Planning electric vehicle bulk charging locations with Open Data
We've recently completed a short project for Connected Places Catapult which considered ways of supporting Integrated Planning for Net Zero. The deliverables for the project were two prototypes of tools which promote the idea of cross-sector and integrated planning. The concept behind both prototypes is to bring together local authorities, energy companies, transport planners and any other organisations with a role to play in achieving Net Zero CO2 emissions by 2050. Our project partners, TPX Impact , focussed on a design-led prototype which aimed to promote adoption of housing retrofit technologies within local communities . At Open Innovations, we focussed on electric vehicle (EV) charging - we started with the data that we knew was available (more or less!), and worked from there. If you want to find out more, you can read about the project on the CPC end-of-project blog post . There's also a webinar on 11th May which will provide an opportunity to learn about both prototypes in more depth. Meanwhile, read on to find out about how we built our prototype.
Our tool, the EValuator , aims to help planners consider the optimal siting of bulk EV charging, and then to understand the impact on the local area of these sites. The prototype has two distinct stages:
Stage 1 - Ranking
The Ranking step presents a series of data layers relevant to planning EV locations. We identified these layers by interviewing people with experience of similar planning work, both in local authorities and in the energy / transport sector. The layers are clustered into Transport, (Energy) Infrastructure, and Commercial, but could be any relevant dataset. We worked at an MSOA ( middle super-output area ) granularity as this is a widely understood geography across the potential users of the tools, is neither too big nor too small (there are 7,201 MSOAs in England & Wales), and is well served by existing datasets.
Having selected a Local Authority (or Combined Authority) for focus, the data in these layers is mixed in different proportions to create a composite score for each MSOA. The user of the tool would select their planning model based on their specific needs - so someone looking to promote bulk charging for a supermarket last mile delivery fleet would emphasise different factors (e.g. number of supermarkets) to someone looking to stimulate the adoption of electric vans for courier deliveries (e.g. brownfield, warehouses). As the model is adjusted, the tool calculates the scores, updating a map and list of the MSOAs under consideration.
Stage 2 - Modelling
The planner can then select an area for further modelling. We present them with a map zoomed in to the MSOA they have selected. The map is overlaid with relevant geographical data such as locations of warehouses, supermarkets, car parks and existing charging infrastructure. The planner can select which of these are displayed.
To model the scheme, the planner enters the size and the breakdown of power types. They can also mark the proposed location of the scheme they are modelling. While this does nothing yet, this precise geographical knowledge could be used for detailed modelling of the impact of the scheme, such as connection cost estimates, assuming such data exists. We’ve added some simple models to the tool to demonstrate the concept: an Energy Demand model which calculates the potential power requirement on the grid, and a Carbon Abatement model which calculates the amount of CO 2 emissions saved, assuming that the scheme is a commercial transport option. Both models are very simplistic, and we’re sure that more detailed and refined models can be made.
A word about the data
The data we used in the two stages comes from a variety of sources. In most cases, we’ve processed this to an MSOA-level granularity. Sometimes this is not precise.
Where are EVs now?
The first task was to try to get a picture of existing EV use. The Department for Transport and the DVLA publish licensed ultra-low-emission-vehicles (ULEV) at postcode district level. After exploring the data we noticed that some areas (e.g. Stockport) appear to be huge hotspots of EVs but that is because there are some large lease processing centres that can distort the picture . So this layer should be treated with care and awareness of local "hotspots". The data is also published at postcode district level and these don't easily map to MSOAs. So we created a tool to estimate MSOA-level values from postcode district values by using the proportion of each postcode district's postcodes which are in each MSOA. Our mapping introduces some level of approximation but we don't think this introduces significant extra uncertainty.
An alternative approach to finding out about existing EVs is to look at chargepoints. The National Chargepoint Registry provides around 21,000 existing public chargepoints. However, records can be incomplete, out-of-date, and contain errors as it is up to the individual providers to keep them updated. In exploring the data we already found around 50 chargepoints with incorrect locations and we've provided that feedback to the National Chargepoint Registry. The Department of Transport and Office for Zero Emission Vehicles (OZEV) recently started publishing experimental statistics at postcode district-level of grants for home chargepoints so we were able to estimate MSOA-level values for home chargepoints too. There is no perfect source of data for where EVs actually are but these different datasets should help give an idea.
How easy is it to get around?
To get an idea of how connected an area was we created an Index of Transport Accessibility based on some previous work we had done. This gives an idea of how well connected any MSOA (in the North of England) is by car in 15 minutes at morning rush hour.
Where are chargepoints needed?
During development of the prototype we decided to focus on bulk EV charging. So we obtained polygons for supermarkets, distribution centres, and warehouses across Great Britain from Open Street Map. Open Street Map can be incomplete but it is a resource that is open, has good coverage, and is improving all the time. We extracted polygons tagged with shop=supermarket and building=warehouse for supermarkets and warehouses. Unfortunately distribution sites do not have a specific type in Open Street Map so we extracted all polygons with
in the name whilst dropping any that are
. We could then calculate the land area of each type - supermarket, warehouse, and distribution centre - by MSOA.
Where can new chargepoints go?
To work out where new chargepoints could go we need to know about potential locations and if the electricity grid can support them.
For this prototype we used Northern Powergrid's Distribution Future Energy Scenario 2021 to find predictions of peak utilisation and demand for every primary substation across Yorkshire and the North East in 2030 under the scenario of "Steady Progression". From that we could work out spare capacity at each substation and then find the largest free capacity available in each MSOA. Note that this data was published at the end of 2021 so is only an indicative estimate - any new connections that get added to the network after that will affect the capacity so it will get out of date. In a fully-functioning version of this tool we would want to make sure that this data was kept regularly up-to-date. Also, to cover the whole of England and Wales we would need the other DNOs to publish their grid capacities in a similar form.
One obvious location for chargepoints is in existing car parks so we extracted polygons from Open Street Map with
. Potential for chargepoints depends on car park capacity. Ideally we would know the capacity of every car park but only a fraction have that data tagged on Open Street Map. So created a scatter plot of car park capacity against car park area for every car park tagged with a known capacity (excluding multi-storey) and fitted a linear trend. There is quite a bit of scatter in the distribution but the trend felt reasonable to approximate capacity for those car parks without a capacity set.
It was suggested that brownfield land may make good locations for chargepoints so we processed the Department of Levelling Up, Housing, and Communities' dataset on Brownfield Land Registers to find the total area of brownfield land in each MSOA. This dataset is brought together from individual local authority datasets which can be out-of-date and contain errors in locations/sizes but gives the best picture we have of brownfield land. Also, brownfield land may be allocated for purposes such as housing so not all of it would be suitable or available for EV charging.
The EValuator was created as a prototype for a project sponsored by Connected Places Catapult. Although the project has concluded, we are interested in testing and developing these in a real neighbourhood or place to make sure they respond to local, specific challenges and use cases. Please get in touch if you have any insights or interest in collaborating. We’d be particularly keen to hear from you if you could help us make the models more realistic and useful. Contact us at @OpenInnovates on Twitter, or email to firstname.lastname@example.org .
Meanwhile, we look forward to seeing you on 11th May at the CPC-hosted webinar .