Conference PaperPDF Available

System dynamics modelling for electric and hybrid commercial vehicles adoption

Authors:

Abstract and Figures

Problems caused by the increasing freight transportation demand in cities call for integrated solutions where all stakeholders’ efforts are coordinated, in order to both reduce the negative impacts of freight transportation, such as pollution and congestion, and carry no disadvantages to public and private operators. Among the solutions that can be implemented for these purposes, one of the most studied and applied one is the partial or total substitution of commercial vans with low emission vehicles. Previous studies have been focused mainly on the vehicle-related factors that make such adoption sustainable for the private stakeholders. However, there is a lack of contributions that also take into account the operational aspects of a city logistics system. In order to contribute to this literature, our work develops a System Dynamics model that assesses the adoption of low emission vehicles by analysing the most important operational factors typical of a freight distribution system. Results of the simulation and the sensitivity analyses demonstrate that the adoption of low emission commercial vehicles is feasible within a reasonable time period if some strategies are put in place. For instance, public contribution including both incentives to low emission vehicles and disincentives to traditional ones could effectively increase the adoption process, along with effective advertising campaigns about the operational benefits given by such distribution model.
Content may be subject to copyright.
System dynamics modelling for electric and hybrid commercial vehicles
adoption
ANNA CORINNA CAGLIANO, ANTONIO CARLIN, GIULIO MANGANO, GIOVANNI
ZENEZINI
Department of Management and Production Engineering
Politecnico di Torino
Corso Duca degli Abruzzi 24, 10128, Torino
ITALY
anna.cagliano@polito.it, antonio.carlin@polito.it, giulio.mangano@polito.it,
giovanni.zenezini@polito.it
http://www.reslog.polito.it
Abstract: - Problems caused by the increasing freight transportation demand in cities call for integrated solutions
where all stakeholders’ efforts are coordinated, in order to both reduce the negative impacts of freight
transportation, such as pollution and congestion, and carry no disadvantages to public and private operators.
Among the solutions that can be implemented for these purposes, one of the most studied and applied one is the
partial or total substitution of commercial vans with low emission vehicles. Previous studies have been focused
mainly on the vehicle-related factors that make such adoption sustainable for the private stakeholders. However,
there is a lack of contributions that also take into account the operational aspects of a city logistics system. In
order to contribute to this literature, our work develops a System Dynamics model that assesses the adoption of
low emission vehicles by analysing the most important operational factors typical of a freight distribution system.
Results of the simulation and the sensitivity analyses demonstrate that the adoption of low emission commercial
vehicles is feasible within a reasonable time period if some strategies are put in place. For instance, public
contribution including both incentives to low emission vehicles and disincentives to traditional ones could
effectively increase the adoption process, along with effective advertising campaigns about the operational
benefits given by such distribution model.
Key-Words: - System Dynamics; Diffusion model; Hybrid and electric commercial vehicles; Sustainable
City Logistics; sensitivity analysis
1 Introduction
In recent years, problems caused by the increasing
freight transportation demand within cities, such as
pollution and congestion, have led both researchers
and public authorities to concentrate their efforts on
City Logistics (CL) initiatives. CL fosters the
development of integrated logistics systems, where
all the stakeholders are coordinated so to reduce
negative impacts on citizens. In this sense, a CL
model should be planned and managed with the aim
of improving the quality of life of communities,
while at the same time carrying no disadvantages to
both public and private operators. In literature, there
is a substantial amount of works focusing on the
positive and negative impacts of urban freight
distribution from an operational and economic point
of view, taking into account the effects on both public
and private stakeholders [1], [2].
Several policies can be implemented to reduce the
negative impacts of CL. For instance, restricting or
even banning commercial vehicles from circulating
in city centres might improve quality of life of
citizens in a considerable way [3]. Other policies
include for instance the installation of Intelligent
Transportation Systems (ITS) for monitoring road
traffic information, the use of reserved lanes for
goods vehicles, load factor control and road pricing
(e.g. congestion charge) for charging the entrance in
restricted areas [4], [5].
CL initiatives often include also the partial or total
substitution of existing commercial vehicles with low
emission ones, mainly electric or hybrid vehicles.
However, in order to make these initiatives
sustainable for private stakeholders, such as logistics
service providers and other freight carriers, it is
necessary to deeply understand the main factors for
the adoption of these kind of vehicles. In such a
context, [6] investigate the case of Amsterdam and
notice that the efforts made by the municipality to
stimulate the diffusion of electric vehicles, even by
ceasing the incentives programme for Euro 6
vehicles, are not necessarily backed up by the private
companies that have adopted both the types of
vehicles (i.e. Euro 6 and electric). This behaviour is
a consequence of the technological gap that exists
between Euro 6 vehicles, which by the way have a
low environmental impact themselves, and electric
vehicles. Moreover, [7] show that investing in
electric commercial vehicles turns profitable only
under certain operational conditions. In particular,
they find that the most profitable strategy would be
to purchase the vehicle while renting the batteries; in
this case in fact the initial high cost of batteries does
not counterweight the advantage of having lower
variable costs for operating the vehicle.
In order to contribute to the existing body of
literature on the factors for adopting low emission
vehicles, we propose a model that assess the diffusion
of a CL system based on electric and hybrid vehicles
in the city of Torino (Italy), by taking into account all
the typical operational factors of a freight distribution
system. As a matter of fact, current literature on CL
lacks studies that analyse the diffusion of low
emission commercial vehicles by focusing on the
operational aspects of the associated logistics
systems. We compare economic and environmental
costs and benefits of the proposed system with the
existing CL system, which mostly uses traditional
diesel powered vehicles. The results of the simulation
and the consequent sensitivity analysis allow us to
identify some factors that might drive the adoption
and diffusion of this distribution system.
We apply System Dynamics (SD) methodology to
develop our model given its proven ability to
represent and simulate the behaviour of systems like
CL ones. The SD approach was originally introduced
in the 1960s at the Massachusetts Institute of
Technology to study the evolution over time of
complex systems composed by numerous and
heterogeneous variables and nonlinear connections
between them [8], [9]. The variables and parameters
of the model are based on reviews of similar case
studies, interviews with the main stakeholders in the
CL system at issue, as well as detailed data on the
characteristics of the vehicles that were provided by
a main manufacturer of commercial vehicles.
