Moisture in the aircon circuit: what the install seals in
A refrigerant circuit is built to hold refrigerant and oil and nothing else. Water vapour sits in the air around every open pipe end, and an install is when those ends stand open. What gets shut inside that day stays inside.
By Team Snowflake | Updated 7 Aug 2026
Every open pipe end is an entry point
Refrigerant and oil are the only two things a circuit is designed to carry. Anything else that ends up inside arrived through an opening, and almost every opening in the life of a system belongs to installation day. That is when copper gets cut, shaped, joined and left standing before the two units are finally tied together.
Air fills that copper until refrigerant replaces it, and air here carries a great deal of water. Warm, damp air is the standing condition in Singapore rather than a seasonal one. There is no drier month to schedule a job into. A pipe end open on a ledge in the afternoon meets much the same load in any week of the year.
Manufacturers put the protection in writing rather than leaving it to site judgement. Daikin's installer reference guide for its split systems is direct about it. The installer is told to keep anything but the designated refrigerant out of the refrigerant cycle. The piping is to be protected so that dirt, liquid and dust cannot get into it. The methods named are a pinched or a taped pipe end, and the indoor side gets protected regardless of the wait.
The shortfall is rarely a decision anybody makes out loud. Sealing every end costs a minute and buys nothing anyone can see, while the schedule is what an installer gets judged on. Where the pipe is run on one visit and the units are hung on another, the run sits open in between. Nobody arriving on the second day can tell what it collected on the first.
The gap between visits is where most of it happens
A single-visit install keeps the window short. Copper gets cut, shaped and joined, and the circuit is closed and dried out the same afternoon. What stands open stands open briefly, and under somebody's eye the whole time.
Renovation work stretches that window without anyone intending it. The pipe goes in with the carpentry, a ceiling closes over it, and the units arrive when the flat is nearly finished. Plastering, screed drying and rain reaching an unfinished ledge all land inside that interval, and the copper is standing open through every part of it.
The acid comes out of the oil more than the refrigerant
Most of the acid formed in a damp circuit comes out of the lubricant rather than out of the refrigerant. That is worth stating plainly, because the usual telling puts refrigerant and water together and stops there. Sporlan's filter-drier bulletin is more careful about it, and the care is the useful part.
Refrigerant and water do react, and what they produce is corrosive. The bulletin calls that reaction hydrolysis and names hydrochloric and hydrofluoric acids as its products, present in the system as a gas and highly corrosive. It also says that in ordinary usage the reaction is negligible, and that meaningful hydrolysis needs a very wet system running at abnormally high temperatures. So the refrigerant route is real, and it is not the everyday one.
The oil route is the everyday one. Current systems run on polyolester lubricant, and the same bulletin names organic acid from lubricant breakdown as another significant source of acidity, formed when polyolester reacts with moisture. It goes further and describes water as always present in refrigeration systems, especially where that hygroscopic lubricant is used. Hygroscopic means the oil draws water in and holds on to it instead of letting it settle out anywhere.
That property is what spreads the problem around. Water dissolved into the oil travels wherever the oil travels, which is every part of the circuit on every run. Nothing about it settles out at a low spot, and nothing about it picks one component to sit against. The bulletin's own summary of the damage names acids, sludge, copper plating and corrosion, and adds that moisture does harm even where no freeze-up ever occurs.
This is not the acid a burnout leaves behind
Two different stories end in acid and they should not be run together. A burnt-out motor destroys its own insulation and the oil around it, and acid contamination then arrives all at once, after a failure everybody can point at. The circuit is dirty because something died inside it.
Moisture works the other way round. Acid forms slowly in a system that has never failed at all, out of water that entered before the machine ever ran. Nothing has broken. What is happening instead is a circuit ageing faster than its design allowed for, and the compressor's own wire is one of the things ageing in it. What acid does to that wire belongs to winding insulation and is set out there.
Ice at the one place the circuit runs cold enough
Free water in the circuit can freeze, and it freezes at the metering stage. Refrigerant is dropped from high pressure to low there by design, and the temperature drops with it. That leaves one point in a running system both cold enough and tight enough for a plug of ice to matter. Sporlan gives protection of the expansion valve against freeze-up as a reason for the liquid line position.
What the room gets from that is cooling which comes and goes. Ice builds at the narrow point, flow drops off, and the room stops getting cold. The obstruction then warms through, the ice clears, and cooling returns as though nothing had happened. Nothing is broken at either end of that cycle, and a technician arriving between episodes meets a system behaving perfectly.
The intermittent shape is the detail worth holding on to. Solid debris lodged at the same point neither clears itself nor comes back on its own. A circuit restriction from scale or a loaded canister behaves steadily. A freeze behaves like a fault with a mind of its own, and neither one is a shortage of refrigerant.
Two neighbouring subjects sit either side of this. The metering device is where the freeze happens and carries its own page. Coil icing is a different event with a similar name, forming on the outside of a coil out of room air rather than inside the tubing out of water sealed in. Both get reported as the aircon icing up, and that shared description is what confuses them.
