The cheap kilowatt is the expensive kilowatt

Here's a position I'll defend in any meeting: price per kilowatt is the least useful number on a generator quote. Not irrelevant — least useful. What you're actually buying with standby power is the probability that 280 kW will be sitting there, unbothered by ice, at 2:47 in the morning on the one night your utility feed drops. That probability is the product. The kVA is just how it's measured.

I coordinate emergency power deployments for an industrial electrical contractor. I've handled 60-plus rush generator jobs in nine years, including same-day turnarounds for cold storage and a couple of hospital auxiliary buildings (the ones where "a few hours late" is a phrase nobody wants to hear). In March 2024 a client called at 4 p.m. on a Friday — their existing 350 kVA unit was down on a low-coolant fault, room temperature climbing, roughly 40 tons of product inside. We had a replacement on a flatbed by 6 a.m. Saturday and running by Sunday afternoon. The expedite and freight premium came to about $6,300 on top of base price. Nobody at that company has ever questioned that fee. The alternative was a write-off north of $300,000.

That's the whole argument. Let me defend it.

The "biggest portable generator" mindset is a unit-conversion error

The biggest portable generator you can buy at retail tops out around 10 to 15 kW of running output (as of 2025 — that ceiling hasn't moved much in a decade). A 350 kVA SDMO generator delivers 280 kW at 0.8 power factor. That's roughly twenty times the capacity, but treating it as "a really big portable" is the wrong mental model entirely.

Different architecture. Air-cooled, 3,600 rpm, 30-amp receptacles, twelve hours on a six-gallon tank — versus water-cooled, 1,500/1,800 rpm, three-phase bus, automatic transfer switch, base tank sized in days rather than hours. Portable units are genuinely good at what they do: site trailers, crews, tailgates, storm cleanup. When someone tells me "we'll just get the biggest portable generator," what they usually mean is "we haven't sized the load." That's the actual problem, and no retail unit solves it.

Everything I'd read about generator selection said to start with the load schedule and work forward. In practice, the jobs that go sideways start with a number someone liked from a catalog. We quoted a laundromat in 2022 off a verbal "about 100 kW." Actual running load was 118 kW, and the inrush from four 25-hp compressor motors pushed us into a 200 kW frame. The 100 kW unit we'd almost bought would have tripped on every motor start. Nobody's fault but ours (note to self: never quote off a phone call again).

Whole-house solar generators: honest about the limits

Whole-house battery systems are legitimately good products now — 10 to 30 kWh of storage, 5 to 15 kW of inverter output, seamless switchover, no fuel. For a home that's real backup for lights, the fridge, internet, and a furnace blower. I recommended one to my brother.

For a facility, the math gets short fast. A 20 kWh battery bank discharging into a 200 kW load gives you six minutes. Not six hours — six minutes (20 ÷ 200 = 0.1 h). Add motor inrush or a chiller start and it's worse. I watched this go wrong in 2023: a client had installed a battery system to ride through outages at a small distribution center and discovered in the first real event that the refrigeration compressors' locked-rotor current tripped the inverter before transfer even completed. They'd been warned. They didn't listen. Now the battery does a beautiful job protecting the server rack, which is not what they bought it for.

Batteries are a bridge; diesel is a duration play. Most SDMO diesel generators in the 300 kVA class will run 24 to 48 hours on a mounted tank and keep running as long as somebody refuels them. That's the difference in one sentence.

Why a 350 kVA SDMO diesel generator gets specified, not shopped

350 kVA at 0.8 PF is 280 kW. Per ISO 8528-1, you need to know which rating you're being quoted: ESP (emergency standby — limited hours per year, no overload capability) or PRP (prime rated, 10% overload for one hour in twelve), or COP for continuous duty. Two quotes that look identical can sit on different lines of the same standard. That's not a trick, it's a spec you have to read.

