Driving electric vehicles towards the early majority : the importance of the public charging infrastructure

(1)

Driving electric vehicles towards the early

majority: The importance of the public

charging infrastructure

Author: Pico van Heemstra Student Number: 10670289 Supervisor: Dr. René Bohnsack Date: 29/6/2015

(2)

Abstract

As the global awareness for carbon emission reduction grows, the transition towards electric mobility has gained increased attention over recent years. However, the limited battery capacities, and the associated range anxiety, of electric vehicles in their current state cause increased complexity and compatibility issues implying that this new technology cannot gain a relative advantage over traditional cars and therefore faces difficulties in achieving widespread adoption.

To overcome these barriers, the erection of an extensive charging infrastructure has become clear, however this development faces a chicken-and-egg dilemma; EV adoption will only grow if there are enough charging stations and charging stations will only be placed if there are enough EVs to make the investment worthwhile. This calls for a rigorous understanding of the demand and supply needs forces behind charging stations.

By applying regressions analyses to investigate the relationship between neighbourhood characteristics and visitors’ use of the public charging infrastructure in Amsterdam, this thesis provides infrastructure developers with valuable insights regarding charging preferences and behaviour and a conceptual tool that can be used to design strategies for public charging infrastructure deployment in urban environments.

(3)

5.1 Location ... 56 5.1.1 Descriptive statistics ... 56 5.1.2 Correlation analysis ... 57 5.1.3 Regressions analysis ... 59 5.2 Time of day ... 61 5.2.1 Descriptive statistics ... 61 5.2.2 Moderation analysis ... 61 5.3 Duration ... 64 5.3.1 Descriptive statistics ... 64 5.3.2 Regression analysis ... 64 6. Discussion ... 66 6.1 Location ... 66 6.2 Time of day ... 67 6.3 Duration ... 68

6.4 Implications for infrastructure development ... 69

7. Conclusion ... 71

8. Recommendations for future research ... 74

References ... 75

(4)

Glossary of terms and acronyms

AE (Amsterdam Elektrisch): Organization responsible for stimulating electric

mobility in Amsterdam.

AFV (alternative fuel vehicle): A vehicle that runs on a fuel other than petrol or

diesel.

CS (charging station): The power outlet that supplies electricity for recharging

electric vehicles.

EV (electric vehicle): A vehicle that runs on electricity only. They are propelled by

an electric motor(s) powered by rechargeable battery packs.

EVSE (electric vehicle supply equipment): Equipment used in charging electric

vehicles.

FCV (fuel cell vehicle): A vehicle that uses fuel cell, generally using oxygen and

hydrogen, to power its electric motor(s). A fuel cell that is fuelled using hydrogen emits only water and heat.

ICE[V] (internal combustion engine [vehicle]): [A vehicle propelled by] an engine

fed with fossil fuel like gasoline or diesel.

LTS (Large Technical System): A system or network of enormous proportions or

complexity

PHEV (plug-in hybrid electric vehicle): A hybrid electric vehicle with batteries that

are recharged by plugging into an electric power source. They are propelled by an electric motor(s) stored in rechargeable batteries with the option to recharge with an on-board, gasoline-powered generator.

POI (Point of Interest): Term used in urban economics to express a potential end

destination

(5)

1. Introduction

1.1 Introduction

In the wake of the 21st century, the challenges related to energy and environmental issues, especially in the field of transportation, have gained significant attention. Worldwide, road transportation contributes about 22% (IEA/OECD, 2013) of greenhouse gas emissions and the global demand for transport is not expected to decrease; the World Energy Outlook projects that transportation fuel demand is to grow by almost 40% by 2035 (WEO, 2013). Global objectives to reduce emissions ask for innovative and sustainable solutions. In recent years this has lead to the electrification of transportation to be an important topic of discussion. Due to the perceived environmental and energy advantages, developments related to electric vehicles (EVs) have gained prioritized positions on the agendas of many policy discussions around the world.

In line with the International Energy Agency agreements, governments all over the world regard substituting the current internal combustion engine (ICE) fleet with environmentally friendlier alternatives as an essential step towards reaching future emission targets. The Netherlands has been at the forefront of this development for the past years. In 2011, the government set itself the goal of having 20,000 EVs on the road by 2015, 200,000 five years later and reach 1 million by 2025. By the end of 2013 there were almost 29,000 EVs, therefore the goal for 2015 has already been realized. Popularity is clearly increasing, as are sales, spurred by a wide array of policies and incentives. However, considering that this amount represents just 0.4% of the 7.5 million personal cars and that EV sales for 2013 represented just a share of 1% of total car sales (RVO, 2013), there are still barriers to overcome and a lot of progress to be made.

At present, EVs still prove difficult to sell due to a number of reasons. Researchers studying the challenges for EVs to penetrate the automobile market generally agree on the following main advantages and disadvantages (Li et al., 2015).

(6)

Table 1 – Advantages and disadvantages of EVs

Advantages Disadvantages

Environmental benefits High purchasing price

Low running cost Limited driving range

Low maintenance Lack of public charging infrastructure

Swift acceleration Battery charging time

Low noise emission

Setting aside the high purchasing costs, the remaining disadvantages are associated with an interrelated problem referred to as range anxiety. Range anxiety arises from users’ continuous concern of stranding with a discharged battery. To relieve potential EV users of this fear and encourage the adoption of EVs, the importance of public charging infrastructure is evident (Beella et al., 2009; Wagner et al., 2013; Gonzalez et al., 2014). However, due to the uncertainty regarding the future success of EVs there is reluctance towards making large investments for the development of a charging infrastructure. This leads to a classic example of the chicken-and-egg dilemma (Huétink et al., 2010).

The chicken-and-egg dilemma is caused by consumers’ and infrastructure developers’ mutual reliance combined with their reluctance to take risks; consumers are influenced in their decision whether or not to purchase an EV by the availability of public charging stations (CS) and the investment decision of infrastructure developers is determined by the amount of EVs on the road (Van der Vooren et al., 2012). Neither is eager to take the “first” step since their interdependence implies that if the other does not follow, the investment is useless.

