4.1. Success or Failure to Complete Trips
The EV model operates by “running” modeled electric vehicles, with differing battery size and charging power, through the trip inventory (333 gasoline vehicles over a year). The battery state of charge of each vehicle is modeled: rising when the vehicle stops where charging is available, falling as the vehicle covers distance. A count of failure to match travel needs is incremented whenever the battery drops to empty during a trip. Trip failures are quantified in two ways: the success fraction is 100% minus trip failure percentage, or days with a trip failure are “adaptation days”.
As an example, the success fraction for EVs charging at Home only is plotted in Figure 3
. This is the count of trip failures, normalized by each vehicle’s total number of trips. For the overall adaptation count in Figure 3
, each vehicle’s fraction of successful trips was averaged across the fleet. The modeled EVs’ range is on the X-axis, and the plug power is on the Y-axis. The surface of all possible EVs is depicted by contour lines, with the number on each contour being the percentage of successful trips. While plug power up to 100 kW was modeled, only those up to 20 kW are displayed in the first several figures. Twenty kW roughly corresponds to the limits of the most common AC charging connectors: the U.S. Society of Automotive Engineers (SAE) standard J-1772, in the U.S. and Japan (up to 19.2 kW at 240 V), or the most common three-phase European plug, the IEC 62196-2 “type 2” (yielding 22 kW at 32 Amp or 11 kW at 16 Amp). Above 20 kW, Figure 3
shows that plug size has very little effect on failure rate—as seen by the near vertical success contour lines at the top of the range (thus the graph’s top is cropped in the figure). This is presumably because Figure 3
is calculated for Home charging only, a situation where a full charge can typically be reached overnight for all but the slowest plug power connections and largest batteries. Plug power is more important when charging time is shorter and/or for batteries that store more energy.
To help interpret Figure 3
, several plug-in vehicles from Table 2
have been marked with an asterisk and labeled on the figure. For example, the original 2011 Nissan Leaf is at the graph location corresponding to its 24 kWh battery, per Equation (1a), its 85-mile range (on the X-axis), and the car’s internal 3.3 kW charger (on the Y-axis). With Home charging only, that vehicle lies between the 95% and the 97.5% success contour lines. In other words, this EV, charged only at home, would successfully complete about 96% of the trips in the trip inventory. By contrast, the asterisk locating a Tesla Model S with an 85 kWh battery (calculated 303-mile range) and a 9 kW Home charger shows that it would successfully complete between 99.5% and 99.65% of the trip inventory even if charged only at Home. Note that the original Chevy Volt, with only 40 miles of all-electric range, accommodates about 85% of round trips, but being a plug-in hybrid, the 15% “trip failures” mean only that the Volt’s built-in gasoline engine then automatically starts in order to complete the trip (and requires time to refill with gasoline later).
The relationships in Figure 3
are illustrated by following lower and higher success rate contours. Suppose only a modest goal is desired, 80% of trip success, that is, the EV makes 80% of trips, but 20% of the time it relies on things like alternative transportation or added stops to recharge. Figure 3
’s 80% contour shows that can be met by a vehicle with only 30 miles range and 2 kW charge power. Above 2 kW (moving up the 80% line), higher charge power has little apparent benefit to this vehicle, since range is its main limitation. By contrast, if the success goal is 99% of all cars’ trips now served by gasoline, following along the 99% contour, two contrasting vehicle configurations can achieve this: either (1) a 19 kW charger with about 160 miles range or (2) a 2 kW charger and a bigger battery with 250 miles range. The cost tradeoff favors the larger 19 kW charger at an incremental cost of roughly USD 250–2000, rather than adding 100 miles of range at an incremental retail cost of USD 6100 (The OEM cost of an EV battery pack in 2021 was USD 157/kWh on a usable-energy basis [28
]. Adding 40% (assembly + vehicle weight support upgrades + retail markup), yields a consumer cost of USD 220/kWh. Thus, adding 100 mi at 280 Wh/mi adds about USD 6160 to the cost).