Fuel is the underrated number. A 350 kVA diesel set burns roughly 50 L/h at three-quarter load. A standard 990-liter base tank gets you 18 to 20 hours. If your outage history includes any multi-day events — and in most regions it does — your real specification is "how many liters," not "how many kVA." We ask that first now.

NFPA 110 treats Level 1 emergency power systems as life-safety equipment: monthly exercising under available load, annual load-bank testing, and 10-second transfer to the alternate source. A generator that has never been loaded isn't a backup generator — it's a diesel-fired space heater.

And the assembly matters: a UL 2200-listed generator set, a listed ATS, correct classification under NEC Articles 700/701/702, and emissions verified against the current EPA tier for your jurisdiction. Most units in this size class land under Tier 4 Final, but stationary emergency provisions vary by state and site — confirm with the dealer, not with a blog post. (That last sentence is me telling you to check. Please check.)

Power strip vs surge protector is an argument about the wrong layer

Spend five minutes in any forum and you'll find people comparing power strip vs surge protector joule ratings, as if the answer to "will our operation survive storm season" lives in a $30 outlet bar. That's the surface illusion: from the outside, surge protection looks like a strip you plug in. The reality is a layered system — a UL 1449-listed SPD at the service entrance, coordination at the panel, and ride-through capacity for the loads that actually care. A power strip is a branch-circuit convenience. A surge protector is a clamping device with a finite energy rating, and both numbers that matter are printed on the box. Neither one protects a three-phase 480 V service. The surge that welds your compressor contactors shut doesn't come through the outlet by the coffee maker.

Facilities people who've been through one bad season get this right. Everyone else argues about strips.

The objection I keep hearing: "the other quote was $14,000 less"

Yes. It usually is.

In 2021 we lost a $210,000 order to a bid that was about 9% cheaper on the equipment line. Sixteen months later that client's unit had been down three times, each incident waiting on a service tech four hundred kilometers away, because the winning bidder had no regional coverage. Their production losses across those three events ran somewhere near $190,000 — they told me the number themselves when they called us back. We now put service response time into every quote as a line item, and yes, that makes us more expensive on paper.

Total cost of ownership for standby power is base price plus freight, rigging, ATS, permits, commissioning, annual load bank testing, scheduled service, and expected downtime cost per hour times hours of exposure. That last term decides the deal, and it never appears on the bid tab. It's also why I tell people to stop shopping the sdmo-generator catalog on kW price alone — you're comparing the wrong column.

What certainty actually looks like on paper

When I'm triaging a rush order, I care about three things in this order: how many hours are left, whether the configuration can physically ship in that window, and what the worst case looks like if it doesn't. Price is fourth. For a planned purchase rather than an emergency, I tell people to get these in writing:

  • A commissioning date with a penalty clause — not an "estimated ship date."
  • A guaranteed service response window in hours, plus the name of whoever actually holds the spare parts.
  • Evidence the unit was load-bank tested at the factory and will be tested again on site.
  • Fuel autonomy stated in hours at a defined load, not tank capacity in liters.

That list is what the premium pays for. The SDMO generator range, backed by the Kohler engineering relationship, is one reason we quote it — the units hold up and the parts network is real. But the brand isn't the argument. The argument is that an outage costs more per hour than a generator costs per year.

So no, I don't think you should buy the cheapest option. I think you should buy the one whose failure mode you've already thought through. If your plan for a 30-hour outage is "we'll figure it out," you've already bought the cheaper unit. You just haven't received the invoice yet.

Marta Kowalska

Marta Kowalska

Marta Kowalska is an electrical equipment procurement and quality analyst covering distribution gear, automation, generators, pumps, test instruments, and engineered assemblies. She reviews IEC 61439 routine-verification records for assembly current and temperature rise, IEC 60034-1 rating and test data for rotating machines, and ISO/IEC 17025 laboratory scopes for reported calibration uncertainty. She helps buyers compare inspection status, document traceability, total installed cost, lead time, service access, spare-parts support, warranty boundaries, and failure consequences across equipment categories.