EV charging loss waterfall
Every kWh a driver wanted and did not get should have a cause and an owner. This interactive demonstration shows how Inerra Impact accounts for them for a charge point operator, from budget to energy delivered and down to the charger behind each loss, using a fictional six-site network in the UK and UAE over 2025.
At a glance:
Against budget
Metered export against the budget at actual weather, of expected energy. Weather moved the budget by .
How much is explained
of the shortfall is explained by counted losses. The other , , is unexplained.
of unexplained losses, more than the whole shortfall. The counted losses are net better than budget, so the gap is a data question first.
of unexplained losses, even though this view is ahead of budget.
of unexplained losses, inside the model and meter tolerance on complete data, so this view is on budget.
How much is actionable
worse than allowance across in hardware, software, payment, load management and bay access: the losses operations, service contracts and host agreements can act on. of it has no contractual owner.
Every controllable category is at or better than its allowance in this view.
Largest variance and next step
at , Main cause: , carried by , and , carried by .
Every category is at or better than its budget allowance in this view, counting paid curtailment as recovered.
Net budget (P50)
Metered against it
Budget at actual weather
Weather
Metered export
against budget at actual weather
Total losses
Expected energy less metered
Unexplained
Not accounted for by any loss
Data completeness
Weighted by expected energy
Battery round trip
against guarantee;
CHP heat delivered
against heat budget at actual demand
From budget to energy deliveredFrom energy demanded to energy deliveredCHP heat, from budget to heat deliveredWhere the CHP gas wentBattery round tripWhere the revenue wentWhere the imported energy went
Each loss bar shows the result against the allowance in the budget. Up is better than budget, down is worse. Click a bar to see what is inside it.Every kWh drivers wanted between demand and the charger meters. Click a loss bar to see what is inside it and who owns it.CHP heat in MWhth, measured inside the energy centre at the CHP heat meters after the dump radiators, so boiler heat and the network beyond the energy centre are outside it. Heat dumped through the radiators is not a bar here, because the network did not need that heat. It burned gas for no heat sale, so it is in the Fuel tab.Every MWh of gas burned, on gross calorific value, split into what came back as electricity and useful heat and what was lost, beside the engine maker's full-load figures converted to gross. A real assessment corrects them to the actual load, ambient conditions and gas quality, and applies the contract tolerance.Every MWh imported to charge the battery and run the site, split into what was exported and what was lost on the way, beside a reference split built from the round-trip efficiency guarantee and the design loss budget.Every pound of energy delivered, at the tariff that applied to each session (ad hoc, member, roaming or fleet contract) net of VAT, split into what was collected and what leaked, against the target in the business plan.Every MWh through the site meters, split into what reached vehicles and what the chargers, the transformer and the site used or lost on the way, against the design figures in the business plan.
- Totals
- Weather (degree days), not controllable
- Heat lost
- Unexplained
- Totals
- Weather and design, not controllable
- Better than budget
- Worse than budgetLoss
- Unexplained
- Largest shortfall
Low dataData completeness is here, below . Bars that look better than budget may simply be under-recorded, with the missing losses sitting in unexplained, so read them with care.
The value axis starts above zero so the loss bars stay readable; total bars carry a break mark where they are cut. Source: fictional sample portfolio, full year 2025.
Inside
Allowed ·Actual ·Against allowance
Items and owners
Click an item to see the events or measurements behind it.
Unexplained has no items, and the budget allows nothing for it. It is what is left once every counted loss and the energy delivered are taken away from demand, so it grows when chargers stop reporting sessions or the CPMS loses records. Compare it with data completeness in the site table: Ashby Gate, with the weakest data, has the largest share of its demand unexplained: 0.4%, against about 0.1% elsewhere.
By site
Bars show the actual loss and ticks show what the budget allowed. Click a site to drill into it. In % view each site is measured against its own expected energy.
Source: fictional sample portfolio, full year 2025.
Who owns the loss
Each loss item mapped to the party that carries it day to day. A charge point operator depends on more parties than a generator: the host, the CPMS provider, the payment provider and roaming partners as well as the charger maker. Operational responsibility and contractual liability can differ, so a real assessment records both.