In more general terms, the efficacy of faster charging compared to an enlarged battery can be seen in Figure 3
as the slope, when tracing up each line of equal trip success: the steeper the slope, the greater the benefit of a larger battery; the flatter the slope, the greater the benefit of added charging power.
The surface topography in Figure 3
suggests some further general principles for charging at Home only. If an EV’s goal is only a majority of gasoline trips (e.g., 80%), a small battery is adequate and faster Home charging does not make much difference. On the other hand, to meet a goal of almost all trips, a bigger battery is required. Nevertheless, a high success fraction, say of 99%, can be met by many different combinations of battery size and recharge power—specifically in Figure 3
, by every combination of X- and Y-axis values lying along the 99% line.
Another general finding is the non-linearity of trip success. Going from 70% to 80% trip success requires less than 10 miles added battery. But incrementing only ½%, from 99% to 99.5%, requires adding over 100 miles. Disproportionally large increases in battery size and/or charging speed are needed to reach 100% success. This will be examined with regard to more charging locations and adaptations as alternatives to very large batteries in Section 4.3
The reader should recall that the trip success contour lines in Figure 3
are based on success for the entire 333 vehicle trip inventory, as will a subsequent analysis of needed “adaptation days”. However, these definitions of trip success or the need for adaptation are analogous to designing a car to meet the needs of all drivers. In Section 4.6
, we will subdivide the trip inventory of all 333 cars’ driving requirements to identify what EV characteristics are needed to meet the needs of different subsets of drivers, that is, we do a market segmentation.
4.2. Charging Availability and “Trip Chains” from EVSE to EVSE
In the preceding analysis a “trip” was defined as travel from one stopped location to another one. In evaluating the effects of varying charging station availability (EVSEs at Home only, Home and Work, or at all stops), the functional definition of a trip is refined to be travel from a stop with EVSE access to the next. Two trips with no EVSE between them are a “trip chain.” For example, when we model EVSEs at both Home and Work, the trip chains possible are from Home to Work, from Work to Home, and from Home back to Home, in each case ignoring intermediate stops since this model specified that they lack EVSEs and thus have no charging during those stops.
illustrates the distribution of trip chain lengths in the trip inventory, comparing the three modeled charger locations. The bars indicate the percentage of trip distances in 5-mile increments. By definition, when charging at Home only, trip chains are longer, and when charging Everywhere (at all stops), trips are shorter. The three lines in Figure 4
show the exceedance, that is, the percentage of trip chains longer than the given distance on the X-axis. For example, with Home charging only (the lightest line in Figure 4
), 14% of trip chains exceed 50 miles. However, if charging is available at every stop, only 4% of trips exceed 50 miles (black line). Looking at the vertical spread of the three exceedance lines, and measuring down from “Home only exceedance”, we see that adding Work charging substantially drops the number of trips (Y-axis) longer than each given distance (X-axis). Then, additionally adding charging Everywhere decreases it about the same amount again. Note that Figure 4
is built only from the data on trips; it does not depend on any modeled EV and chargers.
Distributions of actual trip distances have long tails. Although the X-axis of Figure 4
is truncated to 150 miles, to meet 100% of trips for all 333 cars, an EV must drive over 1000 miles (1610 km) without recharging (per Figures 3 and 4 in [5
]). An EV designed to meet slightly less than 100% of trips, or a driver adapting their vehicle behavior a few times per year, can significantly reduce the vehicle’s range requirement. We will quantify this subsequently. Of course, gasoline vehicles cannot drive 1000 miles without stopping either; they “adapt” by refueling, not reflected as stops in our trip inventory since gasoline filling is shorter than ½ h.
4.3. Metric for Adjusting to Limited Range and “Adaptation Days” per Year
Because the locations of available EVSEs affect the length and count of trip chains, we must define a metric insensitive to different EVSE deployments. Rather than a percentage (trip successes over the total trips each vehicle takes), the next metric is the number of days per year on which alternative travel arrangements would be necessary. This metric corresponds to what Pearre et al. [5
] refer to as “adaptation days,” giving example adaptations including
substituting another vehicle (e.g., swap for a gasoline vehicle in the household or rent one);
making an additional or longer stop to recharge during the day or en route;
delaying some travel (e.g., instead of three side errands after work today, only do two today and the third tomorrow); or
choosing a different mode of transport (commuter rail, bus, air, etc.).
