Public charging choices of electric vehicle users: A review and conceptual framework

Studying electric-vehicle public-charging choices is an important aspect of accelerating electric-vehicle adoption. Understanding which factors would entice existing and potential users to charge their electric vehicles at public locations would provide important evidence for policy-makers, vehicle manufacturers and charging providers. The academic and grey literature has now matured to a point where a critical synthesis of the current state of knowledge is necessary. To fulfil this, this review provides a synthesis and critical discussion of the most up-to-date evidence on public charging choices based on which a conceptual framework to match choices and their determinants is devised. Research gaps and further empirical evidence required in this area, as these emerged from this review, include better understanding of the temporal patterns of public charging, users ’ preferences for different types of charging locations, payment models (monthly subscription, pay as you go) and payment methods.


Introduction
Increasing the market share of electric vehicles (EVs) is a potentially effective route to decarbonise road transport and improve air quality in cities (European Environment Agency, 2022).To successfully achieve this, reliable and sufficient charging infrastructure needs to be developed (Coffman et al., 2017;Liao et al., 2017).Charging points and associated services are currently divided into home, work, and public charging, with home charging being the most popular worldwide (Baresch and Moser, 2019;Chakraborty et al., 2019;Delmonte et al., 2020;Hardman et al., 2018).In 2018, for example, private chargers accounted for over 90% of global charging points (International Energy Agency, IEA, 2019).Although, home and workplace charging plays a fundamental role in fulfilling most potential and existing EV users' charging requirements, this does not mean public charging is not important (Hopkins et al., 2023;Jochem et al., 2022).
The study of public charging choices regarding the availability of public charging is important for building confidence in future EV purchases and in terms of addressing range anxiety and thus facilitating a faster transition to electric vehicle adoption (Greene et al., 2020;Kester et al., 2018;Santos and Davies, 2020).It is also crucial to study public charging choices for infrastructure planning in order to accommodate the existing demand for charging, and especially for EV owners without access to a driveway or a parking space with an EV charger.A better understanding of the public charging preferences of potential and existing EV users is of utmost importance for: the deployment of public chargers, such as where they should be placed; what charging speeds and public infrastructure designs would be required; how they should be priced; how many and when they should be deployed; and what public charging information system should be made available to customers.Such knowledge would greatly benefit government decisionmakers to tailor policies to regulate and support the deployment of public charging infrastructure, vehicle manufacturers to design and produce EV models that meet customers' preferences and maximise the efficiency of charging infrastructure, and public charging providers to plan future infrastructure and make related investment decisions.All of these together will support the development of an adequate public charging infrastructure for the future.
The inadequate provision of public EV charging infrastructure is a pressing problem, especially in the light of the announcements by several governments to bring forward more stringent CO 2 emission targets and restrictions on conventional petrol and diesel vehicles in urban areas.For example, the UK Government plans for at least 50% of new registrations to be low emission vehicles by 2030.This target requires the number of charging points to increase tenfoldi.e., from 25,000 public charging points currently available to 280,000 by 2030 (Competition & Markets Authority, CMA, 2021).More generally, this is an emerging challenge, particularly for highdensity urban areas.For example, nearly half of households in Europe live in multi-family buildings and would face significant challenges to install a home charging point due to the type of dwelling they live in (Azarova et al., 2020).For many of these households, workplace and public charging would be essential.As a result of all this, it is imperative that the coverage of public charging infrastructure is maximised.
The literature on public charging choices has now reached a mature stage with a substantial amount of empirical evidence being published in the academic and grey literature (see Tables A1 and A2 in the Appendix A).This literature review is designed to provide a synthesis of the current-state-of-knowledge and do so beyond anecdotal evidence or an unstructured pull of references.The synthesis generated in this study provides solid evidence for researchers entering this field of study and would benefit charging point operators, vehicle manufacturers and decision-makers to help them plan, design and supply public charging.Better knowledge of public charging choices in turn will: (a) attract potential EV users who have less knowledge about EV and EV charging, (b) attract existing and potential EV users who are risk-averse towards range, and (c) attract potential EV users who would be unable to charge their vehicles at home because of the type of dwelling they live in.
To the best of our knowledge, only the review by Zhang et al. (2018) focused on examining the determinants of the economic aspects of public charging points, which differs from the subject of enquiry in this present paper.Hardman et al. (2018) reviewed studies on consumers' charging preferences more generally rather than for public charging.The usage habits of home charging and public charging are completely different in terms of temporal patterns (e.g.time and frequency of use).This is not trivial, and therefore conducting a literature review to understand the factors of existing and potential EV users' choices regarding public charging is in order.This is particularly important and it needs to be examined separately from home charging considerations.In addition, Zhang et al. (2018) and Hardman et al. (2018), although very informative, are now five years old, which, in a context of a rapidly developing field, calls for an update on the 'stock of knowledge'.This is exactly what this article intends to do, by providing an up-to-date source of reference to an emerging and increasingly important aspect of the electrification of road transport and the deployment of public charging infrastructure for electric vehicles.
This critical review of the literature is aimed at synthesising the published academic and grey literature on public charging choices and answering the following questions: (1) How were public charging choice studies designed, implemented and analysed?(2) What factors are likely to determine public charging preferences?(3) Are preferences heterogeneous?If yes, what are the factors that may explain heterogeneity in preferences?(4) What are the areas for future research?
In the remainder of this paper, Section 2 provides a summary of studies including study design and analytical approach, respondent profiles, countries of study, and type of choices studied.Section 3 discusses the explanatory factors found to be associated with public charging choices.Section 4 presents findings regarding observed and unobserved heterogeneity in choices for public charging.Section 5 summarises the findings and offers a conceptual framework, which captures how explanatory factors are linked with public charging choices.Finally, Section 6 summarises the research gaps in the area of public charging choices and provides recommendations for future research.

Table 1
Review articles related to electric vehicle charging infrastructure.

