GIS for EV Charging and Fibre Site Selection: Grid Overlays and Data Packs

Site selection is a spatial problem with a lot of competing inputs. Where does an EV charging site have power, traffic, and a place to actually put the chargers? Where does a fibre route have premises to pass and a legal path to reach them? Both questions get answered — or should get answered — in a GIS before anyone commits to detailed design.

This article covers how to run that analysis at a practitioner level: how to score candidate sites against multiple criteria, how to overlay grid capacity so you don’t shortlist a site the network can’t feed, what data packs you need, and how the chosen site moves from GIS into CAD for design. It’s written for the engineer doing EV or fibre site assessment, not for the academic literature on location theory.

The problem: too many candidates, too many criteria

A typical EV programme starts with a long list of candidate sites — owned property, leaseable forecourts, retail car parks, depot land. A fibre programme starts with a service area and a set of possible routes and node locations. In both cases you have more candidates than budget, and each candidate scores differently across a set of criteria that pull in different directions.

For EV, the criteria usually include: grid headroom, traffic volume, dwell time (does the use case give drivers time to charge), competing chargers nearby, parking availability, accessibility, and planning constraints. For fibre, it’s premises density, existing infrastructure to reuse (duct, poles), permit friction, and distance from the nearest aggregation point.

You can’t optimise all of these at once by eye. That’s what multi-criteria analysis is for.

Multi-criteria suitability, done practically

The two methods worth knowing are AHP and TOPSIS. Both are standard, both are published, and both are simple enough to run in a spreadsheet against GIS-derived attributes. You do not need specialist software.

AHP (Analytic Hierarchy Process) is a way of deriving weights. Instead of guessing that grid capacity is “twice as important” as traffic, you compare criteria in pairs — grid vs traffic, grid vs planning, traffic vs planning — and score each comparison on a fixed scale. The method turns those pairwise judgements into a consistent set of weights that sum to one. Its value is that it forces the weighting to be explicit and defensible, and it gives you a consistency check that flags when your judgements contradict each other. For a site-selection exercise where a client will ask “why did you rank this site above that one,” an AHP weight table is the answer.

TOPSIS (Technique for Order of Preference by Similarity to Ideal Solution) is a way of ranking. Once you have weights and a normalised score for each site on each criterion, TOPSIS ranks sites by how close each one sits to the best-possible profile and how far from the worst-possible profile. It handles the reality that no single site wins on everything — one has great grid headroom but poor traffic, another the reverse — and produces a single ordered list.

In practice the workflow is: build the criteria layers in GIS, extract a score per candidate per criterion, normalise, apply AHP weights, then rank with TOPSIS (or a simpler weighted sum if the client doesn’t need the extra rigour). The GIS does the spatial heavy lifting — proximity, density, overlay — and the scoring maths sits on top.

A few cautions. Normalise consistently, and be explicit about direction: for some criteria higher is better (grid headroom), for others lower is (distance to substation, planning risk). Keep the weight set small — five to eight criteria is usually enough; twenty criteria produce a false sense of precision and hide the two or three factors that actually decide the site. And treat hard constraints as filters, not scores: a site inside a flood zone or with no viable grid path should be removed before ranking, not left in the list with a low score that a strong traffic number can offset.

The grid-capacity overlay

For EV, grid capacity is the criterion that most often kills a site — and the one that GIS handles worst if you treat it naively. A charging hub is a large, concentrated load. The question isn’t just “is there a supply nearby” but “can the local network feed this load without a reinforcement project that blows the timeline.”

The overlay you want combines a few layers:

Existing network assets — substations, feeders, and their nominal ratings where you have them. In the UK, the DNOs publish network capacity data and heat maps (available-capacity and constraint layers) that you can bring in as a grid overlay. In the US, capacity data is utility-by-utility and far less consistent; you often work from substation locations and known feeder territory rather than published headroom.
Distance to a viable connection point — a candidate 200m from a substation with headroom is a different proposition from one 2km away needing a new run.
Known constraints — areas already flagged as capacity-constrained, or where reinforcement is planned or in progress.

The honest output of this overlay is a headroom estimate that is conditional. Published capacity data is indicative, not a guarantee — the only firm answer comes from a formal connection enquiry to the DNO or utility. Treat the GIS overlay as a way to shortlist and de-risk, flagging sites as “likely has headroom,” “marginal,” or “reinforcement risk,” rather than as a confirmed connection. This mirrors how a proper EV site assessment reports grid feasibility: headroom as conditional on confirmation, not certainty.

Where the load itself is the deciding factor, the overlay pairs naturally with an EV load calculation — the site’s realistic peak demand (after diversity) is what you test the local network against, not the nameplate sum of every charger.

