Reliability of public charging: what field data says

What field studies and rules report about failed public charging attempts, how they count, which causes they name, and how each cause maps to the failure patterns this analyzer recognises in a capture.

What the studies count

Field reliability figures answer different questions, and none of them is the same as a failed session in a capture.

  • Driver surveys (J.D. Power) count visits to a charger that ended without a charge. That includes a broken charger, but also a queue, a blocked bay or a driver who gave up.
  • Uptime rules count the time a charger or port is available, by the definition of each rule. A charger that is online but fails with one car model may still count as up.
  • Site surveys test connectors in person and record whether a session starts.
  • A capture records one attempt from the cable. It shows why that attempt failed, but only when a car was plugged in and the capture ran.

Read a reported rate with its definition, its region and its date. A rate from one study cannot be compared with another without them.

J.D. Power EVX public charging studies

J.D. Power surveys owners of battery electric and plug-in hybrid cars in the United States each year about public charging. The share it reports is the non-charge visit rate: visits where the owner came to charge and did not.

StudyNon-charge visitsDetails reported
2023 EVX Public Charging Study (published August 2023)20 percent of visitsReasons ranged from an inoperable charger to long queues. The rate varied by region: 35 percent in the Miami area, 29 percent in Seattle, Denver and Dallas, 12 percent in the Cleveland area.
2025 study14 percentAs stated in the 2026 release
2026 EVX Public Charging Study12 percentThe lowest in the six years of the study. Fielded from January to June 2026 among 6,594 owners. J.D. Power attributed the improvement largely to new DC fast charging networks backed by car makers.

The breakdown of the non-charge visits by cause (broken or damaged charger, payment, connection, waiting) and the figures of the 2024 study were not confirmed for this page. Add them from the J.D. Power releases.

Other studies and rules

The ChargeX consortium of the US national laboratories (INL and NREL) published a list of Minimum Required Error Codes and, in December 2024, a Seamless Retry recommended practice with a list of session ending errors. The verdict categories of this analyzer follow that list (Session end verdicts), so a verdict can be counted the same way across sites.

Still to be confirmed from the primary sources before they are stated here: the 2022 study of Rempel and others on the functionality of open DC fast chargers in the San Francisco Bay Area (reported as about 23 percent of connectors not working); the reliability reports of ChargerHelp; the findings of the ChargeX consortium on causes; the UK Public Charge Point Regulations 2023 and their 99 percent reliability requirement for rapid charge points; the 97 percent uptime requirement for chargers funded under the US NEVI program; and the reliability indices of Paren.

Causes and where they show

The table sorts the causes into families and maps each family to the failure patterns this analyzer recognises and the troubleshooting page for each. Shares per cause are left out until they are confirmed from the studies.

Cause familyVisible on the cable?Failure patternsRead
Hardware damage: connector, cable, pilot circuit, isolationPartly: as a pilot fault, a failed cable check or no currentcp-state-ef, cable-check-failed, cable-check-ongoing, no-current-in-cControl Pilot problems, Stage: DC cable check and isolation test, Communication works but no current flows
Communication and software: modem, discovery, TLS, protocolYesslac-no-match, slac-slow, sdp-unanswered, tcp-refused, tls-unknown-ca, tls-cert-expired, no-common-protocol, parameters-refused, precharge-stall, response-timeout, connection-lost, unclean-endCar plugged in, no SLAC match, A weak or disturbed powerline link, SLAC matched, but no SDP answer or TCP connection, The TLS handshake fails, No common protocol in the app handshake, The charger answers with a FAILED response code, Stage: DC pre-charge, The session does not end cleanly
Payment and authorizationPartly: the waiting shows on the cable, the reason in the OCPP logauthorization-waiting, ocpp-authorization-rejected, pnc-contract-expiredThe session stops at authorization
Grid and power: outage, limited supply, shared powerPartly: as a charger stop or low limitsevse-shutdown, and low power limits in the Curve toolThe session stops during charging, Charging is slower than expected
Vehicle sideYesev-stops-early, overcurrent, pwm-not-hlc (when the car never starts SLAC), chademo-incompatibleThe session stops during charging, A CHAdeMO or GB/T session stops
Queue, blocked bay, charger offline before plug inNo: no session startsNoneSite data and the OCPP log

In this tool

  • One capture gives one attempt: the verdict names the category and the likely party, and the finding cards name the failure pattern with its steps (Troubleshooting by symptom).
  • A batch of captures gives field data of your own: how sessions ended, per charger (EVSEID or modem) and per car the share completed and the most common problem, and the checks that fail most often (FR-TOOL-12).
  • The batch trends hold the share of completed sessions next to the figures on this page (the J.D. Power studies, the NEVI uptime rule, the UK reliability rule), each with its source, and say that a survey of visits and an uptime rule measure other things than a batch of captures (FR-TOOL-18).
  • An OCPP log beside the capture adds what the backend saw, which is where payment and authorization failures show their reason (demo ocpp-rejected).
  • What happens before a car is plugged in, and a charger that is offline, needs site data; see Recording a charging session for analysis for what a capture can hold.

Sources