Evidence · one real mission
Predicted.
Then
validated.
This is a single mission operated by the founder in a Vauxhall Vivaro-e 75 kWh panel van, assessed by the One Fleet physics engine before it ran and then recorded afterwards. It is one data point, not a study — but it is a real one, and it is the shape every assessment takes.
The mission
- 3.7 miles, urban and A-road mix, 30–50 mph.
- Vivaro-e 75 kWh (64.8 kWh usable), unladen, Eco drive mode.
- 20°C, dry, climate control off.
- Indicated range at departure: 23 miles.
Predicted versus observed
What the gap tells us
The van finished the mission with the same indicated range it started with. Eco mode, a mild 20°C, an unladen body and speeds below motorway cruise put real consumption under the rate implied by the manufacturer figure. The engine was conservative — which is the correct direction for a purchase or dispatch decision, but it is still an error, and it is exactly the error the learning loop is built to close.
Indicated range is the vehicle's own rolling estimate, not a measurement. On a mission this short it can absorb the whole result. That is itself a finding: short urban missions need energy-meter or telematics data to be validated properly, and the validation record now says so.
Provenance
- Vehicle specification — manufacturer published figures.
- Mission conditions — entered by the operator.
- Observed result — driver-recorded, not independently verified.
- Prediction — One Fleet physics engine, assumptions listed in the assessment.
One recorded mission. No fleet-scale validation, accuracy rate or deployment is claimed.