Topic Reference
The comparison of infrastructure across countries

Methodology
This targeted critical literature review identified relevant academic publications via three repositories: Web of Science, Scopus and Google Scholar.The keywords used to identify these studies were: 'charging preferences' OR 'choices' and 'electric vehicles'.The searches were limited to journal articles published in English from 2016 onwards to obtain an up-to-date picture of the literature and its emerging findings.In addition to academic sources, the study targeted relevant grey literature from government, research institutes, and national laboratories.

Overview of the literature related to public charging choices
Table A1 in the Appendix A summarises relevant studies on public charging choices and preferences.The majority of studies utilised stated choice experiments, including experiments where preferences were elicited in a hypothetical electric-vehicle charging scenario.There were also a few of studies that employed revealed preferences, where actual charging choices were observed.For example, some studies collected data from EV trials, such as Chakraborty et al. (2019), Kim et al. (2017), Lee et al. (2020), Sun et al. (2016), Xu et al. (2017), and Yu and MacKenzie (2016).
In terms of type of charging, the majority of published articles did not specify whether it was en route or destination charging.Only four studies focused on en route charging (Ge and MacKenzie, 2022;Li et al., 2023;Sun et al., 2016;Visaria et al., 2022), while five articles looked at destination charging (Latinopoulos et al., 2017;Ma et al., 2022;Pan et al., 2019;Wang et al., 2021;Wen et al., 2016).Regarding charging power, only two articles focused specifically on fast charging (Sun et al., 2016;Visaria et al., 2022), which was generally adopted for charging, while the five studies that looked at destination charging, three of which mentioned the power of public charging infrastructure, were mainly focused on both slow and fast public charging (Ma et al., 2022;Wen et al., 2016).
Only three studies specified the type of trips made by EV users, with two studies focusing on long-distance trips (Ge and MacKenzie, 2022;Visaria et al., 2022), and one study looking at commuting trips (Chakraborty et al., 2019).Four studies did not specify the type of trip in their charging scenario, but did mention that these trips were to home (Ma et al., 2022;Pan et al., 2019;Wang et al., 2021), workplace (Ma et al., 2022;Wang et al., 2021;Zhang et al., 2022), shopping (Pan et al., 2019;Wang et al., 2021) or recreation (Zhang et al., 2022).The type of trip (commute, leisure, long-distance) was closely associated to the type of charging, with long-distance trips mainly associated with en route charging (Ge and MacKenzie, 2022;Visaria et al., 2022), and commute and leisure trips in the context of a shopping centre/workplace/home mainly associated to destination charging (Ma et al., 2022;Pan et al., 2019;Wang et al., 2021).
In general, very few studies extended their analyses to derive the relative value of attributes such as willingness-to-pay (WTP).As shown in Table A1 (see, Appendix A), only five studies, which focused exclusively on EV users (Ge and MacKenzie, 2022;Nienhueser and Qiu, 2016;Visaria et al., 2022;Wang et al., 2021;Wen et al., 2016), and three studies on current and potential EV users (Ma et al., 2022;Sheldon et al., 2019;Wolff and Madlener, 2019) estimated the WTP for different attributes related to public charging choices, although many studies involved 'payment vehicles' (e.g.charging cost) as part of their stated choice experiment (Ge et al., 2018;Latinopoulos et al., 2017;Moon et al., 2018;Pan et al., 2019;ten Have et al., 2020;Wolbertus and van den Hoed, 2019).
Many binary logit and MNL models captured respondents' taste heterogeneity, i.e. how choices varied across respondents given observed characteristics (e.g., socio-economic or demographic characteristics).Some MXL and LC models captured the effect of unobserved characteristics (e.g., level of risk aversion or charging concerns) assuming continuous or discrete distribution of related parameters (weights) of charging attributes.Similar to MXL, RPL and RPLEC models also examine unobserved heterogeneity in individual preferences for route and charging-or-not choices en route, by modelling the random parameters and error components.
Trip chain choices and charge-or-not choices were analysed using the RSBP model, which is a statistical model used to analyse the relationship between two binary dependent variables that may be jointly determined (Zhang et al., 2022).To take into account the effect of time, DDC models were used to analyse how EV users made charging decisions over time with a sequence of charging opportunities en route, which allows to understand how past decisions affect future charging choices (Ge and MacKenzie, 2022).Kim et al. (2017) estimated a latent class hazard duration (HD) model, which allowed for the inclusion of duration dependence, unobserved heterogeneity and the effects of time-varying covariates, and for the treatment of charging regularity as a latent variable.There was also a study in which the heterogeneity of respondents' attitudes towards risk was captured using "Expected Utility Theory", "Rankdependent Expected Utility Theory", and "Prospect Theory" models (Latinopoulos et al., 2017).These approaches, as the authors argued, were suitable for choices with uncertain outcomes (e.g.presence of dynamic charging prices; charge now or wait for lower price).
In parallel, there have been several initiatives across government departments and agencies to capture and report aspects of public transport infrastructure choices for potential and existing EV users.Examples include the California Air Resources Board (2019), the California Energy Commission (Bedir et al., 2018), and the National Renewable Energy Laboratory (Wood et al., 2017) in the United States.In the UK, there have been several studies undertaken by the Department for Transport (Department for Transport, 2022b; Department for Transport, 2022a), the Electric Vehicle Association, England (Hink, 2021), Transport Scotland (2021), and the National Grid ESO (Dodson and Slater, 2019).Also, another study was undertaken on behalf of the Energy Efficiency and Conservation Authority in order to understand public charging choice and improve the user experience in New Zealand (Burroughs et al., 2021).The determinants examined in these studies, the number of respondents, and the modelling approach are summarised in Table A2 in the Appendix A.
Most of these reports conducted stated preference surveys involving potential and existing EV users (Department for Transport, 2022b;Department for Transport, 2022a;Hink, 2021).Only the New Zealand (Burroughs et al., 2021) and the National Grid ESO (Dodson and Slater, 2019) studies analysed actual charging choices (releveled preferences) and charging events of existing EV users, respectively.The survey-based samples of potential and existing EV users across these studies were around 1,000 whereas those involving the study charging events of EVs reached several million records.The reports presented descriptive statistics instead of statistical models to identify the factors driving public charging choices.