The data pack: what feeds the analysis

Site selection is only as good as its inputs. A workable EV data pack pulls from:

Parking and land — car park locations, capacity, ownership, surface area. This decides whether the chargers physically fit and whether there’s a queueing and manoeuvring envelope.
POI and traffic — points of interest that generate dwell (retail, services, leisure), road classifications, and traffic-volume data where available. Dwell time matters as much as raw footfall: a supermarket gives a driver 40 minutes, a motorway services gives 20, a workplace gives 8 hours.
Demand signals — EV registration density, fleet locations, population and commuting patterns. These proxy for where charging demand exists or will.
Existing EVSE — current public chargers and their power levels. A new site next to an underused rapid hub is a weaker bet than a gap in coverage.
Hazards and constraints — flood zones, ground conditions, conservation and planning overlays, listed-building proximity. Most of these act as filters.
Equity layers — where the programme has a public or grant-funded remit, deprivation and coverage-gap layers matter, both for scoring and for justifying the shortlist to a funder.

For fibre, the base pack overlaps but tilts toward premises and route: premises points (UPRNs in the UK), existing duct and pole records, PIA data, road and rail crossings, and the aggregation-point locations you’re designing back to.

Keep the pack in a spatial datastore, not a folder of shapefiles. Storing the candidate layers, criteria scores, and results in PostGIS — or an enterprise geodatabase — means the analysis is repeatable, the scoring is versioned, and multiple people can work the same programme without emailing files around. Shapefiles will also silently truncate your attribute field names and cap you at 2GB, both of which bite on a large candidate set.

Where EV siting meets fibre base-mapping

This is the overlap that makes the two workflows one competency rather than two. The base map you build for EV site selection — road network, parcels, POI, hazard layers, utility assets — is largely the same base map a fibre OSP designer works from. The GIS discipline is identical: accurate, current, correctly-projected layers as a system of record, with new candidate and design features built on top.

The permit-pack work is where this pays off most directly. A chosen EV site needs planning and connection paperwork; a fibre route needs wayleaves, street-works notices, and permit applications. Both are produced from the same GIS: the site or route geometry, the affected parcels and highways, and the constraint layers that determine which consents apply. If your base-mapping is clean and your candidate geometry is captured properly, the permit pack is a report off the database rather than a fresh manual exercise. If it isn’t, you rebuild the spatial context every time.

The broader framing — GIS as the operational system of record across siting, design, and O&M — is covered in the GIS operational layer pillar. Site selection is the front end of that lifecycle.

GIS-to-CAD interop: taking the site into design

Site selection ends when you have a ranked shortlist and a chosen site. Design begins in CAD. The handoff between them is where projects lose time, and it’s worth getting right.

The chosen site’s spatial context — parcel boundaries, existing utilities, access geometry, the connection route to the grid or the aggregation point — lives in GIS. The design — charger layout and civils, or the OSP construction detail — happens in CAD (Civil 3D, MicroStation, or AutoCAD). Moving cleanly between them means agreeing a few things up front:

Coordinate reference system. Fix the CRS on both sides before anything moves. A datum mismatch (for example a missing NAD83↔WGS84 transformation, or a British National Grid site handed off as WGS84) lands the design in the wrong place, sometimes by metres, sometimes subtly enough that nobody notices until construction.
Layer and attribute conventions. Agree what comes across as background reference and what is live design geometry, and keep naming consistent so the CAD side isn’t guessing.
What is reference vs what is editable. GIS layers usually go into CAD as read-only context; the design is authored in CAD and, if needed, comes back to GIS as an as-designed or as-built record. Don’t let design geometry accumulate in the GIS system of record — that bloats it and complicates later analysis.

Handled properly, the site-selection GIS becomes the starting canvas for design rather than a static PDF someone re-traces. The candidate that scored highest arrives in CAD with its parcel, its access, and its grid route already in place, correctly projected. That’s the point of doing site selection in GIS in the first place: the analysis and the design run on the same spatial truth.

The short version

Score candidates against a small, explicit set of weighted criteria — AHP for the weights, TOPSIS or a weighted sum for the ranking. Keep it defensible, not elaborate.
Overlay grid capacity early and treat the headroom as conditional; the firm answer is a connection enquiry, not a heat map.
Filter on hard constraints before you rank, so a strong site doesn’t score its way past a flood zone.
Store the candidate layers and scores in a spatial database, not loose shapefiles.
Build the EV base map so it doubles as the fibre OSP base map and feeds the permit pack.
Fix the CRS and layer conventions before the GIS-to-CAD handoff, or the design lands in the wrong place.

GIS as the Operational Layer for Utility and Telecom Networks — the pillar this sits under.
PostGIS for Network & Infrastructure Data — storing siting and design layers in a spatial datastore.
EV site assessment — the assessment service this analysis feeds.
EV load calculation — sizing the demand you test the grid overlay against.

Informational content, not engineering advice. Grid-capacity data from DNOs and utilities is indicative; confirm any connection through a formal enquiry.