Market forces alone cannot be relied upon to solve chicken-and-egg dilemmas and this especially holds for innovations that require new infrastructure due to the high and irreversible investments (Huétink et al., 2010). For this reason, public policymakers play an important role in the early stages of EV adoption by administering the construction of public charging stations, thereby connecting and managing the short terms steps with the long-term goals of transition (Van der Vooren et al., 2012). Such an infrastructure must fulfil EV users’ charging demands in a convenient manner on the one hand and needs to be technologically and commercially viable for infrastructure developers on the other hand (Jia et al., 2012). In this way,

(7)

the deadlock situation can be overcome and adoption becomes attractive. Once a certain amount of users adopt new technologies, Rogers (1962) estimates this to start from about 16% of the potential market, a critical mass is achieved. From this point onward, theory suggests that technologies become increasingly self-sufficient and can successfully diffuse further without any market interventions (Rogers, 1962; Sahin, 2006). Therefore, the speed and effectiveness of setting up the supporting charging infrastructure is an essential determinant of the rate at which this critical mass can be achieved and the chances of EVs replacing traditional ICEVs successfully.

Much academic work has been done to determine how cities should design a public charging infrastructure that fulfils both demand- and supply side needs or requirements. On the one hand this has been done from an EV user perspective, aiming to minimize the costs of accessing and using charging stations (Ge et al., 2011; Andrews et al., 2012; Xu et al., 2013; Donna Chen et al., 2013; Gonzalez et al., 2014). On the other hand, the perspective of infrastructure developers has been taken, aiming to maximize demand coverage and utilization to minimize construction costs (Wang et al., 2010; Frade et al., 2010; Liu et al., 2013; Xi et al., 2013; Chen & Hua, 2014). Furthermore, various authors have attempted to combine the demand and supply perspectives (Ip et al., 2010; Jia et al., 2012; Sweda & Klabjan, 2012; He et al., 2012; Jin et al., 2012; Ge et al., 2012; Wagner et al., 2014; Helmus & Van den Hoed, 2015). Accordingly, various optimization techniques have been applied based on selected demand and supply variables and budgetary constraints. However, to date, there is no general agreement as to which influencing factors and associated models are most appropriate. For this reason, empirical evidence from cities that have already started with setting up a public charging infrastructure can provide valuable insights.

One of the cities that has pioneered in the development of charging infrastructure and data logging is Amsterdam (Van den Hoed et al., 2014). Due to a valuable cooperation between the municipality of Amsterdam and the University of Applied Sciences of Amsterdam a select number of researchers have gained access to an extensive database consisting of almost 500,000 registered public charging sessions within the city. For this particular research, access was also granted to this database; therefore, the focal city of this thesis is Amsterdam.

Amsterdam started building public charging stations in 2009 and has since developed one of the most advanced public charging infrastructures in the world. The

(8)

city’s charging station deployment strategy can be divided into roughly three stages. In the first stage, which started in June 2009, a few charging station were installed in central and highly visible locations (Amsterdam Elektrisch, 2014). These stations not only covered the demand for the first handful of EVs - mostly owned by the municipality - but also had a promotional functionality to stimulate the early group of private EV users. In the second - and present - stage, Amsterdam implemented a policy allowing EV owners, living or working in the city, to apply for a public charging station. In most situations where no other charging stations have already been installed within a 300-metre radius, these applications are granted. Furthermore, the city also places additional charging stations in areas where it clearly identifies excess demand. At present, in some areas, for example Amsterdam South, enough CSs have been placed to completely cover the neighbourhood according to the 300-metre radius policy (Figure 1). Therefore, a third stage can be expected. In this stage a more dynamic and specific approach should be taken to placing additional CSs in a cost effective manner. This is done according to more sophisticated measurements of demand along different user type characteristics rather than basing choices mainly on aggregated data or on the residential or working location of CS applicants (Helmus & Van den Hoed, 2015; Amsterdam Elektrisch, 2014).

Figure 1 - Map showing the 300-metre radius coverage (pink circles) of charging stations in Amsterdam South (Source: Stadsdeel Zuid, 2014)

(9)

By analysing the utilization of Amsterdam’s charging infrastructure, valuable information van be obtained about demand patterns, charging behaviour and user preferences. For cities planning on developing their infrastructure in the years ahead, understanding the demand factors makes it possible to better predict where, when and how users are likely to charge. In this way an effective charging infrastructure can be built within a shorter timeframe and at a lower cost, encouraging the adoptions of EVs and increasing the rate of diffusion.

1.2 Research question

The goal for the majority of EV charging infrastructure studies is to identify technically and financially viable locations for public charging stations that will decrease overall range anxiety of current and future EV users in order to encourage more widespread adoption of EV. Thereby, range anxiety among (potential) EV users is reduced when they are able to charge their EV conveniently when it is not needed for driving, i.e. when it is parked. In identifying areas where charging demand could exist, two questions arise:

a. Where do urban EV users live? b. Where do EV users drive to?

With respect to the first question, studies relating demographic characteristics to EV owners are conclusive that this group typically has a higher income and education level, and generally owns more than one car (Econ analyse, 2006; Rødseth, 2009; Pierre et al., 2011; Campell et al., 2012). Areas with a higher concentration of inhabitants that correspond to the aforementioned criteria are therefore likely to have more EV owners and a higher charging demand.

Concerning the question where EV users drive to, a distinction should be made between daily commuting to work and other trips. Research suggests that EV users who use their EV to drive to work every day, will generally choose for an EV due to the availability of private or semi-private charging stations located at their work (Pierre et al., 2011; Xi et al., 2013; Wagner et al., 2013). This assumption is to some extent supported by the relative low number (6) of public charging station in Amsterdam’s “Zuidas”, the city’s business hub. Accordingly, it may not have the highest priority for public infrastructure developers to consider this group of EV users.

(10)

Explaining the demand derived from EV trip destinations, other than driving home or daily commutes to work, is less clear. In line with charging infrastructure literature and urban economics, individual charging demand, similar to parking demand, is likely to arise in close proximity to a trip destination to reduce walking time (Rowe et al., 2013). Therefore, being able relate charging data to area characteristics can provide valuable insights into demand factors for EV users (Wagner et al., 2013; Helmus & Van den Hoed, 2015). Until now however, this has been very challenging due to the limited availability of real charging data. This has forced much of the charging infrastructure literature to be based upon proxy variables, consequently reducing the reliability of the resulting models.

This thesis builds upon Wagner et al. (2014), who were the first to use a large database of real charging records. The authors have made an important contribution to the research in this field by attempting to measure the impact of 92 different specific points of interest (POI) categories on charging station utilization, using real charging data, to create a predictive model. However, instead of using all charging sessions, as Wagner at al. do, this research focuses on a specific sub-group of users. As Helmus and Van den Hoed (2015) suggest, creating a better understanding of the charging behaviour and preferences of sub-groups of homogenous users is essential. By combining these different user profiles, more comprehensive decisions can be made on demand-fulfilling and cost-effective roll out of charging stations. In this research, the focal group shall be visitors, comprising of EV users who connect to a public charging station in Amsterdam no more than once every two weeks.