The paper is structured as follows. In Section 2 we
review relevant literature in SD modelling, in order
to build significant knowledge on which aspects
should be represented in a model of a CL system. In
Section 3 we depict the main aspects of the
methodology and provide the theoretical background
for the selected diffusion model. The development of
the model is presented in Section 4 and its calibration
is proposed in Section 5. The results of the
simulations and of the sensitivity analysis are
discussed in Section 6. Then, we propose our
interpretation of the results and some policy
implications in Section 7. Finally, we draw some
conclusions and identify further research
opportunities in Section 8.
2 Literature review
We review SD models along two research fields that
are relevant for our purpose: i) traffic related issues,
such as congestion and pollution, along with
mitigating strategies and policies, and ii) adoption
factors for low emission vehicles. Our aim is twofold:
first of all to highlight the advantages and limitations
of the SD methodology to our field of research;
second, to identify the main CL variables and
relationships among them available in literature,
which will form the background for developing our
model.
Some authors have focused on traffic congestion
and on the consequent problem of polluting
emissions. [10] developed a casual loop diagram (see
Section 3) for the city of Accra, to investigate the
congestion factors and their mutual relationships,
along with the associated levels of emission. [11]
simulate the behaviour of the parameters influencing
pollution levels in Teheran and assess the
effectiveness of some environmental policies.
Among the policies investigated, the most effective
ones are deemed to be technological improvement of
vehicles and fuels and construction of public
transportation infrastructures. Several SD models
have been developed specifically with the aim of
evaluating CO2 mitigating policies and strategies.
Some models consider intercity private transport as
their study object [12], [13]. In particular, [13] focus
on American highways and test different policy
scenarios aiming at reducing CO2 emission levels.
Three policy-making strategies are investigated and
found to be effective when combined together:
increasing fuel efficiency, subsidizing the use of
public transportation, and stimulating the adoption of
electric vehicles. Strategic choices of private
stakeholders are also examined through SD models.
[14] qualitatively estimate the effectiveness of
incentives to the use of alternative fuels vehicles by
considering a timespan equal to the average vehicle
lifetime. Besides the strategic decisions of
manufacturers, the model also includes consumers’
preferences, industry dynamics, and the
environmental impacts during the life cycle of a
vehicle.
We have found a wide presence in literature of SD
models on the diffusion of low emission vehicles.
[15] focus on the adoption of low emission heavy
goods vehicles. The authors highlight the importance
of having both a potential market and an efficient
refuelling network for the adoption of such vehicles.
[16] investigate the Colombian market and show that
good communication is more effective than fiscal
policies to encourage low emission private
transportation. [17] study the diffusion and
competition between low emission vehicles, in
particular electric and hydrogen vehicles. They find
that a critical mass should exist for adopting
alternatives technologies and that this critical mass is
dependent on economic and behavioural factors.
Among them the word of mouth appears to be crucial
in order to stimulate diffusion.
Some authors have focused specifically on the
diffusion of electric and hybrid vehicles. [18] build
on the work of [17] to examine the adoption factors
for hybrid plug-in vehicels and electric vehicles in the
United Kingdom, considering a 40 year time span.
The sensitivity analysis reveals that word of mouth,
average life of the vehicles and emission rates could
influence the adoption of such vehicles more
consistently than other aspects such as incentives or
specific features of vehicles. Lastly, the model
developed by [19] takes into account fuel prices
fluctuation, incentives, network effects (e.g. word of
mouth), operational costs, and ownership costs in
order to model the adoption of light hybrid and
electric vehicles.
However, we find a lack of works that investigate
the diffusion of low emission vehicles by taking into
account the main operational factors of the CL. In
fact, SD models in this field usually focus on the
impact of policies, operating and acquisition costs of
the vehicles and other traditional adoption factors
such as word of mouth or advertising. We aim
therefore at integrating these factors together with the
aspects that define urban freight distribution systems,
such as freight demand, daily vehicle routes and
distance travelled.
3 Modelling diffusion with System
Dynamics
Several diffusion models can serve the purpose of
developing a framework for assessing and identifying
the socio-economic and cultural drivers that explain
the adoption of an innovation, such as the Gompertz
model, the logistic model, the Fisher-Pry model and
the Bass model [20], [21], [22]. Among them, the
Bass model [20] has been applied to various fields,
such as retail, industrial and consumer goods,
agriculture, education, and pharmaceutical.
Our own model is based on the SD representation
of the Bass diffusion model developed by [9], which
provides also the theoretical background for other
existing models in the CL arena, mainly aimed at
studying the adoption of low emission vehicles [17],
[23]. Moreover, the Bass model has been chosen
because of its qualities, namely simplicity and great
capacity of predicting the behaviour of a system [24].
From a methodological point of view, three main
elements compose a SD model: Causal Loop
Diagrams, Stock and Flow Diagrams, and equations
representing the relationships between the variables.
The Causal Loop Diagram (CLD) is a qualitative and
graphical representation of variables and their mutual
connections. These connections are depicted through
feedback loops, both negative (balancing) and
positive (reinforcing) ones. Feedback loops, or causal
loop, are best defined as closed sequences generated
by causes and effects triggered between variables. In
particular, reinforcing loops connect variables that
are positively linked: for each increase in one
variable within the loop, the growth generated in the
linked variables originates an additional increase in
the first variable. The opposite process happens for
balancing loops: the increase in the value of one
variable causes changes in the values of the linked
variables that then result in a decrease in the value of
the first variable. It is worth noting that CLDs do not
comprehend equations. Stock and Flow Diagrams
(SFD) are made up of four funding elements: stocks,
flows, auxiliary variables, and connectors. Stocks are
cumulated quantities given by the difference between
the inflow and the outflow of a process. They can
represent accumulations of goods, money, customer
orders, etc. over time. Flows can be physical,
economical or informational quantities that either
increase (inflows) or decrease (outflows) the value of
a stock. Auxiliary variables can be either constant or
variable over time. In the second case they are
functions of stocks, flows or other auxiliary
variables. Connectors represent the relationships
between the previous mentioned three elements.
Finally, the equations of a SD model can be either
algebraic or differential in nature, they are
independent from one another, and are functions of
the state of the system in the previous time steps.
They can define for instance the values of flows
connecting two stocks or the stock levels.
