Driving Pattern Analysis for Electric Vehicle (EV) Grid Integration Study

Transcription

1 1 Driving Pattern Analysis for Electric Vehicle (EV) Grid Integration Study Qiuwei Wu, Member IEEE, Arne H. Nielsen, Senior Member IEEE, Jacob Østergaard, Senior Member IEEE, Seung Tae Cha, Student Member IEEE, Francesco Marra, Student Member IEEE, Yu Chen and Chresten Træholt Abstract In order to facilitate the integration of electric vehicles (EVs) into the Danish power system, the driving data in Denmark were analyzed to extract the information of driving distances and driving time periods which were used to represent the driving requirements and the EV unavailability. The Danish National Transport Survey data (TU data) were used to implement the driving data analysis. The average, minimum and maximum driving distances were obtained for weekdays, weekends and holidays to illustrate the EV users driving requirements in different days. The EV availability data were obtained from the driving time periods to show how many cars are available for charging and discharging in each time period. The obtained EV availability data are in one hour time periods and one quarter time periods for different study purposes. The EV availability data of one hour time period are to be used for optimal EV charging study in energy power market. The EV availability data of quarter time periods are to be used to investigate the possibility of utilizing EVs for providing regulation power. The statistical analysis software, SAS, was used to carry out the driving data analysis. Index Terms Electric Vehicles (EVs), Driving Distance, EV Availability, EV Charging and Discharging, Energy Power Market, Regulation Power, SAS I. INTRODUCTION In order to cope with the climate change challenge, the governments over the world have made ambitious plans for using renewable which comprises wind power, solar power, biomass power, etc. With the fast growth of renewable energy integration into power systems, the intermittence of renewable energy will put more challenges to the operation and control of power systems with high renewable energy penetration because the electricity production and consumption have to be balanced in real time to ensure the secure operation of the power systems. With more and more renewable energy integrated into the power systems, it has become an attractive option to use electric vehicles (EVs) to balance the uncertainties introduced This work was supported by the project of Electric vehicles in a Distributed and Integrated market using Sustainable energy and Open Networks (EDISON) which is funded by the ForskEl program (ForskEL Project Number 81216). The authors are with Center for Electric Technology, Technical University of Denmark, Elektrovej 325, 28Kgs. Lyngby, DK ( qw@elektro.dtu.dk ). by renewable energy. At the same time, replacing conventional internal combustion engine (ICE) vehicles with EVs can help reduce the green house gas emission from the transport sector and the dependence of transport on petroleum. Denmark is a unique place for the renewable energy utilization and the use of EVs. At this moment, the wind power penetration level in Denmark is around %. The Danish government has set an aggressive target of using wind power which is that 5% of electricity consumption will be supplied by wind power in 25.The average driving distance in Denmark is 42.7 km per day [1]. It is possible to meet the driving requirement for one day with a fully charged kwh battery. Therefore, from the perspectives of balancing the fluctuation introduced by renewable energy and meeting the driving requirements, it is a very attractive option to implement the idea of integrating EVs into the power systems. At present, the high cost of EV batteries is the barrier to the wide use of EVs. But the potential of using EV batteries for balancing the intermittency of renewable energy and providing ancillary service has made replacing conventional ICE vehicles with EVs to be a realistic scenario. A lot of research work has been done regarding integrating EVs into power systems. The possibility of using vehicle to grid (V2G) to improve wind power integration was studied in [2]. The traffic data were used to calculate the vehicle fleet availability. It was concluded that it is possible to have EVs providing instantaneous disturbance and manual reserve to help integrate more wind power. The feasibility study of implementing V2G scenario in Denmark was done in [3]. The system constraints for integrating EVs into power systems were examined and the technical and economical viability of various possible V2G architectures were studied. It was concluded that the V2G technology can assist in realizing the Danish government goal of 5% of the total energy consumption to be met by wind power in 25. In Ref. [4], the potential of using EVs in Denmark was investigated to identify the benefits for power systems with high wind power penetration with intelligent EV charging management. A vehicle to grid demonstration project was implemented in AC Propulsion Inc. to evaluate the feasibility and practicality of EVs providing regulation service [5]. A test vehicle was fitted with a bi-directional grid power interface and wireless internet connectivity to carry out the demonstration. It was shown that it is feasible for EVs to

