Critical risk factors of electric road uptake on motorways: a Swedish Delphi study

ABSTRACT This paper presents the results of a ranking-type Delphi study on the critical risk factors for the adoption of an electric road system (e-road) for trucks on the main motorways in Sweden. The investment cost of such a system is high, necessitating an upfront evaluation of the adoption risk factors to reduce the likelihood of budget overruns and project delays. Participating Swedish e-road experts (N = 52) from the public sector, private sector, and academia identified 32 unique risk factors, which were divided into five categories. The three most critical risk factors, as ranked by the experts, were ‘low expansion rate,’ ‘low utilization rate,’ and ‘lengthy public-sector evaluation.’ Overall, market and financial risks were ranked as more important than institutional, technological, and sustainability risks. This study has important implications for policymakers in countries considering e-road adoption.


Introduction
This study aimed to identify the critical risk factors of large-scale implementation of electric road (e-road) systems for trucks on motorways in Sweden.An e-road operates by means of dynamic power transfer from the road to the moving vehicle, thereby propelling the vehicle forward (Sundelin, Gustavsson, and Tongur 2016).This decarbonization option for road transport has been considered by several countries, including Sweden, Germany, the UK, and the US.To meet the goals of the UN Paris Agreement, greenhouse gas emissions from the transport sector must be reduced.This requires a shift from fossil fuel to low-CO 2 -emitting fuel use (Kluschke et al. 2019).Road transport electrification is reported as a promising alternative for heavy-duty vehicles (Börjesson, Johansson, and Kågeson 2021;Osieczko et al. 2021).Heavy-duty vehicle producers have begun producing (plug-in) battery-electric long-distance trucks, and their share of the global vehicle fleet is projected to grow (Bloomberg 2021).
For long-haul trucks with static charging (e.g. at the depot or destination), large batteries are required: they prevent unnecessary stops for static charging, avoiding additional driver and time costs (Nykvist and Olsson 2021).However, the high density of such battery-electric trucks negatively affects either payload capacity or fuel economy of trucks (Börjesson, Johansson, and Kågeson 2021).Furthermore, equipping trucks with large batteries leads to high purchase prices (Basma, Saboori, and Rodríguez 2021).In the cost-competitive road freight business, haulers may be sensitive to truck purchase prices; a high price discourages a rapid shift to electrified transport.Moreover, forecasts of battery improvements and pack prices diverge, creating uncertainty regarding future purchase prices (Anculle, Bubna, and Kuhn 2021;Bloomberg 2021).Future static charging prices are also uncertain because of the need for overcapacity instalment and electric grid investments (Ainalis, Thorne, Cebon 2020;Osieczko et al. 2021).
E-roads are another electrification option for trucks.One advantage of large-scale eroad implementation on motorways is that trucks with much smaller batteries would suffice (Taljegard et al. 2020).Specifically, Börjesson, Johansson, and Kågeson (2021) estimate that for a medium-sized truck to have a range of 400 km, a quite large battery of 600 kwh would be required.By contrast, Ainalis, Thorne, and Cebon (2020) and Coban, Rehman, and Mohamed (2022) estimate that with a vast e-road motorway system a battery capacity of 100 kwh would suffice for such trucks.Long and medium haul trucks are thus used primarily on motorways and make up a large share of the total truck vehicle kilometers and ton kilometer freight transported per year (Natanaelsson et al. 2021;Börjesson, Johansson, and Kågeson 2021).Battery size reduction of such trucks thereby enables avoiding adverse payload effects and high truck purchase prices associated with reliance on static charging (Nykvist and Olsson 2021).Commercial delivery trucks, however, would not benefit in this way from such e-roads as they mainly operate in urban areas (e.g.Natanaelsson et al. 2021).E-roads on motorways for trucks are nevertheless an important decarbonization option to consider since the heavier long-and medium haul trucks stand for most of the yearly total ton kilometer freight transport (Karlsson 2019;Swedish Transport Administration 2021).A main advantage of smaller battery size is reducing the global carbon footprint from battery production and disposal, which represents a worldwide problem (Lai et al. 2022).Another advantage of e-roads over static charging of heavier trucks is the reduction in hold-up costs (Börjesson, Johansson, and Kågeson 2021); that is, the avoidance of additional charging stops or detours that add to driver costs and the total delivery time.Avoiding such transport delay is crucial considering that delay costs are more significant than transport costs (Izadi, Nabipour, and Titidezh 2020;Kurri, Sirkiä, and Mikola 2000).Therefore, the freight transport road electrification is the main charging alternative and a crucial complement to static charging infrastructure (e.g.Shoman, Karlsson, and Yeh 2022).
The Swedish Transport Administration and its counterparts in Germany and the Netherlands recently highlighted the importance of adopting multiple simultaneous measures to curb heavy-duty traffic CO 2 emissions (Plötz 2023).Thus, development and implementation of different fuel technologies to meet the divergent needs of haulers (e.g.long-haul, short-haul, and order traffic) is necessary.One such pathway is the implementation of an extensive e-road system on motorways in Sweden.Several other countries are considering the implementation of large-scale e-road systems for trucks on motorways.Pilot projects have been conducted in the UK (Ainalis, Thorne, and Cebon 2020), Germany (Wietschel et al. 2019), and the US (Haddad et al. 2022) over the past five years.Moreover, the Swedish government has declared its intention to build a 3000 km e-road on motorways by 2035, the world's first permanent e-road for trucks.As such, e-roads for trucks are 'a reality' to private-and public-sector actors in the country.As the infrastructure investment cost of a large-scale e-road is highestimations range from €1.2 M to €2.6 M per road kilometer (Ainalis, Thorne, and Cebon 2020;Börjesson, Johansson, and Kågeson 2021;Coban, Rehman, and Mohamed 2022) the upfront identification of the critical risk factors is imperative.By identifying the critical risk factors for private-and public-sector stakeholders, these factors can be monitored to prevent cost overruns and project delays, which are often experienced in infrastructure projects (Odeck 2019).In-depth knowledge of the relative importance of risk factors is necessary for cost-benefit evaluations and comparisons of e-road system and other decarbonization alternatives.Therefore, this study of critical risk factors in eroad implementation is imperative.
The key aim of this study is to identify the critical risk factors for e-road adoption on motorways in Sweden.Another aim is to analyze how the public-and private-sector actors and representatives from academia differ in their perceptions of risk-factor criticality.Therefore, we conducted a Delphi-type ranking study.Previous studies either implicitly assumed or modeled specific risk factors associated with e-road uptake.To the best of our knowledge, our study is the first to systematically identify and compare the overall relative importance of critical risk factors for e-road stakeholders.By identifying the critical risk factors for large-scale e-road uptake in Sweden, important lessons can be learned.

