Skip to content
Sign inGet Started

How Fast a Legitimate Refuel Actually Is, and Why the Rate Still Cannot Judge It

By Saeed Baatiah · Published:

Short answer

A retail dispensing nozzle is legally capped at 10 gallons per minute, or 37.9 litres. A 4-inch API bulk-loading coupler is rated at 600 gallons per minute, about 2,270 litres. Legitimate refuelling therefore spans roughly sixty to one, and which rate you observe depends on how often you sampled.

Key takeaways

  • 40 CFR 1090.1550 caps a retail dispensing nozzle at 10 gallons per minute, 37.9 litres per minute.
  • A standard 4-inch API bottom-loading coupler is rated at 600 gpm, 135 m³/h, roughly 2,270 litres per minute.
  • The two legitimate rates differ by about sixty times, so "fast" is not by itself evidence of anything.
  • A fill at 2,270 L/min moves 180 litres, NIST's minimum test draft, in under five seconds, which many sampling configurations never observe.
  • Because the observed rate is volume divided by the sampling interval, the interval is what your rate calculation actually measures.

01The retail ceiling is a legal number

Start with the slower of the two legitimate cases, because it is the one with a hard published limit. In the United States, 40 CFR 1090.1550 requires that retail gasoline dispensing nozzles not exceed 10 gallons per minute, which is 37.9 litres per minute. This is not a manufacturer's typical figure that varies by installation; it is a regulatory ceiling on the equipment. It gives a very useful anchor. A truck taking on 400 litres at a retail pump cannot physically complete the transfer in less than about eleven minutes, and in practice it takes longer because a nozzle rarely runs at its rated maximum for the whole draw. So a level rise of 400 litres spread over twelve or fifteen minutes is exactly what a legitimate retail fill looks like in telemetry, and anything much faster than that did not come from a retail nozzle.

02The bulk end of the scale is sixty times faster

Now the other end. A standard 4-inch API bottom-loading coupler, the fitting used to load and discharge road tankers, is rated by OPW at 600 gallons per minute, or 135 cubic metres per hour, approximately 2,270 litres per minute. That is sixty times the retail ceiling, and both figures describe entirely legitimate, entirely routine fuel transfer. The consequence is worth stating plainly: there is no single flow rate that characterises a legal refuel. The same 400 litres that takes eleven minutes at a pump takes about ten seconds through a bulk coupler. Any rule that says a rise faster than some rate is suspicious will therefore either flag every bulk delivery or miss every fast theft, depending on where the line was drawn, and no choice of line avoids both outcomes.

03Metrology draws its own boundary in the same range

Interestingly, the standards bodies place their own dividing line inside this span, which confirms that the two regimes are genuinely different classes of operation rather than two ends of one continuum. NIST Handbook 44 §3.31 separates vehicle-tank meters into accuracy classes by flow rate: Class 0.3 applies above 115 litres per minute, 30 gallons per minute, while Class 0.5 covers 4 litres per minute through 115. OIML approaches dispensers the same way, defining its minimum-measured-quantity requirement for devices with a maximum flowrate up to 3.6 cubic metres per hour, which is 60 litres per minute. So metrology itself recognises a boundary somewhere around 60 to 115 litres per minute, above which a device is doing bulk work and below which it is doing retail work. Rate does distinguish something real; it just distinguishes the type of equipment, not the legitimacy of the transfer.

04Why sampling, not physics, decides what you see

Here is where the spread becomes a practical problem rather than a curiosity. A rate is a volume divided by a time, and in telemetry the time is not the duration of the physical event, it is the gap between two records. If a bulk delivery moves 1,000 litres in thirty seconds but your records are five minutes apart, the rise appears as 1,000 litres over five minutes, an apparent rate of 200 litres per minute. That figure describes neither the coupler nor the theft; it describes your sampling interval. And the direction of the distortion is always the same: a coarse interval understates a fast event and can never overstate it. So a rate threshold set to catch fast transfers is being evaluated against a number that your own configuration flattened before the rule ever saw it.