4 Model development
In the next sections we present in detail the structure
of our SD model with its main feedback loops.
It is worth mentioning that since the SD approach
does not allow flows of different elements (e.g.
different kinds of adopters) to be easily modelled and
simulated as flowing together out of the same stock
(e.g. the total number of potential adopters), we
assume that any commercial unit (C.U.), that is any
retail store operating in the city of Torino, that adopts
the new distribution system makes an exclusive
choice on the type of vehicle. For this reason, two
configurations of the model have been developed: the
first one for the adoption of electric vehicles
(variables marked with the prefix E) and the second
one for the adoption of hybrid vehicles (variables
marked with the prefix H). A second assumption has
been made on the type of adopters. In fact, we
investigated the adoption by the C.Us as a direct
consequence of the adoption by logistics providers.
Hence, the population stock of the diffusion model is
composed by the potential C.Us that could be served
by the new CL system.
For developing the model we used Vensim DSS
by Ventana Systems and we performed simulations
over a time period of 120 months, with a time step
equal to one month.
4.1 The general structure of the model
Our model presents a general structure subdivided
into four parts:
Number of vehicles in the system and associated
number of kms travelled, which are estimated on
the basis of some operational and demand factors
depicted in section 4.2.
CO2 emissions savings. We consider only CO2
emissions since the level of PM10 emissions is
significantly lower. In fact, the PM10 emissions for
traditional vehicles are on average 0.03 g/km,
while the CO2 emissions are approximately equal
to 275 g/km.
Total vehicle costs savings. They include the
acquisition cost (amortization), the fuel cost, the
maintenance cost (e.g. tire substitution) and the
insurance cost. These savings stimulate the
adoption of the new distribution system.
Charging station costs. The charging stations are
not part of a public infrastructure because we
assume that they are located within the premises
of the logistics providers or the C.U. suppliers.
The model also takes into account a possible
public contribution for purchasing the vehicles and
the charging stations. This contribution is dependent
on the savings in the level of CO2 emissions
generated by the CL system.
The dynamics of the four parts of the model are
represented via three main feedback loops which are
detailed in section 4.5. Due to space constraints the
present paper only describes the main aspects
characterising the developed SD model. The
complete model structure as well as the associated
equations are available from the authors.
4.2 The sub-models
We developed three sub-models in order to provide a
detailed and thorough representation of the general
structure of the model. The first one is named
“Electric/Hybrid TO BE” sub-model and assesses the
vehicles diffusion by comparing the new system with
the traditional one, whose operating variables are in
turn estimated in the “AS IS Model (DIESEL)” sub-
model. Then, the “C.U. adoption Electric/Hybrid”
sub-model studies the adoption process of the C.Us,
and is directly linked to the first one.
4.2.1 “Electric/Hybrid Model TO BE”
As mentioned above, this sub-model aims at
representing causes and effects that lie behind the
diffusion of electric and hybrid vehicles within the
new distribution system.
The number of vehicles depends on a variety of
factors such as:
Quantity of goods delivered, equal to the average
monthly freight demand of each C.U. multiplied
by the total number of adopters. The latter is
taken from the “C.U. Diffusion Electric/Hybrid”
sub-model.
The carrying capacity of the vehicle.
The monthly utilization factor of the vehicle,
calculated in the model as the reciprocal of the
number of monthly routes to serve the C.Us.
The increase in the number of vehicles generates
both a reinforcing and a balancing loop.
As the number of vehicles in the new distribution
system increases, the total number of kilometres they
travel increases as well. If we consider a lower
operating cost for hybrid and electric vehicles than
for traditional vehicles, we can say that for each
increase in the total amount of kilometres travelled
savings are generated in comparison with the
traditional system (from the AS IS sub-model).
Consequently, such savings generate more adoptions
of electric and hybrid vehicles, closing a reinforcing
loop.
On the contrary, the more the vehicles the more
the total investment in charging stations leading to
increased investment costs, which negatively affect
the adoption (balancing loop). The higher the initial
investment costs, for instance because of higher
acquisition costs or lower public contributions, the
higher the effect of the balancing loop and the
disincentive to the adoption of the new distribution
system.
4.2.2 “C.U. Adoption Electric/Hybrid”
This sub-model studies the dynamics of the adoption
process of the C.Us.
The diffusion sub-model is an elaboration of the
SD representation of the Bass model developed by
[9]. In our model, the adoption process takes place as
a consequence of different factors:
The advertising performed by the vehicles
themselves, which will carry a sign stating that
they are part of an eco-friendly distribution
system.
Formal advertising campaigns.
Word of mouth actions between adopters and
non-adopters.
Observation of the cost savings generated by
the new distribution system.
As a matter of fact, non-adopters are stimulated to
adopt in order to take advantage of the lower
operating costs comparing with the traditional
distribution system. In this way, they are able to offer
lower distribution fares to their customers, avoiding
the possibility of decreasing their market share
because of customers turning to the adopters of the
new CL model.
4.2.3 “AS IS Model (DIESEL)”
The present sub-model was developed to make
comparisons between the new and the traditional
logistics system. It is indeed the simplest part of the
SD model.
In each time step the operating costs of a
traditional system are calculated for the same number
of vehicles and kilometres travelled as in the TO BE
sub-model. Likewise, we calculated the taxation
costs for traditional vehicles by adding a carbon tax
and the ownership tax. Operating costs and taxation
costs makes up for the total costs of the AS IS system.
4.3 Analysis of the main feedback loops
The adoption of the new distribution system through
the obtained savings, in terms of both CO2 emissions
and operating costs, gives rise to interesting feedback
loops involving all the sub-models. For example,
Figure 1 shows the positive impact of the savings in
polluting emissions on the adoption. As Adoption
from Savings in Cost increases, the number of
adopting C.Us increases (C.U. Adoption Rate)
generating more freight and transportation demand
in the distribution system (Total C.U. Demand; Total
# Monthly New km). Consequently, when the number
of kilometres travelled increases also the value of
Total CO2 Saved grows in comparison with a
traditional distribution system, increasing in turn the
value of the variable Initial Public Contribution for
Plugin. The higher this contribution the lower the
cost carried by private operators to buy charging
stations (Plugin Total Cost) and the higher the
savings (Savings in Investment Costs and Total Cost
Savings). As a consequence of these economic
benefits generated by the positive impact of the CO2
emissions, Adoption from Savings in Cost increases,
closing a reinforcing loop.
