DataPDF Available
1
Appendices
Appendix A: Entity types and the reference variables
In an entity-based modeling, while all entity types have their own dynamics, they can also use variables of the other entities. These variables indeed connect
the entity types to each other and integrate the disaggregated system. In Ventity software, these variables are called ‘reference variables’. In the following
table, you can find the main references of each entity type.
Entity Type
Related entities and variables
1
EV purchasers
E-mobility innovation system. EV knowledge development and diffusion1
Charging point. Facilitating conditions
EV pricing Collection. Average Price value of EV2
2
Charging points
EV purchasers Collection. Sum EV adopters
E-mobility innovation system. Entrepreneurial activities
3
E-mobility
innovation system
EV purchasers Collection. Sum EV adopters
EV related subsidies. Total subsidies for EVs
4
EV related subsidies
EV purchasers Collection. Sum EV adopters
Charging points. Installed charging points
5
EV pricing
EV purchasers Collection. Sum EV adoption rate
1
This phrase indicates that ‘EV purchasers’ entity type utilizes the variable ‘EV knowledge development and diffusion’ from the ‘E-mobility innovation system’ entity type.
2
This phrase indicates that ‘EV purchasers’ entity type uses the average value of variable ‘Price value of EV’ from the ‘EV pricing’ entity type. In Ventity software, aggregate
values of the variables (e.g., Min, Max, MEAN, MEDIAN, STDEV, and SUM) in each entity type can be accessed through the related collections.
2
Appendix B: The overall variables and equations of the model
We used Ventity software to build the model. For more information on the functions used in the model, see the following link: http://ventity.biz/
For complementary information about the variables and the equations of each entity type, see the related section in the paper.
Element
(Variable)
Element
Type
Expression (Equation)
Note
Entity Type: Model
Final Time
Auxiliary
360
We ran the model for 30 years.
Initial Time
Auxiliary
0
2016 is the earliest time that we have relevant
information on the e-mobility development for the
Netherlands.
Time
Auxiliary
1
Time Step
Auxiliary
0.0625
Entity Type: EV Related Subsidies
Cars per adopter
Auxiliary
1
We assume that every person buys only one car.
Sold cars
Auxiliary
Cars per adopter*EV purchasers
Collection.Sum EV adopters
Subsidy per charging point
Auxiliary
1000
The cost of a public EV charging station can vary
significantly based on the requirements and existing
electrical infrastructure, but we considered an
average of 1,000 Euros.
Subsidy per EV
Auxiliary
TableFunction1(Sold cars)
TableFunction1
Table
Function
([0,8200000],[0,5000]) [(0,5000),
(412020.95,4983.56), (681064.48,4886.88),
(1065412.37,4790.2), (1257586.32,4693.51),
(1500000,4500), (2000000,4000),
(2466006.12,3412.45), (2794066.7,2784.01),
(3273539.85,1913.85), (3853954.72,1357.92),
(4409134.15,971.19), (4989549.02,560.28),
(5771847.32,318.57), (6579381.04,173.55),
(8200000,0)]
As we mentioned in the paper, we assume that the
government offers a changeable direct subsidy to the
purchasers (5000 to 0 Euros as EV adoption
increases over time). As such, we assume that after a
period, direct subsidy of the government will
decrease till approach zero. This is because e-
mobility will be more affordable for the purchasers
and as such there is no need to pay subsidy.
Total subsidies for the
purchase of EVs
Auxiliary
Sold cars*Subsidy per EV
3
Element
(Variable)
Element
Type
Expression (Equation)
Note
Total subsidies for
charging points
Auxiliary
Subsidy per charging point *Charging
point.Installed charging points
Total subsidies for EVs
Auxiliary
Total subsidies for the purchase of EVs +
Total subsidies for charging points
Entity Type: EV Pricing
Cost reduction per
doubling of Experience
Auxiliary
0.3
Heuristics (see e.g., Argote, 1996; 2013; Morrison,
2008)
Initial Experience
Auxiliary
90275
This is the same as the number of EVs at the initial
time (RVO, 2017)
Effect of EVA on L
Auxiliary
1
Effect of EV adoption on learning
Effect of experience on
price
Auxiliary
(Cumulative experience/Initial
Experience)^(Strengths of learning curve)
See Sterman (2000: 338)
Expected annual growth of
EV price
Auxiliary
(- 0.005)/12
We assume that EV price decreases by a steady rate
of (0.005) every year (rule of thumb).
Expected annual growth of
ICEV price
Auxiliary
0.01/12
We assume that ICEV price increases by a steady
rate of (0.01) every year (rule of thumb).
Expected annual growth of
Tax on ICEV
Auxiliary
0.2/12
We assume that Tax on ICEV increases by a steady
rate of (0.02) every year (rule of thumb).
Initial EV price
Auxiliary
30000
We assume that EV price is 10,000 Euros more
expensive than ICEVs in the same model (see e.g.,
Axsen et al., 2016).
Initial ICEV price
Auxiliary
20000
Initial tax on ICEV
Auxiliary
75
We assume that initial average tax for ICEV is 75
Euros
Price of EV
Auxiliary
Initial EV price * ((Effect of experience on
price) + ((1+ Expected annual growth of EV
price)^ Model.Time)) - EV related
subsidies.Subsidy per EV
See the ‘EV pricing entity type’ in the paper.
