Short answer
Monitor two things: accumulated operating hours and fuel actually present in the day tank. Cummins schedules service at 8, 50, 100, 250 and 500 operating hours, whichever comes first, and Caterpillar fits a dedicated hours-run meter. Neither number can be inferred from the other, and both drift silently between visits.
Key takeaways
- Cummins expresses its Onan generator schedule in operating hours, 8, 50, 100, 250 and 500, with "whichever comes first" against calendar time.
- In dusty or dirty conditions Cummins shortens the oil change to every 50 operating hours, which is a Gulf-relevant clause rather than a footnote.
- Caterpillar fits a dedicated hours-run meter and states it "assists with maintenance", with service every 12 months or 500 hours.
- Readiness has a second half the hours meter cannot see: whether the day tank holds the fuel the run will need.
- A generator idle for months accumulates no hours, so the calendar half of "whichever comes first" is the only clock still running.
01Readiness is two variables, not one
A standby generator is bought for a moment that has not happened yet, and that changes what monitoring means. On a working truck, use produces evidence: the odometer climbs, fuel is consumed, and faults surface because the machine is exercised daily. A standby set produces almost no evidence at all. It sits, and the two things that decide whether it will start are both invisible from outside the enclosure. The first is how many operating hours it has accumulated, because every manufacturer schedule is written in those hours. The second is whether the day tank actually holds fuel, because a set that cranks correctly and then starves after twenty minutes has failed just as completely as one that never turned over. Neither variable can be derived from the other, and neither is legible on a walk past.
02The manufacturer already told you the unit of measure
The service intervals in a generator manual are not written in months, and that choice carries information. The Cummins operator manual for the Onan HDKAG set lays out its schedule at 8, 50, 100, 250 and 500 operating hours, and pairs each interval with calendar time under the phrase "whichever comes first". Caterpillar takes the same position from the hardware side: its generator sets carry a dedicated hours-run meter, described in the operation manual as assisting with maintenance, and the manual sets service at every 12 months or 500 hours. Read together, the two manufacturers are saying that an hour of running is the wear unit and a month of sitting is only a fallback. If the site has no reliable record of accumulated hours, the schedule it is following is the fallback clock by default, whether or not anyone chose that.
03Dust moves the interval, and the Gulf has dust
One clause in the Cummins schedule deserves separate attention because it is routinely read as a footnote and it is not. In dusty or dirty conditions the manual shortens the oil change to every 50 operating hours. That is a substantial compression of a normal interval, and for a set installed at a desert telecom site, a remote pumping station or an unpaved yard it is the applicable line rather than the exception. The consequence for monitoring is direct: if the site is running the shortened interval, the hours figure needs to be accurate to a much finer resolution than a quarterly visit provides. Fifty hours of running can accumulate in a single bad week of grid instability. A number read once every three months cannot tell you which side of that threshold the machine is on.
04The fuel half of the answer
An hours meter says nothing about fuel, and a full tank says nothing about hours. Both halves need their own instrument. On the fuel side a level sensor in the day tank supplies the missing variable continuously, and the specifications that matter are the ones that survive a static outdoor tank rather than a moving vehicle. The OMNICOMM listing reads to ±0.5 per cent across a 0 to 4,000 millimetre range, is rated IP68 and −40 °C to +85 °C, responds in under a second, and runs from 9 to 40 V DC over RS-485 Modbus RTU. Those entries describe a probe that can be top-mounted in a standby day tank and left there. Paired with an hours count, the site finally has both readiness variables, and each is answered by the instrument that can actually see it.
05Why a monthly test run is a measurement, not a ritual
Most standby installations run a periodic test, and most treat it as a box to tick. It is better understood as the only opportunity the site gets to measure both variables under load at once. A test run that is instrumented produces four facts a visual check cannot: the hours counter advanced by the expected amount, fuel fell at a rate consistent with the load applied, the level was high enough to sustain the run in the first place, and nothing tripped before the run completed. A test run that is not instrumented produces one fact, it started, and that is the least informative of the four. Given the manufacturer schedules are denominated in hours, an uninstrumented test also quietly consumes interval budget without recording that it did so.
06What the two clocks do while nothing happens
The interesting failure mode of a standby set is not wear. It is the long quiet period in which the hours meter stands still and the fuel level does not. Diesel in an outdoor tank responds to ambient temperature, and a level reading taken at three in the afternoon is not the same reading as one taken before dawn even when nothing has been added or removed. Sediment and water settle. Batteries self-discharge, Caterpillar calls for battery checks every 200 hours in high ambient temperature, which on an idle set means the calendar carries that item too. So while the hours clock is frozen, several other conditions are moving. A monitoring design that only watches hours will therefore report a healthy machine right up to the moment it is needed.