- Owned by a party or contract
- No contractual owner
- Unknown (unexplained)
Source: fictional sample portfolio, full year 2025.
When they happened
Each dot is one event, sized by what it lost. Click a dot or a row to open it.
Events, largest first
| Start | Site | Asset | Root cause | Duration | Lost | Status |
|---|
By asset
By month
Click a month to set the period.
How it is measured.
Times are site local time. Source: fictional sample portfolio, full year 2025.
Losses by month
Every loss category, month by month, with the total the budget allowed marked on each bar. Click a month to set the period, and click it again to go back to the whole period. Click a category in the key to open it.
Source: fictional sample portfolio, full year 2025.
Sites
Sorted by result against budget at actual demand, worst first. Click a row to drill into the site. Utilisation is energy delivered against rated power over every hour in the period.
| Site | Chargers | Rated power (MW) | Budget at actual demand (kWh) | Delivered (kWh) | Against budget | Losses | Unexplained | Data completeness | Utilisation |
|---|
Source: fictional sample portfolio, full year 2025.
Method, definitions and sample data
- Period
- The calendar year 2025, January to December, at every site. Times are local site time.
- Net budget
- Energy the business plan expected each site to deliver, after the losses it allows for in each category.
- Demand
- The change in budget because drivers wanted more or less energy than planned: traffic, new vehicles on the road, the host's footfall, fleet schedules, price changes and drivers avoiding a site with a poor record. It is not an operational loss, so it is shown in grey, although the last of those is not entirely outside the operator's control.
- Energy demanded
- What drivers would have drawn if every bay had been working and unrestricted: the sessions that happened, plus demand that went unserved. That covers drivers who found no working bay free and left, failed starts that were not retried, and energy not taken because power was capped or derated and drivers unplugged at their usual time. Demand that moved to another bay on the same site is not a loss. So a charger fault loses only the share of its demand that found no other free bay: the chance that every other bay was busy, from the site's hourly occupancy (an Erlang loss model), plus a small share of drivers who need that bay or do not look further. That share is a few per cent at the big hubs and higher at busy hours, at smaller sites and at the depot overnight. A real assessment builds it from charger status, session logs, failed authorisations and site occupancy, and states how uncertain it is.
- Budget at actual demand
- The net budget re-run on the demand that actually turned up, with every allowance kept at its budget share. This is the fair target for the period.
- Energy delivered
- Energy measured by the charger meters and confirmed against the session records.
- Allowances
- Each category's allowance is a share of demand. For hardware, software and card terminals it is the downtime the plan accepts, at each site's displacement share: at UK sites 0.55% of EVSE hours for hardware, 0.25% for software and 0.1% for card terminals where there is no free vend, which with supply faults stays inside the 99% reliability rule; at UAE sites 1.0%, 0.4% and 0.4%, from the business plan. Failed handshakes and authorisations, bays blocked and derating have their own budgets.
- Reliability rule
- In the UK, the Public Charge Point Regulations 2023 require an operator's network of public rapid charge points (50 kW and above) to be reliable 99% of the time on average over each calendar year. The first annual report, for calendar year 2025, was due by 14 January 2026. It is measured per EVSE, the part of a charger that charges one vehicle at a time, from the OCPI EVSE status. Faults in the operator's own hardware, software and communications equipment (router, SIM, back office) count as downtime, and so does a failed contactless terminal unless the charger falls back to free vend. OCPI also reports UNKNOWN when a charger is offline, but the guidance treats UNKNOWN as exempt time, reserved for evidenced exceptional events outside the operator's control, and says it must not be used for hardware or software faults. So the tile counts offline time as downtime unless such an event is evidenced. Evidenced exceptional events (grid or communication network failure, vandalism, severe flooding) and evidenced blockages (a parked vehicle, roadworks, planned maintenance not linked to a fault) can be exempt; OPSS decides whether the evidence is a reasonable excuse. Sites open only at set hours are measured over those hours. The tile applies the GOV.UK formula to the sample's 36 UK EVSEs, all 50 kW or more, over calendar year 2025. It treats a grid outage or vandalism as exempt only where the event carries a reference, a DNO outage without one as downtime, and a host's car park closure as an evidenced blockage where the host's notice is held. A Kempower Power Unit trip takes its four Satellites out at once; a single failed power module does not, because the unit keeps running.
