Short answer
A published accuracy figure describes the sensor in laboratory conditions, not the reading you get from a real tank. Mounting position, tank geometry, calibration and temperature dominate the installed result. For theft detection, a stable and repeatable reading matters more than a small headline error figure.
Key takeaways
- A ±1% datasheet figure is a laboratory result; the installed error is dominated by mounting and calibration.
- Tank geometry breaks the height-to-litres relationship, a horizontal cylinder is not linear.
- Repeatability beats absolute accuracy for theft detection: you are comparing a tank against itself.
- Temperature moves the reading without any fuel moving, which is why volumes are corrected to a reference temperature.
- An accurate sensor installed badly performs worse than a modest one installed and calibrated properly.
01What a published accuracy figure actually describes
When a manufacturer states a maximum error, they are describing the sensing element measured against a controlled reference, typically over the full length of the measuring part and across the rated temperature range. The Eurosens Dominator RS, for instance, publishes a maximum error of ±1% across an operating range of −40 to +85 °C.
That figure is real and useful, but it is a statement about the device in isolation. It says how faithfully the element converts a physical level into an electrical value. It says nothing about whether the level it is measuring corresponds to the volume you care about.
The distinction matters because the two can diverge enormously. A perfectly accurate element mounted in the wrong part of the tank, or feeding an uncalibrated volume conversion, produces a confident number that is wrong, and it is wrong consistently, which makes it harder to notice than random noise would be.
02Why tank geometry breaks the height-to-litres assumption
A sensor measures height. An operator needs litres. Converting between them requires knowing the shape of the vessel, and the two are only proportional in a vertical tank with constant cross-section.
A horizontal cylindrical tank, the most common shape on a yard, has a cross-section that is widest at the middle and narrows towards the top and bottom. The same centimetre of height therefore represents very different volumes depending on where in the tank it sits, and a linear conversion overstates the change near the ends while understating it near the middle.
This is why calibration is not optional. The system needs a mapping from measured height to actual volume, built by recording readings at known fill levels. Without it, the sensor delivers its full rated precision against the wrong scale, which is a more expensive failure than a slightly noisy reading.
03What degrades a reading once it is on a real tank
Mounting position is the largest single factor. A probe placed where the fuel surface moves most, near a return line discharge, or at the end of a tank that pitches on a slope, reports the movement as level change. Software smoothing reduces the visible noise but cannot recover information the placement destroyed.
Contamination is the second. Water and sediment accumulate at the bottom of a diesel tank and can sit around the lower part of an immersed element, which is why probes intended for this duty are usually supplied with a filter around the sensing section.
Electrical environment is the third and most often overlooked. Sensors on a generator site frequently share a supply with the machine being monitored, which is exactly the least stable bus available. The Dominator RS carries reverse polarity protection and external voltage protection up to 170 V for this reason.
04Repeatability versus absolute accuracy
For theft detection specifically, absolute accuracy is less important than most buyers assume. The system is not producing a custody-transfer measurement; it is comparing a tank against its own recent history to decide whether a change is normal.
That comparison depends on repeatability, the property that the same physical level always produces the same reading. A sensor that reads two percent low but reads two percent low every time is perfectly usable for detecting an unexplained fifty-litre drop. A sensor that reads correctly on average but wanders by three percent between samples is not.
This reverses the usual purchasing instinct. The right question is not "what is the smallest error figure I can buy," but "which installation will give me the most stable reading over months." Those often point at different decisions, and the second one is the one that determines whether alerts are trustworthy.
05How temperature moves the number
Diesel expands and contracts with temperature, so a tank can report a changing volume with no fuel entering or leaving it. This is not a sensor defect; it is a real physical change in the volume occupied by the same mass of fuel.
The petroleum industry handles it by correcting observed volumes to a standard reference temperature, under ASTM D1250 and API MPMS Chapter 11.1, rather than treating a measured volume as final. Any monitoring system that compares readings across a day without accounting for temperature is comparing two things that were never directly comparable.
On a Saudi site the effect is not marginal. A tank in direct sun during the afternoon and the same tank before dawn sit at materially different temperatures, and the apparent level difference between them can be large enough to resemble a small theft, or large enough to mask one.
