It records that volume left the tank. It cannot record which vessel received it — and that gap is where every fuel loss lives. Detecting theft means binding volume to identity at the moment of transfer, then reconciling both against what the vehicle actually consumed. Everything else is an estimate with a variance column.
Most operations cannot distinguish theft from measurement error, which is why the number in the variance column is argued about rather than acted on.
| Loss | What it looks like | What detects it |
|---|---|---|
| Unauthorised dispensing | Fuel drawn into a vessel that should not have received it | Identity at the nozzle, checked against an authorisation list |
| Ghost transactions | A dispensing event recorded with no delivery, or delivery with no record | Nozzle-side confirmation for the duration of the draw |
| Siphoning | Tank level falling with the engine off, away from any dispenser | Vehicle tank level telemetry and movement state |
| Meter drift | A slow, consistent bias that reads as steady low-level loss | Calibration schedule and delivery reconciliation at the bulk tank |
| Temperature variance | Volume that appears and disappears with ambient temperature | Temperature-compensated volume, not raw litres |
Uncompensated volume and an uncalibrated meter can generate a phantom variance large enough to hide real theft inside it. Removing the measurement error first is what makes the remaining variance worth investigating — and worth putting in front of an operations director.
Each layer answers a question the one before it cannot. The system is only as good as its weakest link, and the weakest link is almost always the nozzle.
Level probe or ultrasonic sensing with temperature compensation, so a delivery can be reconciled against the invoice and a slow leak or a short delivery is visible rather than absorbed.
Driver or vehicle credential checked before the pump enables. This establishes that a valid transaction was started — which is a weaker claim than most operators assume it to be.
An NFC tag on the receiving tank, read by a reader in the nozzle assembly for the duration of the draw. This is what converts intent into delivery: the volume is bound to the vessel that actually received it, continuously, not to whoever swiped a card at the start.
Pulse metering with a calibration interval and temperature compensation, so recorded litres mean the same thing in February and August.
Telematics reading odometer, engine hours and tank level from the vehicle network rather than from a driver-entered figure. Consumption per kilometre is meaningless without an odometer nobody can type.
Deliveries in, dispensing out, consumption implied — with variance flagged as an exception with evidence attached, routed to whoever can act on it. Not a monthly spreadsheet argued over in arrears.
Reading a tag is trivial. Reading it reliably in that location, for years, inside a classified area, and turning the result into a defensible number, is not.
A credential at the pump establishes that someone entitled to draw fuel started a transaction. It says nothing about which vessel the hose went into. In operations where that distinction has financial consequence, the entire control rests on closing that gap at the nozzle.
It is handled, dropped on concrete, dragged, exposed to fuel and solvent, and left in weather. The assembly has to survive that mechanically and chemically while the antenna still couples reliably through whatever the housing is made of.
Fuel dispensing zones are hazardous areas. Electronics near the nozzle fall under intrinsic safety requirements, and the applicable classification and certification route are established at scoping rather than discovered at approval. This is the constraint most often glossed over.
Diesel expands and contracts with temperature by an amount comparable to the losses being hunted. Uncompensated volume produces a variance that appears and disappears seasonally — and comfortably conceals genuine theft inside its own noise.
No amount of instrumentation at the pump detects fuel removed from a parked vehicle overnight. That requires tank level telemetry correlated with ignition and movement state, and a threshold tuned to avoid alarming on slosh, gradient and thermal contraction.
Variance reports that arrive monthly, unattributed, get argued about and filed. The output has to be a specific exception, with the transaction, the identity and the evidence attached, routed to a named role while the event is still recent enough to investigate.
The same telematics that supplies odometer truth for fuel reconciliation carries the rest of fleet management, which is why the two are almost always specified together.
A control introduced before the measurement is trustworthy produces accusations that do not survive scrutiny — and a workforce that stops cooperating.
| Stage | What happens | Indicative duration |
|---|---|---|
| Baseline | Meter calibration, temperature compensation and bulk reconciliation, with no accusations made | 3–6 weeks |
| Identity at the nozzle | Tank tagging and nozzle readers on one site, dispensing bound to vessel | 4–8 weeks |
| Vehicle telemetry | Telematics fitted across the mix of makes and model years actually in the fleet | By fleet size |
| Reconciliation | Deliveries, dispensing and consumption joined; variance characterised before alerting | 4–6 weeks |
| Exception workflow | Thresholds set, exceptions routed to a named role with evidence attached | 2–4 weeks |
| Rollout | Remaining sites and vehicles, provisioning at scale, fleet management and staged updates | By estate size |
Mining and resources, construction and plant hire, haulage and logistics, agriculture, marine bunkering, bus and municipal fleets, and any remote or unattended site where fuel is stored in bulk and drawn without supervision.
By binding volume to identity at the moment of transfer. A flow meter records that fuel left the tank but not which vessel received it, so detection requires reading an NFC tag on the receiving tank from the nozzle for the duration of the draw, then reconciling that against what the vehicle actually consumed.
No. A credential at the dispenser establishes that someone entitled to draw fuel started a transaction. It says nothing about which vessel the hose went into. Closing that gap at the nozzle is what turns intent into evidence of delivery.
Not at the dispenser — it happens nowhere near one. It requires tank level telemetry from the vehicle correlated with ignition and movement state, with thresholds tuned so slosh, gradient and thermal contraction do not generate false alarms.
Diesel expands and contracts with temperature by an amount comparable to the losses being investigated. Uncompensated volume produces a seasonal variance that can conceal genuine theft inside its own noise, which is why compensated volume is used rather than raw litres.
Electronics near the nozzle are in a classified area and fall under intrinsic safety requirements. The applicable classification and certification route are established at scoping rather than discovered at approval — this is the constraint most commonly overlooked.
Yes, and that is the normal case. Faststream engineers the sensing, identification, connectivity and reconciliation layers and integrates with the fuel management, maintenance or ERP systems already in place.
The same telematics supplying odometer truth for fuel reconciliation also carries location, utilisation, driver behaviour, maintenance codes, geofencing and compliance records — which is why fuel and fleet are almost always specified together.
Delivery confirmed, not just authorised.
PRODUCTIdentity on the receiving vessel.
PRODUCTOdometer, tank level and behaviour from the vehicle.
SOLUTIONThe wider fleet solution.
SOLUTIONPlant and equipment beyond the vehicle.
INDUSTRYRemote and unattended sites.
Tell us the specification, the constraint and the deadline. Programmes that cross silicon, radio, embedded and AI are where Faststream is strongest.