Short answer
Diesel expands and contracts with temperature. Its volume correction factor is 1.0000 at 15 °C and 0.9873 at 30 °C, roughly 0.00085 per degree, so a 400-litre tank appears to change by about 6.8 litres across a 20-degree day. Size the day tank for runtime, and set thresholds above that thermal band.
Key takeaways
- Diesel VCF is 1.0000 at 15 °C and 0.9873 at 30 °C, about 0.00085 per degree, per the published correction tables.
- On a 400-litre tank a 20 °C day-night swing therefore shows roughly 6.8 litres of apparent change with nothing removed.
- A day-tank alert threshold set below that thermal band will fire daily, and a daily alert is a disabled alert.
- Measurement Canada treats temperature correction as a required reconciliation step, not an optional refinement.
- Record fuel or ambient temperature alongside level, otherwise no rule can separate a thermal swing from a real loss.
01What a day tank is actually for
A standby generator is usually fed by a small tank close to the set rather than directly from bulk storage, and that tank exists to answer one question: how long can this site run without anyone arriving. Its size is therefore a runtime decision expressed in litres. Work backwards from the load, the consumption at that load and the number of hours the site must survive unattended, and the required capacity falls out. What the tank is not for is inventory. Bulk storage holds the reserve; the day tank holds a measured, bounded quantity that is consumed and replenished. This distinction matters for monitoring because the two tanks ask different questions, bulk asks how much is left, and the day tank asks how many hours are left, which is the same number divided by a rate.
02The number that moves while nothing happens
Diesel is a liquid, and liquids change volume with temperature. The published correction tables put the diesel volume correction factor at 1.0000 at 15 °C and 0.9873 at 30 °C, which works out to approximately 0.00085 per degree Celsius. Apply that to a 400-litre day tank across a 20-degree swing between a cold dawn and a hot afternoon and the level reading appears to change by about 6.8 litres. Nobody added anything. Nobody removed anything. The mass of fuel in the tank is identical at both readings, and the volume the sensor sees is not. On an outdoor tank in a Saudi summer that swing is not a rare event, it is Tuesday. Any monitoring rule that treats level change as fuel movement will therefore report movement daily.
03Why the daily alert is worse than no alert
Set a day-tank loss threshold at five litres and the thermal band described above crosses it every clear day. The technical outcome is a stream of alerts; the human outcome is that within two weeks nobody opens them, and the one alert that describes a genuine siphon arrives into a mailbox that has already been trained to discard the subject line. This failure is not a tuning inconvenience, it is the destruction of the capability the sensor was bought for. The correct order is to establish the thermal band for the tank in question first, from its capacity and the local day-night temperature range, and only then choose a threshold that sits above it. A threshold chosen before that calculation is a guess, and on an outdoor Gulf tank it is usually a guess in the wrong direction.
04Temperature correction is a required step, not a refinement
This is not an unusual position invented for telemetry. Measurement Canada instructs that an uncompensated meter reading be multiplied by the volume correction factor for the measured temperature, treating correction as a mandatory reconciliation step in trade measurement. The same discipline transfers directly to a day tank: a level reading is an uncompensated volume, and comparing two of them taken at different temperatures is comparing two different things. The practical requirement that follows is modest. Log a temperature alongside every level sample. Many capacitive probes report an internal temperature already, and where they do not, an ambient reading at the tank is enough to bound the effect. Without a temperature the correction cannot be applied at all, and the platform is left comparing incomparable numbers and calling the difference a loss.
05What the sensor has to survive on a static tank
A day tank is a harsher home than a vehicle tank in one specific way: nothing about it moves, so nothing about it is ever inspected. The probe is installed once and expected to read for years without attention. The OMNICOMM listing describes that duty cycle plainly, ±0.5 per cent accuracy, a 0 to 4,000 millimetre measuring range, IP68 sealing, an operating range of −40 °C to +85 °C, a response under one second, and a 9 to 40 V DC supply drawing under 30 milliamps over RS-485 Modbus RTU. The IP68 rating matters here because a top-mounted probe on an outdoor tank meets rain, washdown and condensation; the wide temperature band matters because the enclosure sits in the same sun as the fuel; and the low current draw matters because a remote site may be running the instrumentation off a small solar supply.
06Sizing the tank from runtime, not from habit
Day tanks are frequently sized by whatever the last site used. A runtime calculation is better and takes minutes. Generator Source publishes its consumption chart by kW and lists a 100 kW set at about 28 litres per hour at full load. A 100 kVA set at the typical 0.8 power factor is about 80 kW, so reading the chart at that size gives roughly 20 litres per hour at full load, about 10 to 11 at half load and about 6 at quarter load. These 100 kVA figures are our reading of the kW chart, not numbers the chart states for kVA. Choose the load band the site will actually sit in, multiply by the hours of autonomy required, and add the thermal band and the sensor uncertainty as headroom rather than treating the last litre as usable. That final step is where most sizing goes wrong: a tank sized to the theoretical minimum has no margin for the apparent volume change temperature will impose on it.
