Air in the System Is Probably Not Your Problem

Non-condensables — air and moisture that got into a sealed refrigerant circuit, usually during a service that skipped a proper evacuation — are a genuine fault. They raise head pressure, because the air occupies condenser volume that should be condensing refrigerant, and they are diagnosed by comparing measured head pressure against the saturation pressure for the measured condensing temperature.

They are also blamed for a lot of things they do not cause, and it is worth knowing what the measurements say.

What DOE found

A field and laboratory study characterising fault prevalence in residential comfort systems, published under DOE's Building America programme, reports that air alone has quite minor impacts on performance.

That is a direct measurement, and it is worth holding against how often non-condensables come up as an explanation. Prevalence is low too — the NREL and Purdue fault prioritization work estimates non-condensables in about 3% of systems, against 72% carrying a nonstandard charge and 44% with a fouled condenser.

Why we retired the alert

We used to raise a non-condensables warning with an estimated running-cost band attached. We removed the cost figure, because the DOE finding does not support pricing it as an energy fault — and an alert whose stated impact is not defensible is worse than no alert.

The detection itself was never the weak part. What was weak was implying the fault was expensive when the measured effect on performance is small.

What still matters about it

Two things, neither of them the energy bill.

First, air in the system is evidence about process. It means the circuit was opened and not properly evacuated. Whatever else that service visit did is now worth a second look, and the moisture that came in with the air is a durability question rather than an efficiency one.

Second, it is a differential diagnosis. A system with high head pressure has several possible explanations — a dirty condenser, an overcharge, a failing fan, restricted airflow, or non-condensables — and they have different fixes, two of which are opposites. Knowing that air is present, or that it is not, narrows that list.

The general lesson

It is easy for a monitoring product to accumulate alerts, because every additional fault type looks like added value. It is not, if the fault is rare and its measured impact is small. The list should be short, and every item on it should be able to answer why it is worth someone's afternoon.