Literature review
As e-road technology is emerging and is yet to be commercialized on a larger scale, few studies have been conducted on the risks associated with e-road infrastructure investments.Previous studies modeled and explored specific adoption risks, including the utilization rate of e-roads, fuel price volatility, electricity supply shortages to e-roads, sustainability risks, technology risks, and public acceptance risks.Börjesson, Johansson, and Kågeson (2021) found that the social benefit of an e-road system for trucks on the Swedish motorway network will depend on the size and location of the system.The larger the system, and the more it is built on motorway sections most used by trucks, the higher the utilization rate, and the higher the benefit generated to society.Moreover, the authors showed that the social value of an e-road system depends on the fuel cost difference between electricity and diesel and user charge level.Similarly, Taljegard et al. (2020) analyzed the effect of scope and location of an e-road motorway system for light-and heavy-duty vehicles and concluded that rightly placed and scaled e-roads would lead to a significant reduction in emissions.Connolly (2017) compared the economics of e-roads with those of batteries and fossil fuels in the Danish motorway network and concluded that it was cheaper to electrify roads than to use batteries or fossil fuels to propel trucks.These findings are sensitive to the predicted prices of fossil fuels, batteries, and electricity.Coban, Rehman, and Mohamed (2022) analyzed the economics of e-road implementation on most of the Turkish road network and found that the country's energy cost could be reduced by up to 8.5% by making 50% of the Turkish vehicle fleet use an e-road system.Their results were sensitive to battery pack costs and cost differences between fossil fuels and electricity.Natanaelsson et al. (2021) reported a considerable uncertainty regarding the measurement and control system costs of a Swedish e-road motorway system for trucks.Effectively, the user charge levels are highly uncertain.
A prerequisite for electrifying road transport is the ability of the electric grid to meet the increased electricity demand.Olovsson et al. (2021) analyzed the effects of e-roads on German and Swedish electricity grids and concluded that investments in solar and wind power were needed to meet the increased demand.Similarly, Jelica et al. (2018) showed how an increase in electricity demand due to road transport electrification affects the investments required in grid capacity.Other studies have reported that e-roads face technological uncertainties, including energy transfer efficiency and infrastructure maintenance requirements (Majhi, Ranjitkar, and Sheng 2022;Natanaelsson et al. 2021).Public acceptance was also considered.Public opposition may make political decision-makers hesitant to adopt e-roads due to political risks.Konstantinou, Gkartzonikas, and Gkritza (2023) revealed in a US survey study that e-road acceptance is affected by the public perception of such systems' innovativeness, people's environmental views, and perceived road safety.
Different risk factors associated with e-road uptake have been investigated across studies.However, the knowledge regarding the relative importance of risk factors for different stakeholders is limited.To avoid incurring budget overruns and avoid unnecessary delays in e-road implementation, critical risk factors must be identified.This includes risk valuation, allocation, and management during an e-road lifetime.Therefore, the remainder of this study focuses on the relative importance of critical risk factors of e-road uptake.

