Short answer
A delivery note records what was declared as delivered, not what the receiving tank actually took. Verifying an offload means comparing the declared quantity against the tank level change over the same window, at the same place, and accounting for temperature and settling before treating any remaining gap as a loss.
Key takeaways
- A delivery note is a declaration, not a measurement of what the tank received.
- Part of every gap is legitimate: temperature, settling time and meter tolerance all contribute.
- A short delivery is invisible to any rule that only watches for falling levels.
- Verification needs three facts together, declared quantity, tank level change, and location and time.
- Reconcile over a period as well as per delivery; a small consistent bias only shows up in aggregate.
01What a delivery note actually is
A delivery note is a declaration made at the point of dispatch or by the carrier: this many litres were loaded, or this many were discharged. It is a commercial document, and it is usually accurate, but it is not a measurement of the receiving tank.
The distinction matters because the two are produced by different instruments in different conditions. A loading gantry meter, a tanker compartment gauge and a fixed tank probe each measure a different thing at a different temperature, and agreement between them is a result rather than an assumption.
Treating the note as ground truth is what makes short deliveries survivable. If the document is the record of what arrived, then any discrepancy must be an error in the tank reading, and the discussion ends there. Verification exists to prevent that assumption from being the default.
02Why a gap is usually not theft
Most discrepancies at offload are legitimate, and a verification programme that treats every gap as loss will spend its credibility in the first month. Three mechanisms account for the majority.
The first is temperature. Diesel changes volume with temperature, which is why the petroleum industry corrects observed volumes to a standard reference temperature under ASTM D1250 and API MPMS Chapter 11.1. Fuel loaded cool and discharged into a hot tank, or vice versa, occupies a different volume for the same mass.
The second is settling: fuel entering a tank moves, and a reading taken during that movement does not correspond to a stable volume. The third is instrument tolerance, two devices each within their stated accuracy can differ from one another and both be correct. None of these is a loss, and all of them look like one on a naive comparison.
03The short delivery, and why nothing catches it
A short delivery is the case where the note declares a full load and the tank receives less. It is commercially significant because the site pays for fuel it never got, and it is structurally invisible to most monitoring.
The reason is that theft detection is built to watch for falling levels. A short delivery is not a fall, it is a rise that stops early. Every alert rule looking for an unexplained drop will pass over it in silence, and the level after the event is entirely plausible because it is simply lower than it should have been.
It is also the loss that compounds most quietly. A single short delivery is within the range a busy site will attribute to measurement. A hundred of them across a year, each individually arguable, is a margin problem that nobody connects to fuel receipt because no single instance ever looked like an incident.
04The three facts verification needs
Verifying an offload requires three things held together, and any one of them alone is insufficient.
The declared quantity comes from the delivery document. The tank level change comes from the receiving tank’s own sensor over the delivery window, not a spot reading before and after, but the shape of the rise, so that a delivery which stopped early is visibly different from one that completed. The location and time come from the vehicle, establishing that the discharge happened at this site during this window rather than being inferred.
With all three, the comparison is arithmetic. Without the third, a site cannot distinguish a short delivery from a delivery made somewhere else; without the second, it has only the carrier’s declaration; without the first, it has a level change with nothing to compare it against.
05Reading the shape of the rise
A single before-and-after pair of readings is much weaker than the sequence between them, because the sequence carries information the endpoints discard.
A completed delivery is a sustained rise over several minutes, ending at a level consistent with the invoiced quantity, followed by a settling period and then a flat line. A short delivery has the same beginning and a shorter middle: the rise stops before it reaches the expected level, and there is no anomaly at all in what happens next.
Two other shapes are worth recognising. A rise that pauses and resumes suggests an interrupted discharge, which may be entirely routine. A rise followed within minutes by a fall, before any consumption could plausibly occur, is a different event altogether and deserves its own investigation.
06What the tanker side can contribute
Verification is much stronger when the discharge itself is instrumented rather than inferred only from the receiving tank. An electronic lock on the tanker’s valve turns opening and closing into located, timestamped events.
The Jointech JT701 acts as the connected master, carrying GSM/3G with dual SIM and its own GPS at a quoted 2.5 m CEP. On a multi-compartment tanker the JT709ex field locks handle the remaining valves, carrying no cellular connection or GPS of their own, linking to the master over Bluetooth or LoRa at up to 1 km in open area, and reporting shackle cut and use of the mechanical emergency key.
What that adds to reconciliation is the discharge window itself. Instead of comparing a document against a tank level over an assumed period, the comparison runs over the interval during which the valve was actually open, which also exposes a discharge that occurred somewhere other than the delivery address.
