Ask a room of facility managers what would take their standby generator out of service and most will describe something happening to the engine. A rod, a bearing, the alternator letting go. It makes sense. That’s the failure people can picture.
It’s not what we see.
Fuel system work was the biggest single group — contaminated diesel, plugged filters, water in the separator, a transfer pump that quit. Batteries were right behind it. Then cooling: coolant past its service life, a cracked hose, a glazed belt, a water pump weeping at the bearing.
None of that is exotic. All of it is on an interval.
Controls and electrical problems came next, and those deserve their own mention, because they don’t announce themselves during a monthly exercise. An AVR that has drifted, a sensor reading wrong, a loose termination in the panel. The unit starts, the panel lights up, everything looks fine, and then it won’t hold load.
Transfer switch and breaker work was a smaller group. It’s also the group where the consequence is worst. A generator that runs beautifully into an ATS that won’t transfer is a building with no power.
At the far end were the failures people imagine: alternator work and major engine repairs. There were only a few. They cost more per job than anything else we did, by a wide margin. And when you read the histories, most of them started somewhere in the first three categories. A cooling problem nobody chased. A unit that ran underloaded for years.
That’s the useful part. The things most likely to leave a facility dark are cheap, unglamorous, and scheduled.
Why September Matters
September is National Preparedness Month, which is a decent excuse to schedule work that keeps sliding. It’s also a genuinely bad month to be caught out.
The August 11 derecho is the recent reminder. It tore through the ComEd, NIPSCO, and Duke Energy corridors and knocked out distribution across Chicagoland, Northwest Indiana, and Greater Cincinnati. ComEd had most customers back inside about five days. NIPSCO, which called it the largest event in its history, took considerably longer, with parts of its territory dark for roughly two weeks. The facilities that struggled mostly didn’t fail on day one. They failed on day four, when fuel ran low and the delivery queue was already hundreds deep.
The longer trend isn’t friendlier. NERC’s January long-term assessment projects summer peak demand growing 224 gigawatts over the next decade, well above what the same report projected a year earlier, with winter growth climbing faster still. Thirteen of the twenty-three regions NERC assesses face elevated or high resource adequacy risk within five years, MISO, PJM, and ERCOT among them.
Hurricane season is quiet, and that’s its own problem. NOAA is holding a below-normal forecast, and the season has produced no hurricanes so far. The forecast is probably right. It’s also removing the thing that normally forces the maintenance conversation. Meanwhile the remnants of Dolly are being watched for redevelopment as they move toward Florida, there’s a second disturbance in the northern Gulf, and September 10 is the climatological peak.
Below-normal seasons still make landfall somewhere specific. So do derechos, ice storms, wildfire shutoffs, and load-shed events in tight capacity regions.
Start With What You Actually Have
This sounds like the easy part. It usually isn’t, because whoever commissioned the system has moved on and the documentation went with them.
- Pull nameplate data for every unit in the fleet: nameplate kW, standby versus prime rating, model year, hours on the meter, last major service.
- Compare nameplate kW against connected load as it exists today, not at commissioning. Added racks, new HVAC, chargers, process equipment. This is why a system that tested clean three years ago can’t carry the building now.
- Confirm system classification. For life safety and critical branch work, is the EPSS a Level 1 system under NFPA 110, and for health care, how do NFPA 99 category assignments land on your branches? That drives your testing obligations.
- Map single points of failure. One transfer switch feeding two critical branches. One day tank. One fuel supplier. One person who knows the sequence.
- Find the paperwork your AHJ will ask for. Test logs, load test records, commissioning documents, the written EPSS maintenance program. Undocumented work didn’t happen, as far as an inspector is concerned.
Fuel
The largest category we see, and rarely just one thing.
- Pull a sample and test it. Diesel that sat through a warm summer is a candidate for water and microbial growth.
- Service filters and water separators, and check the lines and the day tank transfer pump while you’re in there.