Methodology
We conducted a ranking-type Delphi study.Figure 1 provides an overview of the methodology used.

Selection of the Delphi technique
The ranking-type Delphi methodology matches the aims of this study because it is designed to elicit opinions from experts and measure their levels of consensus (Schmidt et al. 2001).Its use offers advantages over workshop or focus-group methods as it helps prevent single-participant response dominance because respondents are anonymous.Compared with the interview method, it eliminates the potential impact of interviewer effects (West and Blom 2017).Compared with a single-survey study, it represents an iterative approach with interspersed controlled feedback for each participant between survey rounds.Thus, aggregated result feedback allows respondents to effectively form their opinions before the next round (Hsu and Sandford 2007).Finally, the data collection is divided into multiple shorter occasions, which reduces the risk of exhaustion and response-style bias (Drumm, Bradley, and Moriarty 2022) Delphi, as a group method, enables the comparison of the private and public sectors in assigning the relative importance of risk factors for e-road uptake.The diverging goals of these two sectors have been found to cause infrastructure delivery and operational problems, such as cost overruns and delivery delays (Anastasopoulos et al. 2012;Herrera et al. 2020).The smoothness of expert identification supports the choice of the Delphi method.Sweden has established pilot e-roads over the past five years.The setup of these roads has involved different public-and private-sector actors who have gained insights into e-road uptake and use.These actors were approached in this study.
The ranking-type Delphi approach followed the steps recommended by Schmidt et al. (2001) and Hirschhorn (2019).Thus, the brainstorming phase was followed by a narrowing-down phase and a final ranking phase.

