Guides for depot managers

EV charger uptime at an AV fleet depot: how to track downtime per port without a spreadsheet

The charger network already reports every port's status. Turn that feed into an outage log: when an outage starts and ends, which minutes you may exclude, monthly uptime per port, and what one dark stall costs the night.

By XoomPark Editorial Team · November 12, 2026

Use the charger network's own port status feed as the log, not a spreadsheet someone updates by hand. Every port reports a status; an outage opens when a port goes out of order, inoperative or unknown, and closes when it is available or charging again.³ Record each outage with its start, end, minutes, cause and owner, then compute uptime per port each month over the previous twelve months, the way the federal NEVI rule does.¹ That rule already requires real-time status by port in OCPI 2.2.1 terms, so the feed is a standard one.¹ The rule's quarterly reports ask for the same things: error codes for failed sessions, outage and excluded minutes, and the duration of each outage.²

OCPI 2.2.1 port statuses and how this log treats each one. The status definitions are from the spec; the outage column is our reading.
Status	What the spec says it means	Counts as outage in this log?
AVAILABLE	Able to start a new charging session³	No
CHARGING	In use³	No, unless it delivers below the minimum power level (see below)
RESERVED	Reserved for a particular driver, unavailable to others³	No
BLOCKED	Not accessible because of a physical barrier, such as a car³	No; it is a yard problem, log it separately
INOPERATIVE	Not yet active, or temporarily unavailable, but not broken³	Yes, unless it is scheduled maintenance you can show
OUTOFORDER	Out of order; parts may be broken³	Yes
UNKNOWN	No status information, also used when offline³	Yes; an offline port cannot be shown to work
PLANNED, REMOVED	Planned and operating soon, or discontinued and removed³	No; take it out of the port count

When is a depot charging port actually up?

When its hardware and software are both online and available for use, or in use, and it delivers electricity at the minimum power level the rule sets.¹ Online is not enough, and neither is "plugged in and charging".

That last clause matters at a depot. A port that accepts a car and then holds it at a fraction of its rating looks fine on a status screen and still sends the car out short in the morning. Our reading of the rule: minutes a port spends below the power you bought it for belong in the outage log, flagged as derated, even if the network calls the port CHARGING. The power floors in the federal rule are written for publicly funded sites; at a private depot, set your own floor per charger model and write it into the vendor contract.

A private AV fleet depot is not a NEVI site, so none of this is required of you. We use the rule because it is the clearest public definition of uptime there is, and because it is the yardstick a charger vendor's own uptime promise is most likely to be read against.

Which outage minutes can a depot exclude?

Five causes outside the operator's control: electric utility service interruptions, a failure to charge that is the vehicle's fault, scheduled maintenance, vandalism and natural disasters, and only when the operator can show the port would otherwise have worked. Hours outside the station's hours of operation are excluded too.¹

Utility outage: keep the utility's outage notice or ticket number with the record.
Vehicle fault: the ChargeX error codes assign every fault to one of four owners, the EV user, the charging station operator, the charger or the vehicle.⁴ An exclusion for the vehicle needs a code that points at the vehicle, and the same port charging the next car without trouble.
Scheduled maintenance: scheduled means booked before it started. A vendor visit to fix a fault is an outage, not maintenance.
Hours of operation: a depot that charges around the clock has no closed hours to exclude. That is our assumption for most AV fleet depots; if yours closes, write the hours down before you start counting.

Everything else counts, including a network that cannot reach the port. The status for that is UNKNOWN, "also used when offline",³ and an offline port is one nobody can show would otherwise have worked.

What goes in each outage record?

The port, the start and end time, the minutes, the cause, the owner, and whether the minutes are excluded. The federal quarterly report asks for the station and port identifiers, the error codes tied to each unsuccessful session, the outage and excluded minutes for each of the last three months, and the duration in minutes of each outage.²

One row per outage. The fields follow 23 CFR 680.112 and the ChargeX owner categories; the last two columns are ours.
Field	Where it comes from	Why you need it
Station and port ID	The network, the same ID every month²	Uptime is per port, not per charger
Start and end time	The status change that opened and closed it³	Minutes are computed, never typed
Error code	The charger or network log for the failed session²	Tells you who owns the fault⁴
Owner	EV user, charging station operator, charger or vehicle⁴	Decides who fixes it and whether it is excluded
Excluded?	One of the five causes, with the evidence¹	Keeps the uptime number honest
Vendor ticket and time to respond	Your ticket system	Checks the vendor against the contract

The point of building the record from status changes is that nobody has to remember to open or close it. A depot manager's job is to fill in the cause and the evidence for exclusions; the clock runs itself.