In terms of total sessions, visitors do not make most use of a city’s charging infrastructure, however, they are very relevant as they represent a large proportion of the individual users that charge in the city. It is therefore important that the public charging infrastructure is also designed to fit the needs of this large group. Furthermore, due to the fact that these users generally come from outside the city, they are likely to have a higher need to recharge once they reach their urban destination and will also be largely dependent upon the public infrastructure. In contrast to the fairly foreseeable charging behaviour and preferences of EV users that live and work in Amsterdam, visitors are much more unpredictable and therefore require more thorough research.

(11)

This research focuses on investigating, understanding and being able to predict charging preference and behaviour from visitors across different neighbourhoods in Amsterdam. Each neighbourhood is thereby characterized according to its specific composition of land used for shops, hospitality, offices and residential. In contrast to Wagner et al. (2014) and traditional public parking research, the analysis is not limited to just explaining geographical preference but also aims to identify differences in demand relating to time of day and disparities in charging behaviour in terms of connection duration. This give rise to the following research question:

To what extent do neighbourhood characteristics have an effect on visitor charging demands and preferences?

This research question is answered along three sub-questions aimed at uncovering the traditional “where (location), when (time of day) and how (duration)”:

SQ1 - Location: Do neighbourhood characteristics have an effect on visitor charging demand?

SQ2 - Time of day: Does time of day have a moderating effect on the relationship between the types of neighbourhood characteristics and visitor charging demand? SQ3 - Duration: Do neighbourhood characteristics have an effect on the duration of charge sessions?

The goal is to apply data-supported analysis to understand and be able to better predict, areas and times of high charging station demand derived from the large group of users that make occasional use of the charging infrastructure. By examining an extensive database with recordings of all charge sessions in Amsterdam in 2014 regression models shall be created to explain the utilization of Amsterdam’s public charging infrastructure according to specific are characteristics across different neighbourhoods in Amsterdam.

1.3 Structure

The following section provides a brief background on EV development, worldwide as well as in the Netherlands and Amsterdam. Section 3 provides the theoretical framework, evaluating the literature related to diffusion of innovations, the issues associated with infrastructure dependent innovations and provides an overview of previous work on EV charging infrastructure planning. Next, section 4 describes the research design, laying out the conceptual models and the data collection methods.

(12)

Section 5 presents that results followed by a discussion of the results in section 6. Finally, section 7 provides the conclusion of this thesis and section 8 discusses recommendations for future research in this field.

(13)

2. Case background

2.1 General

The first electric car was developed as early as 1834. Until the end of the 19th

century, manufacturers in America as well as in Europe produced thousands of cars powered by an electric motor. In 1900 this even accounted for one third of the total number of vehicles produced (Chan, 2002). However, as from 1910, the internal combustion engine became dominant as the size and costs decreased whilst range, performance and reliability increased.

During the 20th century, car travel made huge leaps and private internal combustion engine vehicles (ICEV) became the dominant method for individual transport. The huge popularity of cars became a central factor in infrastructure planning and design for cities. Officials involved in such planning operations faced an extensive set of considerations and challenges, not only in building an efficient road network but also ensuring that there was sufficient space for cars to be parked and that there were enough, easily accessible, gas stations (Grübler, 1990).

Since the turn of the millennium, increased environmental awareness has pushed for international agreements concerning the reduction of carbonemissions. Along with growing concerns about fossil fuels running out and the perceived advantages of energy independence, the development of electric vehicles has gained momentum and, in addition to manufacturers themselves, an increasingly large group of different potential stakeholders are focussing on how to best manage the process of introducing this new technology and replacing ICEVs with EVs as quickly as possible.

From a commercial perspective, innovative business models have emerged over the past decade attempting to take advantage of market opportunities to increase the adoption of EVs. Start-ups such as Better Place, Car2go and Fastned are three diverse examples that were set up with the goals of gaining their share in the upcoming market. Better Place believed that instead of have charging stations, battery swapping stations would be a much more effective way of reducing range anxiety as empty batteries could easily be swapped for fully charged batteries. Despite their ambitious idea and the large investments Better Place received, among many factors,

(14)

the timing was not ideal and Better Place went bankrupt in 2012. Car2go, a Daimler AG subsidiary, focuses on shared EVs and seems to be performing well. According to a press release from December 2014, Car2go provides one million members and has established itself in 30 cities with 12,500 Smart Fortwos (Car2go, 2014). Finally. Fastned focuses solely on commercial fast-charging stations where customers can recharge their batteries in 20 to 30 minutes. Additionally, they have also gained a licence to contribute to the “European corridor” along Europe’s main highways to make travelling between countries more accessible for EVs.

In addition to entrepreneurs stimulating EV adoption, governments and policy makers play an essential role in the diffusion process too. Stimulating measures include policies and regulations towards standards, direct subsides to car manufacturers, fiscal incentives for consumers and expenditures on infrastructure. Denmark, the USA, the UK, Norway, Japan and China are among countries that offer over €5,000 in total incentives when buying an EV (ICCT, 2014). Moreover, many alternative and/or complementary incentives have been introduced: Denmark offers free parking in downtown Copenhagen (Schwartz, 2009), the UK supports the Plugged-in Places programme to install public charging stations and in Ontario special licence plates were introduced for EVs giving users access to free parking spaces as well as allowing them to use carpool lanes regardless of the amount of passenger in the vehicle.

Although the majority of the world may not yet be eager or ready to stimulate widespread adoption of EVs, there is a clear trend among European, South East Asian and North American countries towards electric mobility. The Netherlands, and specifically Amsterdam, is considered as one of the pioneering and most ambitious cities to replace its ICEV fleet with EVs. The following section takes a closer look at recent developments in national policies as well as local policies and strategies that aim to accomplish the goal of minimizing the amount of carbon emitting vehicle in Amsterdam.

2.2 The Netherlands and Amsterdam

In 2008, the Rijksdienst voor Ondernemend Nederland (RVO), Netherlands Enterprise Agency, submitted the Electric Vehicle Action Plan to the Lower House (Parliament Documents, 2008-2009, 31305, no.145). The aim of this plan was to pave

(15)

the way for EV development in the Netherlands with the goals, in 2011, set at having 20,000 EVs on the road by 2015, 200,000 by 2020 and reach 1 million by 2025.