07Measuring the receiving tank properly
The tank side has to be capable of the measurement before the comparison means anything. A capacitive probe such as the Eurosens Dominator RS publishes a maximum error of ±1% over an operating range of −40 to +85 °C, with a 10–60 V supply and IP67 ingress protection, and RS-485 alongside RS-232 and an ISO 9141 K-Line service interface.
The published figure is not the operative one. Because a horizontal cylindrical tank does not lose or gain litres linearly with height, the system needs a calibration mapping from measured height to actual volume, built from readings at known fill levels. Without it, the probe applies real precision to the wrong scale, which is discussed at length in the accuracy article in this series.
For delivery reconciliation the requirement is stricter than for theft detection. Theft detection compares a tank against itself and tolerates a consistent offset; reconciliation is an absolute measurement against an invoice, so a calibration error translates directly into a commercial dispute.
08Setting a tolerance that is defensible
A verification programme needs a threshold below which a discrepancy is accepted without investigation, and choosing that number arbitrarily undermines the whole exercise in both directions.
The defensible method is to derive it from the measurement chain rather than from a policy preference. Combine the instrument tolerances involved, add the expected temperature effect for the site’s conditions, and observe the actual distribution of discrepancies over a baseline period before setting the line. The resulting figure is specific to that site and can be explained to a supplier.
Then treat exceedances as questions rather than accusations. A discrepancy beyond tolerance means the delivery needs explanation, a temperature difference, an interrupted discharge, a meter due for calibration, or a genuine shortfall. Recording which of those it turned out to be is what makes the tolerance improve over time.
09Reconciling over a period, not just per delivery
Per-delivery checking catches large single discrepancies and misses the pattern that costs the most. A consistent one-percent shortfall on every delivery is inside almost any sensible tolerance and is invisible one delivery at a time.
Aggregating over a month exposes it. Total litres invoiced against total litres received, across the site or the whole estate, turns a series of individually acceptable gaps into a single number with a direction. A random distribution of small discrepancies averages towards zero; a systematic one does not, and the difference is the finding.
This is also the figure a finance team will ask for, and being able to produce it is often what keeps the monitoring programme funded. A reconciliation statement that closes to within a stated tolerance is a much stronger argument than a list of alerts that were investigated.
10What a Saudi site should account for
The temperature differential between a loading terminal and a receiving tank standing in afternoon sun is the largest single contributor to apparent discrepancy, and it is a real volume change rather than an error. Correcting to a reference temperature under ASTM D1250 is the standard treatment and it should be part of the reconciliation rather than an afterthought.
Delivery timing interacts with it. The same physical quantity received at 06:00 and at 15:00 produces different apparent volumes in the same tank, so comparing deliveries to each other without correction will show a pattern that is nothing but the time of day.
Connectivity is the third factor. Remote sites lose coverage, and a delivery that occurs during a gap must not lose its level sequence, the device should buffer readings and resend them in order, so the shape of the rise is available afterwards rather than replaced by a flat line across the exact window that mattered.
11Building the verification into the process
Verification is only useful if it happens close enough to the delivery that something can be done. A reconciliation produced weeks later supports a claim; one produced before the driver leaves supports a conversation.
That argues for an automated comparison at the point of delivery: declared quantity entered or received electronically, tank level change measured over the discharge window, and a flag when the two differ beyond tolerance. The person receiving the fuel is the person best placed to ask about it.
Then keep the record regardless of the outcome. A delivery that reconciled cleanly is evidence too, and a supplier relationship in which every delivery is measured and most are confirmed is a healthier position than one in which measurement only appears when there is a complaint.
12Raising a discrepancy with a supplier
A verification programme is only worth running if the findings get used, and using them means having a conversation with a supplier or a carrier about a specific delivery. That conversation goes better when the position is prepared.
Bring three things: the declared quantity from their own document, the measured level change with the shape of the rise rather than just two endpoints, and the calibration record for the receiving tank showing when it was last verified and against what. The third is what pre-empts the obvious counter, that the tank reading is wrong, and its absence is why many discrepancies are never pursued.
Frame the first approach as a question rather than a claim. Most discrepancies have an innocent explanation, and a supplier asked to explain a specific gap will often identify a meter due for calibration or a discharge procedure that changed. A relationship in which every delivery is measured and most are confirmed makes the occasional real shortfall much easier to raise.
13How ENTEK.AI approaches offload verification
ENTEK.AI holds the receiving tank, the delivering vehicle and the valve lock as assets in one system, which is what allows the three facts this article describes, declared quantity, level change and location with time, to be compared without moving data between products.
