Ourpaper on electric vehicles, data and distribution network planningis now available to download. Big data and electric vehicles (EVs) are two hot topics in the energy sector in 2019 and this paper explores what one can tell us about the other.
Download the full article here
The increasing presence of electric vehicles on our electrical networks “ both in future projections and those already on the road “ is certainly a hot topic for the energy sector. But it poses a challenge for networks, especially low voltage (LV) networks, as they might struggle to meet the power demands from more electric vehicle customers. It is therefore essential that distribution network operators (DNOs) are able to predict where and when EVs are likely to appear, in what quantities, and understand their aggregated demand.
This type of modelling and forecasting can require a substantial amount of data – more than is currently available. We think that determining the network impacts of EVs, especially on LV networks, is for the moment actually a “small data” problem. That is, there is relatively little data about EV consumption behaviour at a national level, so it is especially sparse at very local geographic areas “ including the level of individual specific LV networks (at most a few hundred customers). The challenge, then, is to draw as much insight as possible from this data using an appropriate model for EV demand.
We believe that a good model should be transparent about the uncertainty in the demand level of a specific set of EVs at a given time. Thankfully, there is an approach to statistical modelling that can do this easily “ Bayesian statistics.
In thepaper, we explore a case study where a network planner is trying to determine the risk that a group of 15 EVs causes an LV network’s headroom of 14 kW to be exceeded. Because they aren’t sure about which types of customers are connected to a specific network, and how they will use their EVs, they are actually faced with a range of possible probabilities of exceeding the headroom. We demonstrate a Bayesian model which can easily account for this. It shows that, for an average group of 15 EVs, there’s around a 13% chance that the EV demand will exceed 14 kW “ e.g. this will happen once every 8 years. But this could easily vary between 4% and 22% depending on which specific customers are connected.
Figure 1: Probability of demand from 15 electric vehicles exceeding 14 kW
Download the full article here
Data science and probabilistic modelling are growing specialisms for TNEI, complementing our existing expertise in network analysis and energy systems innovation. We think that now is the right time for the electricity industry to start thinking more deeply about risk and probability, to make sure the power system is ready for RIIO-2 and the electrification of heat and transport. Pleaseget in touchfor further information or if you have any questions or comments.
