Standby Generator Commissioning Record

Why this matters

The commissioning record is the only document that says what the generator did on the day it was handed over. Every later service visit, warranty claim and "it worked last year" argument is settled against it. A sheet of checkmarks settles nothing: "gas pressure OK" cannot be compared with next year's reading, and "transfer tested" does not say how long it took. A sheet of numbers becomes the unit's baseline. This card defines what goes on the record, what each reading proves, and where it lives. The transfer test steps are 'Automatic Transfer Switch Commissioning Technique'; gas pressure and CO measurement methods have their own howtos, cited below.

The record, filled in

One real-shaped record for a 22 kW air-cooled unit set up on natural gas, rated 19.5 kW on that fuel. Values are this example's.

STANDBY COMMISSIONING RECORD (example)
Job 24-118   Date 10/14   Ambient 58 F
Wind from W, about 5 mph
Model 22 / 19.5 kW air-cooled  Ser 30871442
Fuel set: NG, selector verified  FW 2.14

GAS AT UNIT TEST PORT, in.wc (range 5 to 7)
  Static ................. 6.8
  Load bank 19.5 kW, 1st . 4.6  LOW, fitter
  Load bank 19.5 kW, 2nd . 5.9  OK
OUTPUT                     Volts   Hz
  No load ................  241   60.2
  Load bank 19.5 kW ......  237   59.8
TRANSFER, utility main opened
  Outage to load on gen .. 14 s
  Utility back to retx ... 2 min (as set)
  Cooldown after retx .... 1 min (as set)
EXERCISE  Tue 10:00, 12 min, no transfer
CO AT OPENINGS, 30 min run, ppm
  Bkgd 0  W1 0  W2 0  Door 0  Soffit 1
NEUTRAL floating, solid-neutral ATS
WCR 22,000 A vs utility 18,500 A
Installer J. Ortiz, lic E-2231, signed
Customer walk-through done, signed

Identity lines: model, serial, fuel set, firmware

Serial. Warranty registration, service bulletins and parts lookups all run on the serial, not the model. Read it off the unit's data plate, not the box.

Fuel set. Most air-cooled standby units ship set for one fuel and are converted in the field by a selector, a jet or a regulator adjustment per the maker's instructions. "Fuel set: NG, selector verified" says someone looked, which matters because a unit run on natural gas while still set for propane runs lean and short of output, and the symptom looks like a supply problem. The kW rating on the record should be the rating for the fuel it is set for, here 19.5 kW.

Firmware. Controller firmware determines exercise options, fault logic and how some timers behave. A service bulletin that applies to versions below a given number is useless if nobody wrote the version down. Read it from the controller's information screen.

Ambient and wind. Gas pressure on propane, engine temperatures and CO readings all depend on conditions. A reading without its conditions is half a reading.

Gas pressure: static and at full load

What it proves. Static pressure, unit off, says the supply regulator is set inside the maker's range. Pressure at full load, read at the same test port while the unit carries its rated kW, says the meter, regulator and piping deliver full-load flow without falling below the minimum. The drop between the two is the measure of the supply's margin. The method is 'How to Measure Gas Pressure at the Generator, Static and Under Load'.

Reading the worked record. Static 6.8 inches, inside 5 to 7. First full-load reading 4.6 inches, 0.4 below the 5-inch minimum, a drop of 2.2 inches from static. The gas fitter traced it to a regulator undersized for the added load and had it replaced (a utility-owned meter regulator is the utility's to change), and the second full-load reading was 5.9 inches, 0.9 above the minimum, a drop of 0.9 from static. Both full-load readings stay on the record. The first one documents why the regulator was changed; deleting it leaves next year's tech a clean history that does not explain a regulator newer than the house.

Why at full load. A unit carrying a house at 12 kW might never show the shortfall. The day it sees full load, after a long outage with every motor calling, it does. A load bank is what gets the reading at rated output on commissioning day; what that test proves beyond the gas reading is in 'What a Load Bank Test Proves and What It Does Not'.

Voltage and frequency, no load and loaded

What it proves. No-load readings show the voltage regulator and governor are set where the maker wants them. Loaded readings show they hold. Here voltage fell from 241 to 237, 4 V or about 1.7 percent of 241, and frequency from 60.2 to 59.8 Hz, 0.4 Hz. Compare them with the maker's stated regulation for the model; this example's figures are not anyone's specification.

Why it is a baseline. Next year's annual reading 236 V and 59.2 Hz at the same load is a change, and the record is what reveals it. A governor or fuel problem drifting in shows up as a growing loaded frequency drop long before it shows up as a failure.