Panel selection
The procedure for selecting participants followed the two recommended steps: defining relevant expertise and identifying and selecting experts (Okoli and Pawlowski 2004).Road infrastructure typically falls under the public sector, although it involves the private sector in delivery, operation, and maintenance.The contrasting goals of these two sectors (profit maximization versus social value maximization) motivated their partitioning into separate panels.Academia's in-depth knowledge of e-road aspects and its neutral position vis-à-vis the private and public sectors made it a suitable third panel.To define the relevant actor roles within each panel, we referred to the study of Wang, Hauge, and Meijer (2020).Academic roles were defined as capturing the business, economics, and engineering of e-roads because of the importance of these domains in technology uptake (Wang, Hauge, and Meijer 2020).The experts were selected based on the inclusion criteria listed in Table 1.Both authors independently populated the three panels with candidates.Next, the panel catalogues were cross-checked by both authors.Thereafter, the revised catalogues were verified by a small focus group of five e-road experts representing academia, the private sector, and the public sector.The procedure yielded a final list of 62 candidates: 16, 27, and 18 from academia, private sector, and public sector, respectively.
All 62 candidates were invited to participate in the study (cf.Paré et al. 2013).Ten candidates declined to participate because of a change in their jobs or recent retirement.Therefore, the final sample comprised 15, 21, and 16 participants in the academic, private, and public panels, respectively.

Data collection and analysis methods
Four surveys were conducted using two online survey tools (Figure 1).This provided consistency to respondents in completing questionnaires across rounds.In the ranking phase, two types of rankings were performed (Hirschhorn 2019).
Phase 1: brainstorming.Experts proposed five critical risk factors (along with short descriptions) for implementing a 2500 km Swedish e-road motorway system for trucks.The experts provided five risk factors, a trade-off between achieving prioritization and avoiding expert fatigue (Hirschhorn 2019).One author coded all risk factors and then consolidated those that were not unique to the same code.Expert descriptions of the proposed risk factors guided the consolidation.Thus, a reasonably sized risk-factor list was obtained, while avoiding excessive consolidation.The other author performed the same coding and consolidation for a random response sample as a reliability check, confirming the initial coding (Hirschhorn 2019).Thereafter, the consolidated risk factors were categorized.The categorization was cross-checked by a focus group of five e-road experts, which led to minor changes.
Phase 2: Narrowing down.Experts shortlisted the 7 most important of the 32 Phase I risk factors.They were presented with a list and a short definition of each risk factor.Respondents shortlisted seven items as a trade-off between forcing prioritization and avoiding respondent fatigue (Hirschhorn 2019).The responses were examined using the majority vote rule (Hirschhorn, Veeneman, and van de Velde 2018).As the three panels were not of equal size, a procedure in which each panel received equal weighting in voting was used as a robustness check. 1 For the 16 most-voted risk factors, the results were the same using either rule.Therefore, these items were retained for the ranking phase, and the criterion of not using more than 20 items for the ranking phase was met (Okoli and Pawlowski 2004).
Phase 3: Ranking.Experts ranked the 16 risk factors retained from Phase 2 in two rounds.In survey round 3, the respondents were asked to allocate 100 integer points to 16 risk factors based on their importance.A high-point allocation to a single factor implied high relative importance.This procedure allows for a simple parametric statistical analysis of the findings (e.g.average points, standard deviation, max, min, and percentage of zeros) and offers a complement to and robustness check of the ordinal ranking procedure (Paré et al. 2013;Hirschhorn 2019).In survey round 4, respondents ordinally ranked the 16 risk factors (1-16) based on their perceived importance.
Feedback from the previous survey round was provided to participants between the phases.Between survey rounds 1 and 2, this feedback included the list of the unique risk factors identified, and between rounds 2 and 3, the risk-factor list retained for ranking.To avoid peer influence, the experts were kept anonymous (cf.Paré et al. 2013).Each survey was pretested with a focus group of five e-road experts, which helped to clarify the survey instructions and tasks.To avoid exhaustion, the Delphi study ended after survey round 4 (cf.Paré et al. 2013).
To test for differences and similarities in rankings between the groups, we followed Shah and Kalaian (2009).Nonparametric tests were used due to the low number of respondents in the groups.To test for differences in rankings, the Kruskal-Wallis test accompanied by Dunn´s post hoc test for differences was performed.Spearman's rank correlation was used as another analytical technique to compare the differences and correspondences in the risk-factor importance between groups.

