Repair, Replace, or Just Clean It: How I Sort Failing HVAC Equipment Before Spending a Dollar

Our operations lead asked me the same question in February that she asks every year: the unit on the third floor is blowing lukewarm air — do we fix it or replace it?

My honest answer is that it depends. But not in the hand-wavy way that sentence usually means. After six years of managing a $340,000 annual HVAC and refrigeration budget for a 200-person facilities services company, I've stopped treating "repair or replace" as one decision. It's three decisions wearing the same coat, and each one has its own cost logic.

Here's the framework I actually use. It's not elegant. It has saved us somewhere around $40,000 in avoided replacements, and it has also talked me out of two upgrades I thought were obvious wins.

The Three Buckets, and Why the Label Matters

Every service call I've logged over the past six years — I want to say several hundred, though I haven't counted precisely — lands in one of three places:

• The machine is fine; the airflow isn't. Filters, coils, blower wheels, condensate drains.
• A component failed; the chassis is healthy. Fan motors, capacitors, contactors, control boards.
• The unit is at end of life, or it was the wrong unit for the job from day one.

Roughly a third of the calls we used to dispatch a technician for turned out to be bucket one. That's not a knock on our techs. It's that "not cooling" sounds like a mechanical failure and usually isn't.

From the outside, a failing unit looks like a single problem with a single price tag. The reality is that the price of "fixing it" swings by a factor of ten depending on which bucket you're in — and the only way to know which bucket you're in is to rule out the cheap one first.

Scenario A: You're Cleaning, Not Fixing

If the unit runs, the compressor engages, and the air coming out is merely weak rather than warm, start here. Every time.

The air filter is the cheapest diagnostic you own

I've had two calls where the entire complaint was a filter that hadn't been changed since the previous tenant. A clogged filter starves the coil of airflow, the coil ices up, and the system behaves exactly like a refrigerant problem. Same symptoms, $400 difference.

One thing I got wrong early on: assuming higher MERV is always better. MERV ratings come from ASHRAE Standard 52.2, and the number describes particle capture — not pressure drop. A MERV 13 filter in a system designed around MERV 8 can choke airflow badly enough to cost you more in compressor wear than the filtration is worth. We run MERV 8 in most of our package units and MERV 11 only where the blower was sized for it. Verify what your air handler is rated for before you upgrade anything.

How to clean an evaporator coil without making it worse

If the filter was clean and the coil still looks matted, you're in coil territory. The mistake I see — and made — is treating the indoor coil like the outdoor one. The evaporator coil sits above a drain pan and near ductboard; it does not enjoy a pressure washer.

The process that has worked for us:

• Kill power at the disconnect, not just the thermostat.
• Pull the access panel and look at the fin face before touching it. If the fins are crushed, cleaning won't fix bypass.
• Use a no-rinse coil cleaner formulated for aluminum. Alkaline-heavy cleaners will eat the fins if you don't flush them completely, and flushing completely means water in a pan that may not drain well.
• Spray low pressure, following the fin direction, from the clean side out when you can reach it.
• Straighten bent fins with a fin comb before you button it up. Cheap tool, 20 minutes, measurable airflow recovery.

If the coil is more than lightly soiled, or if you can't reach the back face, this stops being a maintenance task and becomes a service call. Coil cleaning done badly is worse than coil cleaning deferred.

Cost reality: a filter change is $12 to $40 in materials. A coil wash by our contractor runs $180 to $260 in our region as of my Q1 2025 invoices. A compressor replacement on the same unit starts around $1,400. Rule out the cheap one first. Always.

Scenario B: A Part Died, and the Frame Didn't

This is where the total cost of ownership conversation actually earns its keep, and also where I've overspent by being too clever.

When an AC fan motor fails, you get a very specific signature: the compressor hums, the outdoor fan doesn't spin, the unit trips on high pressure within a few minutes. The machine isn't broken. A $200–$500 component is.

The real decision inside that decision is motor type. A permanent split capacitor (PSC) motor is what most older units shipped with — cheap to buy, less efficient to run. An electronically commutated motor (ECM) costs two to three times more up front and draws meaningfully less power at part load.