Transfer, retransfer and cooldown

What it proves. The utility main was opened to simulate an outage, and the record times three events: from outage to load on the generator, from utility return to retransfer, and from retransfer to engine shutdown. Each one proves a timer or sensing function works and is set the way the customer and the maker agreed.

• Outage to load, 14 s. Includes the controller's start delay, which rides through brief utility blips, plus crank and warm-up. A much longer time points to a slow start; a much shorter one may mean the start delay was set to zero, which transfers on every flicker.
• Utility return to retransfer, 2 min, as set. The controller waits to confirm the utility is stable before moving the house back.
• Cooldown, 1 min, as set. The engine runs unloaded after retransfer to shed heat before stopping.

"As set" means the value on the record matches the value programmed, and both are written down. Steps for opening the main, watching the switch and restoring the utility are in 'Automatic Transfer Switch Commissioning Technique'; the record only keeps the results. Opening a service main is switching an energized disconnect: it is done by the qualified person the technique names, in the PPE the shop's NFPA 70E program sets for that operation, never by reaching into an open panel.

Exercise schedule

What it records. Day, time, length and whether the exercise transfers load. Here Tuesday at 10:00, 12 minutes, no transfer. The customer chose the time, which matters for two reasons: it is when the unit will make noise every week, and a weekday mid-morning exercise is one the customer or a neighbor will hear and notice if it stops happening. Setting rules and tuning are in 'Generator Exerciser Settings and Tuning'.

CO at openings

What it proves. With the unit running for 30 minutes, a CO meter read at each opening near the unit: windows, doors, soffit vents, any fresh-air intake. Background was 0 ppm; every opening read 0 except the soffit at 1 ppm. That shows the exhaust, on that day, in a 5 mph west wind, is not reaching the house in any meaningful amount. It does not prove it never will in a different wind, which is why the wind is on the record. Method and action levels are in 'How to Survey Carbon Monoxide at Openings During a Generator Run'; siting distances are in 'How to Lay Out a Standby Generator Site for Clearances'. A reading that rises at an opening is a siting problem, not something to note and leave.

Neutral, fault rating and signatures

Neutral. "Floating, solid-neutral ATS" ties the generator's neutral configuration to the switch type, so nobody later bonds it. See 'Reading a Transfer Switch Nameplate'.

Fault rating. The switch's withstand and closing rating against the utility's stated available fault current: 22,000 A against 18,500 A, the comparison NEC 110.10, as adopted by your jurisdiction, in substance requires equipment to pass. If the utility later upsizes the transformer, this line is what says the switch needs re-checking.

Installer and license. Who did the work, under which license, where the locality licenses it. The signature makes the record a statement rather than a form.

Customer walk-through. The customer was shown the controller, what an alarm looks like, how to clear it or call, and what the exercise sounds like. Their signature closes the handover.

Where the record lives

• In the job file and on the equipment record. In a field service system that means attached to the install job and to the generator's equipment record at the customer's site, so the next tech who opens the equipment sees it before arriving.
• A copy to the customer, with the warranty registration.
• The inspector's copy, where the jurisdiction asks for test results at final.
• Photos alongside: the data plate, the transfer switch label, the controller's information screen and the gas test port reading at full load.

How the record gets used, and how it fails

On the annual visit. The tech reads static and full-load pressure, loaded voltage and frequency, and transfer times, and writes them beside the commissioning values. Change is the finding, not the absolute number.

On a warranty claim. The maker's first question is usually whether the unit was installed and commissioned per its manual. A record with numbers answers it.

Where records fail:

• Ticks instead of numbers. Unusable as a baseline.
• A corrected reading with the original erased. The history no longer explains the work.
• No conditions. A propane full-load reading in October cannot be compared with one in January unless both carry the temperature.
• Stored only on paper in the truck. It leaves when the tech does. Attach it to the equipment record the day it is written.

References

• Generator maker's installation and commissioning instructions for the readings, ranges and settings the record holds.
• NFPA 70, National Electrical Code, 110.10 (fault current ratings), as adopted by your jurisdiction; check the current edition.
• NFPA 70E, current edition, for operating a service disconnect during the transfer test.
• See related: Automatic Transfer Switch Commissioning Technique; How to Measure Gas Pressure at the Generator, Static and Under Load; How to Survey Carbon Monoxide at Openings During a Generator Run.
• See related: What a Load Bank Test Proves and What It Does Not; Reading a Transfer Switch Nameplate.