Validation
To validate the final list of risk factors, ten road freight companies were interviewed.The rationales for this choice are as follows.First, the final list of risk factors was based on supply-side expert views: public-and private-sector actors with extensive knowledge of electric road systems.Second, if demand-side actors perceive other risks as critical, handling the risks in the final list of the Delphi study will not support the use of the technology required for economic viability.Third, although the Delphi study captured experts views on e-roads, a vast majority of road freight companies were not experienced with this technology.Conducting semi-structured interviews enabled us to clarify the critical risk factors for potential users.
The characteristics of the ten road freight companies are presented in Appendix 1. Interviews lasted between 30 min and 1 h and were recorded and transcribed.Qualitative content analysis was performed by the authors, independently and then together, with the main aim of identifying whether the critical risk factors on the demand side corresponded to those reported in the Delphi study (supply side).

Results
The following sections present the results of the Delphi study and the validation interviews.

Identification of critical risk factors
Table 2 presents the results of the brainstorming phase, the 32 unique risk factors divided into five categories.Public-sector experts proposed technology and sustainability more than privatesector experts did.Conversely, private-sector experts proposed market and institutional risk factors to a greater extent.No corresponding pattern was observed for the academic experts.
Table 3 presents the 16 most shortlisted risks overall and by panel from the narrowingdown phase.Institutional risk factors were frequently shortlisted by all panels, indicating their overall criticality.By contrast, sustainability risk factors were rarely shortlisted, except for 'cost-inefficient CO 2 emission reduction.'The private panel displayed the highest level of concordance in the shortlists.Some differences in shortlisting were observed between panels.Academic experts shortlisted technology risks more than experts from the other two panels.Private-sector experts shortlisted institutional risks, whereas public-sector experts shortlisted financial risks.The 16 shortlisted risk factors (Table 3) were retained for the ranking phase.