Determinants of individual public charging choices
The thematic analysis of the literature identified four categories of determinants associated with public charging choices: (1) temporal, (2) vehicle, (3) charging infrastructure, and (4) individual attributes.These are discussed in detail in the following subsections.

Temporal attributes
The temporal characteristics of public charging for electric vehicles (EVs) refer to the patterns of charging-point usage over time, including when and how often charging occurs.Understanding these temporal characteristics is essential for designing and managing efficient and accessible public charging to meet the needs of existing and potential EV users (Gellrich et al., 2022).A summary of temporal attributes and their levels is listed in Table A4 in the Appendix A. The value that potential and existing EV owners place on the 'time of the day' and 'day of the week' can inform the pricing strategies of charging service providers.
Table 2 presents a summary of positive and negative effects of temporal attributes.These temporal attributes can be further categorised into three groups: (a) time of day, (b) weekday (vs.weekend) and (c) charging interval/frequency.In the corresponding studies, the 'time of day' either considered day vs. midnight charging choices or different time periods for public charging during the day.'Weekday vs. weekend' attribute refers to public charging preference variations between weekdays and weekends.The 'charging interval/frequency' attribute was used to capture the effect of environmental conditions and on charging choices.
Unlike private home charging, which is more likely to occur overnight (Langbroek et al., 2017;Sun et al., 2018), public charging was the preferred method of EV charging during the day (Ma et al., 2022;Moon et al., 2018).This is evident by both observed charging data at charging points and preference-based studies.For example, Helmus and Wolbertus (2023) examined over 2 million charging sessions at 1,689 public charging stations in Amsterdam, the Netherlands, and confirmed that public charging was more popular between 8:00 and 23:00, with a peak between 16:00 and 19:00.Also, the National Grid ESO report in the UK (Dodson and Slater, 2019) analysed 8.3 million charging events across the country and found that slow/fast public charging contributed to a smaller secondary peak in the morning between 9:00 and 10:00 on weekdays.
Another source of real-world charging data is the BEVs themselves.For example, Märtz et al. ( 2022) studied 2.6 million charging sessions from approximately 21,000 BEVs across Germany for over a year.The analysis of the vehicles' charging patterns provided insights on the distance driven between charging sessions, charging frequency and energy requirements per BEV.Based on these data, the authors were able to identify different vehicle-user groups ('clusters') according to their vehicles' temporal charging patterns and inferred the demand for potential charging points according to identified electricity load curves.Observed charging sessions can complement revealed and stated preference studies when investigating potential demand for public charging points.
As shown in Table 2, Xu et al. (2017) modelled actual (revealed) charging choices across three options: home-slow, public-slow and public-fast charging of EV users in Japan.Their study found a positive association between home charging and a negative association with fast public charging during midnight.Also, Ma et al. (2022) introduced two different charging plans, as part of a stated choice experiment for public charging, with each plan proving the time period during which charging would occur, the cost of charging, 'walking distance to home/work', and the 'type of charger (slow/fast)'.Ma et al. (2022) showed that WTP for public charging was highest between 17:00-22:00 on weekdays and lowest for that same time period at weekends (see, Table A3 in the Appendix A).
In Japan, Xu et al. (2017) found no significant difference in choices made by EV users across slow-home, slow-public and fast-public charging between working days and weekends.However, relative to weekends, business EV users exhibited a significantly higher preference for fast charging at public stations and then slow charging at the workplace during working days.These findings were consistent with the report by Dodson and Slater (2019) for the UK, which found that the overall demand for charging at weekends was approximately 25 lower %, on average, than on weekdays.
The difference in public charging preferences between weekdays and weekends could be associated with travel purpose/type of travel (Xu et al., 2017;Zhang et al., 2022).For example, long-distance travel usually takes place at weekends and local commute/ business travel normally occurs on weekdays.Longer and more complex trip chains involving two or more destinations are more likely to occur on working days, and this observation may explain the corresponding preferences for public charging (Zhang et al., 2022).As a result, it is important to consider travel purpose when studying public charging choices.
The study of 'inter-charging times,' which can be defined as the time-interval between charging sessions, also offers interesting insights.For example, Kim et al. (2017) reported data from charging sessions at public charging stations over a four-year period in the Netherlands and found that 90% of their sample charged their EV randomly at public charging stations, and 10% of their sample did so regularly.The inter-charging times were, on average, 5.65 days and 2.75 days, respectively, with the regular users being more likely to charge their vehicle at a specific charging station (Kim et al., 2017).Charging intervals were significantly associated with weather conditions such as high temperatures, strong winds, and heavy precipitation.These weather conditions might have caused EV users to postpone charging at a public charging point.The authors suggested that providers might mitigate the negative impact of weather on public charging choices by 'minimising exposure' to harsh weather; for example, by reducing walking distance and/or improving shelters (Kim et al., 2017).

