Cummins Generator Maintenance: 9 Costly Mistakes And A Checklist To Avoid Them

I've spent 7 years handling generator maintenance for a commercial property group, and I've made some expensive mistakes. Six significant ones, to be exact, totaling roughly $18,000 in wasted budget. Not a record I'm proud of, but it's the reason I now maintain a checklist that catches problems before they turn into invoices.

If you're responsible for a Cummins generator, whether it's a 20 kW standby unit at a small business or a 120 kW backup system at a hospital, this checklist will save you the pain I already went through.

Step 1: Look Up The Actual Oil Capacity Before Ordering Parts

This sounds embarrassingly basic. I learned it the hard way.

In March 2021, I ordered oil filters and oil for a unit we'd just taken over, assuming the specs matched our previous generator. They didn't. The Cummins 20kW generator oil capacity is right around 3.5 quarts with a filter—ours is an RS20A, but I'd have to check the manual to be sure. That's kind of the point. Don't trust my memory.

I bought 5 quarts out of habit, and the wrong filter. $180 in restocking fees and a full day of delays later, I learned: look it up. Every time. The manual is the only source of truth.

I don't have hard data on how many maintenance delays come from wrong parts, but based on our 5 years of orders, I'd guess it's around 15% of unscheduled work.

Step 2: Size By Real-World Load, Not Nameplate

When we installed our 120 kW Cummins backup generator in 2022, I thought we were being smart. The building's electrical load had grown, and 120 kW seemed like the obvious next step up.

We sized it based on nameplate ratings of our equipment. That was mistake number two.

Nameplate ratings are theoretical maximums. They don't account for how equipment actually runs—startup surges, power factor, harmonics, all of it. Our real-world load was about 25% lower than the nameplate totals suggested, which meant we spent roughly $8,000 more than we needed to on capacity we'll never use.

The fix: do a load study. Run data loggers on your main feeders for a week, and use real numbers to size your generator. Or ask a dealer to do it as part of a quote. It's worth the extra week of effort.

This worked for us, but our situation was a fixed commercial building with predictable occupancy. If your facility has large motor loads or process equipment that cycles, your numbers will look different. Your mileage may vary.

Step 3: Fuel Type Is A Systems Decision, Not A Preference

I went back and forth between natural gas and propane for months. A natural gas whole home generator seemed more convenient—no tanks to monitor, no refill schedules. But our site's gas line pressure was marginal for the 120 kW unit, and a meter upgrade would've run another $4,200. Maybe $4,800, I'd have to check the invoice.

Propane meant two 500-gallon tanks, ongoing fuel management, and the risk of running dry during a multi-day outage. Different tradeoffs, both real.

The natural gas vs. propane decision kept me up at night. On paper, natural gas won. We stayed with natural gas after the utility confirmed we could get the pressure with a meter upgrade. But the process taught me: get written confirmation from the utility and fuel vendor before committing to anything.

Lately I'm getting more questions about propane and solar generator hybrids. I can only speak to conventional setups—solar plus propane transfer switch logic gets complicated fast. If you're considering that route, get a controls engineer involved early.

Step 4: Test Batteries Under Load, Not Just Voltage

I was guilty of this for years. I'd check a battery with my multimeter, see 12.4 volts, and mark it good. Then the generator would fail to start during an actual outage, and I'd be starring at a battery that couldn't deliver enough cranking amps.

People ask me, "can you test a battery with a multimeter?" Yes, you can—but voltage alone is misleading. A surface charge can look fine while the battery is failing internally.

What actually works:

• Use a load tester that applies a controlled draw and reads the voltage drop.
• Watch the voltage in real time while someone hits the starter button.
• Check specific gravity with a hydrometer if your batteries are serviceable.

The second method—watching the drop during an actual crank—caught 3 failing batteries in our fleet that a simple voltage check had cleared.

We now replace every generator battery at 3 years, no matter what the test says. Period.

Step 5: Run A Full-Load Test Every Quarter

No-load testing is better than nothing. But not by much. A generator running without a load never reaches operating temperature, moisture stays trapped in the windings, the cooling system never gets exercised. You're not testing anything, really.

We now rent a load bank four times a year, around $400 per test. When I first pitched this to my manager, I got pushback—felt like spending money to prove something that should already work. Then in Q2 2024, a load test revealed a cooling fan clutch that was slipping under load. That $400 test saved us from what would've been an overheating failure during a summer outage.

The clutch repair cost $900. A generator failure during a two-day outage would've cost us ten times that in lost tenant revenue.

Step 6: Document Everything

In my first year, I kept maintenance records in my head. Terrible idea. Now I record every service visit, every filter change, every test result, every odd noise. I note the serial number, the hours meter reading, and any controller error codes.

If you're buying a used Cummins generator, get the serial number before you commit. I didn't once, and ended up with a unit that had been flooded—the seller conveniently forgot to mention that. The Cummins serial number lookup confirmed its history, but only after we'd already bought it.

Oh, and keep your dealer's phone number written down somewhere visible. Obvious, I know, but you'd be surprised how many people spend an hour hunting through email while a generator won't start.

Common Mistakes I Still See

These are the ones I spot at other facilities, even from people who should know better:

• Forgetting coolant. Oil gets checked constantly, coolant gets ignored. The generator overheats, and it's a $3,000 repair instead of a $50 top-up.
• Trusting someone else's records. A note said "service completed 02/15/2024." The filter had been ordered but never installed. Trust, but verify.
• Assuming new parts are good parts. A brand-new fuel filter failed within a week once. It happens. Test, don't assume.

The Bottom Line

Look, I'm not saying my checklist is the only way. Every facility is different, and nothing replaces knowing your specific equipment.

But this checklist has caught 47 potential problems in the past 18 months—and each one would have been a costly surprise. Maintenance is about data, not guesses. Measure things. Document things. Verify things.

I wish I'd known this from day one. That $18,000 in early mistakes? It bought me a system I've been using ever since. Now you get the lessons without the bill.