Why Air in Steam Is Worse Than Air in Water

Why this matters

Every tech who has bled a hot water coil has a working model of air in a heating system: it collects at the top, it makes noise, you open a vent and it is gone. That model is right for water and actively misleading for steam, because the two fluids do completely different things with the same gas. Air in water blocks a flow path you can find. Air in steam degrades the temperature the system delivers while every gauge keeps reading normally, with no gurgle to point at it. Shops lose whole seasons of coil output to this and bill it as fouling.

This article runs one gate against two systems and lets it resolve opposite ways.

The gate

Put the same volume of air into the system, and ask what it displaces.

In water, air displaces liquid. It occupies volume that flow would otherwise use, and everything outside that volume is unaffected.

In steam, air displaces the working fluid's share of the pressure. It does not simply take up space; it changes what the pressure in that space means, and it does so everywhere it is present rather than only where it has collected.

That single difference generates everything below.

Outcome one: the hot water coil

A hot water coil takes a pocket of air at its high point. The pocket occupies part of the header, flow through the affected circuits drops or stops, and the coil goes cold in the region the pocket blocks. Three properties make this an easy fault.

It is local. The circuits that still have water in them run at full water temperature, because the air has done nothing to the water. Supply at 180 F is still 180 F.

It announces itself. Air moving through a water circuit gurgles, and a blocked circuit shows up as a temperature difference you can feel across the face of the coil.

It is removable at the point of the fault. A manual vent at the high point solves it. Open that vent slowly with the discharge directed into a container and away from your body and face, wearing heat-rated gloves; a loop at 180 F scalds on contact and a vent left alone for years may come out with the plug rather than open cleanly.

Bound the whole thing: a pocket occupying a tenth of the coil's water volume costs you roughly the circuits it sits in, and the rest of the coil performs.

Outcome two: the steam coil

Put the same air fraction into a steam coil and none of those three properties holds.

It is not local in the same way. Air mixed with steam travels with it, and every square foot of surface the mixture touches is affected.

It does not announce itself. There is no flow to interrupt and nothing to gurgle. The pressure gauge reads exactly what it read before, which is the dangerous part, because on a steam system pressure is how everyone reads temperature.

It is not removable where the fault appears. The air has to be pushed to a venting point, which means that point has to have existed since installation.

And the mechanism splits in two, which is where the numbers stop matching anyone's intuition.

Mechanism one: the gauge stops telling the truth about temperature

The whole of steam practice rests on the pressure and the temperature being locked together, so that a gauge reading is a temperature reading. Air breaks that link, because pressure in a mixture is shared.

The total absolute pressure is the sum of what each gas contributes. If air makes up a fraction of the volume, the steam's share of the pressure is the total absolute pressure multiplied by one minus that fraction, and the temperature in the space is the saturation temperature for the steam's partial pressure, not for the total the gauge shows.

Work it on a coil at 15 psig, which is 29.7 psia. With ten percent air by volume, the steam's partial pressure is 0.90 times 29.7, or about 26.7 psia. Reading a steam table at 26.7 psia gives about 244 F, against about 250 F for the pure-steam case at the same gauge reading. The coil is delivering about 6 F less than the gauge implies.

Now put that in proportion. Against room air entering the coil at 60 F, the coil's driving temperature difference has gone from about 190 F to about 184 F, a loss of roughly three percent. That is real and it is not what wrecks a coil's output.

Do not stop here. The most common analytical mistake with this subject is to run exactly the calculation above, find a few degrees, and conclude that air is not the problem. The bulk depression is small. The mechanism that costs a quarter of a coil's output is the next one, and this calculation gives no hint of its size.

One condition attaches to the arithmetic: it assumes the air is uniformly mixed through the space at that volume fraction. That is the assumption the next section demolishes.

Mechanism two: condensation concentrates the gas where it hurts

Here is the structural difference between the two fluids, and it is the load-bearing point of this article.

Condensation is a pump that concentrates non-condensable gas at the heat transfer surface. Steam arrives at the cold tube wall and condenses, leaving the volume. Whatever air came with it does not condense, so it stays. The next parcel of steam arrives, condenses, and leaves its air behind too. The air fraction in the thin layer against the wall climbs far above the fraction in the bulk, and it keeps climbing as long as the surface is condensing.

Nothing in a water loop does this. Water flowing past a wall does not disappear at the wall, so there is no process sweeping air toward it and holding it there. Air in water is carried along and ends up wherever buoyancy and geometry put it, which is a high point, which is why it is easy to find.

The layer that builds up costs you twice.

It conducts badly. Air's thermal conductivity is on the order of 0.017 Btu per hour per foot per degree F near 200 F, against roughly 0.39 for water and roughly 30 for steel. The 0.015 a lot of people carry is air at room temperature, and conductivity climbs with temperature, so it belongs to a different condition. Per unit of thickness, still air resists about 22 times water and about 1,700 times steel. So a stagnant layer a thousandth of an inch thick resists like about 22 thousandths of water or 1.7 inches of steel. Hold the conditions on those ratios: pure conduction through still air near 200 F, no convection inside the layer, so they describe a genuinely stagnant one.