Vehicle attributes
Several studies reported that public charging demand was significantly associated with driving range due to EVs' different battery capacities (Chakraborty et al., 2019;Li et al., 2023;Xu et al., 2017) and charge status, such as low battery state of charge (SOC) (Pan D. Potoglou et al. et al., 2019;Wang et al., 2021;Xu et al., 2017;Zhang et al., 2022), available/remaining range (Li et al., 2023;Zhang et al., 2022), and insufficient excess range to the planned destination (Pan et al., 2019;Wang et al., 2021;Wen et al., 2016) or the next opportunity to charge the vehicle (Wen et al., 2016).Whilst the capacity of the battery and driving range are fixed, the SOC and the excess range are not.The positive and negative effects of these determinants on public charging choices across countries or cities in the reviewed studies are shown in Table 3 and a summary of attributes and their levels is presented in Table A5 in the Appendix A.
As shown in Table 3, Chakraborty et al. (2019) found that a higher driving range of PHEVs might entice drivers to use public charging (vs.home or workplace), which is a counter-intuitive result.The reason behind this counter-intuitive result could be one of PHEVs in the study included in the study, namely the BMW i3s (with the range extender).This particular car model can potentially benefit from fast (and possibly free) charging sessions at public locations.On the other hand, the authors found that multiple stops for public charging along the route were less likely to occur when the driving range of the PHEV was higher, which is an intuitive result, as PHEV drivers can start and finish their trips on electric mode with a single charge.The study also showed that owners of a Tesla with a relatively larger battery capacity were more likely to use public charging stations than owners of other EVs available on the market (Chakraborty et al. 2019).In Japan, although Xu et al. (2017) reported no significant association between public charging and the availability of fast chargers, they did find that company EV car users were less likely to use a fast public charger when a slow charger was available at the workplace.
Insufficient charge status of EVs can trigger range anxiety and incentivise users to opt for public charging (Pan et al., 2019;Wang et al., 2021;Wen et al., 2016;Xu et al., 2017).The charge status of an EV is commonly expressed by three attributes, including the current SOC, the remaining range/ available range and the excess range.SOC is the current charge level displayed on the dashboard of an EV whereas the remaining range / available range shows the distance that the vehicle can travel at the existing SOC, and the excess range is defined as the difference between the range at SOC and the distance to the destination.
Previous studies showed users' decisions to use public charging were negatively associated with SOC (Pan et al., 2019;Wang et al., 2021;Xu et al., 2017), initial available range at the origin, average remaining range at the destination (considering any uncertainties such as traffic conditions) (Li et al., 2023), excess range to the destination (Wang et al., 2021;Wen et al., 2016) and next charging opportunity (Wen et al., 2016).In terms of driver characteristics, BEV users with little driving experience (less than1.5 years) were more concerned about average remaining range at the destination and therefore were more likely to use public charging than those with more driving experience (Li et al., 2023).
In terms of the relative valuation of vehicle characteristics, Wang et al. (2021) reported: (a) when the SOC decreased by 10%, users were willing to pay an increased charging rate of £0.083/kWh (or 0.7 yuan/kWh), and (b) when the excess range decreased by 10 kms, users were willing to pay additional charge of £0.041/kWh (0.346 yuan/kWh).Finally, Li et al. (2023) reported that higher uncertainty regarding the remaining/available range of a BEV at destination would entice users to charge their BEV at a public charging point and the effect was higher for women and those on lower incomes.

Charging infrastructure attributes
Important attributes of charging infrastructure as reported in the reviewed studies included: (1) physical attributes, (2) charging price, (3) speed, (4) accessibility, (5) convenience, and (6) charging-point information.As shown in Table 4, several studies suggested that EV users would prefer public charging points that are cheaper to use, equipped with high-power chargers and thus shorter   The 'driving range' is the distance an EV or a PHEV can travel using the electricity stored in its battery, thus a higher battery capacity will result in a higher average driving range.In the case of PHEVs, of course, the vehicle will be powered on fuel once the battery has depleted.
D. Potoglou et al. charging times, offering convenience and accessibility to other activities and amenities and enhanced information systems (see, also Tables A6 -A11 in the Appendix A).

Physical attributes
The physical infrastructure attributes of a public charging station include location, power, number of charging points, and nearby