Timing is what gives this one away
A freeze needs the metering point to get cold enough for ice, and that condition is not constant. It tends to arrive on a long afternoon run and to stay away on a short evening cycle. An owner reporting that the flat is fine at night and gives up by mid-afternoon is describing something an attending technician may never witness.
Write the pattern down before anybody attends. What time the cooling starts to go, how long the room stays warm, and whether switching off for a spell brings it back. Answers to those three do more for the diagnosis than a visit booked at whatever hour happened to be free.
Why does moisture damage take years to surface?
Moisture damage takes years to surface because none of its routes act quickly, and because the fastest of them clears itself. A circuit carrying water still cools normally at handover. Acid has had no time to form, and ice has had no run of hot afternoons to build against. The system passes the only test an owner can apply, which is whether the room gets cold.
That interval is the whole problem. By the time a compressor gives up or a metering device sticks, the install is years behind, the paperwork is gone, and whoever attends has no reason to ask about a day they were not present for. Age is the available explanation. Age is also usually the right one, which is precisely why this cause is so rarely reached for.
Broken attribution then protects the practice that caused it. A shortcut whose consequence lands the same day gets corrected quickly, because the person who took it is still standing there. A shortcut whose consequence lands long after the relationship has ended never gets fed back to anybody. None of that requires bad intent. It only requires the feedback loop to be missing, and it is.
The table below sets each late symptom against its usual explanation, and against the version moisture would produce.
| What surfaces later | What it usually gets put down to | What a wet circuit would look like |
|---|---|---|
| What surfaces laterCooling that fades and returns on no pattern | What it usually gets put down toA gas leak, or a hot spell | What a wet circuit would look likeIce forming and clearing at the metering point |
| What surfaces laterA compressor gone earlier than the unit's age suggests | What it usually gets put down toA bad batch, or plain bad luck | What a wet circuit would look likeInsulation worn by acid as well as by heat |
| What surfaces laterA canister that has turned into an obstruction itself | What it usually gets put down toA blockage nobody traces to a cause | What a wet circuit would look likeDesiccant loaded faster than the design assumed |
| What surfaces laterDiscolouration or plating on the inside of the copper | What it usually gets put down toNothing, since nobody ever sees it | What a wet circuit would look likeOne of the effects moisture is specifically named for |
| What surfaces laterA second compressor failing the way the first one did | What it usually gets put down toA poor replacement part | What a wet circuit would look likeA circuit never dried out between the two |
Moisture is one route in, not the only one
Overstating this would be its own mistake. Compressors fail from heat, from starting stress, from electrical faults upstream, and from ordinary wear at the end of a working life. Circuits obstruct because of scale left by brazing, because of debris from a failure elsewhere, and because a canister reached its capacity. This belongs on that list rather than at the top of it.
What earns it a place is that it is preventable and that almost nobody checks. Every other item on the list has a symptom somebody can go looking for. Water shut into a circuit at install has no symptom whatsoever until it has already done what it does, and by then the only moment it could have been dealt with is long gone.
What to ask about the days the pipework stood open
The step that removes water is evacuation, and it happens at the end of the install just before the gas goes in. Pulling a vacuum is where that step belongs, along with what a straight answer about it sounds like. This page stops short of it on purpose, because the question underneath is a different one. How much there was to remove got decided long before the pump was ever connected.
Whatever the vacuum leaves behind falls to the filter drier. It holds residual water as refrigerant passes through it, and its appetite for the job is finite. A run that stood open through a renovation hands that part more to deal with than a run closed the same afternoon. What the component is, and why a quote carries a separate item for it, is set out elsewhere.
Exposure is the one piece an owner can still find out about, and almost nobody raises it. Ask about the sequence rather than about the standard. An installer who ran the pipe and charged the system on one visit has a short answer ready. One who piped a flat during renovation and came back later has a longer one, and the length of the gap is not the interesting part of it.
None of the questions below need refrigeration knowledge to put, and none of them are traps. Each one is settled by a fact that whoever did the work either remembers or does not. An installer who works this way will say so without hesitation, because sealing a pipe end is ordinary practice rather than an extra.
- Whether the piping and the charging happened on one visit, or on visits with a wait in between
- What was done to the open ends while the run was waiting for its units to arrive
- Whether the run was ever rained on, or sat through plastering and screed work on the same floor
- Whether existing copper was reused, and what that pipe had been standing open to beforehand
- Whether a new drier went in on any job that reopened a circuit already in service
If the install is already behind you
Nothing about the exposure can be measured after the fact. Saying so beats selling a check that does not exist. Once refrigerant is in the pipework, no gauge reading taken outside the machine says how wet the circuit was when it was closed.
What remains available is inference from the state the circuit is in now. Compressor oil can be tested for acid, which describes where things stand rather than how they got there. On a young system that has never faulted, that answers a question nobody has thought to ask. On one that has already lost a compressor, it is the difference between replacing a part and understanding why the part died.
Ready to get started?
Tell us what’s going on. Symptoms, setup, photos, anything we should know. We’ll assess and come back with the right next step.