2 2 provide regulation service from technical and economical point of view. The naturalistic driving schedules obtained from field operational tests of passenger vehicles in southeast Michigan were used to predict energy usage as a function of trip length [6]. The analysis of naturalistic driving schedules can provide the times spent at given locations as well as the likely battery SOC at the time of arrival. Data from a fleet of PHEVs under normal operation were collected and analyzed to assess the impact of usage patterns on vehicle performance [7]. In references of [6] and [7], the driving pattern analysis and driving data collection and analysis are focused on getting accurate energy usage as a function of trip length and assessing the impact of driving pattern on the vehicle performance. The driving requirements and the vehicle availability for charging and discharging have not been studied, to the best knowledge of the authors. Therefore, the driving data in Denmark were analyzed to obtain the driving distances per day and the available time periods for EV charging and discharging. The rest of the paper is arranged as follows: the driving pattern information needed for EV grid integration study is described in Section II, the description of the available driving data is provided in Section III and the information of the used driving data is also presented, the driving distance analysis and the EV availability analysis are presented in Sections IV and V, respectively, in the end, a brief conclusion is drawn. II. DRIVING PATTERN INFORMATION NEEDED FOR EV GRID INTEGRATION STUDY In order to ensure that the EV users driving requirements are met, it is necessary to study the EV users driving pattern to get the driving distance each day and the available time periods for charging the EVs. At this moment, the EVs on the road are very few. It is difficult to get the real EV driving pattern. But it is acceptable to assume that the EV users will more or less have the same driving pattern. Therefore, the analysis of the driving data of conventional cars can be a good estimate of the EV driving pattern. Beside the driving distance information and available time periods for EV charging, the initial state of charge (SOC) of EV batteries, the destination of each trip and energy consumption per km have to be obtained for the EV integration study. In reference [8], the initial SOC is suggested to be 85% in order to avoid the floating charging stage for EV batteries and the possibility of overcharging. Therefore, the initial SOC of EV batteries can be assumed to be 85%. The destination of each trip gives the extra information of the availability of EVs for charging and discharging. Depending on the availability of charging facilities in the destination, the available time periods for EV charging and discharging can be revised on top of the available time periods from the driving time period analysis. The destination information also indicates which node of the distribution system the EV is connected. The energy consumption per km can be used with the driving distances to calculate the SOC change of EV batteries for specific trips. The energy used per km for a home passenger car is between 1 Wh/km and 18 Wh/km. If there is no detailed information for energy consumption, 15 Wh/km can be used to calculate the energy consumption with the driving distances. With the driving pattern information, the availability of EVs for charging and discharging and SOC change of each trip can determined. The driving pattern information needed for EV grid integration study is illustrated in Table 1. Table 1 Driving Pattern Information Needed for EV Grid Integration Study Information Initial SOC of the first trip Energy used per km Starting time of the kth trip Ending time of the kth trip Location Value or Parameter 85% 15 Wh/km FA(t) III. AVAILABLE DRIVING DATA OF CONVENTIONAL VEHICLES In Denmark, there are three data sources for the driving pattern data of conventional vehicles. The three data sources are listed below. Danish National Transport Survey (TU data) GPS-based data that follow the vehicles (AKTA data) Database of Odometer readings (MDCars) The TU data are interview data collected daily for over 15 years. The number of interview is quite large which is more than 1, interviews. The interviews contain a lot of information on driving behavior. The drawback of the TU data is that the information follows the respondent s instead of the car s behavior. In addition, the interviews only obtain the respondent s behavior on a particular day. Moreover, the TU data are only representative for private vehicles. The AKTA data are collected by means of GPS where a total number of 36 vehicles were followed from 14 to 1 days in 1 3. In this way, the information of the driving pattern of the vehicles and the variation during a week and partly during a month was obtained. However, the car number is not big enough to draw a general conclusion from the obtained data and the detailed information of the trips is limited apart from the exact geographical positions. The main disadvantage of this data is that it only collected the data from cars which belong to families with one car, living in great Copenhagen and attached to the labor market. The AKTA data are not representative and can only be used to supplement the TU data. The MDCars is a database that collects data from the reading of the mileage recorded by the odometer at the vehicle