Relative importance of risk factors
The results of survey round 3, the fixed-point sum allocation survey, are presented in Tables 4a and 4b.The average points allocated to each risk factor were used to rank the 16 risk factors.The top 3 risks overall were 'lengthy public-sector evaluations,' 'low expansion rate,' and 'utilization limited to trucks.' Differences in magnitude measured by the average points allocated to risks were modest for the top 8 risk factors, ranging from 10.3 to 6.5 points.Thus, several risk factors are equally critical for e-road uptake; this is further supported by no risk factors obtaining points from more than two-thirds of panelists.Moreover, important risk factors were not concentrated in a specific risk category.Two-thirds of market risks were of low relative importance, and the only sustainability risk factor, 'cost-inefficient CO 2 -reduction,' was ranked as having the lowest importance.
The differences in risk ranking between panels in the point allocation survey are displayed in Table 4b.The risk factor 'utilization limited to trucks' received a high ranking only from the public panel.Conversely, 'lengthy public-sector evaluations' were ranked   first by the academic and private panels while ranked seventh by the public panel.Moreover, only the public panel gave 'low international interoperability' a low rank.The rankings indicate that institutional risks are of concern to experts in private and academic panels.By contrast, financial risks are of concern to public panels.The high share of zero points awarded to each risk factor across panels indicates the need for a complementary ordinal ranking round.
In round 4, experts ordinally ranked the 16 risk factors.The results are presented in Table 5.The top three risk factors were 'low expansion rate,' 'value-for-money uncertainty,' and 'low utilization rate.'All panels ranked financial risk as high.To provide insight into the differences between the panels, the results of the Kruskal-Wallis test accompanied by Dunn's multiple comparison test are shown in Table 6.The results indicate that the private panelists ranked 'political uncertainty' and 'insufficient grid capacity' (i.e. the risk factors that are the public sector's responsibility) significantly higher than the public panelists.Conversely, the public panel ranked the market risks 'low utilization rate' and 'low user willingness-to-pay' pertaining to the private sector significantly higher than the private panel.These findings indicate that the risk-factor criticality differs between the private and public sectors.The academic panel ranked 'high investment costs' and 'lengthy public-sector negotiations' significantly higher than the private panel.This suggests that academia is more concerned about the opportunity cost of public fund use than the private sector.'Low international interoperability' was ranked significantly higher by the academic than by the public-sector panel.Therefore, academics seems to be concerned with technological e-road standards that are not being reached.In the ordinal ranking survey, the risk factors within a specific category were not particularly important.
The consensus in rankings among participants was moderate in survey 4. Thus, the interquartile ranges for the single risk factors presented in Table 5 are mainly greater than four, indicating dispersed rankings by experts on the 16-point ordinal scale (cf.Von der Gracht 2012).Moreover, Kendall's W of concordance measures (Table 4b supplement) indicate low levels of agreement among experts (cf.Von der Gracht 2012).The results of consensus measurements and an inspection of individual responses suggest that risk-factor importance is dependent not only on sector belonging but also on the actor role within the sector.
The results of the Spearman's rank correlation analysis are presented in Table 7.No single risk factor had statistically significant correlation coefficients among all three expert groups.This is consistent with the concordance level, which is at best moderate according to Kendall's W measure.For few of the risk factors, a positive significant correlation coefficient between two expert groups was found, which is consistent with the Kruskal-Wallis test findings indicating significant differences between expert groups for specific risk factors.For academia and the public sector, rankings of 'high investment cost' displays a positive significant correlation coefficient, indicating a consensus of these two groups on the relative importance of this risk factor.Similarly, a positive correlation coefficient for 'low utilization rate' was reported for the private and public sectors.For rankings of 'low complementary value to battery' a negative correlation coefficient is reported between public sector and academia, as well as between private sector and public sector.Inspecting rankings, high within-group (e.g.public sector) differences in the rankings of this factor are likely attributable to this correlation finding.The median score for Notes: rank is based on the median rank (ordinal ranking by experts, where 1 = most important risk factor and 16 = least important).For comparison, risk factors are presented in the order in which they were ranked overall in survey round 3.
this factor was similar for the three groups, explaining why the Kruskal-Wallis test did not reveal significant differences in rankings between groups for this factor.

Validation of Delphi results
Of the 16 final risk factors, 13 were critical for the interviewees.This suggests that the risk factors on the supply and demand sides of e-road systems correspond.However, 'lengthy public-sector negotiations,' 'unattractive operator role,' and 'ineffective procurement' were not identified as critical risk factors by interviewees.Therefore, these factors may be supply-side-specific risks.Road freight company interviewees (RFC) expressed concerns about single risk factors.Regarding 'utilization limited to trucks,' RFC2 emphasized the importance of electric road systems supporting passenger car use.This contributes to lower user charge levels and motivates complementary grid-capacity infrastructure investments required for e-road functionality (RFC2).Several interviewees raised concerns about the risk of a 'low expansion rate' of e-roads.The larger the electrified motorway network, the lower the risk for road freight companies of investing in e-road compatible vehicles because of more enabled use (RFC1).RFC8 highlighted that e-roads are not viable on roads with lower traffic volumes because they would require higher user charges to recoup infrastructure investment costs.Regarding 'low utilization rate,' RFC8 expressed that e-roads are inflexible, forcing haulers to select specific e-road routes even when other routes are preferred.Therefore, e-roads are useful for the truck traffic segment that is predetermined rather than order-based (RFC2).
Institutional risks pertain to the demand and supply of e-roads.Regarding 'political uncertainty,' several of the interviewees mentioned the sudden and unexpected removal of change in subsidization of biogas.Faced with competition, road freight companies cannot afford such sudden shifts (RFC4).A new truck is to be used for six to seven years; hence, policy conditions need to be constant over that time (RFC6).'Unclear regulations' and responsibilities related to driver safety and damages to goods and vehicles represent other e-road user uncertainties (RFC8).Technological risk is another concern.RFC2 stated that e-road solutions must be compatible across nations to support use and that standards at the EU level are necessary to stimulate large-scale uptake.Regarding 'high investment cost' in e-road infrastructure, one interviewee expressed that user charges may be negatively affected if the public sector has difficulties recouping the investment cost (RFC8).Given the challenges pertaining to the production and disposal of batteries (RFC1), concerns about e-roads as a cost-efficient CO 2 reduction solution have been raised.
A main concern of the road freight companies interviewed was transport buyers' willingness to pay for e-road transport.The competition in the industry, with profit margins of 2-4%, makes the bargaining power low for negotiating prices for transport with buyers (RFC7).Thus, e-road transport must be cost-advantageous for transport buyers to be a viable decarbonization alternative.
Additionally, several interviewees highlighted the risk of an insufficient secondary market for e-road-compatible vehicles.Although specific to this user segment, such risks may impede e-road technology adoption unless resolved.

