Four signals that separate fuel theft from inefficiency
Fuel loss and fuel inefficiency produce the same number at month end and require completely different responses. Telling them apart is a data problem before it is a management problem.
Why one sensor is not enough
A tank level sensor on its own generates false positives constantly. Fuel sloshes on gradients and in cornering. Temperature changes volume. Sensor drift accumulates. A fleet that acts on raw level drops will accuse honest drivers within the first fortnight and lose the room to act when something real happens.
The problem is not sensor quality. It is that a single measurement has nothing to be checked against, so every anomaly looks equally credible.
The four signals worth reconciling
Tank level tells you how much is present. Flow rate tells you how much passed into the engine. Distance tells you how much work was done. Engine hours tell you how long the vehicle was running, including while stationary.
Read together, these four constrain each other. Fuel that leaves the tank while the flow meter is idle and the engine is off has not been consumed — it has been removed. Fuel that is consumed with high engine hours and low distance is idling, not theft. Consumption that rises with distance but not disproportionately is a laden run, not a problem at all.
Baselines have to be per vehicle, per duty
Comparing a loaded truck on a hill route to an empty van on a motorway produces noise. Every asset needs its own consumption baseline, segmented by load state and route type, before an anomaly means anything.
That baseline also has to age. A vehicle at 400,000 kilometres genuinely consumes more than it did at 40,000, and a system that treats the drift as suspicious will cry wolf continuously.
Hold the event, attach the evidence
The operational difference between a system that works and one that gets switched off is what happens at the moment of detection. An alert that says fuel anomaly is an invitation to an argument.
A held event that carries the timestamp, location, tank level curve, flow reading, ignition state and the trip it occurred on is a document. It can be reviewed calmly, discussed with the driver fairly, escalated if warranted, and closed either way with a record. That is what makes the difference between a fuel programme that survives its first month and one that quietly stops being used.
- A single tank sensor generates false accusations; it has nothing to be checked against.
- Reconcile level, flow, distance and engine hours — together they constrain each other.
- Baselines must be per vehicle and per duty type, and must age with the asset.
- Detection should produce a held event with evidence attached, not a bare alert.
More on running a fleet on evidence.
Practical writing on fleet economics, fuel integrity, driver risk and applied AI in transport operations.