On a national scale, financial incentives were put forward as an important tool to reach the goals and mainly apply in the form of tax breaks for both privately owned EVs as well as lease cars. Similar subsidies are also granted to company-owned lorries and trucks as well as taxis. Furthermore, local policy makers were given the authority to define more specific stimulation methods (RVO, 2011).

Certain cities, including Amsterdam, were assigned as “Focus areas”. These cities in particular had the task of linking small-scale projects and testbeds from the preliminary phases so that collective learning would speed up the process of EV penetration into the Dutch automobile market.

A 2011 report, evaluating the first three years, concluded that although progress had been made, there were still significant obstacles related to the following matters:

(a) The development of the charging infrastructure, the lack of a market model and a business case for public charging stations and innovative charging methods;

(b) The availability of vehicles, particularly for specific focus groups; (c) The range of electric vehicles is still limited;

(d) The prices of electric vehicles are still too high. (RVO, 2011)

Points (b), (c) and (d) were considered to be issues to be tackled on a national and international level. In contrast, individual cities were assigned to focus on the challenges concerning point (a).

On behalf of the municipality of Amsterdam, Amsterdam Elektrisch (AE) is the organization responsible for stimulating electric mobility in the Dutch capital. Among its goals, setting up a “well-equipped charging infrastructure” was considered as one of the most important factors contributing to large scale EV adoption (Amsterdam Elektrisch, 2013).

Due to the fact that Amsterdam is a densely populated city and almost all parking occurs “on street”, the need for a well laid out public charging infrastructure is very large. To achieve this goal, AE identified a number of important steps.

Firstly, a choice had to be made concerning the parties that would be involved in the planning and implementation process, including the important decision to what

(16)

extent the market should be open and whether the municipality should be involved. Amsterdam opted for a mixed approach, ensuring that the municipality had a say in policy issues such as where to site charging stations. However, for the physical development of the charging infrastructure, the city contracted with Nuon/Heijmans and Essent to construct and exploit all of the public charging stations until the end of 2015 (Amsterdam Elektrisch, 2013).

Secondly, the charging equipment needed to be standardized so that all EV users could make use of electric vehicle supply equipment (EVSE). In 2011, partly due to efforts by Amsterdam Elektrisch, the Dutch government opted for the ‘Mennekes Plug’ as the standard plug. The ‘Mennekes Plug’ is suitable for both AC and DC charging and avoids EV users from having to carry a multitude of charging cables to charge their vehicles in different cities or regions. Two years later, in January 2013, the European Commission also announced the ‘Mennekes Plug’ as the common standard for charging electrified vehicle across the European Union.

Thirdly, in line with the charging equipment, a decision had to be made whether to focus on ‘normal’ charging or ‘fast’ charging. A study performed by ING (2011) states that most people travel less than 50 kilometres by car each day, a distance that almost all current EVs can cover. Despite Hacker et al. (2009) stating that a charging infrastructure is likely to improve customer acceptance, fast-charging would not be essential for the vast majority of EV users and would be difficult to integrate in a business case due to high costs and substantial risks. Therefore, in Amsterdam, almost all public charging stations are ‘normal’, with a few exceptions mainly targeted at EV users that cover a lot of kilometres per day, such as taxi drivers.

Next, an operating system needed to be developed, for example, to unlock CSs, keep track of availability and record power outflows. This has been achieved by providing a standardized ‘charging pass’, allowing users throughout the whole country to use all types of CSs with a single card. This system also gives the energy providers the opportunity of having a clear overview of the usage per CS for individual sessions. Among other interesting development in the system, an overview of real-time availability for each individual CS is accessible on the Internet.

With respect to issues concerning the grid capacity, 2009 marked the start of a five-year renovation of the power grid. According to an article published in Het

(17)

Parool, by making this €150 million investment the municipality ensured that the power grid would be capable of recharging 20,000 electric vehicles (Stil, 2009). Although this is sufficient for the coming years, the capacity of the grid remains an issue that should be kept in mind during the planning process.

Finally, Amsterdam Elektrisch needed to decide on policy with respect to the scale and siting of CSs. This approach can roughly be divided into three stages. In the initial stage, a few charging stations were installed throughout the city to provide the first few EVs with charging opportunities. These were mainly located close to municipality buildings, that used EVs, and on sites where they would be visible to the public, therefore partly acting as a promotional tool.

In the second stage, in which the city finds itself at present, a gradual, mainly reactive approach is being taken. AE allows owners of an EV to apply for a public CS, either in front of their home, office or any other preferred location. Together with the two CS developers, Nuon/Heijmans and Essent, AE decides whether or not to grant the application. In general, if no other public CS exists within a 300-meter radius, a new CS is placed. However, this is not a rule, and factors such as the grid capacity may also influence this decision. Moreover, AE can also decide to place an additional charging irrespective of an application from an EV user (Amsterdam Elektrisch, 2013).

Finally, the third stage is the next challenge that awaits AE. The goal for Amsterdam is to have installed 2000 CSs by the end of 2015 and as things are looking now the programme is on track. However, as is the case in Amsterdam South, a number of areas are likely to reach the density quota determined by the 300-metre radius policy. After this point, the planning role for the municipality will probably become less prominent and the charging station market is likely to become more reactive to market forces, with the goal of maximizing the profitability of each CS and the utility for EV drivers. It therefore becomes increasingly important to identify determining factors for local charging demand to ensure that the profitability of charging stations is maximized.

(18)

3. Theoretical framework

This section guides the reader through diffusion of innovation theories and the challenges associated infrastructure dependent technologies: the chicken-and-egg dilemma. Subsequently, an overview is provided of related literature on creating an optimal charging infrastructure for successful EV diffusion, followed by an identification of the literature gap and the contribution of this thesis. Finally, the reader is presented with the research question, the sub-questions and the hypotheses.

3.1 Adoption and diffusion of innovations

An innovation can only contribute to economic growth once it is widely diffused and used and it is the diffusion of a technology rather than the invention that determines the economic success (Hall, 2003). Diffusion of technology can be considered as the aggregate result of a series of individual calculations that weigh the incremental benefits of adopting new technology against the costs of change. This occurs in an environment characterized by uncertainty and limited information (Rogers, 1962)

As early as 1903, French sociologist Gabriel Tarde, curious why only ten percent of innovations had a chance of succeeding whilst ninety percent failed to spread, wrote about diffusion of innovations (DOI). He described an S-shaped diffusion curve, suggesting different rates of adoption throughout an innovation’s lifecycle. Although it was only in 1962 when Rogers presented his Innovation Diffusion Theory that this field gained more attention, this first S-curve was an important insight for future ideas about technology acceptance. The S-curve suggests that the pattern of a new technology follows a cumulative normal distribution and that consumers can be segmented into different categories according to their timing of adoption. Furthermore, consumers within each category are influenced according to certain dimensions belonging that characterize the innovation.