Because the tank is an asset with its own calibration and telemetry history, the level change is read against that specific tank rather than against a platform-wide assumption, and the lock events around the discharge can be reviewed alongside the tank readings rather than relying on the delivery paperwork alone.
The platform’s contribution here is holding the three records together and making the comparison possible; the measurement itself comes from the sensors and the meters, whose published specifications are attributed to their manufacturers throughout.
| Outcome | Shape of the level rise | Caught by a falling-level rule? | What resolves it |
|---|---|---|---|
| Complete delivery | Sustained rise to the invoiced level, then flat | n/a | Reconciles within tolerance |
| Short delivery | Rise stops early; nothing anomalous after | No, it is not a fall | Declared quantity vs level change |
| Interrupted discharge | Rise pauses, then resumes | No | Valve event sequence from the lock |
| Discharge elsewhere | Little or no rise at the delivery address | No | Location and time of the valve opening |
| Temperature artefact | Level differs with no transfer | Sometimes falsely | Correction under ASTM D1250 |
Figures and sources
- Petroleum volumes are corrected to a standard reference temperature under: ASTM D1250 / API MPMS Chapter 11.1 (Source: ASTM D1250, Temperature and Pressure Volume Correction Factors, as of 2026-08-30)
- Eurosens Dominator RS published maximum measurement error: ±1% (Source: Wialon sensor catalogue, Eurosens Dominator RS, as of 2026-08-30)
- Eurosens Dominator RS operating temperature range: −40 to +85 °C (Source: Wialon sensor catalogue, Eurosens Dominator RS, as of 2026-08-30)
- Eurosens Dominator RS supply voltage range: 10–60 V (Source: Wialon sensor catalogue, Eurosens Dominator RS, as of 2026-08-30)
- Eurosens Dominator RS ingress protection rating: IP67 (Source: Wialon sensor catalogue, Eurosens Dominator RS, as of 2026-08-30)
- Eurosens Dominator RS digital interfaces: RS-232, RS-485, K-Line (Source: Wialon sensor catalogue, Eurosens Dominator RS, as of 2026-08-30)
- Standard governing the Dominator K-Line service interface: ISO 9141 (Source: Wialon sensor catalogue, Eurosens Dominator RS, as of 2026-08-30)
- Jointech JT701 quoted GPS accuracy, which places the discharge: 2.5 m CEP (Source: ENTEK.AI product specification, Jointech JT701, as of 2026-08-30)
- Jointech JT701 cellular connectivity: GSM / 3G, dual SIM (Source: ENTEK.AI product specification, Jointech JT701, as of 2026-08-30)
- Jointech JT709ex link to its master, and range: Bluetooth / LoRa, up to 1 km (open area) (Source: ENTEK.AI product specification, Jointech JT709ex, as of 2026-08-30)
- Jointech JT709ex alert types, relevant to an unauthorised discharge: tamper, mechanical key used, shackle cut (Source: ENTEK.AI product specification, Jointech JT709ex, as of 2026-08-30)
- IP67, second digit: protection against temporary immersion: immersion up to 1 m for 30 minutes (Source: IEC 60529, Degrees of protection provided by enclosures (IP Code), as of 2026-08-30)
- SOJI LIGO Pro published accuracy, as an alternative receiving-tank probe: up to 99.5% (Source: SOJI Electronics, LIGO Pro fuel level sensor, as of 2026-08-30)
Questions
Why does the delivery note rarely match the tank exactly?
Because they are different measurements taken by different instruments at different temperatures. Diesel changes volume with temperature, hence correction to a reference temperature under ASTM D1250, and settling time and instrument tolerance add more. A small gap is expected; a consistent directional gap is the finding.
Why do theft alerts never catch a short delivery?
Because theft rules watch for falling levels and a short delivery is a rise that stops early. Nothing about it looks anomalous afterwards, the level is simply lower than it should have been. Catching it requires comparing the declared quantity against the measured level change, not watching for drops.
What tolerance should we accept without investigating?
Derive it rather than choose it: combine the instrument tolerances, add the expected temperature effect for your conditions, and observe the actual distribution of discrepancies over a baseline period. The result is site-specific and can be explained to a supplier, which an arbitrary percentage cannot.
Is per-delivery checking enough?
No. A consistent small shortfall sits inside any sensible per-delivery tolerance and is invisible one delivery at a time. Aggregate invoiced against received litres over a month: a random spread of small gaps averages towards zero, a systematic one does not, and that direction is the finding.