The shift from conventional internal combustion engine vehicles to electric vehicles (EVs) hinges on both the quantity and quality of EV charging infrastructure. While much research has focused on the importance of and challenges to increasing the quantity of EV chargers worldwide, less has been devoted to assessing the quality of existing EV chargers. Here, we analyze a year’s worth of real-world EV charging data from 132 EVs driven in California, to provide insights about the reliability of EV chargers as experienced by EV drivers. By quantifying and qualifying the level of disruption associated with both real and hypothetical charge failures, we find that EV chargers are not all equally important to EV drivers, highlighting the need for more nuanced charging reliability standards to effectively meet consumer charging needs. While the level of disruption associated with a charge failure for over 90% of charging sessions was fairly low, the consequences of a charge failure for the remaining sessions were catastrophic - 3% of hypothetical charge failures and 0.3% of the real charge failures led to drivers being stranded. These highly disruptive charge failures are associated with long distance travel (LDT), low charger density areas, DC Fast charging, and lack of access to home charging – chargers linked to these conditions should ideally be held to much higher standards than chargers linked to lower levels of disruption.
Article
How Disruptive are Unreliable Electric Vehicle Chargers? Empirically Evaluating the Impact of Charger Reliability on Driver Experience
https://doi.org/10.21203/rs.3.rs-2592351/v1
This work is licensed under a CC BY 4.0 License
Version 1
posted
You are reading this latest preprint version
The transition from conventional internal combustion engine vehicles to electric vehicles (EVs) is considered one of the most promising pathways for decreasing greenhouse gas emissions from the road transportation sector. Several countries have developed official targets to attain a specific number of EV sales in the near future. The aggressive targets to phase in EVs coupled with the severe shortage of charging infrastructure, motivated a myriad of studies on the importance of and the challenges to expanding EV charging infrastructure [1]–[5]. While it’s important to add more EV chargers on the map, it is equally important to ensure that installed chargers are functional. In the 2022 U.S. Electric Vehicle Experience Public Charging survey, J.D. Power found that while the growth of public EV charging infrastructure is making it easier for EV owners to locate public chargers, one out of every five respondents could not charge their EVs at public charging stations. Of those who couldn’t charge, 72% indicated that it was due to the charger malfunctioning or being out of service [6]. Recognizing consumer dissatisfaction with public EV charger reliability, California, in its released Zero-Emission Vehicle Market Development Strategy, and the Northeast States for Coordinated Air Use Management are calling for more stringent charger reliability requirements [7].
In an electrical system, reliability measures the degree to which the performance of the system results in electricity being transferred to the consumer in the amount desired [8]. Therefore, from the perspective of an EV driver/charger consumer, a reliable charger successfully charges their EV, for the expected duration, at an expected rate, after accepting an appropriate payment method. However, from the perspective of most EV charging service providers, a reliable charger is one that meets the minimum uptime requirement of its jurisdiction. Uptime, the most commonly used charger reliability metric, is a measure of the time during which a machine is in operation. This metric does not take into account all the possible technological and logistical challenges within the EV charging ecosystem that ultimately determine the true consumer-facing reliability of chargers. Since the definition of charger reliability differs between consumers and providers, it is no surprise that there is a contradiction between the high average uptime reported by charging providers and the low user satisfaction scores reported by consumers. The California Air Resources Board (CARB) recently conducted a survey in which they asked 11 charging service providers to report their chargers’ average uptime. Four of surveyed providers responded, claiming that they have a national uptime of 95-98% [9]. This response seems to contradict a simultaneous survey of EV drivers in California who reported mixed experience with existing EV chargers. They reported experiencing broken plugs (9%), unexpected shut off during charging (6%), charging station not functioning (22%), payment problems (18%), and the need to contact customer service (53%) [8].
The degree of consumer dissatisfaction with unreliable EV chargers varies depending on how disruptive a charge failure is to the consumer. In this case, disruption is a measure of how quickly the charge failure is discovered by consumers and how easy it is for them to locate and reach the next nearest operational charger. Unreliable chargers can be especially disruptive during long-distance travel (LDT) within low-charger density regions. Many EV drivers meticulously plan their LDT route based on their EVs’ estimated electric range and scope out compatible charging stations in order to sufficiently meet their perceived charging needs. However, in many cases, drivers’ perceived charging needs may not reflect their actual charging needs. The EPA determines an EV’s estimated electric range based on a number of scenarios which mostly cover warm-weather, a mix of speeds, multiple trips, HVAC needs, and some starts and stops [10], [11]. However, this estimated range can be inaccurate and volatile in driving scenarios that were not accounted for. Since actual EV charging needs can deviate from drivers’ perceived EV charging needs, using an EV for LDT is often framed as a challenge over just a means to an end. There are numerous blogs on the internet charting EV road trips that vary in levels of enthusiasm for the technology but all share the same general theme of “it was a struggle, but we made it!” [11]–[16]. A handful of these blogs document instances of EV drivers being stranded in the midst of LDT due to unreliable chargers [12], [13]. These concerns could dissuade EV drivers from using their EVs for LDT which accounts for a significant portion of total vehicle miles traveled [17]–[19]. In the worst case, unreliable chargers may hinder EV adoption rates. While several studies have noted that the expansion of charging infrastructure can significantly alleviate EV range anxiety, a major barrier to EV adoption, the presence of unreliable chargers could undermine this effect [20]–[24]. The sheer size of a charging network may not be enough to engender the consumer confidence required to overcome range anxiety, if a large proportion of its chargers are unreliable. Furthermore, Hardman et al. discovered that consumer dissatisfaction with the convenience of charging is a significant contributing factor to the discontinuance of EV ownership; consumer dissatisfaction with unreliable chargers could prevent current EV owners from continuing to purchase EVs over the incumbent internal combustion engine vehicles.
