
number of private vehicles should be controlled; on the other hand, the percentage of
EV should be increased. Basically, the research is focused on these two aspects.
The first aspect is about solving the traffic problems in terms of private vehicles:
What policy can be applied to control the total number of private vehicles?
What policy can be applied to control the total number of private vehicles on road?
The other aspect is aimed to solve the air pollution caused by conventional vehicles:
What policy can be used to increase the percentage of EVs? What policy can be used
to decrease the number of CVs on road?
2. Problem qualitative analysis
In this part, the research questions raised up above are first analyzed with basic logic
and brainstorming. Then causal loop diagram is built based on the analysis, which in
retum will be the foundation of the system dynamics (SD) model.
2.1 Qualitative analysis of the research questions
(1) Reasons why there are so many private vehicles?
Beijing is facing a high speed of economic growth and a fast trend of urbanization,
thus more and more people in Beijing can afford to buy a vehicle. Even though
problems such as air pollution and traffic congestion are severe in Beijing, the total
population in Beijing might continue increasing in the near future due to its economic
and political attractiveness.
(2) What policy can be applied to control the total number of private vehicles?
The basic policy used by the Beijing government to control the total number of
vehicles is to restrict the license plates issued per year. 240,000 license plates are
issued per year from 2011 to 2013, 88% of which are for private vehicles. From 2014
to 2017, the Beijing goverment plans to decrease the number from 240,000 to
150,000 per year (Wei, 2013). Data shows that the growing rate of private vehicles in
Beijing becomes slower in recent years, proving the usefulness of this policy. Further
discussions will be made to decide how many private license should be issued per
year in the future to act optimally.
(3) What policy can be applied to control the total number of private vehicles on
road?
In order to decrease the total number of vehicles on road, only part of the vehicles are
allowed to be on road on a certain week day. The current policy in Beijing is that only
80% of the private vehicles are allowed to be on road everyday as mentioned in the
introduction part. It is assumed in this paper that EV generates no pollution. In
practice, however, most the electricity energy comes from fossil fuels, which indeed
pollutes the air. However, these power plants are far away from Beijing, thus the
impact on Beijing air can be neglected. The increase proportion of EV, therefore, will
lead to less air pollution.
(4) Reasons why EVs are not popular in Beijing?
There are several reasons why EVs are not popular in China. The first one is that the
EV price is relatively higher than that of CV due to the immaturity of the EV
technology. The second reason is that the driving range is relatively smaller than CV.
Furthermore, the infrastructures for EVs are not sound yet, adding to inconvenience
for people using EVs (Wikipedia, 2012).
(5) What policy can be used to increase the percentage of EVs?
First, the government will subsidy the EV purchase while tax on CV; Second, the
government need to build more charging stations to provide enough access to EV
battery charge. Besides, the idea of environmental friendliness is suggested to be
propagandized to raise the consciousness of the public. As to the driving range, it is
quite uncertain since it depends on the technology level. Scenarios analysis about
technology level will be made in later part of the research.
(6) What policy can be used to decrease the number of CV on road?
As mentioned in (3), the policy to decrease the total number of private vehicles on
road is to allow only part of the private vehicles every day. For CV specifically, a
more restrict policy can be applied to restrict the CVs on road.
2.2 Causal loop diagram based on qualitative analysis
The development of the vehicles is fundamentally influenced by factors in three
aspects: market discipline, goverment policies and population growth. The factors
acting on market discipline include technology level, vehicle price, oil and electricity
prices, and maintenance cost. The goverment policies include the installation of
charging stations, subsidy for EV, additional tax on CV, and the license-plate lottery,
etc. The population also plays an important role in the increase of private vehicles. An
increase in population will lead to an increasing need for private vehicles, leading to
worse traffic congestion and air pollution. The worse air pollution and traffic
congestion, in feedback, might reduce the growth rate of the population.