Price of ICEV
Auxiliary
(Initial ICEV price * ((1 + Expected annual
growth of ICEV price)^Model.Time))+ Tax
on ICEV
See the ‘EV pricing entity type’ in the paper
Price value of EV
Auxiliary
Price of ICEV/Price of EV
Strengths of learning curve
Auxiliary
log(1 - cost reduction per doubling of
experience)/log(2)
See Sterman (2000: 338)
Tax on ICEV
Auxiliary
Initial tax on ICEV *(1 +Expected annual
growth of Tax on ICEV)^Model.Time
See the ‘EV pricing entity type’ in the paper
4
Element
(Variable)
Element
Type
Expression (Equation)
Note
Time to forget
Auxiliary
5*12
Heuristics (see e.g., Argote, 1996; 2013; Morrison,
2008)
Cumulative experience
Stock
INTEG (Learning Forgetting)
[Default initial value = Initial Experience]
See Sterman (2000: 338)
Learning
Flow
EV purchasers Collection.Sum EV adoption
rate*Effect of EVA on L
See Sterman (2000: 338)
Forgetting
Flow
Cumulative experience/Time to forget
See Morrison (2008)
EV Pricing[]
Average Price value of EV
Aggregate
Average
This variable calculates the average prices of EV
Entity Type: EV Purchasers
Adoption fraction
Auxiliary
RandomUniform(0.0001,0.0005)
This variable denotes the proportion of contacts that
are sufficiently persuasive (if positive WoM) or
incredible (if negative WoM) to induce the potential
adopter to purchase/or not purchase an EV. In this
model we assume that positive and negative WOMs
have the same strength (see e.g., Berger and
Milkman, 2012; East et al., 2008; Fiedler, 2007;
Mizerski, 1982; Park and Lee, 2009).
Using RandomUniform we specify a min (0.0001)
and a max (0.0005) for different contacts that might
be happened within the population.
Adoption from WoM
Auxiliary
Contact rate * Adoption fraction * ((EV
purchasers Collection.Sum EV adopters)-(EV
purchasers Collection.Sum dissatisfied
adopters))*(EV purchasers Collection.Sum
Potential adopters of EV/Initial potential
adopters)
Adoption from Word of Mouth: as we argued in
chapter 4, we formulate both negative and positive
WoM as the recommendations made, subsequently,
by (dis)satisfied EV adopters to the prospective EV
purchasers (see e.g., Berger and Milkman, 2012; East
et al., 2008; Fiedler, 2007; Mizerski, 1982; Park and
Lee, 2009).
Advertising effectiveness
Auxiliary
0.01
This constant represents how successful the
advertising of EVs is in communicating the message
whether potential EV adopters saw or liked it, or be
motivated to think of EV adoption (rule of thumb).
AT delivering EV
Auxiliary
5
The average time required for delivering an EV to a
purchaser (rule of thumb).
5
Element
(Variable)
Element
Type
Expression (Equation)
Note
AT DoP
Auxiliary
2*12
Determines how rapidlyDesirability of EVs price
for purchasers’ will be perceived (rule of thumb).
AT EV purchase intention
Auxiliary
5
Adjustment time to decide (intent) an EV adoption
(rule of thumb).
AT EVP
Auxiliary
12
The average time required for manufacturers to
produce EV based on the demand (rule of thumb).
AT FCE
Auxiliary
4*12
Determines how rapidly available facilitating
conditions are perceived by potential purchasers (rule
of thumb).
AT HE
Auxiliary
2*12
The adjustment time determines how rapidly
available hedonic benefits of EVs are perceived by
potential purchasers (rule of thumb).
AT KH
Auxiliary
5*12
The average time required for ‘EV knowledge
development and diffusion’ to convert into ‘Applied
knowledge’ (rule of thumb).
AT PE
Auxiliary
2*12
Determines how rapidly ‘performance of EVs’ and
‘degree of ease associated with EV adoption’ are
perceived by potential purchasers (rule of thumb).
Average EV performance
Auxiliary
Effect of EV KD and Dif on
performance*Applied knowledge
Average lifetime of EV
Auxiliary
8*12
See e.g., Struben and Sterman (2008)
Car demands
Auxiliary
EV purchasers Collection.Sum Intent to buy
EV *Cars per adopter
Cars per adopter
Auxiliary
1
We assume that every person buys only one car
Contact rate
Auxiliary
randomuniform(0,15)
The monthly rate, with which potential adopters
come into contact with adopters.
Using RandomUniform we specify a min (0) and a
max (15) as a monthly rate, with which various
potential adopters come into contact with adopters.
Effect of AEVs P on HM
Auxiliary
0.75
We assume that 0.75 of hedonic benefits can be
conveyed to the purchasers (rule of thumb).
Effect of AEVS P on PEE
Auxiliary
0.75
We assume that 0.75 of performance can be
conveyed to the purchasers (rule of thumb).
6
Element
(Variable)
Element
Type
Expression (Equation)
Note
Effect of DoP on IRPE
Auxiliary
0.3
For the base run, we assume that potential purchasers
consider (0.3) as weight of importance for the
‘Desirability of EVs price’ on ‘Intention rate to
purchase EV’ (rule of thumb).
Effect of EV KD and Dif
on H benefits
Auxiliary
0.25
We assume that 0.25 of applied EV knowledge are
allocated to hedonic benefits (rule of thumb).
Effect of EV KD and Dif
on performance
Auxiliary
0.75
We assume that 0.75 of applied EV knowledge are
dedicated to performance of EVs (rule of thumb).
Effect of FC on IRPE
Auxiliary
0.3
For the base run, we assume that potential purchasers
consider (0.3) as weight of importance for the
‘Facilitating condition expectancy’ on ‘Intention rate
to purchase EV’ (rule of thumb).
effect of HE on IRPE
Auxiliary
0.1
For the base run, we assume that potential purchasers
consider (0.1) as weight of importance for their
‘Hedonic expectancy’ on ‘Intention rate to purchase
EV’ (rule of thumb).