Of the above adaptations, the current article explicitly models and quantifies adaptations (1) and (2). It also analyzes the use of existing stops in the trip inventory as charging opportunities when stopped for other purposes and quantifies the resulting improvement in trip success.
The “adaptation days” metric quantifies potential driver inconvenience, valid to compare across changes in other variables such as charging locations. The prior Figure 3
results measured in success fraction of Home charging only are presented in Figure 5
using this measure of adaptation days.
Again, the axes are vehicle range and charging power, thus each point on the space represents an EV of given range and charging power. Figure 5
shows, for example, that the original Nissan Leaf, with 86 miles of range and a 3 kW charge power, would require driver adaptation 26 times a year, or every other week. As another illustrative example, the GM Volt would need either its built-in gasoline range extender, or some other adaptation, about 100 times a year (about twice a week). In contrast, an 85 kWh Tesla Model S would require adaptations slightly over 3 days a year. All these Figure 5
adaptation counts are based on charging only at Home.
shows that if EVSEs are available at Work locations as well as Home, fewer adaptations are needed. This is seen by the lower numbers of adaptation days for the six example vehicles shown in both figures. It is also seen by the leftward and downward movement of the adaptation day contour lines in Figure 6
compared to Figure 5
If we allow the modeled EVs to charge Everywhere, that is during all stops longer than ½ h, the adaptations required are in Figure 7
indicates that for EVs with any levels of range and plug power, charging at all stops produces another reduction in adaptation days. We can compare the effect of more AC charging at stops on a less capable EV, the original 2011 Nissan Leaf, versus the 85 kWh Tesla Model S. For the Nissan Leaf, going from charging only at Home to charging Everywhere reduced adaptations from 26 to just over 12 per year (compare Figure 5
versus Figure 7
). Note this substantial increase in functionality is achieved simply by increasing availability of low-cost, low-power AC charging stations at stops that the car is making anyway. A smaller improvement is realized for the fast charging (18 kW) 85 kWh Tesla S, which goes from just over three to just over two adaptations per year. That is, having more AC chargers in community locations to charge where drivers stop anyway provides a large improvement to a small-battery vehicle, while providing a much smaller improvement to a more expensive large-battery vehicle. We’ll return to this finding in the discussion.
Intuitively, two adaptations per year seems a small inconvenience (depending on the form of adaptation), whereas 12 and certainly 50 adaptations per year seems unacceptable. To quantify the comparison of adaptation counts, we can compare the range adaptation counts in Figure 5
, Figure 6
and Figure 7
with other forms of adaptation. For example, any vehicle (gasoline or electric) requires a form of “adaptation” in that it must be occasionally taken in for regular maintenance or repairs. Service visits (excluding collisions) average 1.67 times per year [29
]. In addition, gasoline vehicles require 50 trips to the gas station and fuel refills per year (calculated in the next section). The reader can consider the inconvenience of various EV range adaptations versus the inconvenience of going to a station to refuel and of taking the vehicle in for service; here, we offer them as quantitative benchmarks to compare with our adaptation day counts.
4.4. En Route Charging Time Required to Complete All Trips
Our EV model and trip inventory can also be used to calculate the kWh shortfall on long trips. In this section, rather than simply counting each day’s unsuccessful trip(s) as one additional “adaptation day”, here we calculate the kWh shortfall for each of those unsuccessful trips. From the kWh shortfall and the charging power in kW, we then calculate the precise time required to recharge the vehicle via an added stop en route and complete each of those trips. That is, here we quantitatively characterize the adaptation of charging en route as an addition to travel time not required by the gasoline cars in our trip data.
