As was discussed before, it is important to take into account how many vehicles a household owns when studying the implementation of EV’s. An electric car is more easily integrated when the household has a conventional car available for trips that go beyond the range of their EV. We hypothesize that households without this luxury are more likely to consider a vehicle equipped with a range-extender (like a PHEV).
In Figure 11, a pie-diagram is shown of the breakdown of ‘cars per household’ from the Vehicles dataset, weighted by ‘WTHHFIN’. Again, only cars, vans, SUVs and pickup trucks were selected from the dataset (294,409 out of 307,956 vehicles).
Figure 11: Number of total cars, vans, SUVs and pickup trucks per household based on 150,147 household surveyed in the NHTS 2009.
From the households that own at least one car, 64% own two or more cars.
Driven distance per vehicle
Now that we found out how many cars households have available, we can assess the travelled distance per vehicle from the Travel Day dataset. The length of individual trips, as well as the total travelled distance per vehicle-day was investigated. The latter is important because electric cars are likely to be charged at least once during the day (over night, at home). Because a significant difference is found between driving patterns of urban and rural households, results were also displayed for both groups separately.
Individual trip distance
The relative occurrence of individual car trips (one way) was found to rapidly decrease with trip distance beyond 3 miles. In fact, almost 10% of all 748,918 recorded individual car trips were shorter than a mile. Figure 12 shows a histogram with 1-mile bins for distance. There is some inaccuracy in the graph, since it appears that participants tend to round up their reported distance to 5 mile increments.
Figure 12: Individual trip distance distribution from 748,918 car trips recorded in the 2009 NHTS. Each trip is weighted with variable ‘WTTRDFIN’.
Plotting the distribution of trips over a Cumulative Distribution Function (Figure 13) gives an indication of the percentage of one-way trips that can be covered by electric vehicles with a range equal to the number on the x-axis. However, underlying trip characteristics like driving speed and the use of climate control can significantly affect electricity consumption (paragraph 4). These factors are not recorded in the dataset and the use of the graph is therefore limited.
Figure 13: Cumulative distribution of driven miles per trip from 748,918 car trips recorded in the 2009 NHTS.
The graph shows that 95% of trips are shorter than 30 miles, and 99% is below 70 miles. The weighted average trip distance is 9.4 miles. Vehicles owned by urban households averaged 8.5 miles and rural vehicles averaged 12.1 miles.
Overall, trips for errands, meals and school are shortest (See Table 3). Trips for recreation and work are on average the longest, with means of 15.4 and 12.1 miles, respectively. As expected, differences are found between urban and rural households: Car trips to work were found to be (weighted) 3.5 miles longer for rural households (14.8 miles) compared to urban (11.2 miles). Another interesting difference is trips to not so widely distributed services like doctors/dentists, which accounted for 2% of all recorded trips: urban cars used for this purpose travel only 8.9 miles on average, while people from rural households tend to travel almost twice as far (16.7 miles) for these services.
Table 3: Overview of purposes of car trips recorded in the NHTS 2009. Means and 95th percentile distances are given for urban, rural and all trips weighted.
As was mentioned before, significant differences between States can be found for driven distances. It is hypothesized that driving distances are influenced by the:
State being bound by regions where cars cannot go (e.g. water);
State’s size (e.g. DC);
number of single dwellings (typically agricultural States);
ratio of urban/rural households, or population density in general.
Studying the impact of these different factors is beyond the scope of this study and perhaps material for further research, but several interesting States can be identified where driven car distances differ significantly from the U.S. mean. One good example is Hawaii, where the State’s biggest island Hawai’i measures 90 miles from tip to tip. The other highly populated islands are O’ahu and Maui, both measure around 45 miles from tip to tip. The longest highway on the Island of Maui is about 60 miles long, looping around the island. From the 1,226 car trips recorded in the NHTS, 99% was shorter than 30 miles and the mean distance was 5.95 miles. Besides the favorable distances in Hawaii for the integration of EV’s, the economics also make sense, as gas prices are the highest found in all of the United States.
The District of Columbia also shows low trip distances, averaging 6.5 miles, as the whole DC area is ‘urban’ and counts only 61 square miles. The only contributor to long car trips would be those that have destinations in neighboring States, for instance commutes to Baltimore, Maryland or perhaps trips to Philadelphia.
Daily driven distance
The one-way trip distance distribution may not be a good indicator of the necessary range for an electric car. This is especially true for the first years of implementation, as the charging infrastructure for EVs is limited to Level 1 and Level 2 chargers at the owners’ homes. For this reason, daily driven distances for vehicles were calculated, assuming the EVs will be charged overnight. Trip distances were summed for unique Vehicle IDs and plotted in the same histogram (Fig. 14) and cumulative distribution graphs (Fig. 15). Note that the graphs do not include cars that were not used on the Travel Day.
Cars that were used on their Travel Day in the NHTS made an average of 4.2 trips, yielding a weighted average daily distance of 39.5 miles. The distribution of the total driven distance on the Travel Day is depicted in Figure 14.
Figure 14: Distribution of daily driven distance for U.S. household cars if the car is used that day. Data source: NHTS 2009
With car trips aggregated for the Travel Day, 93% of all vehicle-days show a total distance below 100 miles. It is important to note that only vehicle-days are included where the cars were used that day. As was mentioned before, 39% of cars owned by the participating households were not used on the Travel Day.
Figure 15: Cumulative distribution curve for daily driven distance by cars that were used on the Travel Day (representing 61% of all cars owned by the participating households). Data source: NHTS 2009.
Again, a significant difference was found between daily driven distance of urban and rural household owned vehicles. Urban vehicles used on the Travel Day averaged 36.5 miles and rural vehicles averaged 48.6 miles.
Vehicles from Hawaii and District of Columbia showed 99.0% and 98.1% of daily driven distances below 100 miles, with averages of 24.5 and 24.3 miles per vehicle-day. Alabama, Kansas, Missouri and Montana showed the highest driven distances per vehicle-day, with averages between 48 and 49 miles. New York, California and Texas averaged 34.8, 36.2 and 41.0 miles, respectively.
The use of electric cars for commuting makes sense, as the daily commute is typically of fixed length and employees’ parking space is likely to be one of the first places where charging stations will be installed. With the NHTS 2009 data, a distribution of commuting distances was made based on the variable DISTTOWK from the Person dataset. Again, only people are selected that commute to work by car, SUV, van or pickup truck.
Figure 16: One-way distance distribution for commutes made by car. Data source: NHTS 2009.
Figure 17: Cumulative distance distribution for commutes made by car. Data source: NHTS 2009.
As can be seen from Figure 17, approximately 95% of car commuters in the U.S. travel less than 40 miles to work (the weighted average is 13.6 miles). Weighted averages for States vary from 7 miles (Alaska, North & South Dakota) to 22 miles (Mississippi), although the sample sizes of these States (around 200) yields a ~10% error on a 95% confidence interval.
Your comments and any other ideas worth researching are greatly appreciated. Also, please share this with others (this would be a great help to us: simply getting more eyes on our project)