Keeping depot chargers working: uptime and maintenance

An annual availability figure averages a fault at two in the afternoon, when nothing is plugged in and nobody notices, against a fault at four in the morning on the last port in the row. Those are not the same event and they should not land in the same average. A depot with 99 percent availability and every outage falling inside the overnight window is worse off than a depot with 97 percent availability and outages that fall during the day.Run the arithmetic on whatever figure your vendor quotes before you accept it as reassuring. A percentage in the high nineties sounds like a solved problem and stops sounding like one once it is expressed in ports and days.The number to put on the wall instead is dispatch readiness: of the vehicles scheduled out this morning, how many left at the state of charge their route needs. That is the measurement the business actually feels, it is computable from telematics you already have, and it catches process failures no charger report will ever show — a port that worked perfectly all night with nothing plugged into it.

The communications path is first because it is the longest chain with the least protection. A cellular modem, a SIM with a data plan somebody has to renew, a router reconfigured during an unrelated IT change, a switch port that came up on the wrong VLAN. None of that stops a charger delivering electricity, but it stops the charger reporting, stops remote authorisation on some units, and on a charger configured to require a session start it can stop charging outright. Ask at design time what each unit does when it loses its backhaul. Fail-open, where the port keeps charging and queues the session records, is what a depot wants. Fail-closed is what a public site wants. They ship configured differently.Connectors and cables are second because they take the physical abuse. The cable gets dropped on concrete, driven over, dragged through a puddle and yanked at an angle by somebody in a hurry. Pins get contaminated, latches break, strain relief pulls out of the handle. Moisture in a connector body is also the most common cause of a ground-fault trip that looks like an electrical fault and is really a housekeeping one.Authorisation is third and it is almost always administrative. A card that was never enrolled, a driver who left and whose card was deactivated while still in the truck, a back-office account that lapsed. It presents to the driver as a broken charger, which is why it eats time.Contactors, power modules and control boards do fail, and when they do the outage is long because the part is not on your shelf. That is an argument for a response-time contract, not for stocking a power module.

The question to settle before you buy hardware, not after, is whether the cable assembly is field-replaceable. On some units a qualified electrician swaps a damaged cable in under an hour with a part from your shelf. On others the cable is part of a factory assembly, and a torn cable means a shipped unit and a two-week hole in the row. That single specification difference changes your annual downtime more than any brand comparison you will read.A reasonable starting shelf for a twenty-port depot: one spare cable assembly per ten ports if they are field-replaceable, a spare communications modem, a couple of SIMs, a dozen blank access cards, spare holsters, and the specific breakers used in the charger distribution panel. Add cable management before you add spares, because most cable damage is a management problem and a retractor prevents the failure instead of shortening it.Keep the spares in a locked cabinet in the yard rather than in a facilities store on another site, and log what was used so the shelf gets restocked. A spare that lives forty minutes away is not a spare.

A five-year parts warranty and a two-week parts lead time is a five-year promise to be down for two weeks at a time. What protects a depot is the response clause: how fast a human with the right part reaches the site, measured from when the fault occurred rather than from when somebody got round to opening a ticket, and what you are owed if that does not happen.The other clause worth arguing over is the boundary of responsibility. The installing contractor owns the wire and the terminations, the charger vendor owns the equipment, the network provider owns the software. When a port will not start a session, all three can plausibly point at the other two, and the argument runs on your time. Write the boundary into the contract and name one party as the single point of contact for a fault report.

The walk takes about two minutes per row and checks six things: every connector is in its holster and not on the ground, every latch operates, no cable shows cut jacket or exposed strands, the screens are lit and legible, no breaker is sitting tripped, and no port is occupied by a vehicle that finished hours ago. Write the result down. A log with dates is what turns an argument with a vendor into a claim, and it is also what an insurer asks for after an incident.The last item on that list is not a maintenance problem, but it produces exactly the same outcome as a dead charger and it happens far more often. It belongs to the yard process rather than to the equipment.

Only the third of those actually predicts the failure, and the charger network cannot generate it. The charger knows a session ended; it does not know whether the van needs 140 miles today or 40. The vehicle telematics knows the state of charge; it does not know whether the port it is plugged into is faulted. Joining the two is the whole value of a fleet charging platform, and it is the question to lead with in a software demonstration rather than the dashboard screenshots.Route the alerts to whoever is physically on site at that hour, which overnight is usually not the fleet manager. An alert that arrives in an inbox nobody opens until seven is a record of the failure, not a chance to prevent it.