06Reading two datasheets side by side
Comparing the Eurosens Dominator RS and the SOJI LIGO Pro shows how little the headline number decides. The Dominator publishes a maximum error of ±1%, a 10–60 V supply, IP67 protection and RS-232, RS-485 and ISO 9141 K-Line interfaces.
The LIGO Pro publishes accuracy up to 99.5%, the same −40 to +85 °C operating range, internal isolation up to 2500 V, an IP69K water-ingress rating, a probe that can be cut or extended up to 6000 mm, and analogue, frequency, RS-232 and RS-485 outputs.
Those are different emphases rather than a ranking. IP69K describes resistance to high-pressure, high-temperature washdown, which matters in some installations and is irrelevant in others; a 6000 mm probe length matters only if your tank is that deep. The specification that decides the purchase is usually the one that matches a constraint you actually have.
07What the interface choice actually changes
An analogue output encodes level as a voltage or a current, which means anything that affects the circuit affects the reading, cable length, connection quality, electrical noise from nearby equipment, or deliberate interference.
A digital bus removes that class of problem. RS-485 is a differential, multi-drop balanced bus under TIA/EIA-485, so several sensors can share a single run to a gateway and each returns an addressed digital value. A value either arrives intact or fails a check; it does not arrive quietly wrong.
For theft detection this is more than an engineering preference. An analogue sender is a component someone can influence without opening the tank, whereas a digital addressed reading either reports or visibly stops. The interface line on the datasheet is often the more consequential specification.
08How to verify accuracy after installation
The only accuracy that matters is the one the installed system delivers, and it is straightforward to test. Add a known quantity of fuel to the tank and compare the reported change against the metered delivery. Repeat at several fill levels, because the error on a horizontal cylinder varies with height.
Then test repeatability separately: leave the tank untouched and observe how much the reported level drifts over a day. Movement without any fuel transfer is the noise floor, and it sets the smallest loss the system can honestly detect. Comparing that floor against your alert threshold tells you whether the threshold is even achievable.
Record both results at commissioning and keep them. When someone later asks whether an alert is credible, the answer depends on numbers measured on that specific tank, not on a figure copied from a manufacturer’s page.
09How accuracy requirements differ by use case
Theft detection needs stability rather than precision, because it is looking for a change that is inconsistent with the asset’s own recent behaviour. A repeatable reading with a known offset serves it perfectly well.
Delivery reconciliation is stricter. Comparing what a tanker invoiced against what the tank received is an absolute measurement, so calibration error translates directly into a commercial dispute. Here the height-to-volume mapping has to be right, not merely consistent.
Consumption analysis sits between the two. It aggregates over long periods, which averages out random noise but accumulates systematic bias, so a sensor with a consistent offset will quietly distort a monthly efficiency figure while looking perfectly stable day to day. Knowing which of the three you are buying for prevents both over-specifying and under-specifying.
10What a Saudi installation should account for
Thermal range is the first specification to check against the site rather than against a catalogue. Both the Dominator RS and the LIGO Pro publish an operating range of −40 to +85 °C, which comfortably covers an exposed yard, but the temperature the sensor tolerates and the temperature correction the volume needs are two separate concerns, and only the first is a hardware question.
Ingress rating should be matched to the actual exposure. IP67 under IEC 60529 means dust-tight with resistance to temporary immersion, which suits an open yard; IP69K adds high-pressure, high-temperature washdown resistance, which is relevant where equipment is regularly pressure-cleaned and irrelevant where it is not.
Dust deserves specific attention because it acts on connectors and seals rather than on the sensing element. A rating protects the enclosure, but a poorly made gland at the cable entry is outside that rating entirely, and it is the most common way a correctly specified sensor fails in this environment.
11What drift looks like over months
A sensor that was accurate at commissioning does not stay that way automatically. Sediment accumulates around an immersed element, seals age, connections corrode, and the calibration mapping slowly stops describing the tank it was built for.
Drift is dangerous precisely because it is gradual. A sudden failure is obvious and gets fixed; a probe that reads one percent low this month and three percent low next quarter keeps producing plausible numbers while the reconciliation figures quietly diverge from reality. Nobody investigates a number that looks reasonable.
The defence is periodic re-verification rather than continuous suspicion. Re-run the known-fill-volume test on a schedule, compare the result against the figures recorded at commissioning, and treat a widening gap as a maintenance trigger. A tank whose reconciliation error grows month on month is telling you something about the sensor, not about the fuel.
