A 30 % Undercharge Steals 15 % of Your Cooling. Nothing in Your House Tells You.

Refrigerant is not consumed. A sealed system that was charged correctly at installation should still hold that charge years later. When it doesn't, something is wrong — a leak, or a charge that was never right in the first place.

The Department of Energy's 2018 review of field research found refrigerant charge faults in a wide band of installed systems: 29–78 % undercharged and 4–50 % overcharged, varying by study and region. So the question isn't whether charge faults are common. It's what they cost you.

The measured penalty

NIST answered this in Technical Note 1848. Working with a residential heat pump in a laboratory and then modelling a full house around it, Domanski, Henderson and Payne quantified each fault in isolation.

Two numbers there, and they describe different injuries. Capacity down 15 % means the system moves less heat — it cannot cool the house as fast. COP down 12 % means it gets less cooling per unit of electricity — you pay more for what it does manage.

Overcharge behaves differently. NIST found a 30 % overcharge "produced little reductions or small increases in capacity with 6 % greater total power." Less dramatic on paper, but it raises discharge pressure, which is its own kind of stress on a compressor.

The finding that matters more

Buried in the same report is something more important than any single percentage. NIST compared its own results against two other test programmes at similar fault levels:

• Southern California Edison measured a 54 % drop in cooling capacity at 27 % undercharge.
• Kim et al. (2006) measured a 17 % drop at 30 % undercharge — a slightly worse fault producing a third of the damage.

These large differences in cooling capacity change for a similar fault level exemplify differences in the effect a given fault may have on different systems.

NIST Technical Note 1848

Read that again, because it undoes a very common piece of advice. There is no universal table that converts "your system is a bit low on refrigerant" into "you are losing X %." The same fault, at the same severity, produced a 17 % penalty on one system and a 54 % penalty on another. NIST suggests differences in internal volume as a possible factor.

So what can you actually do with that?

You cannot look up your loss in a chart. That leaves two options. You can have someone put gauges on the system periodically and infer condition from a single moment. Or you can measure the system continuously and watch for the change — because whatever the absolute numbers mean on your particular unit, a drift away from that unit's own established normal is meaningful.

The second approach has an advantage the first cannot match: it does not require knowing what the right answer is in advance. It only requires noticing when today does not look like last month.

The occupant makes it worse

One more finding from the NIST work, and it is the most human part of the report. Installation faults can leave a house humid. People respond to a humid house by lowering the thermostat, because damp air at 76° feels worse than dry air at 76°.

In Houston, NIST modelled exactly that: dropping the cooling setpoint by 1.1 °C (2.0 °F) to chase comfort increased annual energy use by 20 % — and the report notes the effect is not linear, so it gets worse the further you push it.

That is a fault you cannot see causing a behaviour that feels reasonable, producing a bill nobody connects back to the original problem. Vetralis exists to break that chain at the first link, by watching refrigerant pressure and line temperature and telling you when they drift.