Discussion
Several countries have considered the implementation of e-roads to curb CO 2 emissions.The upfront identification and treatment of critical risks in transport infrastructure projects may prevent unnecessary cost overruns and infrastructure delivery delays (Odeck 2019;Herrera et al. 2020).Therefore, the present ranking-type Delphi study investigated the critical risk factors for the implementation of an e-road motorway system for trucks in Sweden.Moreover, to the best of our knowledge, this is the first study to identify and compare the importance of critical risk factors for e-road implementation between the public and private sectors.
Overall, we found that 'low expansion rate,' 'low utilization rate,' and 'lengthy publicsector evaluations' are the three most critical risk factors for stakeholders.The risk of 'low expansion rate' can partly be remedied once the public sector determines which motorway stretches will become e-roads and when (the public sector owns all motorways in Sweden).This declaration may be necessary for vehicle producers to commit to supplying e-road-compatible trucks.Moreover, e-road technology providers may need to scaleup their operations over time.The risk of a 'low utilization rate' can be mitigated by eroad technology providers displaying the advantages of the technology.The reported low expansion and utilization rates are consistent with the findings of previous studies (Börjesson, Johansson, and Kågeson 2021;Taljegard et al. 2020).Previous work found that Swedish e-road motorway system requires at least ten years of use to generate a positive net social present value (Börjesson, Johansson, and Kågeson 2021).Thus, 'lengthy publicsector evaluations' represent a double-edged-sword risk: sufficient a priori evaluations are required for assessing the social value of using significant public funds to implement e-roads; however, for the benefits of e-roads to be realized, implementation in the next few years is imperative (e.g.Börjesson, Johansson, and Kågeson 2021).Even publicsector panelists ranked 'lengthy public-sector evaluations' high.This suggests a process problem in the public sector requiring remedy.'Lengthy public-sector evaluations' have not been modeled or explored in the literature on e-road risk factors.
Some critical risk factors obtained significantly different ranks depending on the panelists' sector.Differences between the public and private sectors are important because of their different goals.The inappropriate risk handling and allocation may make contracts costly.Although ranked high by both private-and public-sector experts, the 'low utilization rate' was ranked significantly higher by the public-sector panel.Thus, it is important that e-road technology providers convince vehicle producers and haulers of e-road advantages.This can help reduce the risk of large public investments in e-road infrastructure not being recouped owing to insufficient user charge revenue.Conversely, 'insufficient grid capacity' was ranked significantly higher by the private sector than by the public sector.Therefore, the public sector should stipulate guarantees of grid capacity to support e-road adoption.This is consistent with the reported challenges of meeting the future grid capacity needs in Sweden (Olovsson et al. 2021).Moreover, academia ranked 'low international interoperability' significantly higher than public-sector panelists.Because of academia's knowledge of technology standards creation and diffusion on a global market scale, we recommend that the public sector consult academia before launching a specific e-road system.For instance, in Sweden, foreign truck traffic is estimated to account for 20% of all truck traffic (Trafikanalys 2018).Achieving high international interoperability increases the incentives for haulers to use e-roads for international truck transport.
The correlation analysis found no complete consensus for any risk factor among the three expert groups.For risk factors that displayed a significant correlation, the relationship to the Kruskal-Wallis test findings are as follows: A significant difference was found in how the public and private sectors rank 'low utilization rate,' according to the Kruskal-Wallis test.However, the significant positive correlation for this risk factor between these two groups indicates that the ranking difference of this factor is negligible.For 'high investment cost,' the positive significant correlation between public sector and academia was consistent with the Kruskal-Wallis test findings.For 'low complementary value to batteries,' the correlation analysis did not confirm the findings of the Kruskal-Wallis test.One explanation is that the rankings within the groups were inconsistent.Overall, the Spearman rank correlation test confirmed the findings of the Kruskal-Wallis test.