3.1.1 Adopter categories

The time dimension of adoption is crucial in innovation diffusion as it recognizes the dynamics of different consumers entering the market at different points in time and being offered an ever-developing version of the innovation (Backarjova et al., 2014). Rogers divides the aspect of time into three sub-elements: consumer

(19)

segments, rate of adoption and the decision stages. Particularly the first two are of interest.

Regarding consumer segments, Rogers (2003) finds that innovation adopters can be segmented into five categories depending on their speed of acceptance: innovators, early adopters, early majority, late majority and laggards. Table 2 provides an overview of the characteristics of these customer segments.

Table 2 – Adopter categories (Rogers, 2003; Sahin, 2009)

Category Characteristics

Innovators (2.5%) • High level of risk-taking propensity • Active information seekers • Advanced technical knowledge

• Able to handle high degree of uncertainty at time of adoption • Enjoy trying out new innovations

• Can easily imagine, understand and appreciate benefits of an innovation

Early adopters (13.5%) • Opinion leaders

• Similar characteristics as innovators but at a slightly lower level Early majority (34%) • Exhibit a deliberate willingness in adopting innovations

• Do not want to be the last to adopt but are not willing to take the risk to adopt ‘too’ early

Late majority (34%) • Sceptical bout innovations

• Adopt innovations when they become an established standard • Will not adopt unless they are comfortable with the ability to handle

the technology

Laggards (16%) • Are cautious about innovations • Adopt when it becomes a necessity

• Have less knowledge and little experience with innovations

Each category reflects the innovativeness, or the propensity to accept a technology, of a homogenous group of adopters. Braak (2001) supports the social character of innovations and states that innovativeness is “a relatively stable, socially-constructed, innovation-dependent, characteristic that indicates an individual’s willingness to change his or her familiar practices”. As validated, both empirically and analytically by Mahajan et al. (1990) this trend gives rise to the innovation diffusion S-curve with the various adopter categories and segment sizes illustrated in Figure 2.

(20)

Initial diffusion is slow, however, as an increasing number of consumers learn about the innovations and are convinced about its convenience, a period of exponential growth sets in and an innovation starts appealing to the wider population. Once the diffusion process reaches the early majority, a critical mass is achieved, and further diffusion becomes self-sustaining and even difficult to stop (Rogers, 2003). Rogers also explains that only adopters of successful innovations generate this curve over time; incomplete adoption and non-adoption do not form this adopter classification. Frenken (2006) adds that the exact shape of the curve is innovation specific and depends on the outcomes of non-trivial interactions between social, economic, behavioural and technical characteristics of the innovation itself and of the adopters.

Considering the characteristics and the typologies of different kinds of users, innovations must be developed and marketed in such a way that the early majority group is reached as quickly as possible so that an innovation has enough critical mass for successful, self-sustaining diffusion. This requires thorough analysis and understanding of the behaviour and preferences of these users.

3.1.2 Innovation characteristics

In addition to the time-related aspect of different user categories, the specific characteristics of innovations play an important role in understanding adoption and

Figure 2 – Cumulative and noncumulative adoption curves (Source: Mahajan et al., 1990)

(21)

diffusion. Rogers (1962) states that innovation-diffusion process is “an uncertainty reducing process” and that this process is mainly determined by the attributes of the innovation and how these attributes are valued by potential adopters. More specifically, Rogers proposes five attributes of innovations that can reduce these uncertainties: (1) Relative advantage, (2) compatibility, (3) complexity, (4) triability and (5) observability. It is the individuals’ perceptions of these attributes that determine rate of innovations.

Relative Advantage

Relative advantage is described as “the degree to which and innovation is perceived as being better than the innovation it supersedes” (Rogers, 2003). The cost and social status of innovations are elements of relative advantage. Therefore, to make relative advantage more effective, direct or indirect financial incentives may be used to support individuals in adopting an innovation.

Compatibility

Although some diffusion research considers compatibility and relative advantage to be similar, conceptually they are different. According to Rogers (2003), “compatibility is the degree to which an innovation is perceived as consistent with the existing values, past experiences and needs of potential adopters”. The adoption of an incompatible innovation often requires the prior adoption of a new value system, which is a relatively slow process.

Complexity

Rogers (2003) describes complexity as “the degree to which an innovation is perceived as relatively difficult to understand and use”. The learning curve associated with how to use an innovation is considered and a potential user must also understand why the innovation is appropriate or beneficial. In contrast to the other attributes, complexity is negatively correlated with the rate of adoption. New technologies that are simple to understand are adopted more rapidly than innovations that require the adopter to develop new skills or understanding.

Triability

“Triability is the degree to which an innovation may be experimented with on a limited basis” (Rogers, 2003). The more an innovation can be tried or tested, the faster it is adopted. Rogers explains that triability has its largest effect in the early stages of adoption and becomes less and less important as the innovation becomes

(22)

widespread; as people feel ensured that the innovations has relative advantages, compatibility and complexity characteristics make it an attractive technology.

Observability

The final dimension described by Roger (2003) is observability, “the degree to which the results of an innovation are visible to others”. For a person to adopt a technology, seeing, hearing about, or otherwise knowing that other individuals are using that technology encourages adoption. Observing increases awareness and discussion among peers and, therefore, it is positively associated with rate of adoption.

Each of these dimensions alone does not provide enough grounds to predict either the extent or the rate of diffusion. However, empirical studies, in particular Moore and Benbasat (1991), show that an innovation with a relative advantage, compatibility with existing practices and attitudes, low complexity, potential triability and high observability will be more extensively and rapidly diffused than innovations with the opposite properties. Furthermore, a meta-analysis on diffusion research performed by Tornatzky and Klein (1982), finds that relative advantage, compatibility and the inverse of complexity had the greatest influence on adoption. In contrast, Agarwal and Prasad (1997) find that complexity can be discredited as a predictor for adoption of innovation.

All in all, it can be concluded that each of these dimensions should be considered when studying the potential rate of diffusion of an innovation, although the specific significance of each dimension may vary from one innovation to another.

3.1.3 Diffusion of electric vehicle

Several authors have investigated the application of Rogers’ diffusion of innovation theory to electric vehicles. Starting with the identification of the five customer segments, various surveys have been carried out to identify characteristics that can be attributed to the target segment in the initial stages: the innovators and early adopters.

