How to test the everyday value of multifamily EV charging

What to take away

• An energized connector proves almost nothing. Test access, billing, sharing, recovery and support separately.
• Measure what residents are charged against what the meter says. Billing errors are common and quietly corrosive to trust.
• Pull the network cable on purpose and watch what happens. That is the test nobody runs and everybody needs.
• Keep the baseline. Without one you cannot tell degradation from a bad night.

The question a test plan answers is not "does it turn on." It is whether a resident can start a useful session at eleven at night without generating work for the property.

Those are very different bars, and equipment clears the first one on the day it is installed.

Record a baseline while the installer is still there

Take the measurements now, in writing, because in a year you will want to know whether something has drifted or whether it was always like this. File them alongside the serial numbers, circuit IDs and firmware versions logged at commissioning.

• Time from arrival to a session starting, for a first-time user and a regular
• Delivered power at a single active port
• Delivered power per car with every port occupied
• Session start success rate over twenty attempts
• Cellular signal strength at each unit, in the place the unit actually sits
• Energy metered by the charger against the building meter over a known session

NIST maintains a plain explanation of how charging is measured and sold. It covers why the number billed to a driver must be a real, verifiable measurement, not an estimate.

If your charger's reported energy and your building meter disagree widely, find out why before residents pay against it.

Test access the way residents will fail at it

Run the awkward cases, not the happy path:

• a brand new resident with no account, at night, in the rain
• an expired card
• a phone with no signal in a below-grade garage
• a guest who is not authorized and should be refused clearly
• two people trying to start the same port within a few seconds
• a session started and stopped immediately

Note what the unit says on screen at each failure. Vague errors generate support calls and unpaid staff time, and they are the most common reason a working installation feels broken.

Check the billing against the meter

Bill accuracy is where trust is lost fastest. Run a set of sessions of known duration, note the energy reported, compare it against the price rule you configured, and check what the resident is actually charged.

Then test the edges: a session interrupted mid-charge, a car that finishes and stays plugged in, an idle fee if you have one, and a refund. What those fees add up to across a year is a separate exercise. And the scenarios worth running a charging budget through cover it.

Rules on what charging providers must disclose and how they may bill are tightening. Read the federal summary of charger billing requirements against your own configuration.

A pricing display that does not match the receipt is a complaint every single time.

Prove the power sharing does what was sold

This is the test vendors dislike, so schedule it deliberately.

Fill every port at once. Measure what each car receives. Then unplug them one at a time and watch the allocation change. Compare the result to the worked example in the proposal.

If the guaranteed minimum per car comes in below the quoted figure, find out during commissioning week. That is when the vendor can still be held to it.

Do not wait until February, when six residents complain at once.

Break the network on purpose

Disconnect the unit from the internet during an active session. Then start a new session while it is still offline. Then reconnect.

You are looking for three answers. Does an in-progress session keep charging? Can a known user still start one? And when the link returns, do the offline sessions arrive in the billing record or vanish?

Each of those is a real operational risk, and each is cheap to discover now.

Test the support path, not the support promise

Call the support number as a resident, at an inconvenient hour, on a weekend. Time the response. Ask a question that requires actually looking at your site.

Then run a fault: report a genuine problem and follow it to resolution, recording who did what and how long the part took to arrive. Note who actually turns up, because the electrician on site usually answers to someone other than the brand on the unit. A four-hour response commitment means nothing if the part is six weeks out.

Keep measuring after launch

Track uptime, sessions, energy delivered, failed starts, support tickets, mean repair time, electricity cost against fees collected, and parking complaints. Review monthly at first.

Those numbers are the only honest basis for deciding whether to add ports, change prices or tighten rules. They also make the next procurement far shorter, because you will know what your building actually does rather than what a survey suggested it might.

If you are still selecting equipment, comparing the specs on the same basis comes first. And the broader planning order explains where testing sits relative to everything else.

Common questions

When should we run these tests? At commissioning, again after thirty days of real use, then annually. The thirty-day run catches what commissioning cannot, because it includes actual residents.

How many sessions make a fair sample? Twenty start attempts gives a usable success rate for commissioning. After that, use the real session log rather than staged tests.

Residents say it is slow but the specs look right. What now? Measure delivered power with every port busy. Almost always the answer is sharing behavior, not a fault.

Is uptime from the vendor dashboard trustworthy? Treat it as one input. Vendors define uptime in ways that can exclude network faults or scheduled maintenance. Your failed-start count is the number residents feel.