To facilitate thinking, a causal loop diagram is shown as in Figure 1. The most
important loop in the figure is the positive loop in the figure. More EV charging
stations cause larger percentage of citizens wishing for EV, which causes more EV
possession; more EV possession leads to more need for EV charging stations.
subsidy for EV
rv air polation possbilty of geting a
electricity cost pice of we ‘pt = plate lisence fora CV
SSS ; traffic jam ——a: *
¢
S peientge wishing 7° r ‘ev Sone ino a)
additional tax on CVF _ for EV vo % JAD ing : y}
+
due to air pollution + [t+ ar of EV total private sansa cal ot
+ vehicles + opulation in applying for a new license
+ ~~
2) + fe Being A
-
price of CV #EV charging ——" &
stations ‘people wasting a
ve
ollpice economic level urbanization
Fig.1 causal loop diagram
3. Model building with vensim
In general, there are two common functions of model building. One is to explore the
plausible further; the other one is to study the implications of different policies (Pruyt
E., 2013). In this paper, the first purpose of model building is to explore the plausible
future trend of private vehicles, both EVs and CVs. Based on that, the policies
mentioned above are tested whether they will actually work to improve the percentage
of EVs while restrict the total number of private vehicles simultaneously. The model
is policy oriented instead of imitating the reality.
3.1 System Boundary
This research is aimed to explore the developing trend of private vehicles including
both CVs and EVs within Beijing with a time horizon of 25 years from 2013 to 2038.
The monetary currency used is Chinese RMB.
The CVs includes the conventional vehicles and normal hybrid electric vehicles
(HEV) since 90 percent of the energy source for HEV is oil. The EVs includes pure
electric vehicles and plug-in HEV (PHEV), whose energy source is mainly form
electricity.
Only private vehicles are taken into account in this paper. Private vehicles account for
about four fifth of total vehicles in Beijing.
The growth of population contributes a lot to the increase of total private vehicles.
However, the population growth rate is not only affected by the air pollution and
traffic congestion, but also affected by many other factors (Li, 2013). In the SD
model, there is no feedback from traffic congestion and air pollution. Instead, scenario
analysis will be done to study its influence on model behaviour. The factors that
influence the population growth rate, therefore, are outside the system boundary.
3.2 Assumptions
Since the Beijing private vehicle system is a very complicated system, we need to
simplify the system and make some assumption to focus on the policy part. The
assumptions made are as follows:
(1) The first assumption is that the needs for EVs by the consumers can always be
met by the EV producers so that the production sector dosen’t need to be
considered. The current situation is, the Beijing Government is making effort to
encourage the use of EV, and the EV producers are also taking the first step to
produce sufficient EVs, so our assumption is reasonable.
(2) The second assumption is that the price of CV is relatively stable since the CV
industry is already mature; while the price of EV is fluctuant which is influenced
by the relevant technology greatly.
(3) The third assumption is that citizens can not apply for a new conventional vehicle
license plate (CVLP) and an electric vehicle license plate (ECLP) at the same
time. This is actually the real situation in Beijing.
(4) The fourth assumption is that EV will not generate any pollution in Beijing. Even
though the generation of electricity may cause pollution, it is far away from
Beijing where is not populous.
(5) The last assumption is that people conceme much more about the price of CVs
and EVs than about the use and maintenance cost due to the long-term uncertainty
and inflation.
3.3 System dynamics model
Based on the qualitative analysis and assumptions above, we built an SD model with
vensim software to show the Beijing private vehicle market. The model is shown in
figure 2.
Fig.2 system dynamic model with vensim
4. Model specification
In this part, first, the major relationships between different factors in the model are
explained; second, some major soft variables in the model are specified. These soft
variables are not well-defined but are very important and influential on the model
behaviour (Eric). Finally, we will give explanations for some important variables and
functions.
4.1 Explanation about the model
From the model, we can see that the Beijing population will increase at a small
growth rate, leading to the increase of population who need a vehicle. These people
have two choices, either to buy EV or to buy CV. There are many factors that
influence the people’s choice. The fundamental factor is the number of charging
stations, which determines the basic percentage of people who choose to buy EV.
Another important factor is the total money gap between keeping a CV and an EV.
The vehicle price, maintenance cost and use cost are considered to determine the total
money.