Fig. 1: Effect of CO2 savings on adoption
Figure 2 depicts the effect on the adoption of the
variable Savings in Operating Costs. If this variable
increases, the variables Total Cost Savings as well as
Adoption from Savings in Cost increase. As
mentioned above, the total transportation demand
grows, together with the number of vehicles
necessary (# Vehicles). If logistics providers and C.U.
suppliers used the same number of traditional
vehicles to fulfil the C.U. demand (D. # Vehicles),
they would bear the costs related to the taxation of
the vehicles, which include ownership taxation and
carbon tax (D. Total Monthly Vehicle Taxes). Since
these types of taxation are not due for electric and
hybrid vehicles, the associated savings increase the
value of the total Savings in Operating Costs, closing
another reinforcing loop.
During the first months of the simulation, savings
in operating costs are lower than investment costs,
hence the sum is negative and the adoption is lagging.
As the number of C.Us and kilometres travelled
increases, then savings turn positive, stimulating the
adoption process.
Adoption from
Savings in Cost
Adoption from
Op e ra tio nal Ut ility
C.U. Adoption
Rate
Tota l C.U.
Demand
# Monthly
Routes
Total # Monthly
new km
Total M onthly
CO2 Saved
Total C O2 S aved
Init ial P ublic
Contribution for Plugin
Plugin total cost
Savings in
Investment Costs
Total C ost S avings
Total C ost
Savings (t- 1)
Adoption Fraction
from C ost Savings
+
+
+
+
+
+
+
+
-
-
+
+
+
+
Environme ntal I mpact on Ado ption
from C ost Savings
Fig. 2: Effect of savings in operating costs
5 Model calibration
In order to carry out the simulation runs, it is
necessary to provide the input values for the
parameters that contribute to define the base case for
the sensitivity analysis (Section 7). The values are
here presented for each sub-model: we defined d
them by crosschecking pertinent literature with data
coming from a van manufacturer and other logistics
operators. Also, the numerical values in the next
sections are related to parcel delivery since this is the
product category we will focus on in the subsequent
discussion of simulation and sensitivity analysis
results.
5.1 Input parameters of the sub-model
“Electric/Hybrid Model TO BE”
A first set of parameters is related to the C.Us and the
necessary routes to serve them. The value of the
variable Average Distance b/w C.U. has been
estimated higher in case of C.Us located outside the
city centre restricted area (ZTL - Zona a traffico
limitato) than for C.Us located within ZTL because
we assume a lower density of commercial
establishments outside ZTL. In particular, this
parameter is equal to 0.04 km/C.U. in ZTL and 0.9
km/C.U. outside ZTL. On the contrary, the value of
the parameter Setup Distance (average distance
travelled by the vehicle from the depot to the first
visited C.U. and from the last C.U. back to the depot)
is higher in case of ZTL than for the outside areas
since warehouses are usually located further from
city centres. Setup Distance is equal to 9.5 km when
talking about ZTL C.Us and equal to 5.5 km for C.Us
located outside ZTL. C.U. Monthly Demand is equal
to 0.44 t/C.U., C.U. Monthly Delivery Factor is a
dimensionless parameter and is equal to 4 for both
type of vehicles.
A second set of parameters relates to the features
of the vehicles. Monthly Vehicle Utilization Factor is
equal to 0.05 for electric vehicles and 0.06 for hybrid
vehicles, while the vehicle load is 1.4 t/vehicle for
both types. Operating Unit Cost is equal to 1.7 €/km
for electric vehicles and 1.6 €/km for hybrid vehicles,
and represents the operating cost of the vehicle before
public contribution. Public Contribution Factor,
meaning the contribution for the purchase of low
emission vehicles, is equal to € 0.005 for each gram
of CO2 saved.
The third set of parameters considers CO2
emissions. CO2 Emissions per km is estimated equal
to 74.7 g/km for electric vehicles and 180 g/km for
hybrid vehicles [25], while CO2 Emissions per km AS
IS is set at 356.5 g/km. In our model we included both
Well-to-Tank (WTT) and Tank-to-Wheel (TTW)
emissions.
The fourth and last set of parameters of this sub-
model is related to the plugin units:
Plugin Unit Cost: 7.000 €/unit
# Vehicle per Plugin Unit: 4 vehicles/unit
(electric) and 8 vehicles/unit (hybrid)
Public Contribution Factor for Plugin: 0,001
€/(g*unit).
5.2 Input parameters of the sub-model “C.U.
adoption Electric/Hybrid”
The input values for the diffusion model have
beenassumed to be the same for both the types of
vehicles. For the definition of some standard
parameters we refer to the Bass diffusion model
representation by [9]. These values are set
intentionally low in order not to overestimate the
impact of the parameters on the adoption process,
which would lead to unfeasible outcomes.
The parameters Contact Rate, Adoption Fraction
and Advertising Effectiveness (see [9] for the
definition) have been estimated equal to 0.1, 0.04 and
0.08 respectively (unit of measure 1/month). An
additional parameter has been introduced: Emulation
Contact Rate defines how frequently a potential
adopter observes the benefits obtained by an adopter,
and it is set equal to 0.14.
The potential C.U. adopters are equal to 2,462 for the
distribution system with electric vehicles and to
9,538 for the one based on hybrid vehicles. These
values are different because in the base case we
assume that the distribution with electric vehicles
takes place only in the ZTL while the distribution
Tota l Cos t
Savings
Tot al C os t Sa vings ( t- 1 )
Adoption Frac tion
from Co st Sa vings
Adoption from
Savings in C ost
Adoption from
Op er at ion al Utilit y
C.U. Adoption
Rate
Total C .U.
Demand
# Monthly
Routes
# M onthly
Routes(t -1)
Vehicle Input
Rate
# Vehicles
D. # Vehicles
D. Total Monthly
Vehicle Taxes
D. To tal C umulated
Mont hly Vehic le Taxe s
D. Total Vehicle
Cumulated Costs
Savings in
Operating Costs
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
Savings in O per ating C os ts
Effec t
with hybrid vehicles is adopted just by the C.Us
located outside the ZTL.