Fast en route EV charging is similar to gasoline refueling; that is, when more onboard energy is needed, the trip is interrupted, and one drives to specific locations to refill with energy. For this analysis, we deviate from this article’s previously discussed charging behavior, all of which assumed that charging only occurs when the driver is stopped for other reasons (Home, Work, etc.). In this section, we assume that en route stations are available along travel corridors (as many countries are installing), and those stations are used by the model only when a trip drains the battery before reaching the next stop with a charging station. Thus, the en route recharge is the only adaptation needed, and it adds to the time required for the trip rather than being counted as a trip failure (More precisely, for each vehicle, it is the additional amount of energy needed to complete any trips that emptied the battery, summed through the year. Dividing energy shortfall per year by the model’s fast-charger power produces a duration, in hours per year, spent at en route charging stations to complete all trips).
Only Home charging is considered in Figure 8
, at both 6 kW and 17 kW. To complete the time analysis, we also estimate the driver engagement time for plugging and unplugging the Home charging station. For Home charging, no time is consumed by waiting for the charge to complete. Based on our own simple stopwatch measures, this is found to be about 7–8 s to plug in and the same to unplug, or 15 s total per charge event. We simplify the time consumed calculation by assuming the typical case of one Home charge per 24 h, thus yearly driver engagement time for Home charging is 15 s/day times 365 days, or 1.52 h/year (Alternatively, if one compared time for charging at both Home and Work, that would increase the daily plug–unplug time but reduce the need for en route charging; we did not consider that additional comparison worth additional space here). Most dedicated Home charging uses an existing reserved space not usable by the public, installed as a simple added circuit on the home electric system. Thus, there is no authorization, credit card swipe, etc. This both simplifies the transaction so it can be completed in the mentioned 7–8 s, and also increases the reliability of the charger.
shows contour lines through the space corresponding to lines of equal hours/year for electric vehicle charging time consumed. Note that we are counting hours to insert and remove the plug daily plus time waiting for en route charging to complete; time at home or at work is not time separately consumed for charging. Moreover, note that although the Y-axis units are kW as before, here the axis only labels the kW of en route charging, whereas all prior graphs showed the kW of normal daily charging. We do compare the daily power of 6 kW versus 17 kW Home charging via parallel grey (6 kW) versus black lines (17 kW) along each contour. Note that the power level of the en route charging has a far greater effect on time required for charging (large increases in time contours as one looks from the top to bottom of the chart) than does the Home charging power level (black and grey lines are very close together). To understand the “en route plug power”, note that during a fast charging session, as the battery fills, the power drops below that reported by the manufacturer; the Y-axis in Figure 8
represents the average power over the charging session.
shows, for example, that the driver of a vehicle with 17 kW at-Home charging, 70 kW en route charging, and 250 miles range, would spend about 5 h per year charging. These five hours are the sum of 3.5 h waiting at 70 kW en route charging stations and about 1.5 h plugging in and unplugging at home. Note that the less time-consuming Home charging accounts for far more charging events and far more total energy added to the EV than does en route charging. As another comparison in Figure 8
, Teslas had a “double AC charger” that could charge at 19 kW; comparing that vehicle with one using the supercharger averaging 80 kW, both with a 60 kWh battery, shows a drop from 20 h/year to under 5 h/year. This substantial improvement quantitatively affirms Tesla Corp’s early decision to provide fast en route charging in order to sell their EVs.
How should we interpret the charging times in Figure 8
—is that a lot of time or a little? The obvious comparison is with how much time we spend fueling gasoline vehicles. As we previously showed in Figure 3
, Figure 4
, Figure 5
, Figure 6
and Figure 7
and the accompanying analysis, Home charging meets most needs, with “adaptation days” rare for large-battery EVs. Gasoline fueling is quicker but always occurs away from home and generally requires driving to a station and waiting during fueling. To calculate total time for gasoline refueling, an average car driving 12,000 miles/year at 27 mpg will consume 444 gallons of gasoline. Gasoline industry statistics give the average refuel as 8.8 gallons [30
], consistent with the research finding that refueling on average adds 2/3 to the tank [31
]. Dividing 444 gallons per year by 8.8 gallons per refuel is 50 visits to a fueling station. Trips to refuel with gasoline require 5.6 min at the gas station [32
]. Time to drive to the gasoline station averages 5 min [33
], of which we estimate 3 min are additional travel time (Although 5 min is the reported travel time to fuel, some of those trips are en route to other destinations, while others are trips for the purpose of refueling. Forty percent of drivers prioritize price in choosing a fueling location (for them, the 5 min may be all added travel), and another seven percent take a round trip just to refuel (for them, 10 min). Considering all these cases, 3 min seems a reasonable approximation for added driving time to achieve a gasoline refill). Thus, total gasoline fueling time is 50 visits times 8.6 min/visit, or about 7 h per year.