When people have made their choice, they need to participate in the license plate
lottery according to the current policy which is aimed to limit the license plates issued
per year. Since there are so many people wanting to buy a vehicle, they have to wait
for a relatively long time before they can get a license plate. It is worth mention that
the lotteries for CVLP and for EVLP are separate, meaning that there are two queues
for lottery. Further on, since the government is encouraging the purchase of EVs, the
chance to get an EVLP is higher. Therefore, people may change their queue from
CVLP lottery to EVLP lottery.
Besides the new buyers, owners whose vehicles are broken also need to replace the
old vehicles with new ones. At this point, the assumption is made that EV owners can
only choose to replace with EV while CV owners can choose to replace either by CV
or by EV without lottery.
In terms of vehicle on road, it depends on the restriction policy by the government.
The current policy is that only 80 percent of the vehicles can go on road per day,
among which only CVs on road affects the air quality.
An important positive loop is between the charging stations and the EVs. The more
EVs in use, the more charging stations will be built, and vice versa.
4.2 Major soft variables
Since there are many factors that should be taken into account, the relations between
which are not well-known and completely logically determined, many soft variables
exist in the model. The major soft variables are as follows:
(1) The first and most important soft variable is “percentage of citizens wishing for
EVs”, because it is this variable that determines the development trend for both
EV and CV. It is evident that the following variables “price gap between EV
and CV”, “maintenance cost gap between EV and CV”,” use cost gap between
EV and CV” and ” number of electric charging stations” will have major
influences on the variable “percentage of citizens wishing for EVs”, but the
exact quantitative relation between them is unknown. After a lot of literature
study, the equation in the model is the optimal one we can come up with so far.
(2) The second one is “citizens from CVLP to EVLP”’. Some citizens cannot wait to
buy a CV because the chance to get a new CVLP is too low. This variable will
be influenced by the chance factor which is determined by the chance to get a
CVLP and the chance to get an EVLP.
(3) Other soft variables include “price gap factor”, “use cost gap factor”, “
maintenance cost factor”, “basic percentage determined by electric charging
station ‘’, etc.
4.3 Other important functions and explanations of the variables
The other important functions or variables in the model are explained in table 1.
Table 1 some important variables
Variables Values or equations Explanations
Vehiclesperl000peo | 1000*total number of) When the valueis low, more
ple people who have a private ‘ F
car/ total population public transport is needed
Percentage of | Basic percentage | This percentage is determined
citizens wishing for
EV
determined by charging
station*0.5“(total money
gap between EV and
CV/20000)
by the money gap and number
of charging station
Basic _ percentage
determined by
electric charging
Station
capacity of the total
chargingstations0.5/10000
The relation between “basic
percentage” and “capacity of
total charging stations” is non-
linear positive
Number of people
who want to replace
by CVs
citizens whose CV out of
work*(1-percentage of
citizens wishing for EV),
0.25, citizens whose CV
out of work*(1-
percentage of citizens
wishing for EV)
CV owners choose to replace
with a CV if they don’t want
EV after 0.25 year on average
Percentage from
CVLP to EVLP due
to chance factor
0.01* Ln (chance factor)
This function reflects the fact
that when the chance factor is
1, the corresponding
percentage is 0; and the
relation is non-linear positive.
Total money gap
between EV and CV
price gap between EV and
CV +(maintenance cost
gap between EV and CV +
use cost gap between EV
Price, maintenance cost and
use cost are considered in the
total money gap
and CV )*uncertainty factor
Percentage of | basic percentage
Citizens wishing for| determined by electric
EV charging station*
percentage due to money
gap factor
Chance factor
chance to get EVLP per
year/chance to get the
CVLP per year
The ratio between chance to
get EVLP and chance to get
CVLP
5. Model simulation and policy analysis
Based on the two base policies mentioned in the introduction and the specification
above, the basic model is built and run to explore the plausible future under the
current policy in 6.1. In 6.2 some new policies are introduced to improve the current
situation. Due to the serious air pollution and terrible traffic, the attraction of the city
Beijing has decreased sharply. In the basic model, the net population growth rate is
assumed to be 0.
5.1 Plausible Future under the current policy
The following figures show the trend of important dependent variables in the coming
25 years. They are: number of CV/EV owners, percentage of people who have EVs,
private vehicles on road per day, traffic congestion factor, percentage of citizens
wishing for EVs, number of charging stations.