5.3 Input parameters of the sub-model AS IS
Model (DIESEL)
The variable D.Operating Unit Cost defines the
operating cost for a diesel vehicle and it is estimated
equal to 1.6 €/km. D. Ownership Vehicle tax and D.
carbon tax are used to calculate the total taxation
costs for a traditional diesel vehicle. The first one is
computed on a monthly basis and it is equal to 6.67
€/vehicle; the second one is dependent on emission
levels (g/km) and is not associated with the actual
use of the vehicle (e.g. kilometre travelled), it is also
computed on a monthly basis and it is equal to 1.0
(€*km)/(Vehicle*g).
6 Simulation
This section shows the results of the simulation runs
of our SD model. In particular, we will focus on the
adoption of the new distribution system in terms of
C.Us and number of vehicles. Two scenarios are
discussed: one scenario considers parcel delivery by
electric vehicles within ZTL and the other shows the
adoption of hybrid vehicles for delivering parcels to
C.Us located outside ZTL.
6.1 ZTL: electric vehicles
The entire stock of C.Us at issue (2,462) adopt the
distribution system in a 51 month period, being
served by a total number of 40 electric vehicles as
shown in Figure 3.
Fig.3: C.Us. and electric vehicles diffusion
Simulation also shows that the total cumulated cost
savings in ten years are around € 2 million (Figure 4).
Fig. 4: Total cost savings for electric vehicles
distribution
6.2 Outside ZTL: hybrid vehicles
In this scenario market saturation is reached in 47
months. The 9,538 C.Us are served by 181 hybrid
vehicles, each of them performing on average 16
monthly routes.
This scenario achieves total cost savings equal to
€ 10 million, turning positive from the fourth month
of simulation on.
7 Sensitivity analysis
The aim of the sensitivity analysis is to reveal how
the outcomes of the model vary when the main input
parameters change. This objective is instrumental not
only to understand the dynamics of the diffusion
process and highlight the most important stimulating
factors, but also to validate the robustness of the SD
model at issue [9].
We will discuss how the main elements of our
model, namely the number of vehicles, the number of
C.Us and the total savings, change as we alter the
input parameters. In the following sub-sections we
therefore propose the results of the three most
significant scenarios, which rely on both univariate
and multivariate sensitivity analysis.
The analysis was performed with Vensim DSS,
which allowed us to vary the input parameters
according to a selected probability distribution. The
software executes a fixed number of simulations,
usually 200, calculating the output variables for each
value of the input parameter. In the next figures the
blue line represents the base case, while the coloured
bands are the confidence bands where the output
values can be found with probabilities equal to 50%,
75%, 95%, and 100%.
7.1 Multivariate sensitivity analysis on
Advertising Effectiveness, Emulation Contact
Rate e Contact Rate
In this scenario we observe how the dynamics of the
adoption process changes as the three parameters
Advertising Effectiveness, Emulation Contact Rate
and Contact Rate vary between 0 and 0.4 [1/month]
according to a standard normal distribution. Figure 5
presents the total cost savings trend for the new
distribution system with electric vehicles. In the first
months of the simulation period the advertising and
the emulation effect drive the adoption and the
associated savings. When a considerable number of
C.Us has already started being served by the new CL
system the word of mouth action becomes relevant in
order to furtherly stimulate the diffusion.
Fig. 5: Sensitivity analysis on Advertising
Effectiveness, Emulation Contact Rate and Contact
Rate on the total cost savings – electric vehicles
7.2 Univariate sensitivity analysis on
Monthly Vehicle Utilization Factor
For this scenario we consider that the parameter
Monthly Vehicle Utilization Factor follows a uniform
distribution varying between 0.015 and 0.06.
This range of values was calibrated so as to obtain a
number of routes per day of between 0.76 and 3,
plausible values for the product category at issue.
With a Monthly Vehicle Utilization Factor equal to
0.06 a total of 46 electric vehicles is necessary to
serve all C.Us. On the contrary, if we consider 3
routes per day (Monthly Vehicle Utilization Factor =
0.015) we reach a total number of around 12 vehicles.
For both the output variables analysed, namely the
number of vehicles in the system and the total cost
savings, we observe significant variation as the
values of the selected input parameters change. For
instance, total cost savings takes values ranging from
around € 500,000 to € 2 million (Figure 6 and Figure
7).
Fig. 6: Sensitivity analysis of Monthly Vehicle
Utilization Factor on the number of electric vehicles
Fig. 7: Sensitivity analysis of Monthly Vehicle
Utilization Factor on the total cost savings - electric
vehicles
7.3 Multivariate sensitivity analysis on Public
Contribution Factor, Public Contribution
Factor for Plugin and D. Carbon Tax Factor
Through this sensitivity analysis we investigate the
degree to which public contribution can support and
influence the adoption of the new distribution system.
All the three input parameters follow a standard
normal distribution. Public Contribution Factor
ranges from € 0 and € 0,009; Public Contribution
Factor for Plugin ranges from € 0 and € 0.003, while
D. Carbon Tax Factor can take values from € 0.1 to
€ 2.
As expected, we observed that the public
contribution dependent on the CO2 emission
reduction is able to lead to a significant increase in
the total cost savings of the distribution system,
because this contribution has a direct impact on the
adoption from savings. Moreover, we noticed
moderate indirect effects of the public contribution
on word of mouth actions.
The same sensitivity analysis was performed
excluding the parameter D. Carbon Tax Factor. We
find out that the positive effects mentioned above are
weakened, meaning that public intervention is more
effective on the adoption if it comprises both
incentives for low emission vehicles and taxation for
traditional vehicles.
8 Discussion of results
With our analysis we demonstrated that a new urban
freight distribution system with low emission
vehicles is feasible both for the city centre restricted
area (ZTL) and for the whole city of Torino.
In fact, by focusing on parcel delivery, in both
areas the market saturation is reached within the
simulation time horizon, and in particular within 51
months for electric vehicles and 47 months for hybrid
vehicles. Moreover, the model simulation reveals that
the new distribution system could bring significant
savings over ten years, equal to around € 2 Mln for
electric and € 10 Mln for hybrid vehicles
Such results are due to two main factors. First, the
involved technology can be considered mature, in
terms of costs (the difference in operating costs
between low emission and traditional vehicles is less
than 10 cents per km) and in terms of operating time
of the batteries that now allow for a whole trip to be
completed without being recharged. Second, the
involvement of the public sector could significantly
support the diffusion of low emission freight
distribution systems. In our model, such involvement
includes both disincentives to traditional vehicles and
incentives to low emission ones. In particular, we
calculated the emission gap between the two types of
vehicles: the higher this gap, the higher the public
contribution. This leads private operators to adopt the
new system.