Despite the salience of consumer dissatisfaction with existing EV chargers and its potential to undermine the benefits of vehicle electrification, very few studies have assessed the consumer-facing reliability of these chargers. There is currently only one major study on EV charger reliability conducted by researchers at UC Berkeley. Rempel et al. (2022) evaluated the functionality of the charging system for 657 EVSE CCS connectors (combined charging system) on all 181 open, public DCFC (direct current fast chargers) charging stations in the Greater Bay Area and found out that 27% of surveyed EVSEs were unreliable or had design failures [8]. 22.7% of EVSEs were inoperable due to unresponsive/unavailable screens, payment system failures, network failure, or damaged connectors [8]. 4.9% of the surveyed EVSEs had cables that were too short to reach the EV inlet [8].
This study complements the study by Rempel et al. (2022) by evaluating the impact of public EV charger reliability on EV driver experience. To accomplish this, we use real-world public EV charging data, collected in California for 132 EVs, spanning three popular EV models (Nissan Leaf Model Years 2011-2017, Chevrolet Bolt Model Year 2017, Tesla Model S Model Years 2012-2017), to quantify how many charging sessions were unsuccessful and qualify how disruptive those unsuccessful charging sessions were to drivers. Due to the limitations of our data collection tool, our charge failure rate is an underestimate of the true charge failure rate as our dataset doesn’t include charge failures wherein EV drivers realized a charger was broken before plugging their EVs into the charger. Therefore, to strengthen our empirical findings, we simulate the level of disruption that would’ve occurred to EV drivers had their successful charging sessions been unsuccessful. We also investigate travel conditions associated with high levels of disruption. The findings in this paper will assist stakeholders to identify charging conditions and locations linked to high levels of disruption in order to have better charging reliability standards to effectively meet consumer charging needs and create a dependable charging infrastructure.
Out of the roughly 10,200 public charging sessions in our dataset, 7% of the sessions were unsuccessful as in no energy or very little energy was transferred from the charger to the cars’ batteries. The stated 7% is an underestimate of the true charge failure rate as it doesn’t include charge failures wherein EV drivers realized a charger was broken before plugging in their EVs. Since unsuccessful charging sessions were logged for a very short duration, we could not effectively extrapolate the charger level for most of these sessions. The identified unsuccessful charging sessions are broken down into categories based on how disruptive they were to EV drivers. Around 96% of unsuccessful charging sessions were ‘Slightly disruptive’ as EV drivers quickly discovered the charge failure and drove to a different charging location immediately. In these cases, the time elapsed between the beginning of an unsuccessful charging session and the beginning of the trip immediately following that charging session is very small, less than 10 minutes with an average of roughly 3 mins. 4% of unsuccessful charging sessions were ‘Moderately disruptive’ as the EV driver discovered the charge failure several minutes or hours after the beginning of the charge attempt. In these cases, the time elapsed between the beginning of an unsuccessful charging session and the beginning of the trip immediately following that charging session is roughly 3 hours on average. Figure 1 illustrates the distribution of the time elapsed between the beginning of an unsuccessful charging session and the beginning of the next trip for slightly disruptive and moderately disruptive unsuccessful charging sessions.
In the previous cases, EV drivers had enough range to complete the next trip despite the failed charge, but 2 unsuccessful charging sessions were “very disruptive” as in the EV did not have enough electric range remaining to complete the next trip. In these two cases, the EV drivers were basically stranded and their EVs needed to be towed to reach the next charging location. Both these charging sessions had a starting battery state of charge of under 7% and occurred in locations with relatively lower charger densities.
Figure 2 shows the car-level distribution of the proportion of unsuccessful charging sessions and Figure 3 shows the proportion of unsuccessful charging sessions by EV charging network. Both Tesla’s cars and charging stations recorded the lowest number of unsuccessful charging sessions.
In order to further understand the impact of charger reliability on driver experience, we simulate the level of disruption that would’ve occurred to EV drivers had their successful charging sessions been unsuccessful.