07A monitoring specification you can hand to an installer
The requirement list is short. Capture a running-state signal and accumulate it into hours on the server rather than the browser, so the total is one number with one owner. Capture day-tank level continuously from a probe rated for a static outdoor tank. Record the ambient or fuel temperature alongside the level, because without it an overnight change cannot be separated from a real loss. Configure the interval thresholds from the manual that ships with the set, 500 hours or 12 months for the Caterpillar case, the shortened 50-hour oil change if the site is dusty, and put the due dates in a maintenance system or calendar that someone owns, so the item is raised on time rather than remembered. Finally, treat each test run as a logged measurement with an expected hours delta and an expected fuel delta, so a deviation is visible the same week rather than the next audit.
| Question | Answered by | Blind to | Interval it drives |
|---|---|---|---|
| How much has it run? | Running-state signal accumulated into hours | Fuel present, battery state | Cummins 8/50/100/250/500 h |
| Will it keep running? | Day-tank level sensor | Accumulated wear | Refill scheduling |
| Is the reading real or thermal? | Fuel or ambient temperature alongside level | Volume itself | Threshold tuning |
| Is a service item due? | Hours plus calendar, whichever comes first | Condition between intervals | Caterpillar 12 months or 500 h |
| Did the test run succeed? | Hours delta and fuel delta together | Reasons for a partial run | Monthly test log |
| Is the battery still good? | Separate battery check | Everything else | Caterpillar 200 h in high ambient |
Figures and sources
- Cummins/Onan generator maintenance schedule, expressed in operating hours: 8, 50, 100, 250, 500 h, "whichever comes first" vs calendar time (Source: Cummins Operator Manual, Onan Generator Set HDKAG, as of 2026-08-31)
- Cummins shortened oil-change interval in dusty/dirty conditions: every 50 operating hours (Source: Cummins Operator Manual, Onan Generator Set HDKAG, as of 2026-08-31)
- Caterpillar generator set hours-run meter, stated purpose, and service interval: "assists with maintenance"; service every 12 months or 500 hours (Source: Caterpillar Generator Set Operator & Maintenance Instruction Manual, as of 2026-08-31)
- SAE J1939 SPN 247, Engine Total Hours of Operation, resolution and PGN: 0.05 hr/bit, PGN 65253 (Source: SAE J1939-71 (Revised MAR2011), Vehicle Application Layer, 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 environmental rating and supply, for a static outdoor day tank: IP68; −40 °C to +85 °C; 9–40 V DC; RS-485 Modbus RTU; < 30 mA (Source: ENTEK.AI product catalogue, as of 2026-08-31)
- Diesel volume correction factors per API MPMS Ch. 11.1, why an idle tank's reading moves: VCF 1.0000 at 15 °C, 0.9873 at 30 °C ≈ 0.00085 per °C (Source: Measurement Canada (ISED), Volume correction factors, diesel fuel, as of 2026-08-31)
- Wialon service-interval implementation, three independent counters tracked simultaneously: mileage, engine hours, and calendar days, each calculated independently (Source: Wialon Hosting User Guide, Service intervals, as of 2026-08-31)
Questions
If the generator rarely runs, why monitor hours at all?
Because the manufacturer schedule uses "whichever comes first". A set that accumulates few hours is being serviced on the calendar half, and you cannot know which half applies without the hours figure. Cummins lists intervals at 8, 50, 100, 250 and 500 operating hours precisely so both clocks can be compared.
Does the dusty-conditions clause really apply to my site?
If the set sits in an unpaved yard, a desert compound or beside a haul road, it is the applicable line rather than an exception. Cummins shortens the oil change to every 50 operating hours in dusty or dirty conditions, which is a large compression and needs an accurate hours count to act on.
Can I just check the fuel gauge on the day tank during the monthly visit?
A monthly reading tells you the level on that afternoon, not whether it held all month. Ambient temperature alone moves an uncompensated reading, diesel VCF goes from 1.0000 at 15 °C to 0.9873 at 30 °C, so a single sample cannot separate a thermal swing from a genuine loss.
Where should the hours total actually be calculated?
On the server, from the stored running-state history, so the total is one number with one owner and the same value appears on every screen and report. A figure recomputed in a browser from a fetched window of data will disagree with itself as soon as the window changes.