April 2024. I'm staring at invoice #21847 for $13,747.29. That's what it cost to "save" a 12-year-old McQuay water source heat pump that was worth maybe $4,000 on a good day.
And the worst part? It started with an $8,500 repair bill I approved without asking the right questions.
Quick background: I manage HVAC maintenance budgets for five commercial buildings in the Boston area—roughly $40,000 a year across everything from filters and belts to full chiller overhauls. I've been in this role for seven years. I'm not a mechanical engineer. I'm the person who signs purchase orders and asks why the line items cost what they cost. This time, I signed first and asked later.
The Problem I Thought I Was Solving
December 2023. A tenant on the second floor calls: no heat. Our in-house tech checks the unit, a McQuay water source heat pump, and reports the compressor won't start. The starter capacitor is fried. This isn't a capacitor-bulge-and-move-on situation though—the oil test shows moisture contamination. The compressor is done.
We get quotes. Replacing the compressor: $8,500. Replacing the entire unit: $13,000–$14,000, installed.
I know what you're thinking. $8,500 beats $13,500 by a mile. That's what I thought too. To be fair, $8,500 < $13,500 is true. It's also a trap.
What I Missed Three Months Ago
Let me walk you through what actually happened, because if you're managing a building with aging HVAC equipment, this pattern will look familiar.
Three months after the compressor replacement, the same unit develops a refrigerant leak in the evaporator coil. Our tech, a guy I trust, lays it out straight: The leak is fixable, but the material is thin, and chasing a pinhole leak in a 12-year-old coil often becomes a game of whack-a-mole. He quoted $5,200 for the evaporator replacement. Or we could replace the whole unit for $13,700—which, by the way, was now higher than the December quote because of a January price increase.
Either way, the $8,500 repair I'd approved was now a sunk cost that didn't improve the unit's outlook. If I'd known in December that this was likely, the decision would have been obvious: replace the unit and skip the compressor repair entirely.
Here's the part that still annoys me. There were three signals that the repair was a bad idea. I saw two of them and ignored both.
Signal #1: R-22 refrigerant
This unit ran on R-22. As of January 1, 2020, the U.S. EPA no longer allows production or import of R-22 (Source: EPA, 40 CFR Part 82). Existing stockpiles are all that's left, and prices keep climbing. A compressor replacement on an R-22 system is a decision to keep a dying refrigerant strategy alive for another 5–10 years.
In 2023, we paid $180/lb for R-22 to top off the system after the compressor swap. In 2020, the same tech charged us $65/lb for the same refrigerant. That's not inflation. That's a controlled substance phase-down. (Prices from our invoices; verify current rates—they've kept climbing since.)
Signal #2: The parts problem in Boston
Anyone booking "McQuay HVAC replacement parts in Boston, MA" knows what I'm talking about. McQuay got absorbed into Daikin Applied years ago, and the supply chain for older McQuay parts has been tightening every year. The compressor itself was a Copeland scroll that wasn't hard to source, but that's the exception, not the rule. When I called around, two of the three suppliers I normally use said they'd have to order the evaporator coil with a 4–6 week lead time, no guarantees. That's not a service network. That's a scavenger hunt with a price tag.
Signal #3: The math I should have done
I keep a cost tracking spreadsheet (yeah, I'm that person). When I audited our 2023 spending, the pattern was right there: across our five buildings, every compressor failure on equipment older than 10 years ended up costing us—repair plus follow-up—more than a preemptive replacement would have. Not sometimes. Every single time.
The reason is pretty simple when you say it out loud: a compressor failure on an older unit is rarely just a compressor failure. It's usually the system telling you it's done. Moisture, contamination, worn valve plates, degraded motor insulation—these aren't isolated events. They compound.
The Real Cost
Let's do the final math, because this is the part I want you to feel.
- Compressor replacement: $8,500
- R-22 refrigerant top-off: $540
- Evaporator leak diagnosis and labor: $350
- Evaporator replacement quote: $5,200 (declined)
- Full unit replacement (final invoice): $13,747.29
- Total spent on the old unit before replacing it: $9,390
- Total avoided if we'd replaced the unit in December: $9,390
That's the "$11,000 mistake." I call it $11,000 because the $9,390 in wasted repairs, plus the emergency upcharge on the rushed replacement, plus the headache of a down unit in February, rounds to somewhere around eleven grand. (We had a tenant whose server room depends on that zone. I don't recommend explaining to an angry IT director why their cooling budget went up 40%.)
What I'd Do Instead (and What I Do Now)
I'm not going to tell you to never repair an older McQuay unit. That's not true for all situations.
But when a compressor fails, here's the quick test I run now, in this order:
- Check the refrigerant. If it's R-22, assume the repair has an expiration date. Budget for replacement within 24 months, and factor that into the repair decision.
- Ask what failed and why. A failed start capacitor is one thing. A failed compressor with contaminated oil is entirely different. The latter means the whole sealed system is compromised.
- Run the TCO comparison. Repair cost + likely follow-up repair (use 50% of the repair cost as a rough guess) + refrigerant refill + increased electricity, versus replacement cost. If the unit is over 10 years old, replacement usually wins.
I also stopped treating service manuals as optional reading. The McQuay water source heat pump manual for that model had a line about compressor contamination that would've taken me thirty seconds to find. It stated—plainly—that moisture in the loop usually means a system-level failure, not just a part failure. I had the manual. I didn't open it until after the second invoice.
While I'm on the subject of what we bought next: we ordered a Milwaukee air compressor for our shop (the old one finally died after eight years of daily abuse) and a Honeywell home thermostat for our front desk area. The thermostat was $89 and paid for itself in a month of smarter scheduling. And even something as basic as knowing which way to put air filter in furnace—arrow toward the blower, always—saves real money over a heating season. This isn't hypothetical: a backwards filter in our building cost about $300 in reduced airflow before anyone noticed. Respect the basics.
Where This Doesn't Apply
I want to be straight about the limits of what I'm saying. This TCO approach assumes you have time to make a call and the unit isn't in a mission-critical role.
If a McQuay chiller is serving a server room or a surgical suite, the calculus changes. In those situations, redundancy and immediate restoration beat cost optimization every time. Do the repair to get things running, then plan the replacement on your own schedule, not on an emergency timeline. That's the responsible move.
Also: I'm talking about water source heat pumps and compressors. Air-cooled chillers, rooftop units, and VRF systems have different failure curves and different parts situations. Don't take a framework built from five Boston-area buildings and force it onto a portfolio floor in Texas. Your mileage may vary. In my case, the mileage was measured in thousands of dollars.
The Bottom Line
I signed an $8,500 repair because it was "obviously" cheaper than a $13,000 replacement. It wasn't. The repair was an expensive detour that ended at the same destination, plus a $9,390 souvenir.
If you're looking at a failed compressor on a unit over ten years old, get the refrigerant type, the failure report, and the TCO in front of you before you approve anything. And if you can't find the manual, that's a signal too. It means your parts situation is going to get worse before it gets better.
The money is real. The math is usually clear. The hard part is slowing down long enough to do it.