Effect of PEE on IRPE
Auxiliary
0.3
For the base run, we assume that potential purchasers
consider (0.3) as weight of importance for their
‘Performance and effort expectancy’ on ‘Intention
rate to purchase EV’(rule of thumb).
EV stock shortfall
Auxiliary
Car demands-EVs in stock
First time intended
Auxiliary
0
In the base run, we assume that whole population
think of EV as a car that they can adopt.
Hedonic benefits of EV
Auxiliary
Applied knowledge*Effect of EV KD and Dif
on H benefits
Ignoring rate
Auxiliary
0.001
We assume that every month, (0.01%) of EV
adopters stop EV adoption. This might be happened
because of their unpleasant experience of EV
adoption (rule of thumb).
Initial potential adopters
Auxiliary
8200000
Intention rate to purchase
EV
Auxiliary
((Effect of FC on IRPE *Facilitating condition
expectancy) + (Effect of PEE on
IRPE*Performance and effort expectancy) +
(effect of DoP on IRPE* Desirability of EVs
price for purchasers) + (effect of HE on IRPE
* Hedonic expectancy))
See the ‘EV purchasers entity type’ in the paper.
7
Element
(Variable)
Element
Type
Expression (Equation)
Note
Percentages of the total
population
Auxiliary
1
For the base run, we assume that the whole initial
population are potential adopters. As such, we set
this variable as 1. However, in the scenario analysis,
we assume that different classes of potential EV
adopters constitutes a subset of whole population.
The effect of EA on EV
production
Auxiliary
0.35
We assume that 35% of entrepreneurial activities will
be dedicated to EV production (rule of thumb).
Time to become neutral
Auxiliary
4*12
The average time that dissatisfied EV adopters might
become neutral about their unpleasant experience of
EV adoption (rule of thumb).
Applied knowledge
Stock
INTEG(Knowledge application rate)
[Default initial value = 0]
Knowledge that can be applied in car industry
(translated in terms of performance and hedonic
benefits)
Knowledge application
rate
Flow
Conversion rate *(Emobility Innovation
System.EV knowledge development and
diffusion - Applied knowledge) / AT KH
The rate that (theoretical) knowledge will be applied
in car industry (translated in terms of performance
and hedonic benefits)
Desirability of EVs price
for purchasers
Stock
INTEG(Change in DPP)
[Default initial value = 0]
This formulation returns a number between (0, 1)
Change in DPP
Flow
(1-Desirability of EVs price for purchasers)*
EVpricing.Average Price value of EV/ AT
DoP
Change in Desirability of EVs price for purchasers
Dissatisfied adopters
Stock
INTEG(Stop EV adoption rate- Becoming
neutral)
[Default initial value = 0]
Stop EV adoption rate
Flow
EV adopters * Ignoring rate
Becoming neutral
Flow
Dissatisfied adopters/Time to become neutral
EV adopters
Stock
INTEG(EV adoption rate - Attrition rate -
Stop EV adoption rate)
[Default initial value = 90275]
(RVO, 2017)
EV adoption rate
Flow
Min(Intent to buy EV,EVs in stock)/AT
delivering EV
Attrition rate
Flow
EV adopters/Average lifetime of EV
Stop EV adoption rate
Flow
EV adopters * Ignoring rate
EVs in stock
Stock
INTEG(EV production rate)
[Default initial value = 0]
8
Element
(Variable)
Element
Type
Expression (Equation)
Note
EV production rate
Flow
(Emobility Innovation System.Entrepreneurial
activities) * the Effect of EA on EV
production*EV stock shortfall/AT EVP
Facilitating condition
expectancy
Stock
INTEG(Change in FCE)
[Default initial value = 0]
This formulation returns a number between (0, 1)
Change in FCE
Flow
(1+ Advertising effectiveness) *(Charging
point.Facilitating conditions)* (1-Facilitating
condition expectancy)/AT FCE
Change in ‘Facilitating condition expectancy’
Hedonic expectancy
Stock
INTEG(change in HE)
[Default initial value = 0]
This formulation returns a number between (0, 1)
Change in HE
Flow
(1+ Advertising effectiveness)*(Effect of
AEVS P on HM * Average EV
performance)*(Hedonic benefits of EV)*(1-
Hedonic expectancy)/ AT HE
Change in ‘Hedonic motivation’
Intent to buy EV
Stock
INTEG(EV purchase intention rate - EV
adoption rate)
[Default initial value = 0]
EV purchase intention rate
Flow
if (model.time> First time
intended){((Potential adopters of
EV*Intention rate to purchase EV)/AT EV
purchase intention) + (Adoption from WoM)}
else {0}
EV adoption rate
Flow
Min(Intent to buy EV,EVs in stock)/AT
delivering EV
Performance and effort
expectancy
Stock
INTEG(change in PEE)
[Default initial value = 0]
This formulation returns a number between (0, 1)
Change in PEE
Flow
(1+ Advertising effectiveness) * (Effect of
AEVS P on PEE *Average EV performance)*
(1 - Performance and effort expectancy)/ AT
PE
Change in ‘Performance and effort expectancy’
Potential adopters of EV
Stock
INTEG(Becoming neutral - EV purchase
intention rate)
[Default initial value = Initial potential
adopters*Percentages of the total population]
Becoming neutral
Flow
Dissatisfied adopters/Time to become neutral
EV Purchasers[]
9
Element
(Variable)
Element
Type
Expression (Equation)
Note
Sum Dissatisfied adopters
Aggregate
Sum
Calculates sum of Dissatisfied adopters belong to
different classes of purchasers
Sum EV adopters
Aggregate
Sum
Calculates sum of EV adopters belong to different
classes of purchasers
Sum EV adoption rate
Aggregate
Sum
Calculates sum of EV adoption rate for different
classes of purchasers
Sum Intent to buy EV
Aggregate
Sum
Calculates sum of potential adopters who intends to
buy EV in different classes of purchasers
Sum Potential adopters of
EV
Aggregate
Sum
Calculates sum of Potential adopters of EV in
different classes of purchasers
Entity Type: Emobility Innovation System
AT EDI
Auxiliary
7*12
Time that Entrepreneurs might be dissuade to
continue in e-mobility system (rule of thumb).