Selected Variables
percentage of people who have EV
200 2008
Time (Y er)
TE
wo
rami
2018
218 20 2008 2055 FT
Time eat)
Fig. 5.1.1
Fig. 5.1.2
1M
1M
900,000,
950,000,
800,000,
goo L-
Selected Variables
traffic congestion factor
2005
3 2008
‘Tn (¥en)
755
Fig 5.1.3
2015
2018 2025 2028 2035
Time (Year)
2038
Fig 5.1.4
percentage of citizens wishing for EV number of electric charging stations
0.05 40
ose oa ie
wit —
002s _- vo
ee:
0.013 oe 100
0 le
2013 2018 2023 2028 2033 2038 0 tt
Tne (ea a a
‘ a Tine (Yew)
Fig 5.1.5 .
g Fig 5.1.6
Fig 5.1.1 and Fig 5.1.2 show that the number of CV will increase at a very small rate,
while the EV will increase fast and its proportion will grow from 0 to almost 0.1 in 25
years. Fig 5.1.3 and Fig 5.1.4 show that the CVs on road will decrease and the traffic
congestion factor will increase slowly, but the total private vehicles will continue
increasing. Fig 5.1.5 shows that the percentage of citizens wishing for EV will
increase from 0 to 0.04. Fig 5.1.6 shows that number of charging stations will also
increase a lot from 0 to around 370.
We can conclude that the current policies to restrict the private vehicles are useful. In
the next part, we will explore the policies that can improve the current situation
further.
5.2 Policy Analysis
In this part, some independent variables are changed to see what influence they have
on the dependent variables. These independent variables include “number of license
plates issued per year”, “the percentage of vehicles allowed on road”, ‘charging
stations built before market needs’, ‘subsidy rate for EV by the government’,
‘additional purchase tax rate for CV’, ‘oil cost per kilometre’, ‘electricity cost per
kilometre ’.
5.2.1 Total private vehicle licenses issued per year’ and ‘percentage of EVLP issued’
The variable ‘total private vehicle licenses issued per year’ and ‘percentage of EVLP
issued’ are studied to see how they will affect the trend of EV and CV.
Under the current policy, 132,000 new private license plates are issued per year,
among which about 13% are for EV. The Fig 5.2.1 shows the number of private
vehicles will increase or decrease about 0.25 million if the ‘total private vehicle
licenses issued per year’ increases 10% or decreases 10%. The Fig 5.2.2 shows the
percentage of people who have EV will almost double or nearly half if the ‘percentage
of EVLP issued’ becomes double or half.
Therefore, the government can adjust the ‘total private vehicle licenses issued per
year’ and ‘percentage of EVLP issued’ according to its expected goal and real
situation to get a good result.
citizens having either EV or CV percentage of people who have EVs
1s 2018 2a 2028
Fig 5.2.1 Fig 5.2.2
5.2.2 ‘Percentage of CV allowed to be on road’ and ‘percentage of EV allowed to be
on road’
The variable ‘percentage of CV allowed to be on road’ and ‘percentage of EV allowed
to be on road’ are studied in this part.Under the current policy, both 80% of CV and
EV are allowed to be on road. Fig 5.2.3 shows the total private vehicles on road per
day will increase about 10% if all CV and EV are allowed to be on road. Fig 5.2.4
shows the ‘CV on road per day’ will drop about 20% if all EV and 50% of CV are
allowed to be on road.
Therefore, the Beijing government should adjust the ‘percentage of CV allowed to
be on road’ and ‘percentage of EV allowed to be on road’ based on the level of
traffic congestion and air pollution.
total private vehicles on road per day CV on road per day
aM 2M
L76M vM
Lom uM
iu ee 11M a
1M 800,000 — ———— }
0201s oss 2038 2032012003 ___b008 aay 2098
Tie ne (a
Fig 5.2.3 Fig 5.2.4
5.2.3 Other independent variables
According to the qualitative analysis, we know ‘charging stations built before
market needs’, ‘subsidy rate for EV by the government’, ‘additional purchase tax
rate for CV’, ‘oil cost per kilo meter’, ‘electricity cost per kilo meter’ will affect the
‘percentage of citizens wishing for EV’ and ‘percentage of people who have EV’.