Econ analyse (2006) carried out a survey among EV owners and find that the typical owner is a male, between 30 and 60 years old, married or living together and has an above average level of education and income. Furthermore, the vast majority, 90%, owns more than one car.

(23)

A study performed in 2009, by Rødseth, included a sample of EV owners as well as a randomly sampled group of driving licence owners. The study finds similar results to Econ analyse, with more EV owners being men compared to the random sample and most owners aged between 30 and 50 and having obtained an education at university level, again relative to the random sample.

Other international studies (Pierre at al., 2001; Campell et al., 2012) also suggest that adopters of electric vehicles have high incomes and high levels of educations.

In studies focusing on hybrid electric vehicles owners, similar results arise, with higher levels of income and higher levels of education characterizing the group of early EV adopters. However, in contrast to studies on EVs, hybrid owners were generally older than the average car buyer (Scarborough, 2007; De Haan et al., 2006; Ozaki & Sevastyanova, 2011).

With respect to the innovation itself, Struben and Sterman (2008) investigate how the five innovation characteristics apply to the EVs. Firstly, they suggest that there are a number of relative advantages for EVs including lower fuel costs, lower maintenance costs, no direct emissions, (possible) parking cost advantages and subsidies. On the other hand however, the high purchase costs compared to ICE vehicles are a significant disadvantage.

Gärling and Thøgerson (2001) also argue that triability is not a large barrier as many rental companies offer EVs for short use and initiatives, such as Car2Go in the Netherlands, offer people the opportunity to experiment with EVs at low costs. Additionally, due to the fact that many EVs have striking designs and are parked and connected to a charging station make these cars very observable.

With respect to complexity, Struben and Sterman (2008) as well as Gärling and Thøgersen (2001) identify a number of issues. Even though the electric vehicle may be easier to use compared to an ICEV since it has no gearbox, the reduced range means that it must be charged more often. Also, the way of charging is different compared to traditional petrol refuelling, which may also be perceived as being difficult. This complexity is reinforced when charging stations are scarce.

In the same line, two compatibility issues arise. Firstly, due to the lower range, an EV may not be compatible with the general perception of what a sufficient range should be, even though most trips are less than the maximum range of an EV. The

(24)

second issue Gärling and Thøgersen (2001) identify is caused by the need to recharge after every trip instead of refuelling after a few hundred kilometres as is the case for ICEVs. Drivers are not familiar with this and will have to develop new routines to overcome this incompatibility. The characteristics of EVs compared to ICEVs along Rogers’ five attributes are summarized in Table 3.

Table 3 – Positive and negative EV characteristics compared to ICEVs along Rogers’ five innovation attributes

Innovation attributes Positive EV characteristics Negative EV characteristics Relative advantage + Lower fuel costs

+ Lower maintenance costs

+ No direct emissions

+ (Possible) parking

advantages

+ Subsidies

− High purchase costs

Compatibility − Limited range

− Charging frequency

Complexity + No gear box − Limited range

− Charging frequency − Difference in manner of

refuelling/charging

Triability No significant difference compared to ICEV

Observability ₊ Striking designs

+ Connected to charging

station

All in all, the challenges for EVs as an innovation itself lie in the higher price and the issues related to range anxiety and charging. Technological developments over time are likely to lead to declining prices as well as increased range. However, to engage consumers at present, solutions to these issues must be found. Focusing on overcoming range anxiety, electric vehicles are dependent upon an adequate charging infrastructure. As shall be explained in the next section, a complex situation arises from this dependency, commonly referred to as the chicken-and-egg dilemma.

(25)

3.2 Infrastructure dependent innovations

3.2.1 The chicken-and-egg dilemma

Generally, market shares of innovations follow the same trend as stated by Rogers and explained above. However, the diffusion of EVs is a more complex process as it not only implies the innovation of a single product, but also encompasses necessary changes in the environment. EVs, like many other transport, as well as energy and communication technologies, fall into the category of Large Technical Systems (LTS) (Mayntz & Hughes, 1990; Geels, 2005). These systems can be characterized as capital intensive and durable and are made up of tangible as well as intangible components; artefacts, organizations, scientific components, legislative artefacts and natural resources (Mayntz & Hughes, 1990; Van der Vooren et al., 2012). These components are highly interrelated and must interact and change in order for the product to be perceived as attractive, along the five dimensions suggested by Rogers (1962), and consequently reach a high level of diffusion. The difficulty in such situations is that multiple actors must simultaneously invest and ramp up technology in order to commercialize an innovation. For EVs specifically, the technology at present is ‘charging infrastructure dependent’; without the necessary charging stations, range anxiety will continue to form a major barrier, making large scale adoption of EVs unlikely. Simultaneously, there is reluctance towards investing in charging infrastructure if uncertainty exists concerning the adoption of electric vehicles. Thus, an indirect network effect,

commonly referred to as the chicken-and-egg dilemma, arises and creates a deadlock situation (Figure 3). This case of market failure is characterized by the inability to reach a certain desired outcome because a necessary precondition is not satisfied, while to meet that precondition in turn requires that the desired outcome has already been realized.

Two types of agents are central to the chicken-and-egg dilemma for the transition towards electric mobility. The first group is made up of end consumers that make the decision whether or not to buy an EV. The second group, however, is Figure 3 – Chicken-and-egg dilemma

(26)

equally essential and consists of actors responsible for setting up the charging infrastructure (Van der Vooren, 2010). Due to the high costs of infrastructure components, the second category is inclined to play a waiting game (Huétink et al., 2010; Van der Vooren, 2010). This in turn implies that the demand for EVs, that are reliant upon the infrastructure, takes off more slowly as the compatibility of EVs remains low and the complexity to use them without being able to charge them remains high. Moreover, these two components reduce the relative advantages that adopters perceive in adopting an EV compared to an ICEV. The relationship between charging infrastructure and the perceived advantages of EV is illustrated in Figure 4.

Linking DOI to the chicken-and-egg dilemma, Grübler (1990) distinguishes between two S-curves separating the diffusion of infrastructure and the diffusion of technology (Figure 5). Based on empirical data, he finds that both curves are interdependent but have different development patterns in time. Although the two S-curves take off at the same point in time, the diffusion of infrastructures precedes the diffusion of the technology that is using the infrastructure. Furthermore, the rate of diffusion is also somewhat lower for the technological component. However, Grübler does not go into more detail concerning the factors that influence the two S-curves. Furthermore, researchers at Cornell (Li et al., 2015) agree, stating that the starting point for EV development is to build enough public charging stations to reassure users that they will be able to use their car with similar ease as ICEVs.

