Key findings

The UK Government has announced its intention to ban the sales of internal combustion cars and vans from 2035. Ofgem’s Decarbonisation Action Plan states that GB electricity network operators should have a network that can power 10 million electric vehicles by 2030. It is widely recognized and acknowledged that stress on current electricity networks can be alleviated with smart technologies, which enable smart demand management using advanced predictive analytics, such as accurate forecasting algorithms, and prescriptive analytics, such as advanced load balancing and optimization algorithms.

To successfully utilize analytical models for charging electric vehicles at scale it is essential for these models to inherently capture vehicle users’ interaction with charging infrastructure, both personal and public. Hence the need for understanding charging behaviors and the factors that influence these behaviors. The aim of this project is to utilize public and home charging data to develop a finer understanding of charging behaviours and influencing factors, and explore algorithmic frameworks that embed these behaviors in realizing large scale smart charging solutions.

Analysis of charging data from 18 public charging stations in Durham revealed that there is a strong connection between usage of charge points and other nearby amenities such as shops, schools, restaurants etc. This shows that placement of charge points has signicant consequences for future urban planning.

A simulation study, embedding advanced forecasts of electric vehicle power demand, on distribution transformers’ impact in a scenario with partial control of home charging shows that a minimum of 60% of users need to be controlled charged to remain within transformer capacity and avoid overloading blackouts.

Clustering analysis of home charging data from Electric Nation project trial 1 shows that, without price interference, vehicles are charged mainly starting from evening to overnight. However, intermittency in charging behavior also makes sizeable proportion of charging data. Similar observations could be made using choice analysis.

Accommodating the behavioral aspects in online scheduling of vehicle charging is necessary for efficient charging at scale. Preliminary experiments with a new approach to embed clustering analysis within online scheduling framework shows more power demand could be satisfied with same amount of resources.

With ongoing COVID pandemic the nature of work and people mobility is bound to change and exhibit a more dynamic yet volatile nature in future. Given this, a sustainable transition to electric vehicles, much needed to tackle climate crisis, requires a substantial amount of research in analytics technologies. For sustainable vehicle charging at scale charging behaviors should be dynamically embedded within optimization models.