AT EI
Auxiliary
3*12
Time that Entrepreneurs will be interested in e-
mobility after receiving signals from ‘Guidance of
the Search’ and perceiving more legitimacy of EVs
(rule of thumb).
AT EKD
Auxiliary
3*12
The time required to convert ‘resource mobilization’
to ‘EV knowledge development and diffusion’(rule
of thumb).
AT PLEM
Auxiliary
5*12
This variable represents how quickly the determined
cause variables leads to ‘Perceived legitimacy of e-
mobility’(rule of thumb).
AT RM
Auxiliary
2*12
This variable represents how quickly ‘Guidance of
the Search’ leads to ‘Resource mobilization’(rule of
thumb).
AT SSE
Auxiliary
1
Time required for sailing ship effect to adjust
TableFunction2
Table
Function
([0,8200000], [0,1]), [(0,0), (825703.24, 0.03),
(1254705.53, 0.05), (1835120.4, 0.08),
(2995950.13, 0.15), (4131544.43, 0.24) ,
(5368080.45, 0.39), (6478439.33, 0.57),
(7538327.34, 0.83), (8200000,1)]
We assume an exponential growth for this table
function.
Beliefs in EVs growth
potential
Auxiliary
(TableFunction2(EV purchasers
Collection.Sum EV adopters)+Perceived
legitimacy of emobility)/2
10
Element
(Variable)
Element
Type
Expression (Equation)
Note
Effect of experience on
Knowledge dev and dif
Auxiliary
(EV pricing.Cumulative experience/EV
pricing.Initial Experience)^(Strength of
learning curve for knowledge)
See Sterman (2000: 338)
Effect size of SSE on
legitimacy
Auxiliary
0.3
Effect size of Sailing ship effect on the legitimacy of
e-mobility (based on rule of thumb; see e.g.,
Rosenberg, 1976; Utterback, 1996; Walrave and
Raven, 2016).
TableFunction5
Table
Function
([0,8200000], [0, 1]), [(0, 0), (1002351.24,
0.03), (1910826.69, 0.06), (2743595.84, 0.11),
(3677306.71, 0.19), (4081073.58, 0.33),
(4207250.72, 0.46), (4510075.87, 0.61),
(4913842.73, 0.76), (5544728.46, 0.84),
(6402733.04, 0.9), (7210266.77, 0.96),
(8169213.07,1)]
We assume an S-shaped growth for this table
function.
EVs effect on legitimacy
Auxiliary
TableFunction5(EV purchasers
Collection.Sum EV adopters)
Guidance of the search
Auxiliary
(Beliefs in EVs growth potential + EV
knowledge development and diffusion)/2
This formulation returns a number between (0, 1)
Knowledge increase per
doubling experience
Auxiliary
0.3
Heuristics (see e.g., Argote, 1996; 2013; Morrison,
2008)
Knowledge decay time
Auxiliary
9*12
The time that ‘EV knowledge development and
diffusion’ is decayed or forgotten (Argote, 1996;
2013; Morrison, 2008).
Sailing ship effect
Auxiliary
Effect size of SSE on legitimacy*SSE
triggers*SSE limit
See the section on ‘e-mobility innovation system’
entity type in the paper.
SSE limit
Auxiliary
IfThenElse(SSE counter<SSE limit threshold,
1, 0)
Binary variable (true/false) that indicates if the SSE
is active (or not).
SSE limit threshold
Auxiliary
7*12
Sailing ship effect limit threshold: Time that sailing
ship will minimally remain active (see e.g.,
Rosenberg, 1976; Utterback, 1996; Walrave and
Raven, 2016).
Strength of learning curve
for knowledge
Auxiliary
log(1 + Knowledge increase per doubling
experience)/log(2)
See Sterman (2000: 338)
11
Element
(Variable)
Element
Type
Expression (Equation)
Note
Threshold percentages for
SSE
Auxiliary
0.15
We assume that if (15%) of potential purchasers
adopt EV, then a ‘sailing ship effect’ will be
triggered (see e.g., Rosenberg, 1976; Utterback,
1996; Walrave and Raven, 2016).
Threshold EVs adopters
Auxiliary
Total potential adopters* Threshold
percentages for SSE
Total potential adopters
Auxiliary
8200000
TableFunction1
Table
Function
([0,36576000000], [0,1]), [(0,0),
(2669978333.71, 0.15) , (6609659641.86,
0.32) ,(10549340950, 0.49) ,
(16177457104.49, 0.67), (21918135582.08,
0.79), (27996501028.93, 0.91),
(36576000000, 1)]
We considered a goal seeking behavior for this table
function. Accordingly, based on rule of thumb we
assume that the effect of subsidies on ‘resource
mobilization’ function will be decreased as e-
mobility develop.