07A rule set that survives a Saudi summer
Three rules cover most of it. First, compute the thermal band for this specific tank, capacity multiplied by 0.00085 multiplied by the expected daily temperature range, and place every loss threshold above it. Second, judge a drop against the generator running state rather than in isolation, because a fall of forty litres is expected during a run and suspicious at three in the morning with the set idle. Third, run a cumulative reconciliation in parallel with the per-event thresholds: deliveries in, consumption implied by run hours, and measured level should close, and a persistent gap is the signal that per-event rules are designed to miss. Together these turn the day tank from a gauge somebody glances at into a number the site can actually act on.
| Tank capacity | 10 °C swing | 20 °C swing | Minimum sensible loss threshold |
|---|---|---|---|
| 200 L | ≈1.7 L | ≈3.4 L | above the 20 °C figure |
| 400 L | ≈3.4 L | ≈6.8 L | above the 20 °C figure |
| 1,000 L | ≈8.5 L | ≈17 L | above the 20 °C figure |
| 2,000 L | ≈17 L | ≈34 L | above the 20 °C figure |
| 5,000 L | ≈42.5 L | ≈85 L | above the 20 °C figure |
| Any size | capacity × 0.00085 × 10 | capacity × 0.00085 × 20 | plus sensor uncertainty |
Apparent volume change from a 20 °C daily swing, by tank size
This article's table, from Measurement Canada diesel volume correction factors (about 0.00085 per °C)
Figures and sources
- Diesel volume correction factors per API MPMS Ch. 11.1, and the apparent change they imply: VCF 1.0000 at 15 °C, 0.9958 at 20 °C, 0.9873 at 30 °C ≈ 0.00085 per °C (≈6.8 L on a 400 L tank over 20 °C) (Source: Measurement Canada (ISED), Volume correction factors, diesel fuel, as of 2026-08-31)
- Measurement Canada instruction that temperature correction is a required reconciliation step: multiply the uncompensated meter reading by the VCF for the measured temperature (Source: Measurement Canada (ISED), Volume correction factors, diesel fuel (PDF), as of 2026-08-31)
- 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 2026-08-31)
- Full-load fuel consumption listed for a 100 kW diesel generator: ~28 L/hour (Source: Generator Source, diesel generator fuel consumption charts, as of 2026-10-18)
- Approximate full-load fuel consumption, 100 kVA diesel generator (derived from the kW chart at 0.8 power factor, 100 kVA ≈ 80 kW): ~20 L/hour (Source: Generator Source, diesel generator fuel consumption charts, as of 2026-08-31)
- Approximate half-load and quarter-load fuel consumption, 100 kVA diesel generator (derived from the kW chart at 0.8 power factor): ~10–11 L/hour at half load; ~6 L/hour at quarter load (Source: Generator Source, diesel generator fuel consumption charts, as of 2026-08-31)
- Typical genset power factor used to convert kVA to kW: 0.8 (100 kVA ≈ 80 kW) (Source: Walt Power, diesel generator fuel consumption, as of 2026-08-31)
- OMNICOMM fuel level sensor accuracy, measuring range and response time: ±0.5%; 0–4,000 mm; response < 1 second (Source: ENTEK.AI product catalogue, as of 2026-08-31)
- OMNICOMM fuel level sensor sealing, temperature band and electrical requirements: IP68; −40 °C to +85 °C; 9–40 V DC; < 30 mA; RS-485 Modbus RTU (Source: ENTEK.AI product catalogue, as of 2026-08-31)
Questions
How do I calculate the thermal band for my own tank?
Multiply the tank capacity by 0.00085 and by the expected day-night temperature range in degrees Celsius. That coefficient comes from the published diesel correction factors, VCF 1.0000 at 15 °C falling to 0.9873 at 30 °C, so a 400-litre tank over 20 degrees gives about 6.8 litres.
Should the platform correct every reading to 15 °C?
It is the cleanest approach where a temperature is available, because it makes two readings taken hours apart genuinely comparable. Measurement Canada takes exactly this position for trade measurement: multiply the uncompensated reading by the VCF for the measured temperature before comparing anything.
What if my sensor does not report temperature?
An ambient temperature reading taken at the tank is enough to bound the effect and set a defensible threshold. Without any temperature at all you cannot apply the correction, so the only safe option is to raise the threshold above the worst-case thermal band for that tank size.
How large should the day tank be?
Work from runtime. Reading Generator Source’s kW chart for a 100 kVA set, about 80 kW at 0.8 power factor, gives roughly 20 litres per hour at full load and about 6 at quarter load; multiply the realistic load figure by the hours of unattended autonomy you need, then add headroom for the thermal band and sensor uncertainty rather than counting the last litre.