05Whether a record exists during the event at all

Beneath the interval question is a starker one: whether the event produced any records to divide by. Teltonika takes records through four simultaneous methods, a time period, a distance travelled, a heading angle and a speed delta, and a zero value disables each of them individually. A vehicle standing at a bulk loading rack travels no distance and turns through no angle, so two of the four triggers are silent by definition, and if the time period is long the fill may be bracketed by two records with nothing between them. A transfer at 2,270 litres per minute moves 180 litres, NIST's own minimum test draft, the smallest volume it will run a compliance test on, in under five seconds. Very few production configurations sample often enough to place a record inside a five-second window.

06What the bus can and cannot deliver

It is fair to ask whether reading the level from the vehicle bus rather than the tracker solves this, since a CAN message arrives far more often than a telematics record. It helps with frequency and not with the rest. The FMS standard specifies PGN 65276, which carries the fuel level, with a repetition rate of 1,000 milliseconds, so the value is available once a second on the bus. But the tracker still decides which of those values it stores and transmits, and the acquisition triggers above govern that decision. So a one-second bus rate becomes whatever the record interval is by the time the number reaches a server. The bus removes the excuse that the data was never available; it does not remove the need to configure the recording side to keep it.

07What to build instead of a rate rule

The conclusion is not that rate is useless but that it is a weak discriminator applied to a strong question. What actually separates a legitimate fill from a loss is context that does not depend on the sampling interval: whether the vehicle was at an authorised location, whether a dispensing record exists to match the level rise, whether the volume reconciles with a delivery note, and whether the direction of change matches the operation claimed. Those facts survive a coarse interval, because a location and a document do not get flattened by averaging. Use rate only where you have first established that your record interval is short enough for the rate to mean something, and record the interval alongside the result, so a reviewer can tell whether the number was measured or merely computed.

Published legitimate transfer rates, and how long 400 litres takes at each
EquipmentPublished rateTime to move 400 L
Retail dispensing nozzle (legal ceiling)10 gpm = 37.9 L/minabout 11 minutes at the maximum
OIML retail dispenser class boundaryup to 3.6 m³/h = 60 L/minabout 7 minutes
NIST accuracy class boundary115 L/min (30 gpm)about 3.5 minutes
4-inch API bulk-loading coupler600 gpm = 135 m³/h ≈ 2,270 L/minabout 10 seconds
Fuel level value on the vehicle busavailable once per 1,000 msnot a transfer rate, an availability rate
Whatever your record interval isset by four acquisition triggersthis is what your rate calculation actually divides by

Published fuel transfer rates, from retail nozzle to tanker coupler

Retail nozzle, legal ceiling37.9 L/min
OIML retail dispenser class, upper bound60 L/min
NIST bulk-delivery class boundary115 L/min
4-inch API bottom-loading coupler, rated2,270 L/min

40 CFR 1090.1550, OIML R 117-1, NIST Handbook 44 and the OPW coupler manual, as cited in this article

Figures and sources

Questions

How fast is a normal refuel at a fuel station?

No faster than 10 gallons per minute, or 37.9 litres per minute, because 40 CFR 1090.1550 caps retail dispensing nozzles at that rate. A 400-litre fill therefore cannot take less than about eleven minutes at a retail pump, and usually takes longer since nozzles rarely run at their rated maximum throughout.

Then is a very fast fill automatically suspicious?

No. A 4-inch API bulk-loading coupler is rated at 600 gpm, around 2,270 litres per minute, and that is ordinary routine tanker work. The legitimate rates span roughly sixty to one, so speed alone tells you which class of equipment was probably used, not whether the transfer was authorised.

Why does my dashboard show a rate that matches neither figure?

Because the rate it shows is the volume change divided by your record interval, not by the duration of the physical transfer. A 1,000-litre fill in thirty seconds, sampled five minutes apart, appears as 200 litres per minute. That figure describes your sampling configuration rather than the equipment.

Does reading the level from CAN fix the sampling problem?

Partly. The FMS standard puts the fuel level on the bus once per 1,000 milliseconds, so the data exists frequently. But the tracker still decides which of those values to store and send, governed by its four acquisition triggers, so the record interval remains the limiting factor unless it is configured deliberately.

Sources

Related insights