A line marked “7” (third down from the top) has been added to Figure 8
for gasoline fueling time to compare with EV charging time. Note that the 7 h gasoline contour was simply drawn for comparison appropriately between 5 and 10 h on the surface, whereas all the EV lines are derived from the range and charging power of the modeled EV.
The EVs in the upper right of Figure 8
spend less time yearly for fueling than the 7 h yearly of a gasoline vehicle. In addition, many EV users do other activities while fast charging en route [1
] not considered here. Older or more moderately priced EVs, for example, the original 2011 Nissan leaf with Home and en route charging, would require 25 adaptations (per Figure 5
). If those adaptations were via en route charging with the Leaf’s optional CHAdeMO charger averaging 38 kW, the driver would consume a total 18 h/year charging (Figure 8
). Those characteristics of the early EVs, and the scarcity at that time of fast en route charging, surely contributed to the perception that EVs require drivers to spend a lot more time refueling than with gasoline. By contrast, drivers of contemporary 2023 EVs, with a range over 200 miles and en route charging power of at least 100 kW, will spend less time refueling than drivers of gasoline cars (Figure 8
). Moreover, those who regularly plug in during long parking opportunities (Home or Work) verbally report little or no time required for fueling and state that low time-and-effort fueling is an advantage
of their EV [1
]. In sum, vehicle capabilities, trip requirements, and whether other activities are done while charging, appear to be major determinants of how likely EV drivers perceive en route charging to be an inconvenience.
4.5. Multi-Vehicle Households
About 60% of U.S. households own more than one vehicle [34
]. When these multi-vehicle households anticipate a long trip or a driving-intensive day, it may be simple to take a fueled vehicle rather than an EV. This concept was described by Kurani et al. [35
] as a “hybrid household”. Today, with increasing numbers of multi-EV households, one might similarly choose their household’s large-battery, fast-charging EV for a longer trip. The fraction of households with multiple-vehicle ownership is given in Table 3
, which shows that our trip inventory sample’s ownership rates are similar to contemporaneous national averages [34
The model realistically required that vehicle exchanges would only happen when both vehicles were at home. For simplicity, we modeled Home charging only, thus the only trip chains are Home–to–Home. We refer to this form of range adaptation as intra-household vehicle substitution, or vehicle swapping for short. For each of the 167 multi-vehicle households, potential EV swaps were extracted, based on a sequence of three times:
finding times when all household vehicles were in use (T1);
identifying the following swap opportunity—with two vehicles at Home (T2);
finding the preceding swap opportunity for those two vehicles (T0); and
assigning to the EV the shorter Home–to–Home trip chains between T0 and T2.
To rephrase this scenario from the household drivers’ perspective, the multi-vehicle model simulates drivers swapping a non-range-constrained vehicle for the modeled EV for long trips. In the resulting set of trips, we count adaptation days. The results are in Figure 9
, which may be compared with Figure 5
. Figure 5
was without intra-household vehicle substitution, and Figure 9
is with it. Figure 9
does not count intra-household substitution as an “adaptation”; if it did, Figure 9
would be exactly the same as Figure 5
Intra-household vehicle substitution significantly reduces adaptations required (compare Figure 9
with Figure 5
). This is with only a Home charger—no fast en route charging and no charging at other locations. For example, in multi-vehicle households with only Home charging, an EV with either a 250-mile range and 10 kW charger, or a 300-mile range with only a 4 kW Home charger, requires one or fewer adaptations per year (see the “1” line in Figure 9
). Without swapping, these EVs require about four adaptation days per year. Swapping does not lower the adaptations to zero because on some days, a few outlying households had all their vehicles’ daily driving exceed the EV’s range.