Table 4
Infrastructure charging attributes associated with public charging choices.
In general, it is hard to derive a universal order of location preferences for public charging due to the geographic differences in data and varied location alternatives considered in the reviewed studies (Anderson et al., 2018;Department for Transport, 2022b;Philipsen et al., 2016;Sheldon et al., 2019;Wolff and Madlener, 2019).Table A6 in the Appendix A summarises the findings regarding public charging preferences of current and potential EV users for different locations.As shown in Table 4, the choice of location involved the choice between public charging, workplace charging and home charging, or a choice across several public locations (Sheldon et al., 2019;Sun et al., 2016;Wolff and Madlener, 2019).For example, using a choice experiment, Sheldon et al. (2019) found that potential EV users in California were more likely to charge their vehicle at home than at work or at a public charging point.Wolff and Madlener (2019) reported that potential EV users in Germany were more likely to choose grocery stores and shopping centres to charge their EVs and less likely to do so at gyms and schools when compared to entertainment venues, which was the reference category.
The relative valuation of location preferences expressed in terms of WTP were reported either at the charging session level (Sheldon et al., 2019) or as monthly payment as part of a subscription with a charging provider (Wolff and Madlener, 2019).Californian respondents were willing to pay up to £2.31 ($2.81) to charge at a grocery store (the highest valuation in the study) whereas they would be seeking compensation, reported as willingness to accept, of up to £1.01 ($1.23) to charge their vehicle at a school.Under a monthly payment model, German EV users would accept between £19.70/month (€22.31/month) to charge their vehicle at work and £40.84/ month (€46.26/month) to charge their vehicle on the road and while en route instead of their home.
Overall, several studies reported a significantly higher preference for faster charging speeds both amongst EV users (Wen et al., 2016) and also potential EV users (Ma et al., 2022).For example, in a stated choice study for the US, Wen et al. (2016) found that owners of plug-in electric vehicles (PEVs, which include BEVs and PHEVs) prioritised charging their vehicles at stations with the highest (50 kW) and second highest power (6.6 kW) over stations with the lowest charging power (1.9 kW).Similarly, in a stated choice study for China, Ma et al. (2022) found that Chinese consumers preferred fast charging compared to slow charging.The willingness to pay (WTP) for fast charging was also estimated in a number of studies.Danish EV drivers, for example, were willing to detour for 0.28 min for an extra km per minute of charging, equivalent to £0.052 (0.44 DKK) using the 2020 official Danish value of travel time (Visaria et al., 2022).US drivers were willing to pay, on average £0.041 ($0.05), for an extra mile per minute of charging (Sheldon et al., 2019), and £1.58/hr ($1.91/hr) to use Level 2 instead of Level 1 charging (Wen et al., 2016).
The number of charging points, also expressed as the ratio of available over unavailable chargers, was an important attribute for BEV users' public charging choices (e.g.Visaria et al., 2022).Similar findings were reported across studies by the National Renewable Energy Laboratory (Wood et al., 2017) and the California Energy Commission (Bedir et al., 2018), and the Energy Efficiency and Conservation Authority in New Zealand (Burroughs et al., 2021).The relative valuation for charging points was expressed as WTP for an available charger.For example, Visaria et al. (2022) estimated that Danish EV users were willing to make a 7.87-minute detour (valued at £1.46 or 12.32 DKK) to reach an available charger, and a 0.53-minute detour (valued at £0.099 or 0.83 DKK) to reach an occupied charger, respectively.
Amenities around public charging stations are expected to compensate for longer waiting times relative to refuelling and, unsurprisingly, have been reported as being important for public charging choices (Ge and MacKenzie, 2022;ten Have et al., 2020;Visaria et al., 2022).As shown in Table 4, examples of preferred amenities included shopping areas around slow public charging points (ten Have et al., 2020) or different attribute-levels with a mix of points of interest such as toilets, supermarkets, and restaurants.Studies also reported significant observed variations in preferences across the surveyed participants.For example, EV users over 60 years of age or households with children were more likely to use public charging services where toilets, supermarkets, and restaurants were available (Visaria et al., 2022).However, there were some counterintuitive findings too.For example, ten Have et al. (2020) reported that the absence of amenities near slow, fast, and ultra-fast charging increased the likelihood of EV users charging at these charging points.
The relative valuation of amenities was expressed in terms of WTP to charge at a location with a specific type of service.Visaria et al. (2022) estimated that relative to 'no amenities nearby', EV users in Denmark were willing to make a 1.23-minute detour (valued at £0.23 or 1.93 DKK) to have toilets at the charging station, and a 9.55-minute detour (valued at £1.78 or 14.97 DKK) to have toilets, restaurants and supermarkets.EV users in the US were willing to pay, on average, £17.25 ($21) for the availability of toilets, dining facilities and Wi-Fi (Ge and MacKenzie, 2022).
In terms of what 'type of payment' for public charging users would prefer, Gutjar and Kowald (2023) found that relative to a flatrate charging option (e.g.unlimited charging at a fixed monthly price), kWh-based payment was the most preferred, while a durationbased fee and a fixed-fee per charging session were the least preferred among German consumers.Also, Visaria et al. ( 2022) found that D. Potoglou et al. among Danish EV users the flat fee pricing model for both public and home charging was the most popular payment method, followed by the 'no contract with the charging provider' (pay as you go) and the monthly subscription fee and kWh-based charging price per session.Danish EV users were willing to trade off 5.59 min of detour (valued at £1.04 or 8.76 DKK) to save £0.12 (1 DKK) per kWh in the price they paid at the charging station (Visaria et al., 2022).Also, EV users who were male, with a higher education qualification, or owned a Tesla, exhibited higher sensitivity to charging prices relative to other EV users (Visaria et al., 2022).
Several studies also included other related costs and fees.For example, Ge et al. (2018) included petrol prices within a scenario involving a decision to charge before or after a day of travel.Lee et al. (2020) specified 'home charging at the electricity rate paid at home' and 'free charging at the workplace' in two models studying the likelihood of public charging by BEV and PHEV users.As shown in Table 4, BEV users in California were more likely to use a public charging point if home charging was paid at the home electricity rate.PHEV users were less likely to charge their vehicle at public charging points when free charging was available at their workplace.
While various public charging opportunities may offer a certain charging fee depending on their payment model, EV owners with network membership can access all public charging stations under the control of the same operator at a special/discounted charging rate.Chakraborty et al. (2019) and Lee et al. (2020) found that EV users who subscribed to a specific EV public charging provider were more inclined to use public charging points.That said, network membership may also pose network incompatibility issues within the public charging market.Visaria et al. (2022), for example, measured the effect of interoperability on the choices of different charging pricing schemes and found that in Demark, BEV users were willing to pay £6.76/month (57.1 DKK/month) to have network access in all of Denmark and a similar price, £7.06/month (59.6 DKK/month), to have access anywhere in the EU, rather than have access to just one network.
Charging prices can be dynamically adjusted by operators to generate more revenue in response to fluctuating customer demand for electricity (Limmer, 2019;Limmer and Rodemann, 2019).This also provides an opportunity for cheaper rates during off-peak periods.Many consumers are risk averse and prefer to charge "now" at a nominal price rather than wait for an uncertain price reduction (Latinopoulos et al., 2017).Visaria et al. (2022) also reported that the monthly subscription model, which included peak and off-peak charging price differences for public charging, was the least preferred option for Danish EV users.Both Latinopoulos et al. (2017) and Visaria et al. (2022) suggested the importance of considering potential and existing EV users' attitudes towards these pricing differences.

Charging station level of service
The level of service at a charging station was expressed in terms of charging and waiting time, certainty of finding an available charging point, parking time and dwell time (see, Table A8 in the Appendix A).Charging time was defined either as 'the time it takes to fully charge the vehicle' (Ge and MacKenzie, 2022;Moon et al., 2018;Sheldon et al., 2019;Wolff and Madlener, 2019) or 'the time it takes to obtain a certain level of battery charge' (Wolbertus and van den Hoed, 2019), which indirectly reflected its charging power.The relative valuation of charging duration was reported either at the charging session level (Ge and MacKenzie, 2022) or as monthly payment as part of a subscription with a charging provider (Wolff and Madlener, 2019).EV users in the US were willing to pay £0.33 ($0.4) for a 1 min reduction of charging time (Ge and MacKenzie, 2022) and under a monthly payment model, potential EV users in Germany were willing to pay £0.14/month (€0.16/month) for a reduction of 1 min in charging time (Wolff and Madlener, 2019).
Queueing or waiting time was negatively associated with public charging choices (ten Have et al., 2020;Wang et al., 2021;Wolff and Madlener, 2019); that is, EV users would prefer to use public charging when waiting times were short (Wang et al., 2021).This was in line with a government report from New Zealand, which examined the preference of 932 EV users (Burroughs et al., 2021).Waiting times were generally measured in the range of 0---30 min (Wang et al., 2021;Wolff and Madlener, 2019).Waiting 30 min to charge an EV was the least preferred option for German drivers, who were willing to pay £0.72/month (€0.82/month) for every 1 min reduction in waiting time.On the other hand, respondents were indifferent between a 5-minute wait, 10-minute wait or no wait at all (Wolff and Madlener, 2019).This finding is also in line with ten Have et al. ( 2020), who reported that a 'certain' waiting time of up to 5 min was preferable to unknown or uncertain waiting times.Finally, parking (Wang et al., 2021) and dwell times (Wen et al., 2016) had a significant positive effect on public charging choices, as longer dwell and parking time induced drivers to actively make full use of that time.