3 3 inspections. The data are nationwide, total and cover all types of vehicles. But the data only comprise the total number of an entire year for commercial vehicles and two years for private cars. Therefore, the MDCars can be only be used to supplement the data on commercial vehicles. For the driving data analysis in this paper, the TU data were used. The TU data were provided by DTU Transport which comprise the survey number, year, month, date, sex of the respondent, vehicle parking starting time, vehicle parking ending time, driving distance before the parking time period, purpose of the trip, parking time period number. The survey data has 134,756 survey results. IV. DRIVING DISTANCE ANALYSIS The driving distance information is very important in order to ensure that the driving requirements of EV users are met. It can also be used to determine the battery size and what the minimum battery SOC has to be for different kinds of customers. The driving distance data from the TU data were analyzed to get the average driving distance and driving distance distribution for different days. The average driving distance data are listed in Table 2. The overall average driving distance is km. For weekdays, the average driving distance of Friday is the biggest, km. The average driving distance of weekends is a little bit less than the ones of weekdays. On holidays, the average driving distance is the lowest which is 18.9 km. Table 2 Average Driving Distance Average Driving Day Type Distance (km) All days Monday Tuesday Wednesday Thursday 3.98 Friday Saturday Sunday 24. Holiday 18.9 The individual and cumulative driving distance distributions are illustrated in Figure 1 and Figure 2, respectively. It is shown that about 75% of the car users drive equal or less than km. Figure 1 Individual Driving Distance Distribution Figure 2 Cumulative Overall Driving Distance Distribution The individual driving distance distributions of weekdays are illustrated in Table 3 and Figure 3. In Figure 4, the cumulative driving distance distributions of weekdays are illustrated. Table 3 Driving Distance Distribution of Weekdays Driving Distance (km) Mon Individual Tue Wed Cumulative Thu Fri Mon Tue Wed Thu Fri Figure 3 Individual Driving Distance Distribution of Weekdays

4 Figure 4 Cumulative Driving Distance Distribution of Weekdays The individual driving distance distributions of weekends and holidays are illustrated in Table 4 and Figure 5. In Figure 6, the cumulative driving distance distributions of weekends and holidays are illustrated. Table 4 Driving Distance Distribution of Weekends and Holidays Driving Distance (km) Mon Tue Wed Thu Fri Sat Sun Holiday Mon Fri Sat Sun Holiday Mon Fri Sat Sun Holiday Figure 6 Cumulative Distance Distribution of Weekends and Holidays The individual and cumulative driving distance distributions of weekdays, weekends and holidays show that the percentage of driving short distance decreases from Monday to Friday and increases from weekdays to weekends and holidays. The percentage of driving short distance is highest on holidays. V. EV AVAILABILITY ANALYSIS In the TU data, the starting and ending time of all parking time periods from one respondent were combined to determine the time periods which are available for EV charging and discharging. The survey number, year, month and date were used to get all the parking data for one respondent. From the parking time periods of one respondent, the availability and unavailability of the EV for this respondent were determined. The time periods between the parking time periods were specified as unavailability time periods. From the raw data, the availability or unavailability of each minute was determined. Afterwards, the 15 min and Hour availability data were calculated based on the minute data. For the EV grid integration study, EV demand will be put into the energy power market to determine the optimal charging scenario to meet the driving requirement with the minimum energy cost. In the energy power market, the time step is one hour. Therefore, the Hour availability data were determined. On top of the optimal charging study, there is a possibility for EVs to participate in the ancillary service market to provide regulation power. For the regulation power market, the time step is one quarter. The quarter availability data were obtained to carry out the ancillary service provision study for EVs. The overall EV hour availability of one day is illustrated in Figure 7. The results in Figure 7 show that the EV availability is quite high if only the driving time periods are considered as the unavailability time periods. The EV availability is 1 percent or very close to 1 percent during the early morning from : am to 6: am. The EV availability is in the range of 94.35% to 97.29% during day time from 7:am to 7: pm. During night from 8: pm to 12: am, the EV availability is in the range of 98.6% to 99.27%. Figure 5 Individual Driving Distance Distribution of Weekends and Holidays

