Most facilities managers in the Denver metro can produce a clean binder of monthly generator tests. Twelve entries a year, thirty minutes each, engine started, load transferred, everything normal. The tests were real. They still do not answer the question an inspector is asking when he asks about fuel.
NFPA 110 splits the job in two. Chapter 8 makes you run the machine. A separate line in the same chapter makes you test the fuel. Buildings do the first religiously and skip the second for years, and the reason is not laziness. It is arithmetic: the required tests burn so little diesel that the fuel in the tank never actually leaves.
What does NFPA 110 actually require you to test?
Five separate things, graded separately. Only one of them is about the fuel.
| Requirement | Section | What it takes |
|---|---|---|
| Inspection | Chapter 8 | EPSS and components inspected at least weekly |
| Monthly operational test | 8.4.2 / 8.4.2.1 | At least 30 minutes at not less than 30% of nameplate kW — or at the manufacturer's minimum exhaust gas temperature |
| Annual supplemental load test | 8.4.2.3 | Where the monthly run cannot reach that loading: 50% of nameplate for 30 min, then 75% for 60 min — 1.5 continuous hours |
| 36-month load test (Level 1) | 8.4.9 | Continuous for the assigned Class duration, not required to exceed 4 hours |
| Annual fuel quality test | 8.3.7 | Fuel tested at least annually using appropriate ASTM standards |
One caution before you build a calendar off that table: the staged percentages in 8.4.2.3 changed between editions — older printings ran 25%, then 50%, then 75% over two hours. Denver-metro jurisdictions do not all adopt the same edition. Confirm which one your AHJ enforces and price the test against that.
The first four rows prove the machine. The last row proves the fuel. Chapter 7 draws the same line from the design side: section 7.9.1.2 calls for a supply of clean fuel to the prime mover. Clean is a fuel condition, and no amount of starting the engine demonstrates it.
Why does a building pass its load test and still fail on fuel?
Because thirty minutes a month is nothing, and the tank is sized for a disaster.
NFPA 110 defines Class as the minimum time, in hours, an emergency power supply system is designed to operate at rated load without refueling. A Class 48 system carries forty-eight hours of fuel. That is the point of the classification: the tank exists to survive an event nobody wants.
Now count what testing takes out of it, converting each required run into full-load-equivalent hours — a half hour at 30% counts as 0.15:
- Twelve monthly tests at 30 minutes and 30% load: 1.8 hours
- One annual supplemental test, where it applies (30 min at 50%, 60 min at 75%): 1.0 hour
- One 4-hour test at 30% every 36 months, spread across three years: 0.4 hours
About 3.2 full-load-equivalent hours a year. If your monthly run does meet the 30% threshold, the annual supplemental test does not apply and the number drops to roughly 2.2.
How long does the fuel in a Class 48 tank actually sit?
| EPSS Class | Hours of fuel at rated load | Years of required testing to burn one tank |
|---|---|---|
| Class 2 | 2 | Under 1 year |
| Class 6 | 6 | About 2 years |
| Class 48 | 48 | About 15 years |
Arithmetic only, at 3.2 full-load-equivalent hours per year, assuming no outages and no refueling. At 2.2 hours a year the Class 48 figure stretches past twenty.
And that is the optimistic version, because nobody drains a generator tank and starts fresh. Tanks get topped off. New fuel mixes into old, and the layer that never moves is the one at the bottom — which is exactly where the problem lives. ASTM D975 says it plainly in its storage appendix: keep tanks free of water and give them provisions for draining water on a scheduled basis, because water promotes corrosion and microbiological growth occurs at the fuel-water interface. The same appendix defines long-term storage as fuel held more than twelve months after the user receives it.
So a Class 48 tank on a testing-only diet is in long-term storage from its first birthday and stays there for the life of the building. The load test keeps passing the whole time. It is measuring the engine.
NFPA 110 saw this coming. Section 7.9.1.3 requires that tanks be sized so the fuel is consumed within its storage life — or that provisions be made to remediate fuel that is stale or contaminated, or to replace it with clean fuel. Hold that sentence next to a Class 48 tank and only the second half of it applies. For most large standby systems, remediation or replacement is not an upgrade someone sold you. It is the design condition the standard assumed you would meet.
Does Denver's elevation change what 30% of nameplate means?
It changes the machine, not the requirement — and it is worth knowing before you rent a load bank.
Diesel engines lose output as air thins. Cummins publishes site derating factors on its spec sheets; for the C1000D6RG standby set, the engine runs up to 2,000 feet and 104°F with no deration, and above that derates 3% per 1,000 feet and 13% per 18°F. Denver sits at roughly 5,280 feet, about 3,280 feet past that threshold — call it a 10% reduction for that model.
Section 8.4.2.1 sets the monthly load against nameplate kW, not against what the set can make at your address. So in Denver, 30% of nameplate is closer to a third of the engine's real local output. That is not a compliance problem — the standard says nameplate and means nameplate — but it changes how much load bank you rent, and how much comfort to take from the phrase "we are only at 30 percent." Derate factors are model-specific. Pull yours off the engine data sheet, not off a rule of thumb.
Are you allowed to run long enough to burn the tank down?
Almost certainly, yes. The air rule is not what is stopping you.
Emergency stationary engines fall under EPA's RICE NESHAP. Under 40 CFR 63.6640(f)(2), an emergency engine may run up to 100 hours per calendar year for maintenance checks and readiness testing. There is no hour limit at all during an actual emergency, per (f)(1). Up to 50 of those 100 hours may go to non-emergency operation under (f)(3) and (f)(4), with real restrictions attached — peak shaving and income generation are out except in narrow cases.
Now add up what NFPA 110 requires: six hours of monthly runs, 1.5 hours of annual supplemental testing, and the 36-month test amortized. Under ten hours a year against a hundred-hour allowance — more than ninety hours of headroom, sitting unused. That headroom is where the fuel problem gets solved, if you want to solve it that way: longer runs at higher load move real volume out of the tank and real fuel back in. Check your Colorado air permit first, with the state air division rather than a blog post. It can carry conditions tighter than the federal floor.
What should a Denver facilities manager do about it?
- Find your Class. It is on the EPSS documentation and it sizes the whole problem. Class 6 is a two-year fuel problem. Class 48 is a fifteen-year one.
- Pull the last fuel quality test. If there is not one, 8.3.7 is already open — on a system whose load-test binder looks perfect.
- Drain the water first. Cheapest item on the list, and it removes the interface where microbial growth starts. On a schedule, not when someone remembers.
- Let the test results decide polishing versus turnover. Section 7.9.1.3 gives you both options. Filtering cleans what is in the tank; turnover replaces it. Dirty fuel gets polished. Chemically finished fuel gets pumped out.
- Make the load test do double duty. If a load bank is already on site and billed for, running longer costs the fuel and little else — and it is the only test that moves volume.
- Put the fuel test on the same calendar line as the load test. Separate lists are how the second one gets skipped for six years.
We handle the fuel side of that list across the Denver metro — scheduled generator top-offs, tank pump-outs, and fuel swapped on a calendar instead of after an outage. The services page covers what that includes; the emergency delivery page covers the other case, the one where the tank matters at 2 a.m. Cities we run are on the service areas page. English y en Español.
None of this makes your load test less important. Run it, document it, keep the binder. Just stop treating it as evidence about the fuel. It has never been that, and the standard has never claimed it was.
General information, not a compliance determination. NFPA 110 section numbers and requirements vary by edition, and Denver-metro fire districts adopt and amend differently. Confirm the adopted edition and local amendments with your authority having jurisdiction, and run-hour limits against your own air permit. Last reviewed August 2026.