We define three levels of disruption, as follows, in order to qualify how much of a hassle driver would’ve faced had a successful charging session failed.
Trip Disruptive: We consider a charging session to be trip disruptive if the next trip cannot be completed had that charging session failed. In this case, the electric range remaining prior to the charging session isn’t sufficient to cover the next trip.
Charge Disruptive: We consider a charging session to be charge disruptive if the next charging location cannot be reached had the charging session failed.
Day Disruptive: We consider a charging session to be day disruptive if all the trips that occur after the charging session on that day cannot be completed had the charging session failed.
Out of the roughly 9486 successful public charging events in our dataset, around 9% of the sessions are trip disruptive and over 35% are charge or day disruptive. This implies that, in 9% of charging sessions, if those charging sessions were unsuccessful, EV drivers would not be able to complete subsequent trips as planned, without locating and using another functional EVSE within the electrical range remaining of their EVs. And in 35% of charging sessions, if those charging sessions were unsuccessful, EV drivers would need to alter their habitual/planned charging behavior and charge their EV sooner than initially intended in order to complete all trips within the day. Figure 4 illustrates that
there is a higher proportion of disruptive charging sessions among DC Fast charging sessions over level 2 charging sessions. This observation especially holds true for charge and day disruptive charging sessions; over 55% of successful DCFC sessions are charge and day disruptive while only around 27% of successful level 2 charging sessions are charge and day disruptive.
Figure 5 shows the vehicle-level distribution of the proportion of disruptive level 2 and DCFC charging sessions, respectively. Each point overlaid on the boxplots represents the proportion of disruptive charging sessions for a single car. There seems to be a negative correlation between electric range and rate of disruption for level 2 charging sessions, but the same trend doesn’t hold for DC Fast sessions. There are more disruptive DC Fast sessions than Level 2 sessions for all car models and this observation is especially pronounced for the Tesla Model S vehicles. For the Model S vehicles, the average proportion of charge/day-disruptive charging sessions is around 7% for Level 2 charging sessions but is over 30% for DCFC sessions. Higher electric range and the access to a robust fast charging network (Tesla Supercharging Network) make Model S vehicles more likely to be used for LDT than the other two vehicle models, rendering these vehicles to be more dependent on public DCFC chargers along highway corridors. Figure 6 shows the share of vehicles miles traveled on LDT days (over 50 miles) as a percentage of total VMT and confirms that the Model S vehicles are more likely to be used for LDT.
successful charging sessions were more prone to be disruptive upon failure than others. Table 1 provides the means of each travel condition for disruptive and non-disruptive charging sessions, along with the results of the Kruskal-Wallis test. The difference in means between trip disruptive and trip non-disruptive charging sessions is very significant for most analyzed travel conditions. All disruptive charging sessions are associated with higher distance traveled in the day and lower starting battery state of charge compared to their non-disruptive counterparts. The average distance traveled in day among trip disruptive charging sessions is around 30% higher than the average distance travelled by charge and day disruptive charging sessions. Since higher battery capacity vehicles, such as the Model S, are more likely to be used for LDT, trip disruptive charging sessions are associated with higher battery capacity whereas charge and day disruptive charging sessions are associated with lower battery capacity, compared to their non-disruptive counterparts. In addition to the five travel behavioral conditions, we also analyzed how access to home charging impacted charge disruption levels. We found that drivers who did not have access to home charging are 9% more likely to experience charge and day disruptive charging sessions than drivers with access to home charging; this is likely because drivers without access to home charging primarily rely on workplace or public charging to meet their daily charging needs whereas drivers with access to home charging primarily rely on home charging and are not as dependent on outside charging sources.