- Visits lost to failed starts
- Failed starts the driver did not retry (card terminal, app, RFID and roaming authorisations, handshakes, firmware faults and back office outages), at each site's average energy per session, against all visits. Time a charger was offline is not counted, because it is not a start attempt. A true attempt-level success rate, retries included, needs the OCPP transaction log.
- Free vend
- Wyvern Park and Calder Moor let drivers charge for free when the contactless terminal is down. That keeps the charger available under the rule, but the energy is not billed, so it shows up in the revenue view instead. The other sites have no free vend.
- Prices
- 2025 prices, net of VAT. UK ad hoc prices are 79p a kWh including VAT at Wyvern Park, 82p at Calder Moor and 65p at Ashby Gate, against the AA's 2025 averages of 78p to 79p for ultra-rapid and 74p for rapid charging. At the UK sites members get 20% off and roaming partners pay a B2B tariff 12% below ad hoc. UAE public sites charge the unified fee for DC charging under Cabinet Resolution No. 81 of 2024: AED 1.20 a kWh plus VAT. It is a minimum price to drivers, so members pay it too. Roaming partners settle with the operator at a wholesale rate 10% below it, and their drivers still pay at least the minimum. The depot pays a fleet contract rate of AED 0.75. Energy costs are the UK non-domestic average for Q2 2025, 24.3p a kWh including the Climate Change Levy, at Wyvern Park and Calder Moor; the host's recharge at Ashby Gate; DEWA's commercial tariff plus fuel surcharge (44 fils) in Dubai; and the 30 fils tariff for EV charging equipment in Abu Dhabi.
- Revenue view
- The value of the energy delivered at the tariff that applied to each session, split into what was collected and what leaked: sessions never billed, roaming charge records the partner rejected, tariff errors, and failed payments or chargebacks. The target is the leakage the business plan allows.
- Site energy view
- Energy through the site meters split into energy delivered, conversion losses in the chargers, standby, heating and cooling, transformer losses at the UK hubs metered on the high voltage side, and unmetered or other use. At the UAE sites the utility owns the substation and meters on the low voltage side, so its transformer losses are not on the operator's meter. Standby follows the season: cabinet heaters in UK winters and cabinet fans in UAE summers. It weighs most where chargers sit idle.
- Margin view
- Each kWh valued at the site's average price across ad hoc, member, roaming and fleet sessions, net of VAT, less the energy cost. It is not lost profit: it ignores roaming and payment fees, the host's revenue share or lease, and the standing and capacity charges on the supply.
- Allocation
- In every interval, lost energy goes to the first cause that applies, in the order the bars run: grid and site supply, charger hardware, software and connectivity, payment and access, load management and set points, design and environment, then bay access. Whatever remains is unexplained. Events never overlap on one charger, and site-wide events (supply, back office, roaming hub and token list failures, closures) block every other event at the site while they last.
- Events
- Each discrete loss is listed one event at a time, with its charger, first alarm, attendance, restoration, root cause, owner, contract position and evidence. The events add up exactly to the item above them. Waiting on spares starts when the visit that diagnosed the fault ends, and a power limit left in place starts when the fault that set it is cleared. Shared systems fail at every site on the same back end at the same moment. An event that runs past a month end is carried into the next month with that month's share of its energy. Event energy is shown in kWh.
- Response target
- The time from the first alarm to someone on site: 24 hours for the maintenance contractor, including payment terminal swaps, and 48 hours for the charger maker, in this sample.
- Data completeness
- The share of charger intervals with status and meter data, weighted by demand.
Sample data. The six sites, their names, contracts, allowances, events and every number on this page are fictional, generated to show how the waterfall behaves. Charger and terminal models are named to make the sample realistic; the faults and figures attached to them are invented. Prices are 2025 ones, net of VAT, and UAE figures are converted to pounds at AED 4.84 to the pound, the 2025 average.