4.6. Segmentation of the EV Market—Need for a Moderate-Capability EV
Recently, more attention has been paid to improving EV usability for underserved, low-income, and urban communities. The lowest income households drive half as many VMT as high-income households, and urban households drive less than rural ones [36
]. Several studies have shown that a smaller-battery vehicle can meet all household needs for a significant fraction of households [5
]. This paper’s data further elucidate this suggestion. Table 4
gives eight example EVs (varying in battery kWh and charging kW), examples that span a range of actual vehicles (per Table 2
). Note that this paper’s prior metrics show trip success or hours spent charging across the entire database of 333 trip patterns. The shortfall of that approach is to ignore that each individual only requires their own specific set of trips; no one individual requires meeting trips of all 333 drivers in our trip inventory. Following that logic, this section segments the set of the 333 cars’ trip inventory, dividing into users with similar needs. Previously, Kempton suggested such segmentation, giving as illustrative examples for EV segments, one EV of 240 km range (250 miles) and one of 1000 km (620 miles), but these numbers lacked quantitative justification [37
]. A more quantitative approach to market segment has been taken in newer studies such as [38
] and is refined in this paper for Table 4
. Here, we tabulate the number of households for which those EV characteristics and availability of chargers meet every single trip taken in the year.
For example, the first line in Table 4
shows that an EV with a 10.4 kWh battery, charging at 2 kW, and only charging at Home, will fully meet the trip needs of less than 1% of vehicle owners, clearly not a viable market share. Nor does 24 kWh with 3.3 kW seem viable at 4.8% potential market. A bigger jump in the fully satisfied market share is achieved by increasing to a 40 kWh battery. An EV of 40 kWh (range of 143 mi/230 km) with 9.6 kW Home charging meets all driving needs of 11.4% of drivers; adding low-cost AC chargers in added community locations, it meets the needs of 25% of drivers (or 38% of those with another car in the household). Again, these are people who meet 100% of their trip needs with no adaptation or added time all year—and 25% is a significant market segment. As noted in the discussion of Figure 7
, adding low-cost AC chargers in more locations helps a modest-battery-sized vehicle more than it helps a large-battery vehicle.
This finding has equity and EV-access implications. The large-battery vehicle (85+ kWh, 9.6 kW charging) with only Home charging meets all trips with zero adaptations for 33% of drivers. More precisely, if every driver had this EV, the whole population would average only 3 adaptation days per year (Figure 5
), which could be met, for example, by using fast charging on only 3 days per year. However, the 85 kWh EV is notably more expensive and the faster DC-charging events also cost several times more per kWh of electricity. In other words, the current predominant configuration is best suited for people who can afford a higher-priced new car, who have dedicated parking at home, and who do not mind paying a premium, perhaps more than gasoline, for en route DC charging. This solution, in its favor, is cost-effective in equipment (Table 1
) and works well as measured by the few adaptations needed and low time spent.
The urban multi-family housing resident, elderly driver, or low-income car owner on average drives fewer miles, may not have a driveway or any dedicated parking, and is likely more sensitive to high vehicle cost and high en route charging cost. Our analysis suggests an alternative: a 40 kWh EV would cost USD 9900 less. (The 40 kWh vehicle is 45 kWh less in battery size than the 85 kWh one. Retail cost delta, including mounting, battery management, etc., is approximately USD 220/kWh, or for 45 kWh reduction, USD 9900 less cost.) Yet with Home and community 10 kW charging, that EV would meet 100% of the needs of a market segment including 25% of drivers (Table 4
). Home charging might not be available to the urban poor; thus, to serve the urban poor, the “charge everywhere” option would need to include street or lot parking in residential areas near home, as also shown by [38
]. The installation of low-cost community charging in residential areas would make EV charging accessible to residents without dedicated parking. As EVs continue their price decline and a used-EV market grows, providing common public charging in dense areas would make the moderate-range EV more affordable than gasoline is today and could also appeal to market segments who make very few long-distance driving trips, prefer a smaller car, and/or are unable to pay for an expensive new EV.