Table 1 Travel Conditions affecting Level of Disruption Means and Kruskal-Wallis Test Results
|
Travel Condition |
Disruption Level |
Disruptive Mean |
Non-disruptive Mean |
Kruskal Wallis Test |
|
Vehicle Battery Capacity (kWh) |
Trip |
53.71 |
49.67 |
F Statistic = 80.41 p = 3.040e-19 **** |
|
Charge |
39.33 |
56.14 |
F Statistic = 727.91 p = 2.552e-160 **** |
|
|
Day |
39.35 |
56.53 |
F Statistic = 777.94 p = 3.383e-171 **** |
|
|
Distance Travelled in Day (miles) |
Trip |
115.33 |
70.37 |
F Statistic = 363.96 p = 3.865e-81 **** |
|
Charge |
89.11 |
72.35 |
F Statistic = 207.25 p = 5.462e-47 **** |
|
|
Day |
90.85 |
70.97 |
F Statistic = 294.62 p = 4.891e-66 **** |
|
|
Number of Charges in Day (#) |
Trip |
2.23 |
1.71 |
F Statistic = 323.64 p = 2.332e-72 **** |
|
Charge |
1.70 |
1.85 |
F Statistic = 81.97 p = 1.380e-19 **** |
|
|
Day |
1.75 |
1.82 |
F Statistic = 33.15 p = 8.522e-09 **** |
|
|
Starting Battery State of Charge (%) |
Trip |
38.40 |
51.74 |
F Statistic = 503.13 p = 1.985e-111 **** |
|
Charge |
32.43 |
58.36 |
F Statistic = 3245.24 p = 0.0 **** |
|
|
Day |
32.73 |
58.81 |
F Statistic = 3350.37 p = 0.0 **** |
|
|
Distance from Home (miles) |
Trip |
49.15 |
31.41 |
F Statistic = 201.73 p = 8.767e-46 **** |
|
Charge |
32.34 |
35.70 |
F Statistic = 1.09 p = 0.296 |
|
|
Day |
32.79 |
35.52 |
F Statistic = 0.04 p = 0.851 |
The trip, charge and day disruptive charging session criteria don’t necessarily imply that an EV driver would be stranded at some point in a day if those charging sessions had failed. For instance, if a trip disruptive charging session had failed, the EV driver could find an alternative operational charger along their route and finish their next trip; in this case, finding another charging location may be very inconvenient to the EV driver, but this isn’t the worst case scenario. This is why we also identified location critical charging sessions without which EV drivers may potentially end up being stranded. A charging session is considered location critical if the EV doesn’t have enough electric range remaining to reach the next nearest charging location had the charging session failed.
We found that in around 2.87% of successful charging sessions, drivers were one unsuccessful charging session away from being stranded. The locations of these critical charging sessions are plotted on the map in Figure 7. In these cases, the EVs did not have sufficient range remaining to get to the next nearest charging station. Many of these critical charging sessions occurred in relatively low charger density areas within the central valley and rural northern California as illustrated in Figure 7. Moreover, these critical charging sessions were more likely to occur during LDT days, as suggested by the total day distance distributions in Figure 8.
This study reveals that EV charging sessions are not all equally important and that not all charge failures have the same consequences. A charging session is more important to an EV with a low state of charge in a low charger density location compared to an EV with a high state of charge in a high charger density location. We simulated the level of disruption that would’ve occurred to the drivers had their charging sessions failed. In 9% of charging sessions, if those charging sessions were unsuccessful, EV drivers would not be able to complete subsequent trips as planned, without locating and using another functional charger within the electrical range remaining of their EVs. And in 35% of charging sessions, if those charging sessions were unsuccessful, EV drivers would need to alter their habitual/planned charging behavior and charge their EV sooner than initially intended in order to successfully complete all trips within the day. There is a higher proportion of potentially disruptive charging sessions among DC Fast charging sessions over level 2 charging sessions, especially for the Model S vehicles which are more likely to be used for LDT. All potentially disruptive charging sessions are associated with LDT, low battery SOC, low charger density areas, DC Fast chargers, and lack of access to home charging. In around 3% of charging sessions, the EVs were one unsuccessful charging session away from being stranded; most of these location critical charging sessions occurred during LDT, in low charging station density areas.
This study also finds that 7% of recorded charging sessions were unsuccessful. Based on how quickly drivers discovered these charge failures, we reasoned that over 95% of them were only slightly disruptive to the EV drivers but in a small number (2) of cases EV drivers did end up being stranded. Both Tesla’s cars and charging stations recorded the lowest proportion of unsuccessful charging sessions. Our loggers only recorded ‘plug-in’ events as charging sessions, so the stated 7% is an underestimate of the true charge failure rate as it doesn’t include charge failures wherein EV drivers realized a charger was broken before plugging in their EVs. Moreover, we cannot extrapolate exactly why these charging sessions were unsuccessful. Rempel et al. (2022) found that unresponsive/unavailable screens, payment system failures, network failure, or damaged connectors were major contributing factors to the charge failures in their study.
While not all EV charging sessions are equally important to drivers, all EV chargers are essentially held to the same reliability standards by stakeholders, revealing that current EV charging reliability standards do not consider the nuances in consumer charging needs. A charger in a low charger density area along a LDT route is held to the same uptime requirement as a charger in a high charger density urban area. This is a problem because the consequences for a charge failure in low charger density areas can be more dire than the consequences for a charge failure in high charger density areas. It’s the difference between being stranded in the middle of nowhere and being a little annoyed about driving less than a mile away to the next functional charger. Moreover, Chargers along LDT routes or in low charger density areas may be more prone to technical, communication and logistical failures due to weak network connectivity. These concerns could dissuade EV drivers from using their EVs for LDT which accounts for a significant portion of total vehicle miles traveled [25]. Or in the worst case, motivate EV owners to give up their EVs and return to gas vehicles [26]. Stakeholders should have higher reliability standards for chargers that are deemed to be more critical for consumers; this can be in the form of a higher uptime or the instillation of more chargers in locations with those critical chargers so that if one charger goes down, there will be an alternative.
