A week in the life of a car: a nuanced view of possible EV charging regimes

Panel: 6. Transport and mobility

This is a peer-reviewed paper.

Authors:
Giulio Mattioli, TU Dortmund, Germany
Jillian Anable, University of Leeds, United Kingdom
Phil Goodwin, University West of England, United Kingdom

Abstract

In thinking about the charging and associated energy requirements of plug-in vehicles, spatial and temporal forecasts of electricity demand tend to rely on analysis of individual car usage. These are derived from travel diary studies or, increasingly, GPS traces to provide diurnal, weekly and seasonal patterns by different people in different places.

More accurate forecasts of electricity demand require knowledge of the patterns of the individual cars themselves – where they will be, when, for how long, and with what likely level of battery charge. We present a two-stage optimal matching analysis of the 2016 UK National Travel Survey (NTS) to classify cars based on their patterns of use over a week. This required a novel reconfiguration of NTS data into a ‘vehicle travel diary’ dataset, to which sequence and cluster analysis of individual vehicle use sequences were applied.

Firstly, each of the seven days of the travel diary was subdivided into 48 half hour time slots with cars recorded either in use or not in use at any point in each slot. From this, six types of ‘car day’ were identified, with less than half of cars found to exhibit day-types with the stereotypical pattern of ‘morning-out and evening-home’. These six rhythms are exhibited by different groups of cars, and in different proportions on different days of the week.

Secondly, each car was attached with their own set of 7 x daily rhythms using the car-day types and then grouped with cars with similar ‘lifestyle’ across the week. Here we found 8 clusters of car-weeks, each with different rhythms within and across weekdays and weekends. We examine how these car ‘lifestyles’ are associated with household and vehicle characteristics using a broad range of variables available within the NTS. Finally we assess likely typical charging regimes and the flexibility of these patterns and contrast these findings to assumptions commonly being made in assessments of the impacts of electric vehicle grid integration.