Subsidies effect on
mobilization
Auxiliary
TableFunction1(EV related subsidies.Total
subsidies for EVs)
Entrepreneurial activities
Stock
INTEG(Entrepreneurial interest -
Entrepreneurial disinterest)
[Default initial value = 0.06]
We estimate the initial value considering the current
development of e-mobility in The Netherlands (to
estimate the values, we use, more specifically, the
recent reports on EV via the following link:
www.ieahev.org).
Entrepreneurial interest
Flow
(1 - Entrepreneurial activities) * (Guidance of
the search) * (Perceived legitimacy of
emobility) / AT EI
Entrepreneurial disinterest
Flow
Entrepreneurial activities / AT EDI
EV knowledge
development and diffusion
Stock
INTEG(Knowledge dev and dif rate -
Knowledge dev and dif decay rate)
[Default initial value = 0.004]
This formulation returns a number between (0, 1); we
estimate the initial value considering the current
development of e-mobility in The Netherlands
Knowledge dev and dif
rate
Flow
(1- EV knowledge development and diffusion)
*(Effect of experience on Knowledge dev and
dif)*(Resource mobilization)/ AT EKD
Knowledge development and diffusion rate
Knowledge dev and dif
decay rate
Flow
EV knowledge development and
diffusion/Knowledge decay time
Knowledge development and diffusion decay rate
12
Element
(Variable)
Element
Type
Expression (Equation)
Note
Perceived legitimacy of
emobility
Stock
INTEG(Changes in PL of EM)
[Default initial value = 0.03]
We estimate the initial value considering the current
development of e-mobility in The Netherlands (to
estimate the values, we use, more specifically, the
recent reports on EV via the following link:
www.ieahev.org).
Changes in PL of EM
Flow
(1-Perceived legitimacy of emobility)* ((EV
knowledge development and diffusion)+ (EVs
effect on legitimacy) - (Sailing ship
effect))/AT PLEM
Changes in ‘Perceived legitimacy of e-mobility’
Resource mobilization
Stock
INTEG(R mobilization rate)
[Default initial value = 0.001]
This formulation returns a number between (0, 1); we
estimate the initial value considering the current
development of e-mobility in The Netherlands.
R mobilization rate
Flow
(1-Resource mobilization)*(guidance of the
search) *(Subsidies effect on
mobilization)/AT RM
Resource mobilization rate (see e.g., Rosenberg,
1976; Utterback, 1996; Walrave and Raven, 2016).
SSE counter
Stock
INTEG(SSE counter rate)
[Default initial value = 0]
Sailing Ship Effect counter (see e.g., Rosenberg,
1976; Utterback, 1996; Walrave and Raven, 2016).
SSE counter rate
Flow
SSE triggers/AT SSE
Sailing Ship Effect counter rate (see e.g., Rosenberg,
1976; Utterback, 1996; Walrave and Raven, 2016).
SSE triggers
Stock
INTEG(SSE trigger rate)
[Default initial value = 0]
Sailing Ship Effect triggers (see e.g., Rosenberg,
1976; Utterback, 1996; Walrave and Raven, 2016).
SSE trigger rate
Flow
IfThenElse(EV purchasers Collection.Sum EV
adopters>threshold EVs adopters && SSE
triggers=0, 1/Model.Time Step,0)
Sailing Ship Effect trigger rate (see e.g., Rosenberg,
1976; Utterback, 1996; Walrave and Raven, 2016).
Entity Type: Charging Point
Actual capacity of
charging points
Auxiliary
EV purchasers Collection.Sum EV
adopters/Installed charging points
AT CPP
Auxiliary
1.5
We assume that average time required to produce a
public charging point after charging points shortfall
(rule of thumb).
AT ICP
Auxiliary
1.5
We assume that average delay to install a public
charging point upon the request (rule of thumb).
Average lifetime of CP
Auxiliary
15*12
We assume that average lifetime of charging points is
15 years (see e.g., Weider and Philip, 2010).
Charging points shortfall
Auxiliary
Planned capacity of CP -Installed charging
points
13
Element
(Variable)
Element
Type
Expression (Equation)
Note
Effect of EA on CPP
Auxiliary
0.35
Based on rule of thumb, we assume that 35% of
entrepreneurial activities will be dedicated to
installing charging points.
Facilitating conditions
Auxiliary
Planned relation between EV and CP/Actual
capacity of charging points
Planned capacity of CP
Auxiliary
EV purchasers Collection.Sum EV
adopters/Planned relation between EV and CP
Planned capacity of charging points
Planned relation between
EV and CP
Auxiliary
TableFunction1(EV purchasers
Collection.Sum EV adopters)
Planned relation between EV and charging point
TableFunction1
Table
Function
([0,8200000], [1,4]), [(0,4), (548113.52,3.51),
(1078057.53,3.03), (1633236.97,2.61),
(2264122.69,2.18), (3122127.28,1.8),
(4131544.43,1.5), (5166197.02,1.23),
(6402733.04,1.1), (8194448.5,1)]
We posit that the desired relation between the
number of EVs and charging points will be decreased
from four to one as the number of ‘EV adopters’
increases (see e.g., Shulock et al., 2011).