Uptime, in general, may be an insufficient metric to measure EV charging reliability. There are numerous electrical, mechanical, software, communication and logistical factors within the EV charging ecosystem that ultimately determine the operational status of an EV charger. On the electrical side, the components within a charger’s ICCB are responsible for temperature monitoring, contact monitoring, leakage current detection, and overcurrent detection - all to ensure safe supply of power to the EV. A failure within even one of these components can make the charger unreliable and potentially unsafe [27]. On the mechanical side, the external components of the chargers are prone to damage from various environmental factors. Consumers tend to drop EV chargers repeatedly, wrap and drive over cables, as well as leave them out in the rain. Animals may chew on the cables. charger design must account for these mechanical challenges [27]. On the communication side, configuration errors, line damage, power loss or traffic spikes, and hardware failure anywhere along the communication network within the charging ecosystem may cause reliability issues as it can interrupt the flow of information between the various actors in the EV charging ecosystem. On the logistical side, membership requirements, payment issues, complicated charger instructions/operations, difficulty locating chargers, lack of charger availability, and poor cell service/Wi-Fi availability make EV charging daunting to current and prospective EV drivers.
Figure 9 extends these technical and logistical factors to all stakeholders in the charging ecosystem where each stakeholder is color-coded with the potential factors that can compromise charger reliability.
The calculation for uptime doesn’t seem to adequately incorporate the aforementioned factors. Moreover, there is currently no standard method to measure uptime across EV charging service providers and no standard minimum uptime requirement across jurisdictions. It is critical to have a common definition and standard for uptime so EV drivers aren’t blindsided by a slew of unreliable chargers from charging service providers claiming high uptimes. Moreover, the complexity of measuring charging reliability calls for standardizing not only the reliability measures but also the test procedures for collecting the data. The stakeholders that are responsible for maintaining EVSEs are also not standardized across jurisdictions - the responsibility can be with the local electric utility, the installer, the site host, the charge point operator, or the servicing company. These stakeholders will likely have different levels of visibility over the system, making it especially challenging to maintain high uptime [8]. Overall, the complex dynamics within the charging ecosystem and lack of a standardized method to measure uptime make it especially challenging to accurately measure and maintain this metric.
In addition to standardizing and increasing uptime requirements across charging service providers and jurisdictions, stakeholders should explicitly enforce frequent and thorough maintenance and servicing of chargers. Maintenance should include repairing damaged charger parts, removing garbage/obstacles surrounding the charger, and ensuing that the software and communication protocols are functional. These routine checks would lower the probability of a charge failure and ensure that any faults within the chargers are quicky resolved. Additionally, charging reliability standards for chargers in a given location should take into account that location’s charger density and the chargers’ associated utilization rates. In a location with a relatively high density of chargers, the reliability of each charger does not need to be very high as if one port is inoperable, EV drivers can easily locate and access another port nearby. On the other hand, in a location with a relatively low density of chargers, the reliability of each charger should be high as if one port is inoperable, EV drivers are less likely to have enough electric range remaining to reach the next nearest charger. This recommendation can lower the costs associate with maintaining charger reliability as chargers in the presence of high charging station redundancy need not be held to very high reliability standards. Charger signage should include instructions to locate the next nearest charging port in case of a charge failure. If reliability standards are lower in locations with high charging station redundancy, signage to more easily locate the next nearest charging point could lower the level of driver disruption associated with more frequently encountering inoperable charging ports in those locations.
Our study uses real-world EV driving and charging data from 132 EVs driven in California, over a one year period. The charging sessions from the EVs in the study were geographically merged with the U.S. Department of Energy’s Alternative Fueling Stations Dataset to get more robust charging station information.
Aggregate Driving and Charging Data
The driving and charging data analyzed in this study is a subset of the data from the Advanced PEV Driving and Charging Behavior project, a California-wide study spanning five years (2015-2020) that aims to understand the driving and charging behavior of EVs [28][29]. This EV study collected on-road data from around 400 households and 800 vehicles (400 EVs). Households selected for the study that contained at least one EV had FleetCarma’s C2 or C5 data logger installed in each of their vehicles for roughly 12 months. The loggers collected key driving and charging attributes such as vehicle speed, battery state of charge, and electrical energy consumption. Table 2 provides an overview of the data analyzed in this study. We are focusing on trip and charging session data, collected over a one year period for 132 EVs, spanning three popular EV models: the Nissan Leaf (Model Years 2011-2017, 24-30 kWh battery capacity), the Chevrolet Bolt (Model Year 2017, 66 kWh battery capacity), and the Tesla model S (Model Years 2012-2017, 60-100 kWh battery capacity). We are only interested in evaluating non-home charging sessions in this study. Figure 10 presents the percent share of non-home charging sessions by charger level.