Here's where I nearly got this wrong. In 2023 I built a spreadsheet comparing a $185 PSC replacement against a $430 ECM swap on one of our rooftop units. Running the numbers at our blended commercial rate and roughly 2,000 fan operating hours a year, the ECM saved something like 25% of motor energy — which came out to about $30 a year. Payback was over eight years on a unit with maybe three years of life left in it. The spreadsheet said no.

But the spreadsheet was answering the wrong question. When the motor has already failed, the PSC price isn't the baseline — the ECM is the alternative you're choosing against, and the true comparison is $245 of incremental cost against a decade of lower run hours on the next unit. We went with PSC on the dying unit and specified ECM on the two replacements we bought that year. That's the whole point of TCO thinking: it tells you what to do, not what to spend.

I should add that motor sizing matters more than motor type. A replacement motor with the wrong horsepower or the wrong speed rating will fail early regardless of how efficient it is. Get the nameplate data. Don't let anyone eyeball it.

Scenario C: The Unit Is Tired, and the Sale Tag Is Distracting You

Now the expensive bucket. Two rules I use here, both learned the hard way.

Rule one: the nominal BTU number is not the cooling number

When our reception-area buildout needed spot cooling in 2024, I priced the Midea 9000 BTU portable air conditioner against two competing units. All three spec sheets said roughly 9,000 BTU. None of them cool like 9,000 BTU of central capacity, and the reason is regulatory, not deceptive: the DOE's portable air conditioner test procedure, effective 2016, produces a Seasonally Adjusted Cooling Capacity (SACC) figure that accounts for duct heat gain and the fact that a single-hose portable unit pulls conditioned air out of the room it's trying to cool.

That last part is the thing nobody tells you at the point of sale. A single-hose portable creates negative pressure in the room and draws hot outdoor air in through every gap in the envelope. SACC typically lands well below the nominal BTU figure, and dual-hose models recover much of the difference. If the room is small and the run time is short, it doesn't matter. If you're cooling against a real load for eight hours a day, it matters enormously.

The Midea unit was fine — good control interface, serviceable filtration. My point isn't the brand. My point is that I compared three products on a spec number none of them were actually rated on.

Rule two: "on sale" is not a cost analysis

This one still stings. When I see Midea dehumidifiers on sale, or any dehumidifier on sale, I now check the Integrated Energy Factor before I look at the price — and I check the test condition, because the labels aren't comparable across years.

The DOE changed the dehumidifier test condition from 80°F/60% RH to 65°F/60% RH in June 2019. A unit rated 50 pints under the old standard and a unit rated 50 pints under the new one are not the same machine. The new number is smaller for identical performance, which means old-stock listings showing generous pint ratings aren't lying — they're answering a different question. I once nearly bought four units off a clearance list on the strength of a pint number that would have been roughly 30% lower under the current test.

That was a gut-versus-data moment. The price per pint looked unbeatable. Something felt off about how the listing had no IEF rating at all. I passed, and later confirmed with the manufacturer that the model was discontinued and the spec sheet predated the 2019 rule change.

One more timing note worth flagging: under the EPA's AIM Act, new residential and light commercial AC equipment sold in the US as of January 1, 2025 must use refrigerants below a 700 GWP threshold. If you're buying new equipment this year, ask what refrigerant it uses and whether your service contractor stocks it. Buying into a refrigerant that's phasing down is a TCO problem disguised as a purchase.

How to Tell Which Bucket You're Actually In

Three questions, in this order. Stop at the first one that gives you a clear answer.

1. Does the unit run but underperform? If yes, you're in Scenario A until proven otherwise. Change the filter, inspect the coil, check the blower wheel. Do this before you call anyone. The cost of being wrong is $20.

2. Did one identifiable component fail while the rest of the system tests normal? If yes, Scenario B. Get the nameplate data, get three quotes, and compare against the remaining service life of the unit — not against the price of a new unit. A $400 repair on a unit with five years left is not a $400 decision against a $6,000 replacement. It's a $400 decision against $1,200 of prorated remaining value.

3. Is the repair cost more than half the replacement cost, or is the unit past 12–15 years for a packaged system? If yes, you're in Scenario C, and the sale price is now the least important number on the page. Look at efficiency rating, refrigerant, service access, and whether your contractor can get parts in under a week.

If your answers land between buckets, err toward the cheaper investigation. You can always spend more later. You can't un-spend on a replacement that was never necessary.

Serviceable, most of the time. Expensive, some of the time. The sorting is what keeps those two categories from trading places.