12Setting the alert threshold from measured noise
A threshold chosen before the sensor is installed is a guess. The defensible way to set one is to measure the installed noise floor first: leave the tank undisturbed and record how far the reported level wanders over a full day and night, including the thermal swing.
That measured wander is the smallest change the system can distinguish from nothing. An alert threshold set below it will fire on the tank sitting still, and no amount of tuning elsewhere will fix that, because the information required to separate signal from noise was never in the data.
Setting the threshold a defined margin above the noise floor makes the alert rate predictable and the reasoning auditable. It also produces a number worth recording per tank, since a tank with a high noise floor is a candidate for re-mounting rather than for a looser threshold that quietly gives up on detecting small losses.
13How ENTEK.AI treats the accuracy question
ENTEK.AI does not publish a field accuracy figure for a sensor it did not manufacture, and does not present a vendor’s laboratory number as an ENTEK measurement. The specifications quoted throughout this article are attributed to the manufacturers who published them.
What the platform contributes is on the other side of the reading: holding each tank as an asset with its own calibration and its own history, so a level is interpreted against that specific tank rather than against a platform-wide assumption. That is where a stable reading turns into a usable alert.
A monitoring platform that overclaims on measurement is the fastest way to lose the confidence of the operations team it is meant to serve, which makes accuracy an area to be conservative about rather than competitive on.
| Specification | Eurosens Dominator RS | SOJI LIGO Pro |
|---|---|---|
| Published accuracy | Max error ±1% | Up to 99.5% |
| Operating temperature | −40 to +85 °C | −40 to +85 °C |
| Ingress rating | IP67 | IP69K |
| Digital interfaces | RS-232, RS-485, K-Line (ISO 9141) | Analogue, frequency, RS-232, RS-485 |
| Probe length | Modular, cut or extended | Cut or extended up to 6000 mm |
| Electrical protection | Reverse polarity, up to 170 V external | Internal isolation up to 2500 V |
Figures and sources
- 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 external power voltage protection: up to 170 V, with reverse polarity protection (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)
- SOJI LIGO Pro published accuracy: up to 99.5% (Source: SOJI Electronics, LIGO Pro fuel level sensor, as of 2026-08-30)
- SOJI LIGO Pro published operating temperature range: −40 to +85 °C (Source: SOJI Electronics, LIGO Pro fuel level sensor, as of 2026-08-30)
- SOJI LIGO Pro published internal isolation voltage: up to 2500 V (Source: SOJI Electronics, LIGO Pro fuel level sensor, as of 2026-08-30)
- SOJI LIGO Pro maximum probe length when cut or extended: up to 6000 mm (Source: SOJI Electronics, LIGO Pro fuel level sensor, as of 2026-08-30)
- SOJI LIGO Pro published water-ingress standard: IP69K (Source: SOJI Electronics, LIGO Pro fuel level sensor, 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)
- 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)
- RS-485 signalling type, which is what allows multi-drop sensor wiring: differential, multi-drop balanced bus (TIA/EIA-485) (Source: TIA-485-A, Electrical Characteristics of Generators and Receivers for Balanced Digital Multipoint Systems, as of 2026-08-30)
Questions
Is a ±1% sensor accurate enough for theft detection?
Comfortably, provided it is calibrated to the tank and mounted where the reading is stable. Theft detection compares a tank against its own recent behaviour, so repeatability matters more than absolute accuracy. A stable reading with a known offset is more useful than a precise one that wanders.
Why does the same sensor read differently on two tanks?
Because the height-to-volume relationship depends on tank shape. A horizontal cylinder is widest at the middle, so the same centimetre of height means different litres at different levels. Without a calibration mapping built from known fill levels, the sensor applies its precision to the wrong scale.
How do I test accuracy after installation?
Add a metered quantity of fuel and compare it against the reported change, repeating at several fill levels. Then leave the tank untouched for a day and measure how much the reading drifts, that drift is your noise floor and it sets the smallest loss you can honestly detect.
Is a higher IP rating always better?
No, it is better for a specific exposure. Under IEC 60529, IP67 means dust-tight with resistance to temporary immersion, while IP69K adds high-pressure, high-temperature washdown resistance. The latter matters where equipment is pressure-cleaned and is irrelevant where it is not.