Accessibility
Accessibility to charging stations has been measured either as the distance between the home or destination and charging stations (Ma et al., 2022;Moon et al., 2018;Xu et al., 2017;Zhang et al., 2022) or as the excess distance (time) required to travel in order to access a charging station (Ge and MacKenzie, 2022;Sun et al., 2016;ten Have et al., 2020).Several reports also confirmed that accessibility is an important factor for public charging choices (Burroughs et al., 2021;Department for Transport, 2022b;Department for Transport, 2022a, Hink, 2021) (see, also Table A9 in the Appendix A).
Overall, EV users preferred public charging stations involving shorter distances (Ma et al., 2022;Moon et al., 2018;Wolbertus and van den Hoed, 2019) and access times (Ge and MacKenzie, 2022).For example, Ma et al. (2022) reported that Chinese consumers were more likely to use public charging when the distance between the charging station and home/workplace/requested locations was shorter.Meanwhile, shorter detours were preferred by EV users when they employed en route charging in Japan (Sun et al., 2016) and Denmark (Visaria et al., 2022).The Danish study also revealed that EV users would opt-in for shorter detour times in order to charge their vehicle at a public charging point (Visaria et al., 2022).
In terms of the relative valuation of accessibility, Ge and MacKenzie (2022) found that EV users in the US were willing to pay £2.05 ($2.5) for a reduction of 1 min to access a charging station.They were also more likely to avoid charging en route if they could reach the next charging station without deviating from the originally planned route and were willing to pay £200.37 ($244) to avoid the D. Potoglou et al. deviation (Ge and MacKenzie, 2022).Finally, an indirect measure of accessibility relates to the coverage (or density) of public charging stations, which was expressed as the number of charging points within an areas.For example, Zhang et al. (2022) found that when the coverage of charging facilities at the parking locations in the trip chain increased, Chinese BEV users were more likely to use public charging as part of the trip chain.

Convenience
Convenience involves a number of features that a public charging point may exhibit and which improve the user experience.These features may also facilitate a more effective use of the infrastructure.The features, listed in Table A10 in the Appendix A, include inductive charging, serviced charging, plug and charge authentication and card-based payments.
Most, although not necessarily all, of these features were found to have a positive association with public charging choices (Gutjar Note: The percentages in the brackets refers to the proportion of the sample being allocated into a specific group.D. Potoglou et al. and Kowald, 2023;Moon et al., 2018;Wolff and Madlener, 2019).For example, consumers in South Korea were more likely to opt-in for self-charging instead of serviced charging by station employees (Moon et al., 2018).German EV users were more likely to use inductive charging than traditional cable charging, and they were willing to pay £7.40/month (€8.38/month) extra for that feature (Wolff and Madlener 2019).Inductive charging only requires a driver to park their EV at a specific location for charging to commence automatically.
EV authentication refers to the ability of the operator to automatically identify an EV and its user to enable faster charging and payment.Gutjar and Kowald (2023) found that German consumers intending to buy an EV perceived 'plug and charge' (i.e., automatic authentication and payment) as the most convenient method compared to app-based solutions, which rely on mobile phones and may therefore not work due to internet connection issues, low battery or cold weather.The 'plug and charge' option also came on top of the RFID-card (Radio Frequency Identity Card) solution, which is often incompatible across different charging networks.Finally, Gutjar and Kowald (2023) also found that potential EV users in Germany preferred web-based (e.g, PayPal) and card-based (e.g.credit card) payment methods rather than automatic debit transfer.

Charging station information
Charging station information may include location, price and availability of charging spaces but also the energy source of electricity.As shown in Table 4, these features were also associated with users' and potential users' decision to charge at a public station.Similarly, open data (websites such as Zap-map in the UK, an app from the charging provider, or vehicle onboard maps) and sustainability of public charging points were identified as important considerations (Hink, 2021).
Offering more information would be likely to attract EV users at those public charging stations (Moon et al., 2018;Wolff and Madlener, 2019).For example, Moon et al. (2018) found that potential EV users in South Korea were more likely to use public charging when station information (e.g.location, charging fees, charging station status) was available online (via Korea's EV Charging Station Monitoring System, an online information platform).Information on energy sources was also important for public charging choices for potential and existing EV users.For example, Wolff and Madlener (2019), Nienhueser and Qiu (2016) and Wolff and Madlener (2019) found that charging stations with a higher share of renewable energy were more likely to be chosen.
The relative valuation of renewable energy preferences expressed in terms of WTP were reported either as a payment per hour (Nienhueser and Qiu, 2016) or as a payment per month (Wolff and Madlener, 2019).PHEV users in the US were willing to pay £0.50 ($0.61) per hour for a 1% increase in renewable energy at Level 2 chargers and £1.49($1.82) per hour for a 1% increase in renewable Notes: The Electric Vehicle Infrastructure Projections (EVI-Pro).
D. Potoglou et al. energy at Direct Current Fast Chargers (DCFC) (Nienhueser and Qiu, 2016).Also, German potential EV users were willing to pay £0.37 per month (€0.42/month) for a 1% increase in renewable energy at public charging points (Wolff and Madlener, 2019).Gutjar and Kowald (2023) also found that potential EV users in Germany, who were young, female, or male but with high environmental awareness, were more sensitive to the share of renewable energy at public charging points.Nienhueser and Qiu (2016) found that for PEV drivers in the US, the WTP for renewable energy at public charging stations increased with income and age, decreased with years of education and was higher for women.

Table A3
Overview of willingness to pay estimates in previous literature (in UK pounds).