Model sensitivity is studied on these variables in this part.
(1) Fig 5.2.5 shows the difference between when there are 10 and 0 charging
stations in the first 5 five years. When we see the first 10 years, there is a
difference between them. The government should build some electric charging
stations even before there are many EV. The availableness of electric charging
stations will increase the competiveness of EV against CV.
percentage of people who have EV
o
2018 D018 2023 2028 7035 2038
Time (¥eat)
Fig 5.2.5
(2) Sensitiveness tests are shown in Fig 5.2.6 and 5.2.7, where the setting condition
is that the subsidy rate fluctuates within the range 90% and 110% of the current
subsidy rate. It turned out that the percentage of citizens wishing for EV is very
sensitive to the subsidy rate. The ‘percentage of people who have EV’ is less
sensitive to subsidy rate because license plate lottery policy plays a major role
here. To make it in a more extreme case, when the subsidy rate is 0, Fig 5.2.8
shows that the ‘percentage of people who have EV’ will decrease by 40 percent
compared to the current subsidy policy. Thus the subsidy for EV is extremely
important for the development of EV.
a or ertage of people who bave EV
Fig 5.2.6 Fig 5.2.7
Fig 5.2.8
(3) We might charge additional tax on CV due to the air pollution caused by them to
encourage citizens to choose EV instead of CV. The figure 5.2.9 and 5.2.10 show
that the influence of the additional tax on CV on the ‘percentage of citizens
wishing for EV’ and ‘percentage of people who have EV’. It showed that the
‘percentage of EV allowed to be on road’ is more sensitive to the additional tax
on CV than ‘percentage of people who have EV’ since the license plate lottery
policy plays a major role for ‘percentage of people who have EV’.
percentage of citizens wishing for EV percentage of people who have EV
3 2018 Ey 7028 2055 78
Tine (Vea)
Fig 5.2.9 “Fig 5.2.10
(4) Sensitiveness analysis is done on the ‘oil cost per kilometer’ and ‘electricity cost
per kilometer’ as shown in Fig 5.2.11& Fig 5.2.12 and Fig 5.2.13 & 5.2.14
respectively when they fluctuate within the range of 90% and 110% of current
cost. The ‘percentage of EV allowed to be on road’ is more sensitive than
‘percentage of people who have EV’ to the factor ‘oil cost per kilometer’ and
‘electricity cost per kilometer’ because the license plate policy plays a major role
for “percentage of people who have EV”.
cmt cae
50% 75% 956 100% Soe 759595 00%
peas of cts wig EV pera of pop who Bae EV
08 oD
005 ou |
oot na on
om -_ 05
a a a ET am ci 78
Tate ea) nay
Fig 5.2.11 Fig 5.2.12
cane came
50% 75% 059% 100% 50% 75% 95% 100%
pet of cn wishing or EV perentage of peopl wo he EV
% 02
one ous
mas a1
0.013 | 0.05 | ee
Soin mais Ed cr 78 os 2018 226 7032 7s
‘Time (Year) Ta oe
Fig 5.2.13 Fig 5.2.14
5.3 Scenario analysis
There is much uncertainty about the future. In this part, scenario analysis is done to
explore plausible future of Beijing private vehicles, both of CVs and EVs. Scenarios
are generated by setting different inputs for the model.
(1) The first scenario analysis is done by setting different inputs for the variable
‘average price of EV. The average price of EV is influenced by the relevant
technology about EV majorly. There might be great technology breakthrough in
terms of EV in the near future and the average price of EV can drop greatly. It
could also be that the in the next near future, no technology breakthrough can be
achieved and the average price will not change much.
percentage of people who have EV percentage of people who have EV
The Fig 5.3.1 and Fig 5.3.2 show that for the variable ‘percentage of people who
have EV’, there is a large difference in terms of model behaviour in these two
different scenarios. In the first scenario when the average price of EV will
decrease 5000 RMB per year, the percentage of EV increases sharply. Even the
subsidy rate decrease from 40% to 20%, the simulation result is still quite good
according to the blue line in the figure. The conclusion about this scenario
analysis is that the subsidy rate should be dependent on the average of EV. When
the average price of EV is high, the government should provide a high subsidy for
it.
