Finding EV fast chargers

Written By Matthew Muspratt

A QGIS and Python project exploring the location of Level 3 (aka DC fast) electric vehicle chargers in North America, with a Massachusetts deep dive.

One in a series of projects highlighting my progress as a self-taught programmer.

I began teaching myself Python, SQL, and GIS in spring 2023 — starting from zero. I therefore welcome feedback on these projects and review for errors. And I’d be interested in taking a crack at your own data and geospatial questions too. Please get in touch.

3 July 2023

Background

In mid-2023, Ford and General Motors announced their forthcoming electric vehicles would be compatible with the Tesla network of fast chargers—a move away from the Combined Charging System (CCS) currently employed by most non-Tesla EVs in North America, including my Chevy Bolt. A Canary Media analysis probed the ramifications, likening the evolving charging-connector landscape to the 1980s VCR battle between VHS and Betamax and the emergence of USB as the common computer port technology.

While Canary wondered whether Tesla had won the charging war—and how easy an industry-wide technology transition would be—I had another question: If, overnight, I gained access to Tesla Supercharger stations, would my Bolt driving experience noticeably improve?

The plugs are different

As it stands in 2023, EV owners searching for a fast charger—which adds ~200 miles of range per hour—need to find a station with a connector compatible with their car. Major station networks like ChargePoint and Electrify America typically co-locate CCS and CHAdeMO types, but Tesla runs its own network with a proprietary plug.

Importantly, the Tesla system also enjoys a reputation for reliability and ease-of-use unmatched by CCS/CHAdeMO networks.

Chart source: EPRI, via Canary Media

For drivers with access to 240-volt Level 2 charging at home or work—which fills the tank overnight or quicker—the only range constraint to EV ownership is long-distance travel. Well-placed Level 3 fast chargers can ensure cross-state and cross-country drives free of range anxiety. (See my wife’s article about our daring, successful 1,000-mile family road trip to Quebec.) Existing phone apps allow drivers to filter for and find available charging stations, but with GIS (and a Python assist) we can do it a bit more.

Project outputs and techniques

By my calculations, nearly 28 percent of Massachusetts residents (represented above by ~7 million tiny black dots) live within a 10-minute drive of a DC fast charging station of some kind (light green catchment polygons). I estimate that 22.6 percent of residents are within 10 minutes of a CCS or CHAdeMO station, while 11.3 percent of the population are similarly close to a Tesla Supercharger.

To produce this map and calculate the mentioned figures—and as discussed below vis-à-vis additional project outputs—I wrote a Python program to pull, clean, and filter data from a National Renewable Energy Laboratory website, and then imported the info into QGIS, where I used vector geoprocessing and analysis tools (e.g. polygon unions and dissolves; count-points-in-polygon; a third-party travel-time plugin; centroid creation) in combination with U.S. Census and Department of Transportation data to estimate proximities to the EV stations. (QGIS’s Random Points in Polygons tool generated resident dots at the Census block group level.)

While some residents do care how far they live from a fast charger (renters and multi-unit housing dwellers face obstacles to installing Level 2 chargers at home), long-distance travelers might want to slice the NREL data differently:

(Click maps to enlarge)

Ignoring population densities, non-uniform postal code extents, and the modifiable areal unit problem that plagues choropleth maps (see my co-authored paper with Green River Data Analysis), the above maps ably provide a snapshot of where fast chargers are most often found in North America. In short: highway corridors, California, Quebec, and those locales’ metropolitan area postal codes.

In terms of Python and QGIS, these maps involved using an API key to ingest a large file of station data; defining a histogram function to calculate the count of unique values in any column of any pandas dataframe I specified (e.g. postal code and state/province); joining those count files to the appropriate geographical shapefile in QGIS; and using graduated symbology and bespoke bins to achieve the heat map (choropleth) effect.

The described process worked well enough for U.S. states and ZIP codes, but I encountered a problem with Canadian data: a discrepancy between the province/postal code abbreviations employed by the GIS shapefile (source: Statistics Canada) and the station location data (source: NREL). To resolve, I used QGIS’s Count Points in Polygon tool to tally the station points in each province and postal code.

In the end, what I personally care about most as an EV driver is how far I have to drive to find a fast charger while cruising a major road. ICE (internal combustion engine) drivers enjoy a widespread distribution of gas stations, including easy-access highway rest stops and stations immediately off exit ramps. But do EV drivers experience similar convenience?

To find out—for Massachusetts drivers—I imported into QGIS the location of highway interchanges (i.e. exits and on-ramps) from a Massachusetts Department of Transportation shapefile and applied a third-party plug-in to display a 5-minute driving “radius” from each interchange. Did those catchment polygons span fast charger locations?

Answer: They often did. On the map above, just over 50 percent of the red dots representing Tesla stations fall within light green 5-minute-drive catchment areas; it’s just under 50 percent for CCS and CHAdeMO stations. And that “five minutes” was conservative: I inputed 5 p.m. as a time-of-day parameter, anticipating a worst case, high-traffic range crisis. Importantly, a similar map investigating a 10-minute reach captured many more stations, though who wants to spend 10 minutes chasing a fuel station? Five minutes seemed to me a reasonable search time for EV drivers.

Despite the alignment of fast charging stations with major roads and interchanges, my personal experience—and the map below—expose a gap. For my occasional round trips between western Massachusetts and Boston, the critical central segment of the Mass Pike (aka I-90) between Springfield and I-95 is barren. I have zero CCS stations to rely on besides the Auburn Mall, which itself is on the periphery of the desired 5-minute range.

But, here, we have the answer to my ultimate question: Access to Tesla Supercharger stations would indeed greatly benefit my Bolt driving experience, for both the Palmer Big Y supermarket and the Pike’s Charlton rest stop boast Tesla stations—right where I typically need them.

Data sources

I obtained shapefiles, csv files, and data in other formats from the following sources:

Matthew Muspratt
Previous

Chess. When do I play my best?