How to design a fleet charging pilot that scales

Electric vans charge and electric vans drive. You do not need a pilot to establish that. What you cannot get from a vendor spreadsheet is how many kilowatt-hours your route actually consumes when it is loaded, when it is cold, when the lift gate runs all afternoon and when the driver idles the cab with the heat on. Nor can you get the shape of your own dwell window, which is a dispatch fact rather than an engineering one and is usually shorter than anybody in the room believes. Write the pilot's objectives as measurements with target precision, get sign-off on them, and resist every attempt to turn the pilot into a demonstration. A demonstration ends with a photo. A measurement ends with a load profile you can hand to an engineer.

Everything else in the pilot exists to produce these. Instrument for them before the first vehicle arrives, because retrofitting measurement halfway through costs you the first month of data.

The reason dwell matters more than charger speed falls straight out of division. Nightly energy divided by available hours is the average power each vehicle needs, and that number is usually far below what a charger nameplate suggests. Add a margin for charging losses and for the way charging slows near a full battery, and add margin again for the vehicles that come back late, but the shape holds: a long dwell converts an apparently large power requirement into a small one.

This is the failure worth reading twice. Pilots get put wherever a spare panel and a short conduit run happen to exist — beside the shop, near the electrical room, at the end of the building where the service lands. The pilot then succeeds cheaply and quickly, and the number everybody remembers is the cheap one. The full build has to go where the vehicles actually park, which is the far end of the yard, across a fire lane, past a drainage swale, and the cost per port triples. Worse, the pilot answered none of the questions that matter for that location: trench length, pavement type, water table, whether the route crosses a truck circulation path, whether the distribution point is on the right side of the site. Put the pilot in the rollout's footprint. If that means a longer trench for four ports, spend it, because you are buying the real number instead of a flattering one.

The pilot is also the cheapest opportunity you will ever have to buy the make-ready. Trenching once for the full eventual port count and populating a fraction of it is documented to cut per-port cost 40 to 60 percent against retrofitting the same site later, and the pilot is the moment when the trench is open anyway. Pull the conduit for the full row, leave pull-strings in every spare, pour the pads for the positions you are not yet using, and specify panel headroom above the pilot's load. The counter-argument is real: if the electrification plan slips by three years, that is capital sitting in the ground. Split the difference the way the depot guidance does — oversize the passive infrastructure, which does not become obsolete, and defer the powered equipment, which does.

Long enough to see your worst season, which for most of the United States means the pilot must run through winter or it has not measured anything. Cold weather changes energy per mile and changes how a battery accepts charge, and a summer pilot that reports comfortable margins will be wrong in February. On vehicle count, three or four of a single duty cycle beats one of each of four duty cycles: you need to see the spread within a route type, not a sampler. If you genuinely run distinct duty cycles — a local delivery loop and a regional run, say — treat them as separate pilots with separate conclusions rather than averaging them into a number that describes neither. And run at least one deliberate stress night where every pilot vehicle returns at the same time on the same ports, because that is the condition the full build lives in every single night.

The pilot's electrical load is small and the utility conversation it should trigger is not. Ask for the existing service capacity at the yard, ask what a new service or a larger transformer would cost and how long it would take, and ask which rate schedules apply to a load that runs overnight. Ask whether the utility has a make-ready or fleet program, what it requires and whether pre-approval before work begins is a condition — several programs will not pay for work already done. Do that while the pilot is running and the answers arrive before the rollout design is frozen, which is the only time they are useful. There is no federal credit to factor in: Section 30C terminated for property placed in service after June 30, 2026, so the utility and state layer is the whole incentive picture now.

A one-page load profile: energy per vehicle per night, distribution rather than average, worst month highlighted. A dwell histogram from real gate times. A plug-in compliance rate with the reasons for the misses. A measured fault rate and time to restore, which is what tells you whether the full build needs spare ports. A verified statement of existing service capacity from the utility. A site plan showing where the rollout goes, with the trench route and the distribution point marked. And a per-port installed cost taken from the pilot's actual invoice, adjusted for the parts of the full build the pilot did not have to do. That package is what turns an RFP from a description of an aspiration into a specification. Depot Level 2 nationally runs roughly $3,500 to $15,000 per port installed, and the difference between the top and the bottom of that spread is almost entirely the things a well-sited pilot has just measured for you.