Aggregation and Disaggreation of Populations of Electric Vehicle Charging

With increasing penetrations of renewable energy sources onto the grid in the UK more flexibility
in demand is required to secure the reliability and resilience of our power systems. Electric vehicle
(EV) uptake is growing and will make up a sizable portion of electricity demand in the near future.
If done correctly smart charging of populations of EVs stands to be highly beneficial by providing
increasing amounts of flexibility to the grid. Aggregators will play an important role in curating
populations of EVs so that they may collectively provide services to the grid, however they require
quantitative methods to control and characterise these populations.


This project aims at providing aggregators of EVs with methods to control and characterise their
populations such that they may provide the most amount of flexibility to the grid, whilst respecting
customer preferences.


A model of a population of EVs that implicitly respects customer preferences is used and by using
set-theoretic methods in control, optimal control policies will be derived for individual EVs so
that the aggregate demand of the population matches a desired value. This will be done by using
geometric approaches to enumerate the number of power trajectories that a population of EVs
might take. Then by maximising the number of power trajectories whilst maintaining a desired
power demand, an ordering of which EVs to charge can be found. Once an optimal control policy
has been derived this will be used to simulate populations of EVs and verify if a specified power
trajectory can be taken by the population. By doing this systematically the set of available power
trajectories that may be taken by the population can be found and the amount of flexibility of the
population may be quantified. Finally populations of EVs will be modelled stochastically so that
they may better resemble situations presented to aggregators in the real world. Techniques from
probabilistic model checking will be used to quantify the risk of certain events happening, such as
the population's power demand moving out of a preset bound. Analysis of this will be used to define
properties that populations of EVs must have, such as size of population and diversity of charging
schedules. This will help aggregators curate populations of EVs and allow them to procure the
most amount of flexibility from EVs under their control allowing them to provide services to the grid.


This project will be in line with EPSRC's research areas in verification and correctness and control
engineering.