This study aimed to calculate the percentage of occupied taxi trips that could be completed in a day if BEVs were used instead of CGVs as the means to evaluate the feasibility of such a replacement. In doing so, we coded a program in Python that determined the portion of daily trips that were feasible, considering the battery capacity as well as the battery driving range (battery autonomy) of the electric vehicle and the time required for a taxi to recharge the battery. For the analysis, we assumed that the complete fleet of taxis used BEVs. Then, we simulated taxi trips based on the actual routes, pick-ups, and drop-offs obtained from the GPS data. For that, the code reconstructed the route of each taxi trip per day. In doing so, a total of 352,555 taxi routes were reconstructed for 634 taxis. Thus, for each taxi, we have the actual routes, travel times, and locations/times of the pick-ups and drop-offs. We followed a methodology similar to [
20], which seeks to replicate whether a taxi should take a passenger or not if it has enough battery to complete the trip. If not, the taxi can recharge based on the availability of recharging facilities and the ability to pick up the next passenger. Otherwise, the taxi will skip the trip to the electric station and wait or drive to the next passenger. Details of this procedure follow, starting with the parameters used in this procedure.
4.1. Parameters
One of the parameters required for the assessment is the autonomy of the BEV. Equation (
1) shows the full battery driving range (the full autonomy)
of vehicle
i, which can be computed as the efficiency of the battery
in km/kWh times the capacity of the battery
in KWh.
The data obtained from the Ministry of Energy of the Government of Chile (
www.consumovehicular.cl (accessed on 1 February 2023) show that the autonomy of electric vehicles currently available in Chile varies between 130 and 250 km. For this study, we included four values of full autonomy, i.e., 150, 175, 210, and 245 km. These values were calculated by considering two battery capacities: 30 and 35 kWh, which match the capacities of batteries available in the Chilean market. Similarly, for each of these values, efficiencies of 5 and 7 km/kWh were included in the analysis. The combination of these values generated the four full autonomies (from 30 kWh ∗ 5 km/kWh = 150 km to 35 kWh ∗ 7 km/kWh = 245 km).
Other parameters were related to the charging stations, as the decision of recharging a battery of a BEV depends on the distance to the closest station, the power of the station, and the percentage of the remaining charge in the battery. Regarding the latter, Reference [
19] pointed out that about 75% of taxi drivers would not recharge unless the battery had less than 50% of its charge. Thus, we considered that the driver would not recharge if the battery had 50% or more charge in the battery. To compute the recharging times, we considered two additional parameters: the distance to the closest charging station (
) and the charging power available in the station (
). To determine the closest charging station, we considered the locations of current charging stations in Santiago de Chile, which consisted of a total of 124 stations (see the x in
Figure 3). Finally, for the charging power, we found that the majority of charging stations had a charging power of 22 kW, while several of them were close to 7.4 kW or 50 kW. These values will also be included in the analysis.
4.2. Daily Trip Taxi Feasibility Procedure
The proposed procedure seeks to compute the percentage of trips that can be completed from the data we obtained. In other words, as we reconstruct the routes and the whole-day information for each taxi from the GPS data (routes, as well as locations of pick-ups and drop-offs), we can simply apply this procedure to obtain the number of trips in a day that a taxi could complete if the CGV is replaced by a BEV.
To determine the proportion of feasible taxi trips, another program was coded, considering the values of the parameters discussed in
Section 4.1. The procedure is similar to the procedure proposed by [
20]. That is, it analyzes whether a trip can be completed with a battery charge at the beginning of the day by evaluating the alternatives of recharging and/or picking up a passenger. However, the main difference is that in [
20], the drivers worked in shifts. That is, they needed to recharge before their shifts were completed, while in our case, this restriction did not apply, as the drivers were the sole owners of their vehicles.
The developed procedure registers the number of trips
k for each vehicle
i. The distance traveled from the initial taxi location to the passenger of trip
k is stored at
, the distance of the trip is stored in the variable
, while the distance between the taxi location and the closest electric station is recorded at variable
. The available time associated with
is
, which represents the minutes available between trips
k and
. Similarly,
denotes the driving time to the closest electric station. Depending on the values of these variables, the taxi can have time—or not—to recharge the battery.
Table 1 shows a summary with the description of the variables, and
Figure 4 shows a flowchart with the general steps included.
As shown in the figure, the procedure starts with
for each vehicle, assuming that the battery is charged at full, at the beginning of the day. This is a valid assumption, considering that, different from [
20], the taxis in Chile are owned by the drivers, who start working at the beginning of the day. The next step is to calculate the distance to the next passenger (
). Then, we compare the total distance of trip
k (
) plus the distance to the closest electric station
, with the current driving range of vehicle
. We included
to avoid the situation where the BEV runs out of battery power. If the total distance exceeds the current taxi battery driving range
, the trip is considered infeasible because it will require the driver to interrupt the trip to recharge, and the driving range of the next trip is updated with the current driving range (
). If the trip is feasible, the current driving range is updated by subtracting the distance to the pick-up location and the distance traveled with the passenger (
).
Once the current driving range is updated, we check whether to proceed with the next trip or attempt to charge the car. Regarding the latter, we check if the driving range of the vehicle is greater than 50% of its original charge (
), as recommended by [
19]. In this case, the taxi moves to the next trip. Otherwise, we check if there is enough time to charge (
). If the taxi does not have enough time, then it moves to the next trip. If there is enough time, then the taxi goes to the closest electric station and charges the battery until we are ready for the next passenger or the battery is full. In the case that we do not know the next passenger, we could assume
, forcing the taxi to go to the charging station. In the flow chart, the increment in the driving range is given by
, where
is the driving distance from the last passenger drop-off to the electric station and
is the driving range added by the current charging station. Either way, after the charging process, the taxi moves to the next trip.
In the case that the taxi does not have enough time to be fully charged, we estimate
using Equation (
2), which is based on [
20]. In this equation, we first calculate the actual time in hours that the taxi will be charging its battery (
). This time is multiplied by the actual power of the electric station (
) and the driving range efficiency of the vehicle (
) in km/kWh.
The next section provides more details on the feasibility assessment of using the taxi data we obtained with the procedure described in this section.