Preference heterogeneity for public charging
As potential and existing EV users are the subjects making public charging choices, it is also useful to examine the personal (observed) characteristics and efforts to capture unobserved heterogeneity of public charging choices (see Table A12 in the Appendix A).
EV users most likely to use public charging were identified across the reviewed studies as younger (Wang et al. 2021), female (Lee et al., 2020;Pan et al., 2019;Wang et al., 2021;Zhang et al., 2022), with lower education qualifications (Pan et al., 2019) and lower income (ten Have et al., 2020).They were also more experienced drivers (Wang et al., 2021) and had owned an EV for more than a year (Pan et al., 2019), lived in a flat (Chakraborty et al., 2019;Lee et al., 2020), did not have a home charging point (Zhang et al., 2022), often made long chained trips (Zhang et al., 2022), and held risk-averse attitudes towards EV range (Pan et al., 2019).EV users with previous experience (Xu et al., 2017) and those who used public charging frequently (ten Have et al. 2020) were more likely to use fast charging.Also, those who placed importance on driving comfort and had higher education qualifications preferred ultra-fast charging (ten Have et al. 2020).
As shown in Table 5, some studies captured unobserved heterogeneity in public charging choices by dividing the sample into different groups via latent class discrete choice modelling (Ge et al., 2018;Latinopoulos et al., 2017;Pan et al., 2019;Wang et al., 2021;Wen et al., 2016).Differences in public charging choices were explained by charging concerns (Wang et al., 2021;Wen et al., 2016) and risk-taking attitudes (Pan et al., 2019).For example, Wang et al. (2021) found that 76.2% respondents in their sample were 'charging-service concerned' and valued SOC, queuing time and charging station satisfaction.These respondents were younger, female, had driving experience, a higher income, and were risk averse.The reference group, which comprised the remaining 23.8% of the sample, were pragmatic about their charging needs and valued SOC, ER, parking time and charging fee.Similarly, Wen et al. (2016) identified three subgroups of EV users based on their charging concerns: those who: (1) 'chose to charge based on price and charging needs'; (2) 'charged at every opportunity'; and (3) 'charged taking into consideration other factors such as SOC, dwell time and home charging price'.
When considering risk attitudes towards SOC, Pan et al. (2019) showed that most EV users belonged a 'risk-seeking group' (70% of the sample), and their choices were driven by price, SOC and parking price.The choices of the 'risk averse group' (30% of the sample) were driven by the ER and its members were males, those earning more than 5,000 yuan/month ($687/£600), and had owned an EV for more than 12 months.
PHEV users may act differently compared to EV users when facing the same charging options, as they can either use a public charging point or refill their cars at petrol stations.Ge et al. (2018) segmented PHEV users into two groups: 'petrol anxiety' and 'costminimising' users.The majority of PHEV users were more likely to be in the 'petrol anxiety' group, and would avoid using petrol as much as possible, and, due to their environmental attitudes, would be willing to pay four times the cost of charging rather than refuel.The 'cost-minimising group', only 34% of the sample, was more pragmatic and placed equal importance on charging and refilling and would charge or refuel depending on the relative costs of electricity and petrol.
EV users charge-or-not decisions for public charging were influenced by risk-averse attitudes towards dynamic pricing (Latinopoulos et al., 2017).The results showed that the majority of current and potential EV users were more likely to be risk-averse and preferred to charge immediately at the prevailing tariff at the time rather than charge strategically, which would have entailed waiting for a better price.

Conceptual mapping of public charging choices
This section proposes a conceptual framework to capture the determinants of public charging choices for potential and existing EV users.Fig. 1 reveals that public charging has and will continue to have an increasingly important role in road transport.The empirical work examined in this study has the odd outlier and counterintuitive conclusion, but there is a clear set of common themes that emerge.The determinants of public charging choices can be categorised into four levels: (1) temporal, (2) vehicle, (3) charging Notes: '+' indicates a significant positive effect on public charging choices, and '− ' represents a significant negative effect, * refers to the results of the descriptive analysis and they are not significant.Notes: '+' means significant positive effect for public charging choices, and '− ' represents significant negative effect; '*' means the result is not significant.

Table A6
Explanatory factors for public charging choices: physical infrastructure attributes.Note: Order of preference in bold imply that the results of the study were significant, while the order of preference not in bold represent the descriptive analysis results and they are not significant, and the superscript F refers to studies focused on fast charging stations, superscript G refers to the grey literature.infrastructure, and (4) individual attributes.Public charging tends to take place during the day rather than during the night and may be different during weekdays or weekends, or holidays.Vehicle attributes include battery capacity and driving range, which are fixed, but also the SOC, remaining range, and the excess range to destination.According to Fig. 1, charging infrastructure attributes are categorised into physical attributes, price, charging station level of service, accessibility, convenience, and station information.Individual attributes refer to the observed heterogeneities reported across studies and include socio-economic and demographic characteristics, and level of risk aversion.
The lines and arrows in Fig. 1 show how the four levels of determinants (temporal, vehicle, infrastructure, and individual attributes) are linked with public charging choices.Temporal attributes are an important consideration as the daily, weekly, and inbetween recharging patterns are essential for understanding public charging preferences of EV drivers.Vehicle attributes are also important because higher fixed battery capacity/mileage creates more opportunities for EVs to use public charging.As some public charging networks are (still) free and have high efficiency compared to home and workplace charging, while constantly changing SOC, remaining range and excess range to destination may be influenced by range anxiety, which drives EV users to use public charging.

Table A8
Charging and waiting times attributes related to public charging infrastructure.Note: '+' means significant positive effect on public charging choices, and '− ' represents significant negative effect.