How do you turn the outage log into monthly uptime per port?

The federal formula: uptime = ((525,600 − (outage minutes − excluded minutes)) / 525,600) × 100, where 525,600 is the minutes in a year, recalculated monthly for the previous twelve months. Public NEVI ports must average more than 97%.¹

For a monthly view our arithmetic is the same formula on one month. A 30-day month has 43,200 minutes, and 3% of it is 1,296 minutes, about 21.6 hours. So a port that loses one full overnight charging window of eight hours in a month is still well above 97% that month, which is exactly why the annual percentage alone hides the problem at a depot.

Report two things side by side: uptime per port, and the hours lost inside your charging window. Uptime is the number the vendor contract speaks to. Hours lost in the window are what the morning dispatch feels.

What is session success, and why track it next to uptime?

Session success is the share of charge attempts that start delivering power and then finish without an error that stops or suspends charging, with the connector unlocking normally. It is one of four interim measures from the ChargeX Consortium, with charge start success, charge start time and charge end success, all calculated from the charger's own protocol messages.⁵

Uptime catches ports that are down. Session success catches ports that are up and still fail cars. In OCPP 1.6J, a plug-in shows up as a status change to Preparing, and power delivery as a change to Charging,⁵ so the same message stream that feeds the outage log also gives you attempts and successes. A port with good uptime and falling session success is usually the next outage; send it to the technician before it fails overnight.

What should the charger service contract say about downtime?

A response time, a time to repair, and an overall uptime requirement. The Alternative Fuels Data Center says maintenance contracts should include all three, and that station owners should estimate average maintenance costs of up to $400 a year per charger.⁶

Define "up" the federal way, including the power level, so a derated port is not counted as working.¹
List the exclusions and the evidence each needs. Anything not on the list counts against the vendor.
Measure from your log, not the vendor's dashboard. Ask for the raw status changes so both sides count the same minutes.
Start the repair clock at the status change, not when someone files a ticket. That choice is ours; it removes the gap between a port going dark at 1 a.m. and the first person noticing.

What does one dark stall cost an AV fleet depot overnight?

Count it in cars, not percent. If each car needs about two hours on a stall and the charging window is eight hours, one stall serves four cars a night, so one stall down for the night is four cars that charge late, charge somewhere else or start the day short.

The two hours and the eight-hour window are assumptions; use your own session times and window. The habit worth building is reading the outage log every morning in cars lost, because that is the number dispatch cares about, and it moves the conversation with the vendor from "99% uptime" to "four cars short on Tuesday".

When the depot is short of working stalls for a night, outside capacity can cover it. XoomPark arranges AV fleet charging at existing stations in the hours their stalls sit unused, never blocks a station's regular customers and never pushes the station past its monthly peak.

When this log is the wrong tool

Your chargers are not networked. With no status feed there is nothing to build the log from; a paper log at each shift change is the fallback.
You need to fix a fault, not count it. That is the charging technician's triage; see the troubleshooting guide.
Your site receives NEVI funds. Then the rule is a requirement, not a template, and the state's reporting format governs.²
The question is whether a car is ready, not whether a stall is. Readiness also depends on cleaning and inspection.

Questions depot managers ask

What is uptime for an EV charger?

The share of time a port is online, available or in use, and delivering electricity at its minimum power level.¹ The federal NEVI rule calculates it per port, monthly, over the previous twelve months.¹

How much downtime is normal for an AV fleet charger?

For public NEVI ports the line is more than 97% uptime averaged over a year,¹ which our arithmetic puts at about 21.6 hours of unexcluded outage in a 30-day month. A depot should judge it in its charging window, where an hour lost costs more.

Does a charger that is slow count as down?

Under the federal definition a port is up only if it delivers electricity at the minimum power level,¹ so our reading is that minutes below that floor belong in the outage log, flagged as derated.

Can I exclude an outage the car caused?

Yes, if the failure to charge was the vehicle's fault and you can show the port would otherwise have worked.¹ The ChargeX error codes help show it, since each one names the owner.⁴

My vendor reports 99% uptime but cars still leave short. Why?

Usually one of three things: the vendor counts a derated port as up, excludes outages that are not on the list, or reports a network-wide average instead of per port. Count from the status changes yourself and compare port by port.

Sources

Sources are public. XoomPark isn't affiliated with the companies referenced in this post.

Authorship

Written by XoomPark Editorial Team in collaboration with Claude Fable 5.2

Role breakdown: Topic selection and publishing by XoomPark Editorial Team; research, source checks against primary sources and draft prose by Claude Fable 5.2.