Load bank testing
A generator load bank test runs a standby generator at known loads, up to its nameplate rating, so the engine, cooling, fuel and exhaust systems are proven before an outage depends on them. This page explains the method and the rules behind it.
Because their routine exercise rarely loads them. Generac's paper notes that during weekly or monthly exercise a generator runs at a lower load than its full-rated output, and that many owners run units unloaded and rely on the occasional utility loss to exercise the whole system, a practice the makers do not recommend. A load bank test runs the set at its advertised nameplate rating and checks that its primary components work under load.
Data centres meet the problem in a particular form. Kohler's data-centre paper explains that generators there were historically oversized for the usual load, and that operators do not want to transfer building load for a monthly exercise, so the exercise needs a load bank. With five to ten generators on a typical site, Kohler notes, that becomes a significant time commitment for operations staff.
Wet stacking is unburned diesel fuel collecting in the exhaust system of an engine that runs too lightly loaded. Kohler explains that below 30% of capacity for long periods the engine cannot hold the temperatures needed to burn fuel completely. Generac puts the exhaust temperature target at about 280 °C and the load needed to reach it at roughly 30%, more in cold weather; the result of running below it shows as black smoke and oily grime in the exhaust stack. Modern aftertreatment (oxidation catalysts, particulate filters, selective catalytic reduction) is sensitive to it, which is why a loaded run, often on a load bank, is part of generator upkeep.
NFPA 110 sets monthly and annual load rules for emergency and standby power systems, and CSA C282 is the Canadian standard for emergency generators that the building code requires. The NFPA 110 clauses below are as reprinted by Cummins from the 2016 edition; check the edition your authority having jurisdiction applies.
| Clause | Requirement |
|---|---|
| NFPA 110, 8.4.2 | Exercise at least monthly, for at least 30 minutes, at the maker's minimum exhaust temperature or at not less than 30 percent of standby nameplate kW |
| NFPA 110, 8.4.2.3 | Diesel sets that cannot meet 8.4.2: exercise monthly with the available load, and annually with supplemental load at 50 percent for 30 minutes and 75 percent for 1 hour, 1.5 continuous hours in all |
| NFPA 110, 8.4.9 | Level 1 systems: a test at least once every 36 months for the assigned class; over 4 hours, the test may stop at 4 hours |
| NFPA 110, 8.4.6 | Transfer switches operated monthly |
| CSA C282 | Design, installation, operation, maintenance and testing of generator sets that supply emergency power required by the National Building Code of Canada, and essential electrical systems in health care facilities; minimum load for monthly testing back to 30% in the current edition |
CSA C282 explicitly excludes emergency power from batteries and UPS, so a UPS is tested under its own maker's procedures; see UPS and battery load bank testing. Codes vary by location and application, and Generac's advice holds in Canada too: confirm the requirements with your local authority having jurisdiction.
The test connects the load bank to the generator output, warms the engine, then raises the load in timed steps while readings are taken. Generac's recommended sequence is:
A resistive load bank tests the engine; a reactive one is needed to test the alternator at its full kVA. ASCO notes that a resistive unit fully loads the engine, cooling, fuel and exhaust systems at 100% of nameplate kW but reaches only 80% of nameplate kVA. Combined resistive and reactive load banks test the set at its rated power factor, in most cases 0.8, and are usually required to calibrate load sharing and voltage regulation on paralleled generators. ASCO's example: a 3000 kW load bank can test one 1500 kW set at full kW, or two 1500 kW sets in parallel at full nameplate kW. More on the difference in resistive vs reactive load banks.
They are separate tests of two links in the same chain. The generator carries the site through a long outage; the UPS carries the load through the transfer and filters what the generator delivers. A generator test proves the engine and alternator; it does not prove the UPS, and a UPS test does not prove the generator. How the two work together (sizing, transfer behaviour and the input the UPS accepts) is covered in UPS and generator compatibility.
Under NFPA 110 the set is exercised monthly; diesel sets that cannot reach 30 percent of nameplate kW (or the maker's exhaust temperature) on the available load need an annual supplemental load test of 1.5 hours. In Canada, CSA C282 sets the testing rules for building-code emergency generators.
Not as a load test. Generac notes that units exercised unloaded are not proven at their rating, and running diesel engines lightly loaded leads to wet stacking.
NFPA 110 is an American standard that many specifications reference. In Canada, emergency generators required by the National Building Code are covered by CSA C282, and the local authority having jurisdiction decides what applies.
Because transferring a data centre to its generators for a test puts the IT load at risk. Kohler notes that data-centre operators prefer not to transfer building load, so the monthly exercise uses a load bank.
Sources, checked 2026-10-04:
Tell us the site, the UPS and the timing. We reply and quote within 4 business hours, or call 1 (438) 800-0382
Brand names are trademarks of their owners and are used only to identify the equipment we service. We are not the manufacturer and are not affiliated with it.