(2) The oil price is also a very important factor in the model. There is some
uncertainty about the oil price. It might stay stable in the next 25 years or
increase a lot due to oil shortage or other political reasons. The figure 6.3.3
shows that in the scenario when the oil price increases to 1.5 RMB per
kilometre, about 60% of the private vehicles will be EV. In this scenario, the
government can lower the subsidy rate to some extent because the
competitiveness of EV is big enough.
percentage of people who have EV
as i=Au |
0B ie 2085 2038
“Tine (eat)
Fig 5.3.3
(3) Another interesting scenario is about the increase of Beijing population. Under
the license plate lottery policy, the increase of the population has little influence
on the number of total private vehicles since every a fixed number of license
plates are issued independent of the population. However, this has great
influence on the variable ‘vehicles per 1000 people’. As shown in the figure
5.3.4, in the scenario when the population net increase rate is 0.01, the number
of vehicles per 1000 people decreases from about 200 to 175. In this scenario,
the government should invest more on public transport infrastructure. It is worth
mention that without the license plate lottery, the number of total private
vehicles will be affected by the population greatly.
vehicles per 1000 people
2015 16 05 208 2035 2038
‘Time (Y ea)
Fig 5.3.4
(4) There is hot debate on whether there should be restriction on EVLP or not. In the
scenario when there is no restriction on EVLP, the total number of private
vehicles (on road) will be larger while the percentage of EV on road will be
smaller. Thus, there should be a trade-off between them. It might be suggested
that in the first 10 years no restriction should be made on EVLP to encourage EV
purchase. A fter that the restriction should be added as on CV LP.
The Fig 5.3.5 to Fig 5.3.8 show that if there is no restriction on EV, the total
number of private vehicles will almost double and the total number of private
vehicles on road will increase greatly in the future. The traffic congestion,
therefore, will be much worse.
citizens having either EV or CV percentage of people who have EV
Fig 5.3.5 Fig 5.3.6
total private vehicles on road per day CV onroad per day
175M mane
. Bee as
10am a =
1. ae ee
——T 00,000 BS Siaomace
2033 a =
Fig 5.3.8
Fig 5.3.7
(5) Some citizens propose to adopt the auction policy for license plate as in Shanghai.
Only if a fixed number of license plates are issued per year, it doesn’t matter
whether the lottery policy or the auction policy is adopted in term of total number
of private vehicles. The advantage of lottery policy is the fairness for everyone.
The advantage of auction policy lies in that those who need a private vehicle most
and are willing to pay extra money for the license plates can get the license plates
without waiting.
Thus a mix policy is proposed in this paper. The number of total license plates
issued per year remains the same, half of which are assigned to those who join the
lottery while the other half are distributed to those who participate in the auction.
Therefore, potential vehicle buyers have the freedom to choose to participate in
lottery or auction based on their situation.
It is worth mention that the government can get money in the implementation of
the plate policy to cover the subsidy for EV as shown in Fig 5.3.9. However, the
percentage of EV will be weakened under the mix policy compared to pure lottery
policy as shown in Fig 5.3.10.
Selected Variables percentage of people who have EV
Fig 5.3.9 Fig 5.3.10
6. Reflection of the model
The model in this paper is a policy-exploration-oriented model. The model cannot
represent reality. The trend of the model behaviour instead of exact quantity
predictions of the future should be stressed. The trend of model behaviour can be used
to study the implications of different policies.
7. Conclusion
The paper first analyzes the reasons why there are so many private vehicles in Beijing
and why EV is not popular currently in China. Based on this analysis a causal diagram
is drawn to explore the relations between many related factors. Both qualitative and
quantitative analyses are conducted further to explore the plausible future about
private vehicles in Beijing and the implications of different policies.
The model cannot be used to predict the exact vehicle number in the future. The
behaviour trend, however, can be used for policy design. The following policies are
recommended to the Beijing government to solve the current problems caused by the
private vehicles.
+ Policy 1: The government should build more electric charging stations before the
demand for them. Lack of infrastructure is a major reason that discourages citizens
from choosing EV.