The sensitivity analyses performed show that the
most determinant aspects for the diffusion process
are the same for electric and hybrid vehicles. To be
more precise, Advertising Effectiveness, Public
Contribution, Initial Public Contribution for Plugin
and Plugin Unit Cost are the most influential factors
for stimulating the diffusion process. As a matter of
fact, the total cost savings deriving from the
distribution with low emission vehicles are moderate,
because of the low gap in operating costs and the
necessary investment in charging stations. This
means that the economic aspect is less relevant to the
diffusion process than the awareness of adopting a
more eco-friendly freight distribution system.
Therefore, we can state that this new freight
distribution system should be implemented based on
structured advertising campaigns aiming at
delivering the real environmental and operational
benefits of such a CL model, on a public intervention
and on consolidated and mature technologies. Only
with these pillars it is in fact possible to reach a
complete diffusion in reasonable times.
9 Conclusion
This work studies the dynamics of the adoption of
electric and hybrid commercial vehicles to perform
freight distribution activities in the city of Torino
(Italy). The analysis has been conducted through the
SD approach since it appears to be very useful to
describe the behaviour of a complex system and its
associated variables. The adoption results to be
influenced by the economic savings, the word of
mouth and the green image that are related to the
proposed sustainable logistics model. The parcel
delivery supply chain has been considered. The
outcomes show that the market saturation is achieved
in about three years and the new CL system leads to
a significant reduction in pollutant emissions. The
financial sustainability is ensured by the mature
vehicle technology and by the public economic
contribution. Thus, it can be stated that the actual
environmental benefits of the systems that are
promoted via advertising campaigns, the
involvement of the public authorities, and the
adoption of suitable technologies are the main
aspects that can stimulate the diffusion. Future
research efforts will be directed towards applying the
SD model to other product categories.
References:
[1] Browne, M. and Gomez, M., The impact on
urban distribution operations of upstream supply
chain constraints, International Journal of
Physical Distribution & Logistics Management,
Vol. 41, No. 9, 2011, pp 896-912.
[2] Figliozzi, M.A., The impacts of congestion on
commercial vehicle tour characteristics and
costs, Transportation Research Part E, Vol. 46,
No. 4, 2010, pp. 496-506.
[3] Witkowski, J. and Kiba-Janiak, M., Relation
between city logistics and quality of life as an
assumption for referential model, Procedia –
Social and Behavioral Sciences, Vol. 39, 2012,
pp. 568- 581
[4] Visser, J., Van Binsbergen, A., and Nemoto, T.,
Urban freight transport policy and planning.
City logistics I, 1999, pp. 39-70.
[5] Teo, J. S., Taniguchi, E., and Qureshi, A. G.,
Evaluation of Load Factor Control and Urban
Freight Road Pricing Joint Schemes with Multi-
agent Systems Learning Models, Procedia-
Social and Behavioral Sciences, Vol. 125, 2014,
pp. 62-74.
[6] Trip, J.J. and Konings, R., Supporting electric
vehicles in freight transport in Amsterdam, E-
Mobility NSR, North Sea Region, Electric
Mobility Network, August the 15th 2014.
[7] Gries S., Witte C., Föhring R. and Zelewski S. ,
Investments in Electro Mobility for Freight
Traffics in the Field of City Logistics: A
Profitability Analysis, Innovative Methods in
Logistics and Supply Chain Management, 2014,
pp. 123-141.
[8] Forrester J.W., Industrial Dynamics. The MIT
Press, Cambridge, Massachusetts, 1961.
[9] Sterman J.D., Business Dynamics: systems
thinking and modeling for a complex world,
McGraw – Hill, 2000.
[10] Armah F.A., Yawson, D.O. and Pappoe,
A.N.M., A Systems Dynamics Approach to
Explore Traffic Congestion and Air Pollution
Link in the City of Accra, Ghana. Sustainability,
Vol. 2, 2010, pp. 252-265.
[11] Vafa-Arani H., Jahani S., Dashti H., Heydari J.,
Moazen S., A system dynamics modeling for
urban air pollution: A case study of Tehran, Iran,
Transportation Research Part D, Vol. 31, 2014,
pp. 21-36.
[12] Han, J. and Hayashi, Y., A system dynamics
model of CO2 mitigation in China’s inter-city
passenger transport, Transportation Research
Part D, Vol. 13, 2008, pp. 298-305.
[13] Egilmez, G. and Tatari O., A dynamic modeling
approach to highway sustainability: Strategies to
reduce overall impact, Transportation Research
Part A, Vol. 46, 2012, pp. 1086–1096
[14] Stepp, M. D., Winebrake, J.J., Hawker, J.S. and
Skerlos, S.J., Greenhouse gas mitigation
policies and the transportation sector: The role
of Feedback effects on policy effectiveness,
Energy Policy, Vol. 37, 2009, pp. 2774–2787.
[15] Seitz, C. and Terzidis, O., Market Penetration of
Alternative Powertrain Concepts in Heavy
Commercial Vehicles: A System Dynamics
Approach. 32nd International Conference of the
System Dynamics Society July 20 - 24, 2014,
Delft, Netherlands.
[16] Ardila L. and Franco C., Policy analysis to boost
the adoption of alternative fuel vehicles in the
Colombian market. 31st International
Conference of System Dynamics Society July 1-
4, 2013 Cambridge, Massachusetts USA
[17] Struben, J. and Sterman, J. D., Transition
challenges for alternative fuel vehicle and
transportation systems. Environment and
Planning B: Planning and Design, Vol. 35,
2008, pp. 1070-1097
[18] Shepherd, S., Bonsall P. and Harrison G.,
Factors affecting future demand for electric
vehicles: A model based study, Transport
Policy, Vol. 20, 2012, pp. 62–74
[19] Gorbea, C., Lindemann, U. and de Weck, O.,
System Dynamics Modeling of New Vehicle
Architecture Adoption. International
Conference on Engineering Design, Iced11, 15 -
18 August 2011, Technical University of
Denmark.
