Year: 2026
Type: article
Abstract: Direct-current fast-charging connectors are a scarce and expensive asset, and how long each vehicle occupies one determines how many vehicles a site can serve. This paper examines where that occupancy actually goes, using 1,876 sessions recorded over fifteen months at a two-connector fast-charging installation, corroborated against minute-level telemetry from the same site. Three quantitative findings emerge. First, delivery is limited at the connector in every session recorded: the median vehicle requests 191 kW against a median delivered maximum of 98 kW, leaving a median of 86 kW of demand unserved, which excludes vehicle-side limitation as an explanation for underutilization. Second, connector utilization averages 0.619 and falls monotonically with the state of charge at which the driver disconnects, from 0.688 for sessions ending below 60 percent to 0.526 for those ending above 95 percent, a relative reduction of 23.5 percent across six bands with no reversal and a session-level rank correlation of −0.462 (95 percent bootstrap interval −0.500 to −0.426); sessions ending below 80 percent average 0.679 against 0.541 for those ending above 90 percent, a difference of 0.138 (Welch t = 20.41, p < 10⁻⁷⁰, Cohen's d = 1.21). A Jonckheere–Terpstra test confirms the ordered decline (z = −20.92, p < 10⁻⁹⁶). Third, the rate of energy transfer degrades over the same range: throughput falls from 1.20 kWh per occupied minute below 60 per cent to 0.74 kWh above 95 per cent, and the minute-level telemetry gives a mean delivered power of 80.2 kW below the 80 per cent threshold against 37.7 kW above it, a reduction of 53.0 per cent. 55.5% of sessions of sessions continue past 80 per cent, and those final percentage points consume between 176 and 215 of the 1,029 connector-hours in the record, 17.1 to 20.9 per cent of all occupancy, to deliver a median of only 7.1 kWh each; the share is sensitive to where the threshold is placed, rising to 32.1 per cent at 70 per cent and falling to 5.7 per cent at 90 per cent. Minute-level data further establish that this occupancy is not idle time: power flows for 99 per cent of the connected period in every band, so the decline is attributable to the charging curve rather than to vehicles left connected after charging has finished. An incentive linked to the state of charge at disconnection rather than to time of day is outlined at the concept level. It is deliberately not evaluated: the record contains no intervention, so no driver response is estimated and no improvement in utilization is claimed. What the record establishes is the size of the target and the two site conditions under which pursuing it would be warranted. The site studied satisfies neither; its two connectors are occupied for 1,029 of 21,513 available connector-hours, a site occupancy of 4.8 percent, so the measurement is offered as a diagnostic that can be computed from records already held by an operator.
DOI: 10.33411/ijist/2005
Authors: Hamza Mehmood, Atif Sardar
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W7214068594