What should be in a fleet EV charging RFP?

Depot charging projects fail in a small number of predictable ways, and each one traces back to a sentence that was never written. The utility work turns out to be somebody else's problem. The site is handed over and nobody ever ran all the ports at once. Adding four more ports next year means cutting the yard again. The network subscription renews at a price nobody quoted. A charger is dead for three weeks and the contract that promised ninety-nine percent uptime turns out not to say what uptime is or what happens when you miss it. A long RFP full of boilerplate does not prevent any of that. Five specific clauses do.

Each row below is a clause you can lift into your document and adapt. The middle column is the dispute it prevents, which is also the argument to use when a bidder asks you to soften it.

Write the sentence in the negative first, because that is the one bidders answer honestly. 'Utility-side work, including the interconnection application, any utility construction charges, easements and transformer supply, is excluded from this scope and will be contracted directly by the owner' is a clear position. So is the opposite. What kills projects is a scope that says the contractor will 'coordinate with the utility', which means nothing, costs nothing and obliges nobody. Then add three sub-items: who prepares and submits the application, what happens to the contract price and the schedule if the utility requires a service upgrade that was not assumed, and what the assumed existing service capacity is. That last one turns an unpriced risk into a stated assumption you can test.

Acceptance should be an event you can fail. Require that every port simultaneously delivers rated current under the live load-management profile for a defined duration, with the results logged and handed over, plus the standard electrical closeouts and a demonstration of every network function you are paying for. Tie retention to it. The reason this belongs in the RFP rather than in the closeout meeting is that it changes how the site gets designed: a contractor who knows the whole depot will be run at once at handover sizes the feeder for that, and one who does not, sometimes does not. There is a fuller version of the test list in our commissioning article, and it is worth attaching as an exhibit rather than summarising.

Every proposal says it is designed for future expansion. Almost none of them says how much. Specify it: the number and diameter of spare conduits from the distribution point to each row of parking positions, pull-strings left in place, the number of spare breaker positions in each panel, and spare capacity in the transformer or panelboard expressed in amps or kVA. The economics are not subtle — building the make-ready once for the full eventual build is documented to cut per-port cost by 40 to 60 percent against retrofitting the same site later, and the civil work is the part you cannot cheaply redo. Two lines of specification is what turns that saving from a slide into a contract obligation.

Networked chargers carry a per-port subscription, and it usually appears in the proposal as a small annual number bundled inside a large capital number. Ask for it separately, and ask three things about it: the price per port per year, the renewal price or the cap on renewal increases after the initial term, and who the contracting party is — many installers sign the network agreement on your behalf, which means the vendor relationship you depend on in year four belongs to a company that finished its work in year one. Add a clause requiring the network agreement to be assigned to you at acceptance. Then do the arithmetic over the life of the asset rather than over the first year, because that is the number that decides whether networking is worth it at all.

An uptime obligation needs four parts or it is worth nothing. The measurement: what counts as a port being up, measured from whose data, over what averaging period, and per port or across the site. The exclusions: utility outages, vehicle faults, damage, and any scheduled maintenance window are the usual ones, and a definition that excludes 'parts availability' has excluded the most common real cause. The response tiers: time to acknowledge, time to a remote attempt, time on site, time to restore. And the remedy: service-credit money, an escalation path, and a termination right if the number is missed repeatedly. Ninety-nine percent uptime with no measurement method and no remedy is a marketing sentence that survived into a contract.

The remainder of the document is not clever, but getting it wrong wastes everybody's time and produces bids you cannot compare. Include the existing service size and the utility name, a recent interval-data export or twelve months of bills, a site plan with parking positions and the location of the electrical room, the vehicle list with battery capacities and the daily miles each one runs, the dwell window between the last return and the first dispatch, the phase plan with port counts and dates, and the constraints you already know about — snow storage, fire lanes, truck turning circles, accessibility routes. Then require every bidder to price the same scope in the same format, with equipment, electrical, civil, engineering, network and commissioning as separate lines. Bids that are not comparable are not bids; they are brochures.

Do not specify a charger brand unless you have a reason that survives being said out loud. Do not require a specific power level per port before you have measured the energy each vehicle actually needs overnight — that is what the pilot is for, and over-specifying power is the most expensive assumption in depot design. Do not ask for references and case studies in place of a technical response; they are the easiest part of a bid to write and the least informative part to read. And do not include a federal tax credit in your budget model. Section 30C terminated for property placed in service after June 30, 2026, so the federal line is zero, and any bid that shows one has an arithmetic error in its business case.