Table 2 EV Data Overview
|
EV Type |
Number of Vehicles |
Number of Trips |
Total Miles Traveled (mi) |
Number of Charging Sessions |
Number of Non-home Charging Sessions |
Total kWh Charged |
|
Nissan Leaf |
57 |
79202 |
532425 |
16091 |
5415 |
120575 |
|
Chevrolet Bolt |
27 |
47760 |
382603 |
9434 |
1137 |
100612 |
|
Tesla Model S |
48 |
46361 |
660619 |
12073 |
3671 |
246597 |
Charging Stations Geographic Data
We used the U.S. Department of Energy’s Alternative Fueling Stations Dataset to get more robust charging station information. This dataset contains comprehensive charging station metrics such as location, charger/connector types etc., for most existing public charging networks in the U.S. Based on a study by Xu et al. (2021), that successfully combined charging station datasets, we used DBSCAN clustering to match charging station locations to charging session locations [30]. DBSCAN typically has two parameters: minPts is the minimum number of neighbors a point must have to be within a cluster and epsilon is the search radius used to figure out if two points are neighbors. The magnitude of these two parameters were based on findings in the study by Xu et al., who determined that an appropriate search radius for DBSCAN for charging station matching should be in the range of 50 to 200 meters. The minPts was set to 2 as we intended to match a charging session to a charging station and epsilon was set to 50 meters, the lower end of the recommended range as this led to the most robust matches. The Alternative Fueling Stations Dataset’s coverage isn’t perfect so we could only identify accurate charging station information for roughly 67% of charging sessions in our dataset.
Level of Disruption from Simulated Charge Failure
We define three levels of disruption, as follows, in order to qualify how much of a hassle drivers would’ve faced had a successful charging session failed.
Trip Disruptive: We consider a charging session to be trip disruptive if the next trip cannot be completed had that charging session failed. In this case, the electric range remaining prior to the charging session isn’t sufficient to cover the next trip. To determine if a given charging session is trip disruptive for an EV driver, we compare the distance covered by the trip immediately following the charging session to the maximum distance the EV could travel given its SOC prior to the charging session. If the SOC derived distance is lower than the next trip’s distance, then we classify the charging session as trip disruptive.
Charge Disruptive: We consider a charging session to be charge disruptive if the next charging location cannot be reached had the charging session failed. To determine if a given charging session is charge disruptive for an EV driver, we compare the distance covered by all trips between that charging session and the immediately following charging session to the maximum distance the EV could travel given its SOC prior to the charging session. If the SOC derived distance is lower than the distance covered until the next charging session, then we classify the charging session as charge disruptive.
Day Disruptive: We consider a charging session to be day disruptive if all the trips that occur after the charging session on that day cannot be completed had the charging session failed. In this case, the electric range remaining prior to the charging session isn’t sufficient to cover all remaining trips within that day, even after accounting for other successful charging sessions within that day. To determine if a given charging session is day disruptive for an EV driver, we compare the distance covered by all trips from that charging session till the end of that day to the maximum distance the EV could travel given its SOC prior to the charging session. If the SOC derived distance is lower than the distance covered from the session to the end of the day, then we classify the charging session as charge disruptive.
The maximum distance that can be traveled given an SOC of SOCX for an EV with an estimated electric driving range of RX is estimated using the following equation:
The time segment in Figure 11 captures all driving and charging events that occur from Chargex at 12AM to the end of the day at 12PM for a given EV– this includes four trips and two charging sessions. We will simulate a charge failure on Chargex i.e., pretend the charge fails. In order for Chargex’s hypothetical failure to be trip-disruptive, the estimated maximum distance given the SOC before Chargex needs to be lower than Distance A which represents the total distance of Tripx. In order for Chargex’s hypothetical failure to be charge-disruptive, the estimated maximum distance given the SOC before Chargex needs to be lower than Distance B which represents the combined distance of Tripx and Tripx+1. In order for Chargex’s hypothetical failure to be day-disruptive, the estimated maximum distance given the SOC before Chargex needs to be lower than Distance C which represents the combined distance of Tripx ,Tripx+1,Tripx+2 and Tripx+3. The above technique was applied to all charging sessions in our dataset to simulate the level of disruption associated with hypothetical charge failures.
Travel Conditions impacting Level of Disruption
In addition to classifying the charging sessions by level of disruption, we also analyze five travel behavioral conditions to better understand why some successful charging sessions were more prone to be disruptive upon failure than others. We compared the following travel conditions between disruptive and non-disruptive charging sessions:
- Vehicle battery capacity
- Total distance travelled in the day in which the charging session occurred
- Number of charging sessions in the day in which the charging session occurred
- Starting battery state of charge prior to the charging session
- Distance between the charging session location and the driver’s home
- Driver’s access to home charging
Since none of the mentioned travel conditions met the normality assumption of the ANOVA test, we opted to use the Kruskal-Wallis test, the non-parametric equivalent of the ANOVA test, to analyze the difference between the means of disruptive and non-disruptive charging sessions. The Kruskal-Wallis test is a rank-based non-parametric test that can deduce if there are statistically significant differences between two or more groups of an independent variable on a continuous or ordinal dependent variable.
