Stochastic Modeling of Battery Electric Vehicle Driver Behavior

Jing Dong, Zhenhong Lin
2014 Transportation Research Record  
A stochastic modeling approach is proposed to characterize battery electric vehicle (BEV) drivers' behavior. The approach uses longitudinal travel data and thus allows more realistic analysis of the impact of the charging infrastructure on BEV feasibility. BEV feasibility is defined as the probability that the ratio of the distance traveled between charges to the BEV range is kept within a comfort level (i.e., drivers are comfortable with driving the BEV when the battery's state of charge is
more » ... ve a certain level). When the ratio exceeds the comfort level, travel adaptation is needed--use of a substitute vehicle, choice of an alternative transportation mode, or cancellation of a trip. The proposed stochastic models are applied to quantify BEV feasibility at different charging infrastructure deployment levels with the use of GPS-based longitudinal travel data collected in the Seattle, Washington, metropolitan area. In the Seattle case study, the range of comfort level was found to be critical. If BEV drivers were comfortable with using all the nominal range, about 10% of the drivers needed no or little travel adaptation (i.e., they made changes on less than 0.5% of travel days), and almost 50% of the drivers needed travel adaptation on up to 5% of the sampled days. These percentages dropped by half when the drivers were only comfortable with using up to 80% of the range. In addition, offering opportunities for one within-day recharge can significantly increase BEV feasibility, provided that the drivers were willing to make some travel adaptation (e.g., up to 5% of drivers in the analysis). Keywords Battery electric vehicle, Range anxiety, Charging infrastructure, GPS-based longitudinal travel data Disciplines Civil Engineering | Transportation Engineering Comments This is a manuscript of an article published as Dong, Jing and Zhenhong Lin. "Stochastic modeling of battery electric vehicle driver behavior: Impact of charging infrastructure deployment on the feasibility of battery electric vehicles. Abstract 1 This paper proposes a stochastic modeling approach to characterize battery electric 2 vehicle (BEV) drivers' behavior using longitudinal travel data, thus allowing more realistic 3 analysis of the charging infrastructure impact on BEV feasibility. BEV feasibility is defined as 4 the probability that the ratio of distance traveled between charges and the BEV range is kept 5 within a comfort level (i.e., drivers are comfortable with driving the BEV when the battery's 6 state of charge is above a certain level). When the ratio exceeds the comfort level, travel adaption 7 is needed-using a substitute vehicle, choosing an alternative transportation mode, or canceling a 8 trip. To account for day-to-day variations, travel distances-in terms of daily vehicle miles 9 traveled or trip lengths-are represented by gamma distributions. The actual range of a BEV, 10 influenced by traffic conditions and atmospheric and environmental factors, is regarded as a 11 Weibull-distributed random variable. By assuming trip lengths following a gamma distribution 12 and the number of trips between charges following a Poisson distribution, the between-charge 13 travel distances are characterized by a Poisson-gamma distribution. Building on these probability 14 distributions, BEV feasibility can be evaluated for a heterogeneous driving population. 15 The proposed stochastic models are applied to quantify BEV feasibility at different 16 charging infrastructure deployment levels, using GPS-based longitudinal travel data collected in 17 the Seattle metropolitan area. In the Seattle case study, the range of comfort level is found to be 18 critical. If BEV drivers are comfortable with using all the nominal range, about 10% of the 19 drivers need no or very little travel adaption (i.e., make changes on less than 0.5% travel days), 20 and almost 50% of the drivers need travel adaption on up to 5% of the sampled days. These 21 percentages drop by half when the drivers are only comfortable with using up to 80% of the 22 range. It is also found that, offering within-one-day recharge opportunities can significantly 23 increase BEV feasibility, provided that drivers are willing to make some travel adaption (e.g., up 24 to 5% in the analysis). 25 26
doi:10.3141/2454-08 fatcat:w34r5ipcovac3gvakqp4inoce4