Charging Station Power Failure Monitoring

The blind spot

Monitoring generally works by receiving messages. A unit that has lost power sends nothing, and nothing is also what a working unit sends when it has nothing to say.

• Detection has to be based on expected reporting: a heartbeat at a known interval, whose absence is the signal
• The timeout has to be tight enough to be useful and loose enough not to fire on normal network variation, which on a mobile connection in a basement is considerable
• A site with intermittent connectivity will produce false outages unless the timeout accounts for it
• Every minute of timeout is a minute of outage the operator cannot see, so the setting is a real trade rather than a default

Partial failures

A total outage is obvious and gets attention. A partial one persists, because the site looks broadly fine.

• One phase lost on a three-phase site: single-phase points on that phase are dead, others work, and the site reports as mostly up
• One circuit tripped: a group of bays out while the rest continue
• One unit failed: invisible in a site-level view
• Reduced voltage rather than absent: units may run and misbehave, which is worse than being off

Partial failures need per-unit monitoring rather than per-site, because a site-level view will report ninety percent availability while a third of the bays are unusable.

What has to stay powered

For a site to report its own outage, something at the site must survive it. This is a small design decision with a large operational effect.

• A gateway or controller with a small battery backup, enough to report the event and confirm the network is still up
• It only needs minutes. The message that matters is the one sent as power is lost
• It must be on a different supply path from the chargers, or it will fail with them
• It should report restoration as well as loss, so the outage has a duration rather than an open end

Without this, an operator is inferring from silence. With it, the site says the power has gone and the ambiguity in the previous section disappears.

When to dispatch

1. 1

   Confirm it is real

   Against the heartbeat timeout, and against whether anything else at the site is reporting. A network blip that resolves itself does not need a van.

2. 2

   Establish whose it is

   A utility outage affecting a whole area is not an operator's to fix, and dispatching into one wastes the visit.

3. 3

   Check history

   A site that has done this three times this month is a different problem from one that has done it once, and it needs a different response.

4. 4

   Tell customers

   Before dispatching, because the app and the map should stop directing drivers to a dead site immediately rather than after the repair.

5. 5

   Then dispatch, with a hypothesis

   A visit without one produces the finding that no fault was found.

What the driver should see

The operational response matters less to a customer than whether they drove to a site that was not working.

A site known to be down should disappear from availability immediately, in the app, on the map and at any signage the operator controls. This is the cheapest part of outage handling and the most often missed, because it lives in a different system from the alert. A driver who is not sent to a dead site does not become a complaint.