Location Critical Charging Sessions
The trip, charge and day disruptive charging session criteria don’t necessarily imply that an EV driver would be stranded at some point in a day if those charging sessions had failed. For instance, if a trip disruptive charging session had failed, the EV driver could find an alternative operational charger along their route and finish their next trip; in this case, finding another charging location may be very inconvenient to the EV driver, but this isn’t the worst case scenario. This is why we also identified location critical charging sessions without which EV drivers may potentially end up being stranded. A charging session is considered location critical if the EV doesn’t have enough electric range remaining to reach the next nearest charging location had the charging session failed. To find the next nearest charging station for a charging session, we used DBSCAN clustering with minPts set to 5 and a search radius equivalent to the estimated maximum distance that can be traveled given the SOC prior to the charging session. We then estimated the road network distance from the charging session to each matched charger; if this distance is greater than the estimated maximum distance that can be traveled given the SOC prior to the charging session for all matched chargers, then we mark that charging session as location critical.
Quantifying and Qualifying Observed Unsuccessful Charging Sessions
In this part of the analysis, we quantify unsuccessful charging sessions from our dataset and then qualify how much of a hassle these unsuccessful charging sessions were to the EV drivers.
In this study, we identify an unsuccessful charge based on the following conditions:
- There is no or very little energy (less than 0.5 kWh) delivered to the EV’s battery by the end of the charging session.
- The charging session immediately following that charging session is at a different charging station (at least 50 meters away from the current charging station).
We qualify the level of disruption associated with identified charge failures based on the time it took drivers to notice the charge failure and move on to the next operational charger; this is based on the time elapsed between the beginning of the unsuccessful charges and the beginning of immediately subsequent trips.
Ethics and Consent
The study has been approved by the Institutional Review Board (IRB) Administration of the University of California, Davis. The study has adhered to all relevant ethical regulations in the study of human subjects for social science research. Prior to participating in the logger study, all participants provided their informed consent. All results presented in the analyses are anonymized to protect the participants’ personal information and maintain confidentiality.
Data Availability
The data used in this study is a subset of a preexisting dataset that is currently confidential and cannot be made publicly available due to legal or ethical restrictions. However, interested researchers may contact the authors to request access to the full dataset, subject to the provider's data sharing policies and conditions. Aggregated and anonymized versions of the dataset will be made available from a digital repository.
- S. Hardman et al., “A review of consumer preferences of and interactions with electric vehicle charging infrastructure,” Transp. Res. Part D Transp. Environ., vol. 62, pp. 508–523, Jul. 2018.
- F. Baumgarte, M. Kaiser, and R. Keller, “Policy support measures for widespread expansion of fast charging infrastructure for electric vehicles,” Energy Policy, vol. 156, p. 112372, Sep. 2021.
- R. J. Javid, M. Salari, and R. Jahanbakhsh Javid, “Environmental and economic impacts of expanding electric vehicle public charging infrastructure in California′s counties,” Transp. Res. Part D Transp. Environ., vol. 77, pp. 320–334, Dec. 2019.
- J. Dong, C. Liu, and Z. Lin, “Charging infrastructure planning for promoting battery electric vehicles: An activity-based approach using multiday travel data,” Transp. Res. Part C Emerg. Technol., vol. 38, pp. 44–55, Jan. 2014.
- B. J. Mortimer, C. Hecht, R. Goldbeck, D. U. Sauer, and R. W. De Doncker, “Electric Vehicle Public Charging Infrastructure Planning Using Real-World Charging Data,” World Electr. Veh. J. 2022, Vol. 13, Page 94, vol. 13, no. 6, p. 94, May 2022.
- J.D. Power, “2022 U.S. Electric Vehicle Experience (EVX) Public Charging Study | J.D. Power.” [Online]. Available: https://www.jdpower.com/business/press-releases/2022-us-electric-vehicle-experience-evx-public-charging-study. [Accessed: 09-Feb-2023].
- “Reliability Blog Series #2: Supporting EV Drivers with a Charging Station Reliability Standard.” [Online]. Available: https://www.flo.com/blog/reliability-blog-series-2-supporting-ev-drivers-with-a-charging-station-reliability-standard-flo/. [Accessed: 26-Jul-2022].
- D. Rempel, C. Cullen, M. Matteson Bryan, and G. Vianna Cezar, “Reliability of EV Direct Current Fast Chargers Reliability of Open Public Electric Vehicle Direct Current Fast Chargers.”
- E. Standards Staff, “ELECTRIC VEHICLE SUPPLY EQUIPMENT STANDARDS TECHNOLOGY REVIEW,” 2022.