CP in stock
Stock
INTEG(CP production rate- Net installing CP)
[Default initial value = 0]
Charging points in stock
CP production rate
Flow
Charging points shortfall*Effect of EA on
CPP*Emobility Innovation
System.Entrepreneurial activities/AT CPP
Charging point production rate
Net installing CP
Flow
CP on stock/AT ICP
Net installing charging point
Installed charging points
Stock
INTEG(Net installing CP- Scrap rate of CP)
[Default initial value = 17786]
(RVO, 2017)
Net installing CP
Flow
CP on stock/AT ICP
Net installing charging point
Scrap rate of CP
Flow
Installed charging points/Average lifetime of
CP
Scrap rate of charging point
14
Appendix C: The stylized overview of the entity types
Figure C.1. Feedback loops in E-Mobility Innovation System entity type.
15
Figure C.2. Feedback loops in Charging Point entity type.
16
Figure C.3. Feedback loops in EV Pricing entity type.
Figure C.4. Feedback loops in EV Related Subsidies entity type.
17
Figure C.5. Feedback loops in EV Purchasers entity type.
18
Appendix D: Initialization of the UTAUT variabless for different scenarios
Following table provides more details about the scenarios which are illustrated in ‘experimental
setup’. The assumed weights of UTAUT variables, initial values of EV adopters, and the first time that
each class of EV purchasers intended to buy EV follow from the logic described for each scenario. The
percentage of the total population for each class of EV purchasers are derived from Maltha et al.
(2017) and Bockarjova and Steg (2014).
Scenario
Class of EV
purchasers
Definition
Initial value
of EV
adopters
Percentage
of the total
population
First time
intended to
buy EV
Performance
& effort
weight
Hedonic
weight
Price
weight
Facilitating
conditions
weight
Scenario 2
Aggregated purchasers
90275
1
0
0,275
0,025
0,4
0,3
Scenario 2-1
Income1
0-10,000 Euros
0
0,024
264
0,29
0,05
0,35
0,31
Income2
10,000-20,000 Euros
0
0,12
216
0,3
0,075
0,3
0,325
Income3
20,000-30,000 Euros
0
0,227
124
0,315
0,1
0,25
0,335
Income4
30,000-40,000 Euros
0
0,214
96
0,4
0,15
0,1
0,35
Income5
40,000-50,000 Euros
0
0,167
36
0,6
0,2
0
0,2
Income6
50,000+ Euros
90275
0,248
0
0,275
0,025
0,4
0,3
Scenario 2-2
Age1
18-19
22569
0,003
0
0,25
0,2
0,25
0,3
Age2
20-29
22569
0,073
0
0,3
0,175
0,25
0,275
Age3
30-39
22569
0,14
0
0,35
0,15
0,25
0,25
Age4
40-49
22569
0,199
0
0,4
0,125
0,25
0,225
Age5
50-64
0
0,315
124
0,45
0,1
0,25
0,2
Age6
65-74
0
0,17
216
0,5
0,075
0,25
0,175
Age 7
75+
0
0,101
264
0,55
0,05
0,25
0,15
Scenario 2-3
Urbanization1
Very high density
60183
0,152
0
0,2
0,03
0,57
0,2
Urbanization2
High density
30091
0,246
0
0,225
0,05
0,5
0,225
Urbanization3
Moderately high density
0
0,2
96
0,28
0,07
0,4
0,25
Urbanization4
Low density
0
0,248
124
0,3
0,1
0,3
0,3
Urbanization5
Very low density
0
0,155
216
0,33
0,12
0,2
0,35
19
Appendix E: Credibility and quality of the model
Testing SD models is a continuous process that is essentially starts from the early stages of modeling
(Barlas and Carpenter, 1990). In this regard, a wide variety of procedures has been conducted to
uncover model flaws and assure sufficient confidence in the structure
3
and associated behavior
4
(e.g.,
Sterman, 2000; Barlas, 1996; Coyle and Exelby, 2000).
Structural validation was performed through unit checks, structural verification and extreme
condition analysis.
In this regard, first the unit names of the variables have been checked in order to see whether
any variable is missing in the equations, and to assure dimensional consistency in the
equations (see the unit names of the variables in Appendix B). Accordingly, we made a few
revisions in the structure of the model to correct the unit errors.
Structural verification addresses whether the model structure is consistent with the
descriptive knowledge about the real system. In this paper, the constructs used in the all
entity types are drawn from the theoretical frameworks and sensible empirical insights.
Therefore, the model can be considered, structurally, verified (see section 2 and section 3.1 of
the paper).
Extreme condition analysis evaluates whether the parameters in the model behave
appropriately under extreme conditions. In this regard, we assessed whether any parameter
takes on values outside the prescribed limits. We also examined the response of the
endogenous variable to changes in the inputs. Accordingly, behaviors that do not make sense
in the extreme points lead to more introspective evaluation of the equations. For instance, we
used the following equation for EV adoption rate:
EV adoption rate= Min(Intent to buy EV, EVs in stock / Cars per adopters ) / AT delivering EV
As such, the model is also robust in case of no EVs in stock, because no EV will be also
adopted accordingly.
Furthermore, in case of no population (no potential adopters), no EV adoption will be
occurred. In addition, in case of no car demand, there will not be any EV in stock, and EV
adopters, and subsequently installed charging points, will be approached to zero eventually
(see Figure E.1). Overall, the model passed this analysis.
3
A structural validation seeks to determine if the model reflects the real world accurately.
4
Behavioral assessment focuses on the model behavior during execution, and assesses the degree of confidence
that can be placed in the results
20
Figure E.1. The outputs of extreme conditions test in case of no (new) car demand.