[20] Bass, F.M., A New Product Growth for Model
Consumer Durables, Management Science, Vol.
15, No.5, 1969, pp.215-227.
[21] Fisher, J.C. and Pry, R.H., A simple substitution
model of technological change, Technology
Forecasting and Social Change, Vol. 3 No.1,
1971, pp. 75-88.
[22] Meade, N. and Islam, T., Forecasting the
diffusion of innovations: Implications for time
series extrapolation, in Armstrong, J.S. (Ed.),
Principles of forecasting: A handbook for
researchers and practitioners, pp. 577-595,
Kluwer Academic Publishers, Dordrecht, The
Netherlands, 2001.
[23] Park, S.Y., Kim, J.W. and Lee, D.H.,
Development of a market penetration
forecasting model for Hydrogen Fuel Cell
Vehicles considering infrastructure and cost
reduction effects, Energy Policy, Vol. 39, 2011,
pp. 3307–3315.
[24] Daim, T. and Suntharasaj, P., Technology
diffusion: forecasting with bibliometric analysis
and Bass model, Foresight, Vol. 11, No. 3,
2009, pp. 45-55.
[25] Element Energy Limited, Ultra Low Emission
Vans study. Final Report, 2012,
http://www.element-energy.co.uk/wordpress
(accessed on 08/05/2015)
... The TIMBER framework refers to (T) technology; (I) infrastructure; (M) market; (B) behavior; (E) energy; (R) regulation, and policies are related to (ECO) economic; (SOC) social; (LEG) legal; and (KNL) knowledge-based instruments. [30] Iran Nation/Region 2009-2034 Vensim [31] Qualitative Nation/Region -Stella [32] Generic Nation/Region 10 years Stella [33] Malaysia Nation/Region 1990-2016-2040 Powersim [34] Latvia Nation/Region 2013-2030 Powersim S and F * [35] Latvia Nation/Region 2016-2030 Powersim CLD [36] Italy Urban 120 months Vensim CLD [37] Italy Urban 120 months Vensim CLD [38] Italy Urban 120 months Vensim [39] South Korea Nation/Region 100 months Vensim [40] Brazil [30] x 4 x x [31] x x x x x [32] x x x x x [33] x ...
... The model does not consider the performance deterioration of battery and fuel cells. It also lacks the analysis of costs of refueling and recharging infrastructure [36][37][38] To assess the diffusion of a city logistics system based on electric and hybrid vehicles Subsidies for alternative technologies and investment in refueling/recharging infrastructure Advertising campaigns, involvement of public authorities, and adoption of suitable technologies are the main aspects that can stimulate the diffusion of alternative vehicles ...
... In Refs. [36][37][38], the authors modeled the diffusion of a city logistics system based on electric and hybrid vehicles. The size of the fleet depends on freight demand, vehicle capacity, and load factor. ...
Article
Full-text available
Freight transport decarbonization is currently an urgent challenge. Decarbonization strategies have a specific time to take effect, and it is essential to consider their time dependence. The system dynamics approach is well suited to represent feedback, lagged responses, and the time dependence of decarbonization strategies. We carried out a systematic literature review of system dynamics models in relation to strategies for freight decarbonization to identify the treatment of relevant dynamics of the system within the models. The 50 studies that fulfilled our search criteria were categorized by decarbonization strategies, the external factors needed to support them, and simulated policy instruments. The results show that no model presented a broad view of the system, addressing a limited combination of strategies. Most importantly, system dynamics models do not clarify how time-dependent behavior is determined, which indicates a significant research gap that can be critical for understanding the policy’s urgency and impacts.
... Flows). Finally, "the equations of a SD model can be either algebraic or differential in nature, they are independent from one another, and are functions of the state of the system in the previous time steps" (Cagliano et al., 2015). ...
Article
Full-text available
In City Logistics (CL) contexts, public stakeholders’ goal is often to implement policies aimed at reducing the negative externalities generated by freight transportation activities, while at the same time maintain and foster the efficiency of the CL system without hindering the profitability of private operators. This task can become quite arduous because of the complexity of CL systems, which are composed by multiple actors with different objectives and driving factors. In this context, we argue that existing literature should explore more deeply the effects of public policies on operation and economic variables that shape the CL context. To this end, we propose a System Dynamics (SD) model that identifies the interconnections between those variables and helps assess the effects of public policies on the system.
... In fact, it allows to analyze in a systemic way all the variables determining EVs diffusion as well as their interdependences and causal relationships [20]. Worth of note is that SD has already been used in other studies on EV impacts, even if many of them are centered on very specific aspects as they consider case-by-case one of the triple sustainability principles as predominant over the others [21][22][23][24]. The present study, instead, aims to leverage the SD method to further elaborate EVs impacts analysis by putting together the social, environmental and economic dimensions. ...
Article
Full-text available
The present paper proposes a model for evaluating environmental, social, and economic impacts exerted by the diffusion of electric vehicles (EVs), which is a phenomenon that can significantly affect the achievement of some of the objectives set by the Sustainable Development Agenda. The impact evaluation is carried out through the System Dynamics methodology, combined with scenario analysis. Considering the Piedmont region (Italy) as a case study, the model forecasts the impacts of EV diffusion using a simulation timeframe of 12 years and leveraging eight EV diffusion scenarios. According to the model, an increase in the number of EVs results in less air pollution and, therefore, minor public health expenditure. These cost savings can be turned into incentives for purchasing new EVs, which make the fleet increasingly greener as part of a self-reinforcing loop. Despite the fact that the model could be improved through additional research on some variables’ definitions, this ex ante evaluation tool represents a valuable instrument for policy-makers. In fact, it provides a comprehensive picture of EV diffusion in view of the triple sustainability principles: System Dynamics, in particular, allows singling out causal relationships among variables, thus anticipating possible effects of planned policy actions.
... Finally, the third stream of the literature analyzes new business models based on vehicles that have a low or very low emission impact, including the effects of replacing traditional commercial vans with such vehicles. In particular, Cagliano et al. (2015) developed a system dynamics (SD) model for a freight distribution system. Using this approach, the authors observed how savings in CO2 emissions and operating costs give rise to interesting feedback loops for the adoption of a new distribution system based on electric or hybrid vehicles. ...