5 Time (Hour) Availability Figure 7 Overall EV Hour Availability Data The EV hour availability of weekdays, weekends and holidays are illustrated in Figure 8 and Figure 9. It is shown that the EV availability from Monday to Thursday is more or less same with small difference. The EV hour availability is a little bit lower on Friday and the low availability in the afternoon is one hour earlier than the ones on Monday to Thursday. The EV hour availability on weekends and holidays is higher. The lowest time periods are around noon on weekends and holidays Time (Hour) Mon Tue Wed Thu Fri Figure 8 Hour Availability Data of Weekday Time (Hour) Mon Fri Sat Sun Holiday The EV quarter availability is illustrated in Figure 1, Figure 11 and Figure 12. The EV quarter availability has the same pattern as the EV hour availability with fluctuation in each hour. The EV quarter availability can be used for the EV ancillary service provision study Time (Quarter) Availability Figure 1 Overall EV Quarter Availability Data Time (Quarter) Mon Tue Wed Thu Figure 11 Quarter Availability Data for Weekday Figure 12 Quarter Availability Data for Weekday, Weekend and Holiday Fri Time (Quarter) Mon Fri Sat Sun Holiday Figure 9 Hour Availability Data of Weekday, Weekend and Holiday

6 6 VI. CONCLUSION The Danish National Transport Survey data were used to implement the driving pattern analysis for EV grid integration study. The driving distance and vehicle availability for charging and discharging are the main focus of the driving pattern analysis. The driving distance analysis results show that the average driving distance in Denmark is km. On Friday, the average driving distance is the biggest and is km. 75% of the cars drive equal or less than km per day. The km driving distance can be used to determine the EV battery size to meet the driving requirements in Denmark. The availability analysis results show that the vehicle availability is very high for charging and discharging when only considering the driving time periods as the unavailability time periods. The lowest percentage of availability is 94.35%. The availability on weekends and holidays is higher than the one on weekdays. The availability on holidays is the highest. The driving pattern analysis results can be improved by using the actual EV driving data. In the future work, the change of available time periods with the number of electric vehicles will be investigated to provide more information for EV aggregator for providing regulation service. REFERENCES [1] Standardv;rider for traffic data til OSPM modellen, Tetraplan A/S, 1. [2] Jean Brassin, Vehicle-to-Grid Improving Wind Power Integration, master thesis, Center for Electric Technology, DTU, 7. [3] D. K. Chandrashekhara, J. Horstmann, J. Østergaard, E. Larsen, C. Kern, T. Wittmann and M. Weinhold, Vehicle to Grid (V2G) in Denmark-Feasibility Study, Center for Electric Technology, DTU, Jun. 8. [4] J. Østergaard, A. Foosnæs, Z. Xu, T. Mondorf, C. Andersen, S. Holthusen, T. Holm, M. Bendtsen and K. Behnke, Electric Vehicles in Power Systems with 5% Wind Power Penetration: the Danish Case and the EDISON programme, European Conference Smart Grids and Mobility, pp. 1-8, Jun. 9. [5] A. N. Brooks, Vehicle-to-Grid Demonstration Project: Grid Ancillary Service with A Battery Electric Vehicle, Dec. 2. [6] B. Adornato, R. Patil, Z. Filipi, Z. Baraket and T. Gordon, Characterizing Naturalistic Driving Patterns for Plug-in Hybrid Electric Vehicle Analysis, IEEE Power Vehicle and Propulsion Conference 9, pp , Sep. 9. [7] S. M. Mohler, S. Ewing, V. Marano, Y. Guezennec, and G. Rizzoni, PHEV Fleet Data Collection and Analysis, IEEE Power Vehicle and Propulsion Conference 9, pp , Sep. 9. [8] Q. Wu, A. H. Nielsen, J. Ostergaard, S. T. Cha, F. Marra and P. B. Andersen, Modeling of Electric Vehicles (EVs) for EV Grid Integration Study, accepted for publication in the proceedings of 2 nd European Conference SmartGrids & E-Mobility.