0
100000
0
15
30
45
60
75
90
105
120
135
150
165
180
195
210
225
240
255
270
285
300
315
330
345
EV adopters (Persons)
Time (Months)
00.0625 0.125 0.1875 0.25 0.3125
0.375 0.4375 0.5 0.5625 0.625 0.6875
0.75 0.8125 0.875 0.9375 11.0625
1.125 1.1875 1.25 1.3125 1.375 1.4375
1.5 1.5625 1.625 1.6875 1.75 1.8125
1.875 1.9375 22.0625 2.125 2.1875
2.25 2.3125 2.375 2.4375 2.5 2.5625
0
2000
4000
6000
8000
10000
12000
14000
16000
18000
20000
0
15
30
45
60
75
90
105
120
135
150
165
180
195
210
225
240
255
270
285
300
315
330
345
360
Charging Points
Time (Months)
0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1
0
15
30
45
60
75
90
105
120
135
150
165
180
195
210
225
240
255
270
285
300
315
330
345
EVs in Stock (Cars)
Time (Months)
21
Behavioral assessment requires running the entire model and examination of the results.
First, parameters are varied systematically to assess whether the entire model is functioning
as expected. In addition, the behavior of individual variables is followed over time, and
analyzed for any aberrant behavior. Accordingly, a closer examination of the unexpected
behaviors have been conducted, and required recalibration and restructuring of the model
have been done.
We, also, halve the time step (0.0625), and make sure that the model behavior does not
change, accordingly.
To further validate the model settings, a sensitivity analysis is also conducted. In this respect,
we investigated whether model output changes substantially when the assumptions
(parameter values) are varied over a plausible range of uncertainty (Sterman, 2000). More
specifically, we concentrated on ‘The effect of EA on EV production’, ‘Effect of EA on CPP’,
‘Cost reduction per doubling of experience’, ‘Effect size of SSE on legitimacy’. As such, these
parameters are varied between (-/+ 5%) while the other variables are kept constant. The
results illustrates that the number of EV adopters will not be changed (significantly)
comparing to the ‘base case (i.e., Scenario 2)’, within the specified ranges of these variables
(see Figures E.2 to E.6). We also consider the simultaneous variation of these parameters
between these ranges. However, the number of EV adopters does not change significantly
(see Figure E.7). Therefore, the model can be considered robust.
Hence, we also formulated the variables of adoption fractionand contact ratebased on a
random function. This function made it possible to consider various random numbers within
a numerical range. However, these results also show that the outputs do not change
meaningfully within the specified ranges.
adoption fraction= RandomUniform (0.0001,0.0005);
Contact rate = randomuniform(0,15)
All in all, we can conclude that the overall model settings, the structure and the corresponding
behaviors are credible in line with the research problem and the specified purpose of modeling and
simulation.
22
Figure E.2. Sensitivity analysis of ‘EV adopters’ based on (-/+ 5%) changes in the value of ‘The effect of
EA on EV production’ that is 0.35; this value has been varied between [0.3325, 0.3675].
Figure E.3. Sensitivity analysis of ‘EV adopters’ based on (-/+ 5%) changes in the value of ‘Effect of EA
on CPP’ that is 0.35; this value has been varied between [0.3325, 0.3675].
0
1,0 M
2,0 M
3,0 M
4,0 M
5,0 M
6,0 M
7,0 M
8,0 M
0
15
30
45
60
75
90
105
120
135
150
165
180
195
210
225
240
255
270
285
300
315
330
345
EV Adopters (Persons)
Time (Months)
EV adopters in case of 'The effect of EA on EV production=0.3325'
EV adopters in case of 'The effect of EA on EV production=0.3675'
EV adopters in case of 'The effect of EA on EV production=0.35'
0
1M
2M
3M
4M
5M
6M
7M
8M
0
15
30
45
60
75
90
105
120
135
150
165
180
195
210
225
240
255
270
285
300
315
330
345
EV Adopters (Persons)
Time (Months)
EV adopters in case of 'Effect of EA on CPP=0.3325'
EV adopters in case of 'Effect of EA on CPP= 0.3675'
EV adopters in case of 'Effect of EA on CPP= 0.35'
23
Figure E.4. Sensitivity analysis of ‘EV adopters’ based on (-/+ 5%) changes in the value of ‘Cost
reduction per doubling of experience’ that is 0.3; this value has been varied between [0.285, 0.315].
Figure E.5. Sensitivity analysis of ‘perceived legitimacy of e-mobility’ based on (-/+ 5%) changes in the
value of ‘Effect size of SSE on legitimacy’ that is 0.3; this value has been varied between [0.285, 0.315].
0
1M
2M
3M
4M
5M
6M
7M
8M
0
15
30
45
60
75
90
105
120
135
150
165
180
195
210
225
240
255
270
285
300
315
330
345
EV Adopters (Persons)
Time (Months)
EV adopters in case of ‘Cost reduction per doubling of experience=0.285’
EV adopters in case of ‘Cost reduction per doubling of experience=0.315’
EV adopters in case of ‘Cost reduction per doubling of experience=0.3’
-0,2
0
0,2
0,4
0,6
0,8
1
1,2
0
15
30
45
60
75
90
105
120
135
150
165
180
195
210
225
240
255
270
285
300
315
330
345
360
Perceived Legitimacy of EVs
Time (Months)
Perceived legitimacy of EVs in case of ‘Effect size of SSE on legitimacy= 0.315’
Perceived legitimacy of EVs in case of ‘Effect size of SSE on legitimacy= 0.285’
Perceived legitimacy of EVs in case ‘Effect size of SSE on legitimacy=0.3’
24
Figure E.6. Sensitivity analysis of ‘EV adopters’ based on (-/+ 5%) changes in the value of ‘Effect size of
SSE on legitimacy’ that is 0.3; this value has been varied between [0.285, 0.315].