Perceived(a*ributes(EV:( ( •  Rela4ve(advantage( •  Compa4bility( •  Complexity( •  Triability( •  Observability( ( ( Charging(infrastructure( ( ! Adop4on(

Figure 4 – The relationship between charging infrastructure and perceived EV attributes

(27)

Clearly, infrastructure development is critical to the success of EVs. However, an important issue remains who should be responsible for constructing such an infrastructure and how the construction should be timed (Grübler, 1990; Van der Vooren, 2010). Although the literature is limited and inconclusive about this issue, looking into historical chicken-and-egg problems gives us a better understanding of the main obstacles in such dilemmas. The following section provides a brief account of similar deadlocks that have arisen during the diffusion process of infrastructure dependent technologies and how these have been overcome.

3.2.2 Overcoming the chicken-and-egg dilemma

Over the past century, numerous technological breakthroughs have faced similar chicken-and-egg dilemmas. Examples include satellites and cellular radio towers that needed to be in place for mobile phone, optical fibre cables necessary for computers to be connected to the internet and, similar to the EV dilemma, gas stations and roads that had to be in place for the adoption of ICEVs (Grübler, 1990). All of these fall into the category of infrastructure dependent technologies for which, in order to serve even a few users of a technology, a significant amount of infrastructure should be available beforehand. Due to this discrepancy between demand and supply in the early stages, the above-mentioned interdependency, high investments and high uncertainties, market forces alone are not sufficient to drive the adoption of such technologies. Governments and policy makers therefore play an essential role in overcoming the deadlock and initiating diffusion (Van der Vooren et al., 2012).

(28)

In all cases above, the transitions to the new technologies are considered desirable as they are expected to increase overall welfare. Although consumers eventually determine the success of a technology, Van der Vooren (2010) states that governments are essential to kick-start adoption and can also influence the speed and success rate of the diffusion process. This can be done in two ways: (1) direct and/or indirect subsides and (2) by setting standards and providing regulations. Respectively, these two mechanisms reduce the issues related to high investments and high uncertainties.

In Geels (2005) the example of the transition from horse-drawn carriages to ICEVs as a means of personal transport has been investigated closely. For cars to replace horse-drawn carriage, a whole new environment for personal transportation had to be created. Geels approaches the issues using the multi-level perspective (MLP) of socio-technical systems, arguing that transitions come about through the alignment and interaction of dynamics at three levels: 1) technological niches (micro), 2) socio-technical regimes (meso) and 3) socio-technical landscapes (macro).

In short, technical niches act as ‘incubators’ for radical, new technologies that are initially unstable configuration with low performance (Schot, 1998). When these niche-innovations are sufficiently stable they will form part of the mainstream market to increase competition and consequently put pressure on existing regimes. On the other side, the landscape level represents the broader exogenous environment, putting pressure on regimes and opening windows of opportunity. The regime is made up of three dimensions: (i) the socio-technical system comprising of all elements pertaining production, distribution and use of technology (ii) actors consisting of organizations and groups involved in the system and (iii) formal (laws), normative (expectations) and cognitive rules (routines and user practices) that guide the activities of the actors (Van Bree et al., 2010). The relationship between the three levels is hierarchical as the niches are embedded within the technological regimes, and the regimes within the landscapes.

For personal transportation, during the first decades of the 20th century, Geels (2005) explains that the internal combustion engine represented a niche technology. ICE technologies had been developing since the end of the previous century; however, it took massive subsidies in public roads, refuelling infrastructure and suburban development to achieve the transition to widespread use of ICEVs as a

(29)

mode of private transportation. These needed to be in place before people were prepared to buy their cars and required heavy government investments. It also holds for mobile phones, Internet and electricity networks that in order to serve a few users of a technology, a significant amount of infrastructure should be available already (Van der Vooren, 2010).

The importance of government expenditure is investigated further by Van der Vooren et al. (2012). They extend their prior research and, for infrastructure-dependent vehicle technologies, investigate the trade-off between bottom-up expenditures relating to technological R&D, or top-down investments in infrastructure development. Government funding therefore can have a bottom up (R&D) as well as top down effect (infrastructure investments) and can play an important role in technologies gaining enough momentum to become self-sustaining. However, they emphasize that governments must work on a restricted budget due to the risk associated with overcoming market failures and are therefore obliged to carefully prioritize expenses. Their findings indicate that policymakers should allocate the majority of the available financial resources to the support of public infrastructure developments. Furthermore, they argue that public spending on R&D is only rewarding when there is enough infrastructure to support the anticipated innovations. Relating this to EVs, it implies that governments should, for example, restrain from supporting R&D for improved battery capacity until there is adequate infrastructure for EVs in their current state.

All in all, private parties are not willing to take the risks of overcoming the chicken-and-egg dilemma. Therefore, government expenditure is crucial in the initial stages to create an environment in which potential customers feel confident and comfortable enough to adopt new technologies. Government expenditure and charging infrastructure expansion remains essential until the “early majority” group starts adopting the new technology. At this point, the innovation becomes self-sustaining and further developments will be regulated through market forces.

Since all cities stimulating EVs have had to start from scratch in designing the charging infrastructure, there still exist many uncertainties concerning the most desirable type of public charging stations, the optimal locations and the necessary scale of the network. Depending on a wide variety of factors, such as consumer behaviour, location characteristics, grid capacities and demographics, a charging

(30)

network must be set up that is technologically viable and mitigates the problem of range anxiety. Furthermore, it must also aim at minimizing costs and maximizing utilization to ensure sustainability in the long run. Developing charging infrastructure blueprints that achieve these objectives should contribute to overcoming the chicken-and-egg dilemma. In turn, this will encourage private parties to invest in further infrastructure development and stimulate the adoption of electric vehicles. The following section provides an overview of the literature that has focused on these issues.

3.3 Charging infrastructure planning

The chicken-and-egg dilemma illustrates the complexity associated with planning an adequate charging infrastructure to ensure that range anxiety among potential EV adopters is relieved and the rate of diffusion amongst the initial groups is increased.

Different views exist regarding the planning and deployment process of a public infrastructure, and generally the main arguments concern (1) who should be responsible for the planning and layout, (2) what type of charging stations should be placed, (3) at what rate the charging infrastructure should be placed and (4) where the charging stations should be located.

