Commissioning a charging depot: the acceptance test

It is not energisation. Turning the equipment on and plugging in one vehicle proves that one port works with no other load on the system, which is the easiest condition the depot will ever see. It is not a punch walk either — a punch walk finds scratched enclosures and missing labels, both of which matter and neither of which tells you whether the site delivers its design energy overnight. And it is not the electrical inspection. The inspection confirms the installation complies with the adopted code; it does not confirm that twenty vehicles will be full by 5am. Those are different questions and only one of them is your problem at dispatch.

Every port, at rated current, simultaneously, for long enough to reach thermal steady state, with the load-management profile configured exactly as it will run in service. Log the results: current per port, voltage at the far end of the longest run, total site demand at the meter, and every event the load controller generates. It is a slightly awkward test to stage — it usually needs either the fleet on site or a set of load banks — which is precisely why it gets skipped, and precisely why the defects it catches survive into service. Put it in the contract as an acceptance condition with retention attached, so that the site is designed and built by people who know the test is coming.

These are the tests a competent commercial electrical contractor already knows how to do, and the reason to list them in the contract is so the results are handed over rather than merely performed. Insulation resistance on the feeders and branch circuits before energisation. Continuity and impedance of the equipment grounding conductors. Termination torque to the manufacturer's values with a calibrated tool, which recent Code editions require rather than recommend. Verification of overcurrent protection ratings against the design, remembering that EV charging is a continuous load and the branch circuit is sized above the charger's continuous current under Article 625. Ground-fault protection functionality where the adopted edition requires it — and the adopted edition varies by jurisdiction, so confirm which one governs your site. Voltage measured at the far end of the longest run under load, against the Code's informational-note guidance on voltage drop, which is a recommendation rather than a requirement but is exactly what decides whether your furthest ports underperform.

For each port, individually, before the whole-site test: it starts a session with a real vehicle or a suitable test device, it delivers rated current, it stops cleanly, it recovers after a simulated fault and after a power interruption without needing a manual reset, its emergency stop and disconnect operate, and its metering agrees with an independent measurement within the tolerance the specification names. That last one matters more than it looks: if you are ever going to allocate energy cost by vehicle, by department or by client, the port meter is the source of truth and a systematic error in it is expensive and invisible. Record the serial number, firmware version, network identifier and physical position of every unit while you are there. Nobody wants to do that later and everybody needs it later.

If you are paying a per-port subscription, commissioning is when you verify you are getting what you pay for. Every port appears in the management portal with the correct identity and physical location. Session records reconcile against port metering. Remote start, remote stop and remote reset work from the portal, on every unit, not on the one the technician demonstrates. The smart-charging profile downloads and is observed to take effect — configured and effective are different states and the gap between them is a common failure. Scheduled charging fires at the right local time, including the day after a daylight-saving change, which is a genuinely common cause of a yard charging on peak. Alarms and notifications reach a real person at your organisation. Data export works in the format your reporting or your utility program actually requires. And the administrator account is yours, with your credentials, not the installer's.

One night, the whole thing, as designed. The fleet returns on the normal schedule, plugs in as normal, and charges under the real profile with the real tariff windows. In the morning you compare the state of charge of every vehicle against what dispatch needs, and you look at the site demand curve against the tariff's peak window. This is where you learn that the profile leaves three vehicles short, or that the yard sets its monthly peak at 6:15pm because everything starts at once, or that four drivers did not plug in and nobody noticed. None of those is a defect in the equipment and all of them are defects in the depot. Do it before final acceptance, while the people who built it are still on site and still holding retention.

As-built drawings including the actual trench and conduit routes, which is what saves you money on phase two. Marked-up panel schedules. The full test record. Equipment list with serial numbers, firmware versions and physical positions. Load-management configuration exported, not described. Network administrator credentials in your name and the network agreement assigned to you. Warranty documents with start dates that begin at acceptance, not at delivery to the yard. Spare parts delivered if you bought them. Operating instructions for the yard supervisor that a person can follow. Sign when the tests pass and the package is complete, not when the ports light up. Depot Level 2 runs roughly $3,500 to $15,000 per port installed nationally, and on a twenty-port site that is a large enough number that a day of testing before you release retention is the cheapest insurance in the project.