0
1 M
2 M
3 M
4 M
5 M
6 M
7 M
8 M
0
15
30
45
60
75
90
105
120
135
150
165
180
195
210
225
240
255
270
285
300
315
330
345
EV Adopters (Persons)
Time (Months)
EV adopters in case of 'Effect size of SSE on legitimacy=0.315’
EV adopters in case of 'Effect size of SSE on legitimacy=0.285’
EV adopters in case of 'Effect size of SSE on legitimacy= 0.3'
25
0
1M
2M
3M
4M
5M
6M
7M
8M
0
15
30
45
60
75
90
105
120
135
150
165
180
195
210
225
240
255
270
285
300
315
330
345
EV Adopters (Persons)
Time (Months)
EV adopters in case of the min values of the selected variables
EV adopters in case of the max values of the selected variables
EV adopters in case of the values of the selected variables in Scenario 2
Figure E.7. Sensitivity analysis of ‘EV adopters’ based on the simultaneous (-/+ 5%) changes in the values of the
following variables:
* The value of ‘The effect of EA on EV production’ is 0.35; this value has been varied between [0.3325, 0.3675];
* The value of ‘Effect of EA on CPP’ is 0.35; this value has been varied between [0.3325, 0.3675];
* The value of ‘Cost reduction per doubling of experience’ is 0.3; this value has been varied between [0.285, 0.315];
* The value of ‘Effect size of SSE on legitimacy’ is 0.3; this value has been varied between [0.285, 0.315].
26
References
1. Argote, L., 1996. Organizational learning curves: Persistence, transfer and turnover. International
Journal of Technology Management 11 (7-8), 759769.
2. Argote, L., 2013. Organizational Learning: Creating, Retaining and Transferring Knowledge (2nd ed.).
Springer, New York.
3. Barlas, Y., 1996. Formal aspects of model validity and validation in system dynamics. System Dynamics
Review 12 (3), 183210.
4. Barlas, Y., Carpenter,s S., 1990. Philosophical roots of model validation: Two paradigms. System
Dynamics Review 6 (2), 148166.
5. Berger, J., Milkman, K.L., 2012. What makes online content viral? Journal of Marketing Research 49 (2),
192205.
6. Bockarjova, M., Steg, L., 2014. Can Protection Motivation Theory predict pro-environmental behavior?
Explaining the adoption of electric vehicles in the Netherlands. Global environmental change 28, 276-
288.
7. Coyle, J.M., Exelby, D., Holt, J., 1999. System dynamics in defence analysis: Some case studies.
Journal of the Operational Research Society 50 (4), 372382.
8. East, R., Hammond, K., Lomax, W., 2008. Measuring the impact of positive and negative word of mouth
on brand purchase probability. International Journal of Research in Marketing 25 (3), 215224.
9. Fiedler, K. (Ed), 2007. Social communication: introduction and overview. Psychology Press, New York.
10. Maltha, Y., Kroesen, M., Van Wee, B., & van Daalen, E., 2017. Changing influence of factors explaining
household car ownership levels in The Netherlands. Transportation Research Record 2666 (1), 103-111.
11. Mizerski, R.W., 1982. An attribution explanation of the disproportionate influence of unfavorable
information. Journal of Consumer Research 9 (3), 301310.
12. Morrison, J.B., 2008. Putting the learning curve in context. Journal of Business Research 61 (11), 1182
1190.
13. Park, C., Lee, T.M., 2009. Antecedents of online reviews' usage and purchase influence: An empirical
comparison of U.S. and Korean consumers. Journal of Interactive Marketing 23 (4), 332340.
14. Rosenberg, N., 1976. Perspectives on technology. Cambridge University Press, Cambridge.
15. RVO (Rijksdienst voor Ondernemend Nederland: The Netherlands Enterprise Agency)., 2017. Electric
transport in the Netherlands; 2016 highlights. (downloaded on 1 May 2018 from
https://www.rvo.nl/sites/default/files/2017/04/Highlights-2016-Electric-transport-in-the-Netherlands-
RVO.nl_.pdf)
16. Shulock, C., Pike, E., Lloyd, A., Rose, R., 2011. Task 4 report: complementary policies. Vehicle
Electrification Policy Study. The International Council on Clean Transportation (downloaded on 1 May
2018 from
https://www.theicct.org/sites/default/files/publications/ICCT_VEPstudy_Mar2011_no4.pdf).
17. Sterman, J.D., 2000. Business Dynamics: Systems Thinking and Modeling for a Complex World.
Irwin/McGraw-Hill, Boston.
18. Struben, J., Sterman, J.D., 2008. Transition challenges for alternative fuel vehicle and transportation
systems. Environment and Planning B: Planning and Design 35 (6), 10701097.
19. Utterback, J.M., 1996. Mastering the Dynamics of Innovation. Harvard Business School Press, Boston,
Massachusetts.
20. Walrave, B., Raven, R., 2016. Modelling the dynamics of technological innovation systems. Research
Policy 45 (9), 18331844.
21. Wiederer, A., Philip, R., 2010. Policy options for electric vehicle charging infrastructure in c40 cities.
Report for Stephen Crolius, DirectorTransportation, Clinton Climate Initiative, Harvard Kennedy
School. (downloaded on 1 May 2018 from
https://www.innovations.harvard.edu/sites/default/files/1108934.pdf )
ResearchGate has not been able to resolve any citations for this publication.
ResearchGate has not been able to resolve any references for this publication.