EV Charger Uptime: Online Is Not the Same as Working

A charger's OCPP heartbeat proves its WebSocket connection is alive. It proves nothing about whether the charger will accept a session, deliver the promised power, or record energy correctly.

In real fleets we see chargers that heartbeat happily while their meter reads zero, their RCD trips instantly on session start, or their relay is welded closed. The CMS's dashboard shows green; the driver's experience is red.

Attempted sessions per charger, per week, per site. Successful sessions where the driver got the energy they expected. Ratio is the uptime.

This aligns with what drivers experience, and with the port-level uptime framing NEVI uses under 23 CFR sec 680. Anything portfolio-only is measuring the wrong thing.

Portfolio-level uptime hides the two chargers dragging your network down. Per-site and per-model rollups surface them. Vendor A models failing at 20% more often than vendor B, that's a procurement signal. Site 7 failing more than the rest, that's a network or civil issue at the site.

Site-host SLAs need the same disaggregation. A mall wanting to know the uptime of the chargers on their property needs the report scoped to that property. A franchise partner wanting to know the same. The reporting engine has to slice cleanly by dimension.

• Online uptime: percentage of time the charger is reachable via OCPP
• Charging-port availability: fraction of time the port is ready to accept a session (NEVI's core metric)
• Attempted-session success rate: percentage of driver attempts that deliver energy
• Energy-delivery success: sessions that hit the requested SoC or duration
• Payment-success rate: percentage of paid sessions with clean settlement

• Attempts started vs successful starts
• Delivered kWh vs requested kWh
• Failed payment starts and their reasons
• Total fault duration per charger per site
• Planned maintenance downtime separated from unplanned