Article
In recent years, the role of freight transportation and parcel delivery in urban areas has increased, supporting the economic and social development of cities. At the same time, the industry is affected by various issues, inefficiencies, and externalities, particularly in the last-mile segment. As such, there is an emerging awareness of a need to improve urban mobility and transportation, making them more sustainable and competitive by mixing traditional and emerging technologies, such as cargo bikes, autonomous vehicles, and drones. In contrast, the complexity of the overall system, characterized by multiple actors with conflicting goals, requires a strategy that harmonizes these actors’ business and operational models. This study contributes in this direction along three axes. First, it defines the main actors involved in urban parcel delivery, and then analyzes their business models and the interactions between them. Second, it investigates the integration of traditional and green logistics (mainly cycle-logistics), from both business and operational perspectives, in order to identify synergies, conflicts, and the operational and economic consequences of adopting green vehicles. Third, it introduces a simulation-optimization decision support system tool capable of assessing mixed-fleet policies for the management of parcel delivery in urban areas. Finally, the decision support system is tested using real data of the city of Turin.
... environmental, economic and social) variables involved, thus allowing to portray their dynamic interdependencies and causal relationships [13]. This method has been used for evaluating EVs diffusion impacts also by other authors, who have examined in their studies specific aspects: Cagliano et al. [14] analysed the diffusion of commercial electric vehicles in an Italian region and estimated its impacts on the city logistics; Garcia and Miguel [15] modelled in System Dynamics the adoption of EVs in Spain and developed different diffusion scenarios focusing on the cost variables (e.g. purchasing cost, maintenance cost); Gorbea et al. [16] proposed a System Dynamics simulation model to predict the market share of electric vehicles over time; finally, Yang et al. [17] integrated the System Dynamics technique with Agent-Based Modelling for predicting the diffusion of electric vehicles. ...
Article
Supply chains in general and last-mile logistics in particular, have been disrupted due to COVID-19. Though several innovative last-mile logistics solutions have been proposed in the past, they possess certain limitations, especially during COVID-19 motivating the need for an alternative last-mile logistics solution. We present a review of literature related to last-mile logistics and supply chain disruptions to identify the limitations of existing last-mile delivery practices during COVID-19. Using a stylized analytical model, we then propose that “mobile warehouse” can be an effective solution to last-mile logistics issues faced during COVID-19 and beyond under certain conditions. A mobile warehouse is a truck dedicated to a particular geographical location and carries the inventory of various products based on the estimated demand requirements for these products in that geographical location. We provide the condition under which the B2C e-commerce providers find it profitable to adopt a truck as a mobile warehouse to sell high demand items quickly.
Article
Full-text available
This paper describes the use of the multi-agent systems (MAS) modelling approach, incorporated with reinforcement learning models, that is experimented on a part of Osaka road network with ITS-based travel time information. The MAS model included an exact solution method to solve the vehicle routing problem of carriers’ delivery jobs and is used to evaluate the short-term impact of distance-based road pricing on the major stakeholders including the carriers, shippers, administrators and customers. The evaluation is extended to consider the load factor control scheme implemented as a joint scheme with the urban freight road pricing. The results from our experiment show that the city logistics joint scheme has the potential of improving average daily load factors and reduce emissions in comparison with no schemes implemented.
Chapter
Electro mobility is considered to be a pioneering trend in the field of road traffic to reduce the emissions of greenhouse gases. Because of technical restrictions, which limit the range and loading capacity of vehicles, electro mobility is mostly discussed with regard to passenger cars. However, recent developments raise the question whether electro mobility can be also used promisingly for local freight traffics of the so-called "city logistics". Interesting are especially transports on the “first” and “last mile” of supply chains with combined transports. However, electro mobility has only then potential as a future trend if the profitability of investments can be demonstrated. The profitability of investments in electro mobility is being examined. The costs of three diesel versus three similar electric commercial vehicles are calculated for a typical case of application in city logistics. The analyses cover three scenarios that represent alternative investment options for the application of electro mobility: the complete purchase of electric commercial vehicles, the partial purchase combined with renting the battery as well as the complete vehicle leasing. Drive-specific features like e.g. the smaller range of electric commercial vehicles compared to diesel commercial vehicles are especially considered. It is shown that already today the regarded electric commercial vehicles prove to be economically attractive in some of the examined scenarios. Moreover, it is elucidated which expectable changes in future within the circumstances of city logistics will promote investments in electro mobility.
Article
The selection of an S-shaped trend model is a common step in attempts to model and forecast the diffusion of innovations. From the innovation-diffusion literature on model selection, forecasting, and the uncertainties associated with forecasts, we derive four principles. 1. No single diffusion model is best for all processes. 2. Unconditional forecasts based on a data-based estimate of a fixed saturation level form a difficult benchmark to beat. 3. Simpler diffusion models tend to forecast better than more complex ones. 4. Short-term forecasts are good indicators of the appropriateness of diffusion models. We describe the evidence for each principle in the literature and discuss the implications for practitioners and researchers.
Article
Purpose The purpose of this paper is to use bibliometric analysis to forecast RFID technology and uses the adoption of barcode scanner to model the RFID scanner adoption in the US retail market. Design/methodology/approach Forecasting emerging technologies and identifying the rate of diffusion of products based on these technologies is difficult because of lack of data. This paper uses techniques such as bibliometric analysis and Bass model based on analogous products. Findings The authors were able to come up with a good Bass model for the adoption of RFID scanners in the retail sector. And according to the Bass model it appears that it will take at least seven or eight years for the retailers to adopt to this new technology at their point‐of‐sale. Originality/value This paper attempts to use bibliometric analysis and Bass model for forecasting technologies and provides a new research discussion as results from these two are compared.
Article
(This article originally appeared in Management Science, January 1969, Volume 15, Number 5, pp. 215–227, published by The Institute of Management Sciences.) A growth model for the timing of initial purchase of new products is developed and tested empirically against data for eleven consumer durables. The basic assumption of the model is that the timing of a consumer's initial purchase is related to the number of previous buyers. A behavioral rationale for the model is offered in terms of innovative and imitative behavior. The model yields good predictions of the sales peak and the timing of the peak when applied to historical data. A long-range forecast is developed for the sales of color television sets.