- “What to Do When Your Electric Car Runs Out of Power - Bloomberg.” [Online]. Available: https://www.bloomberg.com/news/articles/2022-05-21/what-to-do-when-your-electric-car-runs-out-of-power. [Accessed: 26-Jul-2022].
- “5 Reasons Battery-Electric Vehicle Road Trip Range Stinks | Torque News.” [Online]. Available: https://www.torquenews.com/1083/5-reasons-ev-road-trip-range-60-or-less-maximum-range-possible. [Accessed: 26-Jul-2022].
- Tim Levin, “How my first road trip in an electric car almost ended in disaster,” Business Insider , 23-Nov-2021. [Online]. Available: https://www.businessinsider.com/ev-electric-car-ford-mustang-mach-e-road-trip-charging-2021-8. [Accessed: 21-Dec-2021].
- Jennifer Sensiba, “Charger Reliability Is The Next Challenge For The EV Industry,” CleanTechnica, 21-Apr-2021. [Online]. Available: https://cleantechnica.com/2021/04/21/charger-reliability-is-the-next-challenge-for-the-ev-industry/. [Accessed: 21-Dec-2021].
- “I rented an electric car for a 4-day road trip. I spent more time charging it than I did sleeping. | Fox Business.” [Online]. Available: https://www.foxbusiness.com/lifestyle/electric-car-four-day-trip-more-time-charging-sleeping. [Accessed: 26-Jul-2022].
- “Woman’s Road Trip Highlights Issues Taking EV Cross-Country.” [Online]. Available: https://www.businessinsider.com/electric-car-road-trip-highlights-issues-driving-cross-country-charging-2022-6. [Accessed: 26-Jul-2022].
- “Can a Big Western Road Trip Be Done in an Electric Car? - Outside Online.” [Online]. Available: https://www.outsideonline.com/adventure-travel/essays/tesla-electric-car-road-trip-test/. [Accessed: 26-Jul-2022].
- L. Aultman-Hall and H. Ullman, “Long-distance and intercity travel: Who participates in global mobility?,” Mapp. Travel Behav. Genome, pp. 187–207, Jan. 2020.
- K. Goulias, “Long Distance Travel in the California Household Travel Survey (CHTS) and Social Media Augmentation,” May 2017.
- G. Tal and M. A. Nicholas, “Exploring the Longest Trip: A New Look on the Impact of Long Road Trips on VMT,” 2016.
- L. Noel, G. Zarazua de Rubens, B. K. Sovacool, and J. Kester, “Fear and loathing of electric vehicles: The reactionary rhetoric of range anxiety,” Energy Res. Soc. Sci., vol. 48, pp. 96–107, Feb. 2019.
- R. Chen, X. Liu, L. Miao, and P. Yang, “Electric Vehicle Tour Planning Considering Range Anxiety,” Sustain. 2020, Vol. 12, Page 3685, vol. 12, no. 9, p. 3685, May 2020.
- J. Neubauer and E. Wood, “The impact of range anxiety and home, workplace, and public charging infrastructure on simulated battery electric vehicle lifetime utility,” J. Power Sources, vol. 257, pp. 12–20, Jul. 2014.
- D. Pevec, J. Babic, A. Carvalho, Y. Ghiassi-Farrokhfal, W. Ketter, and V. Podobnik, “Electric vehicle range anxiety: An obstacle for the personal transportation (r)evolution?,” 2019 4th Int. Conf. Smart Sustain. Technol. Split. 2019, Jun. 2019.
- D. Pevec, J. Babic, A. Carvalho, Y. Ghiassi-Farrokhfal, W. Ketter, and V. Podobnik, “A survey-based assessment of how existing and potential electric vehicle owners perceive range anxiety,” J. Clean. Prod., vol. 276, p. 122779, Dec. 2020.
- “DC Fast as the Only Public Charging Option? Scenario Testing from GPS-Tracked Vehicles.” [Online]. Available: https://trid.trb.org/view/1130070. [Accessed: 26-Jul-2022].
- S. Hardman and G. Tal, “Understanding discontinuance among California’s electric vehicle owners,” Nat. Energy, vol. 6, no. 5, pp. 538–545, May 2021.
- “Quality and Reliability Matter When It Comes to EV Chargers.” [Online]. Available: https://www.aptiv.com/en/insights/article/quality-and-reliability-matter-when-it-comes-to-ev-chargers. [Accessed: 26-Jul-2022].
- G. Tal et al., “Advanced Plug-in Electric Vehicle Travel and Charging Behavior Final Report (CARB Contract 12-319-Funding from CARB and CEC),” 2020.
- G. Tal et al., “Emerging Technology Zero Emission Vehicle Household Travel and Refueling Behavior (CARB Contract 16RD009),” 2021.
- B. Xu, A. Wilkinson Davis, and G. Tal, “Estimating the Total Number of Workplace and Public Electric Vehicle Chargers in California,” vol. 2675, no. 12, pp. 759–770, 2021.
There is NO Competing Interest.