- Verify on-site volume against required runtime at actual load. Not nameplate tank hours at half load.
- Get the emergency delivery agreement in writing. Priority language, response time, minimum volume. Signed while nothing is happening.
- Confirm a fuel truck can actually reach the tank under storm conditions. Downed trees, a flooded approach, and badge access on a dead panel have all stopped deliveries that were otherwise on their way.
Batteries
The most common reason a standby unit doesn’t start, and among the cheapest parts on the machine.
- Load test them. A voltage reading on a resting battery tells you very little.
- Check age against replacement interval. Most starting batteries in standby service get replaced on a three-to-five-year schedule whether or not they still seem fine.
- Look at terminals, cables, and hold-downs. Corrosion and looseness are constant findings.
- Verify charger output and float voltage. A bad charger shows up as a dead battery months later, and the battery gets blamed.
Cooling
Cheap to address, and the shortest path to an expensive engine repair if ignored.
- Test coolant condition and freeze point, not just level.
- Inspect hoses, belts, and clamps for cracking, glazing, and soft spots.
- Check the water pump for weep-hole seepage and bearing play.
- Clean the radiator core, and look at the fan, shroud, and louvers.
- Verify block heater function and jacket water temperature. An engine that starts cold is not the same as an engine that accepts load cold.
What Only Shows Up Under Load
A monthly no-load exercise proves the engine turns over. That’s all it proves.
- Schedule load bank testing on anything that hasn’t run under real load this year. Chronic underloading causes wet stacking, and wet stacking is why a unit that “runs fine every month” quits when it finally matters.
- Check your testing interval against the NFPA 110 edition your AHJ actually enforces. Annual load testing is the general requirement, but supplemental testing provisions tied to monthly exercise load thresholds vary by edition. Confirm rather than assume.
- Test the transfer switch under load. Verify pickup and dropout settings, time delays, and the retransfer and cooldown sequence. This is the one that turns a healthy generator into a dark building.
- Push the controls. AVR drift, ECU faults, bad sensors, and loose panel terminations mostly show up under load and stay hidden at rest.
- Run the whole sequence: utility loss, start, transfer, load acceptance, sustained run, retransfer, cooldown, shutdown. Partial tests find partial problems.
Who Shows Up When It Fails
Response capacity in any region is finite and shared, and it gets allocated before an event, not during one.
- Get service coverage in writing with a response time attached. Everyone is oversubscribed inside 72 hours of a regional event, and your place on the triage list comes from your contract, not from how urgent your call sounds.
- Ask where you sit in the queue. OEM service networks have been steering technician capacity toward large data center commitments, and that’s a lasting pattern rather than a seasonal one. If the answer to “what’s my response time in a regional event” is vague, that’s your answer.
- Price the rental fallback now. Size, source, lead time, cable and connection requirements, and whether the site has a docking station or a manual tap ready. A rental unit with no way to connect is an expensive paperweight.
- Set your bridge power trigger in advance, so it isn’t a debate at hour thirty of an outage.
- Walk the escalation plan with the people who’ll run it. Names, mobile numbers, after-hours spending authority, who calls whom at two in the morning.
The Takeaway
Standby generators mostly don’t fail dramatically. They fail because a battery nobody load tested was four years old, because fuel sat through a hot summer, because a hose that looked fine let go on hour six of a sustained run. The expensive repairs, the ones people picture, usually arrive later as the bill for the cheap ones.
None of this work is complicated. It just needs to get on a calendar, and September is as good a month as any to put it there.
Generator Source sells, services, rents, and buys commercial and industrial generators from 20kW to 4MW+, with headquarters in Brighton, Colorado and branches in Jacksonville and Pensacola, Florida, and Austin, Texas. Services include preventive maintenance, load bank testing, transfer switch verification, fleet assessment, and rental and bridge power.
Contact our team today to discuss your preparedness plan.