Future studies should address the limitations of this study.First, the identified risk factors should be considered given the characteristics of the transport sector in Sweden: the relative importance of critical risk factors may differ in countries with other characteristics.Furthermore, consistent with Sweden's plan, the use of e-roads was assumed to be limited to trucks.Allowing light traffic to use e-roads may alter the relative importance of the critical risk factors.Sweden is geographically large and sparsely populated country, which may affect the risk factors; hence, comparative studies in countries with other profiles are required.
Second, this study was limited to a scenario involving the build-out of a vast national motorway e-road system.However, several transitional risks during the build-out, likely taking years (e.g.Ainalis, Thorne, and Cebon 2020), should be investigated, including the need for a complementary static charging infrastructure or battery-swapping options.Alternatively, a supply of hybrid-fuel vehicles rather than pure electric vehicles may be required during the transition period.A validation user interview revealed that the incentive to invest in e-road-compatible vehicles depends on the system size.The interviews also revealed that uncertainty in the development of a secondary market for electric trucks may create uncertainty in initial investments in such trucks, thereby inhibiting e-road use.Therefore, paths to resolve such uncertainties on the demand side are an important research area.As the e-road stretch for trucks is built in Sweden, lessons about risk factors should be learned.In addition, how e-roads on motorways matter for truck fleet management is worthy of scrutiny.This includes e.g.truck fleet size and configuration, route selection, and traffic scheduling, which are aspects reported crucial to haulers (e.g.Guerrero 2014;Moghdani et al. 2021).Fleet management consequences' importance for e-road utilization is thus important to analyse.
Third, this study was limited to investigating risk factors for one decarbonization option.Therefore, the identification of risk factors and their magnitudes for alternative decarbonization paths for truck traffic is necessary.The validating interviews indicated that e-road use may be plausible for predetermined-route truck traffic but less plausible for order-based truck traffic.The fit of various road freight transport characteristics with the use of different alternative fuel technology options should be researched to understand the combinations of fuel technology solutions that are effective in curbing CO 2 emissions in a cost-efficient manner.This is consistent with the Swedish Transport Administration's recent statement on the need to adopt simultaneous measures to curb long-term truck transport emissions (Plötz 2023).
Future research may benefit from using the identified risk factors as inputs for analysis, for example, how to optimally allocate the identified critical risk factors between the sectors and actors involved in the implementation of e-roads.A related inquiry concerns the handling of critical risks to spur e-road-compatible truck production required for infrastructure utilization.
This study identified 16 critical risk factors for implementing e-roads for trucks in Sweden.Validation interviews with 10 road freight companies corroborated these findings, indicating the correspondence of risk-factor criticality on the supply and demand sides.Moreover, this study compared the importance of these risk factors

Figure 1 .
Figure 1.Overview of the use of the ranking-type Delphi method in the study.

Table 1 .
Summary of candidate inclusion criteria for the e-road Delphi study.

Table 3 .
Round 2: Retained risk factors based on experts' selections (N = 51: N academy = 16, N private = 21, and N public = 14).Ri denotes a risk factor i, n denotes the number of experts shortlisting the risk factor Ri, and N is the total number of experts participating in survey round 2.
Note: rank is based on the average points allocated to risk factors.