+ Policy 2: To encourage more EV buyers, the govemment should provide subsidy to
EV buyers since the high EV prices another major reason that EV lacks
competitiveness against CV. Furthermore, since oil and electricity price are another
two important money factors. The Beijing government should adjust the subsidy rate
for EV based on EV price, oil price, and electricity price altogether.
+ Policy 3: To control the fast increase of private vehicles, there should be restriction
on new private vehicle license plates. However, there will be more serious traffic
congestion if the restriction is only on CV in the long run. A suggested policy design
is that in the first 10 years, no restriction is on EVLP to increase the proportion of
EV; then after 10 years, EVLP should also be restricted to control the increase the
total number of vehicles.
* Policy 4: To control the total private vehicles on road per day, only some
percentage of vehicles are allowed to drive on a certain day. The percentage of
CV allowed to be on road can be decreased gradually to reduce to control the air
pollution.
+ Policy 5: A mix policy should be adopted to get the advantages of both lottery
policy and auction policy. Half of the new license plates are assigned to those who
join the lottery while the other half are given to those who participate in the auction.
+ Policy 6: With the restriction on new license plates, the increase of population has
little influence on the total number of private vehicles. However, the variable
‘number of vehicles per 1000 people’ will decrease a lot in the scenario where the
population increase greatly. In this scenario, the government should invest more on
public transport.
All these policies can be mixed to achieve the expected percentage of EV and number
of total vehicles.
Reference
Lim, L. (2011).License Plate Lottery Meant To Curb Beijing Traffic. Retrieved
March 18, 2014, from: http://www.npr.org/2011/01/26/133231061/license-plate-
lottery-meant-to-curb-beijing- traffic
Pruyt, E. (2013). Small System dynamics Models for Big Issues: Triple. Delft: TU
Delft Library, Delft, the Netherlands.
Tie, L. (2013). NDRC Study Addresses Beijing’s Population Boom. Retrieved
March 18, 2014, from: http://english.caijing.com.cn/2013-01-01/112404657.html
Vaughan, A. (2013). Nine Chinese cities suffered more days of severe smog than
Beijing. Retrieved March 18, 2013, from:
http://www. thequardian.com/environment/2014/mar/12/china-smog-pollution-
beijing
Wei, X.(2013). China Daily. Beijing cuts number of new cars. Retrieved
March 18, 2014, from: _ http://usa.chinadaily.com.cn/china/2013-
11/29/content_17138933.htm
Wikipedia. (2012). Electric vehicle industry in China. Retrieved March 18, 2013,
from:
http://en.wikipedia.org/wiki/Electric_vehicle industry_in China#Challenges to_the
indus
Yue, L. (2013). Wall Street Joumal. In Shanghai, A License Plate Can Cost as Much
as a Car. Retrieved March 18, 2014, from: _http://blogs.wsj.com/drivers-
seat/2013/01/25/in-shanghai-a-license-plate-can-cost-as-much -as-a-car/
Akhtar, M. et al. (2011). Water resource research report: An integrated system
dynamic model for analyzing behavior of the social-energy-economics-climate system.
London: Western Engineering.
Barlas, Y. (2002). System dynamics: Systemicfeedback modeling for policy analysis:
Encyclopedia for life support systems. Oxford: UNESCO Publishing.
Forrester, J. (1969). Urban Dynamics: Pegasus Communications. MT,USA: Waltham.
Hamed, V. A. et al. (2014, 8). A system dynamics modeling for urban air pollution: a
case study of Tehran, Iran. Transportation Research Part D , Vol.31 pp. 21-36.
Kim, D. H. (1999, 2). Guidlines for drawing causal loop diagrams. The systems
thinker, Vol 3, No.1, pp. 5-6.
Springael, J. et al. (2002, 4). A multicriteria-based system dynamics odeling of traffic
congestion caused by urban commuters. Central European Journal of Operations
Research, pp. Vol.10 Issue 1, pp81-97.
Sterman, J. (2000). Busiess dynamics: System thinking and modeling for a complex
world. Boston: McGraw.
Ventana. (2010). Vensim DSS Software. Retrieved July 28, 2014, from Vensim
Systems: http://vensim.com/docs/
Appendix: Vensim model (big size)