Table A9
Accessibility attributes related to public charging infrastructure.Note: '+' means significant positive effect for public charging choices, and '− ' represents significant negative effect; '*' means the result is not significant.
D. Potoglou et al.However, this does not mean that they will necessarily charge at public stations, as temporal and vehicle attributes, and the characteristics of the public charging infrastructure in terms of physical aspects, price, level of service, accessibility, convenience, and station information also influence charging choices.At the same time, individual attributes, including socio-economic and demographic characteristics and level of risk aversion, contribute to preference heterogeneity.Vehicle manufacturers are all too aware of what EV users want, but this review is a reminder that range is absolutely crucial when it comes to electric vehicle choice (Potoglou et al., 2020;Song and Potoglou, 2020).The indirect evidence provided in this review is clearly saying that range is an important determinant in charging choices as well.
The individual and temporal attributes identified determine public charging choices, but they cannot be changed with policy.Policy, can, however, seek to adapt public charging infrastructure to better satisfy EV users' requirements.All in all, public infrastructure needs to be easily accessible, frequent, and with enough fast charging points to satisfy demand at peak times, especially morning and afternoon.It needs to be sheltered somehow, to protect drivers from harsh weather conditions, and amenities, especially toilets, are important.The slower the charging speed, the more important extra amenities, such as shopping and leisure facilities, are.All information regarding the public charging station, including pricing and methods of payment needs to be clearly and quickly accessible, and payment needs to be hassle-free.There is also incipient evidence that EV users prefer electricity generated in a clean manner, which is not surprising, as they are likely to be environmentally conscious.Clean electricity generation is, in any case, a must when it comes to road transport electrification, as there is plenty of evidence that shows that emissions will increase, rather than decrease, otherwise (Cox et al., 2022;Gómez Vilchez and Jochem, 2020;Liu and Santos, 2015;Woo et al., 2017).An open area for further investigation here is also how potential preferences may be affected by a wider grid integration (e.g.vehicle-to-grid) (e.g.

Table A10
Convenience attributes related to public charging infrastructure.Note: '+' means significant positive effect for public charging choices, and '− ' represents significant negative effect.

Table A11
Information attributes related to public charging infrastructure.Yes, no (Moon et al., 2018) Note: '+' means a significant positive effect for public charging choices, and '− ' represents a significant negative effect.

Table A12
Explanatory factors for public charging choices: user (incl.potential) attribute.

Respondents Personal characteristics EV users Who prefer public charging
Young (Wang et al. 2021), female (Pan et al. 2019;Lee et al. 2020;Wang et al. 2021), with lower level of education ( Pan et al., 2019) and lower level of income (ten Have et al., 2020), purchased EV more than a year (Pan et al., 2019), had rich driving experience (Wang et al., 2021), living in an apartment (Chakraborty et al., 2019) and not having a house (Lee et al., 2020), without home charging point (Zhang et al., 2022), with long and complex trip chains (Zhang et al., 2022), and holding risk-averse attitudes towards EV range (Pan et al., 2019) Who prefer slow public charging Lower level of income (ten Have et al. 2020) Who prefer fast charging Young (Wang et al., 2021), and rich fast charging experience (Xu et al., 2017), lower level of income (ten Have et al. 2020), higher current frequency of using fast charging (ten Have et al. 2020) Who prefer ultra-fast charging Have lower level of income, put higher importance of driving comfort, and have higher level of education (ten Have et al. 2020) PHEV users Who prefer public charging Less vehicle ownership (Lee et al., 2020), long time of using PHEVs and buy PHEVs not only by financial benefits (Ge et al., 2018) Potential users

Who prefer public charging
White, believe local air quality is poor, and politically liberal leaning (Sheldon et al., 2019) D. Potoglou et al.Ensslen et al., 2020;Sachan et al., 2020).The recommendations above are blindingly obvious and intuitive, and one may wonder what the point of even listing them is.The point is threefold: (a) as of 2023, no country has public charging stations that fulfil the requirements set above, so, as obvious as they may seem, action has not been taken yet; (b) academics and policy makers have devoted substantial time and resources over the last ten years trying to identify what these requirements are, and the present study synthesises them all in one place; and (c) ensuring these requirements are met will guarantee that EV users' demand is satisfied.

Conclusion
This paper provides an up-to-date critical review of the academic and grey literature on public charging choices and their determinants.Findings from this critical analysis and synthesis of the literature show what determinants contribute to the public charging choices of potential and existing EV users.This evidence will not only benefit public charging providers in the planning of and investment decisions on charging infrastructure, but will also facilitate government policy measures to further deploy public charging infrastructure and promote electric vehicle uptake.The synthesis of the empirical evidence is also integrated into a conceptual framework that illustrates the complexities and dimensionality of public charging choices and shows how these are linked with the range of determinants identified.This framework explains the importance of temporal, vehicle, charging infrastructure, and individual attributes for current and potential EV users, who make or will potentially make those choices.
On the basis of the reviewed material there are also a number of research gaps that can be identified.Firstly, the evidence regarding temporal charging preferences is far from conclusive, with different patterns across and within different countries, which in itself is not a problem per se, except that it is difficult to find any reasons for these differences given the number of studies, the different sampling strategies and the different approaches used.Temporal charging preferences therefore should be further explored, especially in relation to travel purpose, trip length and charging speeds.Secondly, further research is needed to fully understand EV users' preferences for charging locations and payment models (e.g.monthly, pay as you go), especially in relation to safety, battery size of EVs, type of charging (destination or en route charging), and travel purpose.Preferences for charging locations are influenced by those factors but the evidence is sparse.Thirdly, more research is needed on preferences for different charging models and compatibility of payment methods.Exploring those themes further would allow to build a more solid evidence base.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 1 .
Fig. 1.A conceptual framework of determinants of public charging choices.

D
.Potoglou et al.

Table 2
Temporal attriutes associated with public charging choices.

Table 3
Vehicle attriutes associated with public charging choices.
Notes: AR: available range; references marked in bold refer to results that are specifically applicable to EV users, including PEVs, BEVs, or PHEVs users, while regular fonts indicate studies of potential users; the superscripts in the table represent: [1] Japan; [2] China; [4] California, US; [5] US; [6] Beijing, China. 1

Table 5
Unobserved heterogeneity in choices and segmented groups.

Table A1
Overview of empirical studies on public charging preferences.

Table A2
Overview of grey literature.

Table A4
Overview of explanatory factors for public charging choices: temporal attributes.

Table A5
Overview of explanatory factors for public charging choices: vehicle attributes.

Table A7
Public charging price levels.Note: '+' indicates a significant positive impact on public charging choices, '− ' represents a significant negative impact.The level '[]' in square brackets and superscript numbers indicate different payment models, 1 -charging time based, 2 -kWh based payments, 3 -fixed monthly charging fee, 4 -pay for a certain range.