The solutions to chicken-and-egg dilemmas presented above imply that the role for governments, especially in the initial phases, is very important to overcome the deadlock arising from EV users and infrastructure developers’ reluctances to taking high risks. Furthermore, it is clear that a standardized system which is accessible to all EV users is important, meaning that plugs, sockets and power have to be standardized so that a charging infrastructure is deployed that is usable for all EV users. This section therefore focuses on the literature concerning points (3) and (4) mentioned above.

Regarding the rate of deployment, in some cases a very reactive approach is preferred, only placing new CSs to suit demand whilst other cases have shown a more proactive approach by placing CSs according to predicted future demand. In extreme cases, a complete charging infrastructure could be installed directly, extensive enough to fulfil demand for the next 20 years. Beella et al. (2009) state that such a supportive infrastructure being in place prior to the introduction of a technology is essential to convince consumers to buy EVs. On the other hand, Katz and Shapiro (1985) believe

(31)

that expected developments in the industry play an important role in the consumers decision, implying that a gradual build up may be a successful approach due to uncertainty concerning the future for EVs, thereby reducing risks by spreading investments over time. Also, if battery technologies were to advance at a much faster rate than expected, an infrastructure that is very extensive may lead to underutilized charging stations and low returns on the investment.

The research done on actual placement and utilization models for EV charging has developed rapidly over the past decade, including both qualitative and quantitative approaches. Furthermore, some authors have chosen to focus on specific cities or areas whilst others have attempted to develop more general models. Finally, a wide diversity of factors believed to influence demand have been tested to determine ideal combinations for rate of deployment, locations and sizing for charging stations considering government budget constraints (Franke and Krems, 2013; Chang et al., 2012; Wagner et al., 2013).

Despite the diversity in approaches, it is possible to divide the literature into three streams, each with a distinct perspective. Firstly, a supply side perspective can be identified that seeks to minimize costs for construction, operation and maintenance of the charging infrastructure. Secondly, a demand side stream of literature focuses on models to minimize costs for EV users. Thirdly, a mixed approach attempts to combine the previous two perspectives and provide a complete model that with optimal solutions for all parties involved. Although the demand side and mixed perspectives provide the basis for this research, all three perspectives have been added to this literature review to provide a broad summary of the status quo regarding charging infrastructure studies.

3.3.1 Supply side perspective

The stream of literature applying optimization techniques from a supply side point of view focuses on minimizing at least one of the following three components of the costs associated with charging infrastructure; construction costs, operational costs and/or maintenance costs. In many cases, these costs fall under one single actor. In these cases, this actor is referred to as the charging infrastructure developer.

Several authors have looked into the possibility of adapting current refuelling infrastructure, i.e. gas stations for future use as EV charging stations. Wang et al. (2010) process data from Chengdu to determine an ideal infrastructure layout using a

(32)

numerical, multi-objective planning model. Their goal is to minimize construction costs by making the best possible use of existing gas stations, whilst being able to cope with current EV fleet energy demands as well as preparing for future EV quantities. Furthermore, the model is built considering the restrictions due to the condition and capacity of the power grid. The model they develop considers a wide range of factors for predicting energy demand that can be expanded, revised and optimized according to the development of EVs.

Similarly, Chen and Hua (2014) also propose a location model taking the existing gas stations as candidate sites in order to save construction costs and land expenses. However, in this case the authors focus on fast-charging stations that can recharge EV batteries in 20 minutes instead of 6-8 hours. This much shorter timeframe implies that fast-charging will be more similar to gas refuelling, making it more understandable to make use of existing gas stations. They conclude that although a network of individual CSs would also be necessary for urban settings, transforming and reusing traditional gas stations for future use as EV fast charging stations, would be a way to reduce high initial construction expenses.

Liu et al. (2012) develop a two-step optimization model aiming to minimize all three expenses; construction, operating and maintenance costs. Taking into account factors such as land price, proximity to power supply and convenience of EV owners as well as the service radius of each CS, the authors firstly identify optimal charging sites. In a second step, a mathematical model for the optimal sizing is developed. The model they develop not only optimizes planning schemes but also reduces network losses and significantly improves the voltage profile. If applied, these improvements can result in cost savings for charge point operators.

Ip et al. (2010) focus their planning model solely on charging locations in urban areas characterized by high congestion, restricted street space and several technical factors such as the distribution of power grids. They propose a two-step model, first converging the road information into ‘demand clusters’ by hierarchical clustering analysis and then applying optimization techniques under certain constraints and cost considerations to maximize usability and minimize operating costs. They believe their model is useful for city planning, and designing a refuelling infrastructure for EVs in an urbanized area from scratch. However, this may not be

(33)

very suitable when the charging infrastructure is rolled out more gradually as this may require a more evolutionary or adaptive approach.

Likewise, Frade et al. (2010) focus on urban areas, taking Lisbon as their case study. Their objective is to minimize the amount of CSs necessary by formulating a discrete maximum covering model to maximize the demand each CS can cover in a certain area. The constraint they apply is that the demand is served by an ‘acceptable’ level of service. However, the usefulness of this research may be questioned as it is not only a subjective constraint, but the constraint is very likely to change over time due to technological advancements or a higher rate of EV diffusion.

In Xi et al. (2013) the objective is also to maximize the service level per EV, thereby minimizing the amount of CSs necessary. In additional to determining locations, Xi et al. also model the combination of level one and level two slow chargers to further minimize costs whilst fulfilling demand.

The importance of considering the grid has been acknowledged by a number of the abovementioned authors but remains questionable. Research done by Beella et al. (2010) focuses not so much on the location problem; however, they do conclude that it is very important for developers to ensure that charging is separated over a 24-hour day, with day and night having different capacities. Indirectly it is therefore important to consider such patterns when placing CSs to ensure that the grid is not excessively strained in certain areas. In contrast, Clement et al. (2009) find that models based on grid capabilities are not very relevant for modern cities. The grid may need reinforcing in some cases but should generally be adept for basic charging infrastructure. Furthermore, by using smart charging schemes, power grid stress can be alleviated under peak demand and meet demand requirement in the regular market (Cao et al. 2012). Botsford and Szczepanek (2009) support these finding and claim that extensive modelling has shown that even fast charging stations, which require much more power, have no negative impact on the grid.

Sweda and Klabjan (2012) state that understanding how EV adoption occurs with respect to geography as well as to demographics will prove critical to determining the most cost effective charging infrastructure deployment strategies. For their research they use the Chicago region as a case study. By applying an agent-based decision support system, identifying patterns of EV ownership and driving activities, strategic locations for new charging infrastructure are determined.

Driving electric vehicles towards the early majority : the importance of the public charging infrastructure