It starves the surface of steam. Every pound of steam that wants to condense has to diffuse through the layer to reach the wall, and the steam's partial pressure right at the wall is far below the bulk value the calculation above used. The local saturation temperature at the surface is correspondingly lower than 244 F, and unlike the bulk figure it is not something you can compute from a gauge reading.

Both penalties run the same direction and they compound, which is why a coil with a few percent of air in it can lose output out of all proportion to that percentage.

The regime that produces the really large gaps

There is a second regime and it is worth naming, because it is where the biggest field numbers come from and it behaves differently from the mixture case above.

When air is not mixed but segregated into a pocket, at the far end of a main or the remote header of a coil, the mixture arithmetic no longer applies at all. In a pocket there is essentially no steam, so there is no condensation, and the metal there is not being heated by anything. Its temperature falls toward whatever is happening on the other side of the wall, which for a coil is the air stream and for a main is the room.

That is why the sibling article on what an air vent is doing in a steam system reports a 70 F gap between the gauge's saturation temperature and the metal at a coil's return header, on a system whose bulk partial-pressure depression would only have been a few degrees. Both numbers are correct and they belong to different regimes. A few degrees is what mixed air does; tens of degrees is what a pocket does, and a pocket is what forms wherever the air has nowhere to go.

One more direction correction while you are placing a pocket. Air's molecular weight is about 29 against water vapour's 18, so in a steam space air is roughly 1.6 times as dense as the steam and it does not float to the top the way it does in a water loop. The advancing steam front pushes it to the point most remote from the inlet, and that is where it settles.

Worked example: running the gate on one air handler

An air handler carries a hot water preheat coil and a steam reheat coil, fed from the same plant, both complained about in the same week.

The water coil. Supply water 180 F, return 160 F, and the bottom third of the coil face reading cold to a probe while the top two thirds read normal. Air-side rise across the coil down about a third from its logged figure. The high-point vent was opened slowly into a container held to the side, producing air for several seconds and then water, and the face evened out inside a minute. One visit, no parts. The water that had been flowing was always at 180 F; the fault was that some of it was not flowing.

The steam coil. Gauge at the coil reading 15 psig, so about 250 F from the steam table. Air-side rise down about a quarter. Nothing audible, nothing visibly wrong, no vent to open. A remote header with no vent is a finding to write up, not a fitting to crack. Do not loosen a plug, union or gauge tapping on a live coil to release a pocket: behind the air is condensate at saturation, arriving at the opening as flash. Fitting a vent is a scheduled job, coil isolated, locked and tagged under 29 CFR 1910.147, drained and cooled below 120 F.

Running the mixture calculation on the assumption of ten percent air gives a delivered temperature around 244 F and a loss of about three percent of the driving temperature difference, which does not account for a quarter of the output. That result is the finding, not a dead end: the loss is not distributed, so it is segregated. A contact reading along the steam path confirmed it, with the inlet header near saturation and the remote header far below it.

Why the two coils resolved so differently. Identical gas, identical quantity, same air handler, same week. The water coil gave up its air at a vent in sixty seconds and lost nothing but flow. The steam coil could not give up its air at all, because the venting path had to have been designed in, and it lost temperature at the surface rather than flow through the tube. That asymmetry is the entire practical content of this subject.

The failure mode if the gate is not run. The steam coil gets cleaned, because a quarter of the output missing with no fault visible reads as fouling. Cleaning changes nothing, and the second visit quotes a coil. The tell that would have prevented both is free: on the water coil, a temperature difference across the face; on the steam coil, a temperature difference along the steam path with the gauge reading normal.

What this changes about how you look for it

Stop looking for air in a steam system and start looking for metal colder than the gauge says it should be. That is the fault's only symptom, and it is a two-instrument test: a pressure gauge, a steam table, and a contact temperature reading taken at the point most remote from the steam inlet. Nothing on the steam side of this test is judged by hand. Seat the probe on metal that is already bare, in gloves rated for that surface and with eye and face protection under 29 CFR 1910.132, and do not lift or brush lagging to reach metal, because thermal system insulation on steam piping is presumed asbestos-containing under 29 CFR 1926.1101 and 29 CFR 1910.1001 until sampled. Use a contact probe or high-emissivity tape rather than infrared on bare steel, which reads low and would manufacture this same signature on a healthy coil.

And treat air's arrival as a system property rather than an event. It comes in dissolved in the feedwater, through packing and joints anywhere the system runs below atmospheric, and through every shutdown, when steam in a closed space condenses and pulls in whatever the vacuum breaker admits. Removing it at the boiler belongs to the water treatment and deaeration articles, the corrosion it causes to the condensate chemistry article. What belongs to the field tech is the venting path, and whether one exists at every dead end.

References

• ASME Steam Tables, or an equivalent saturated-steam table, for saturation temperature at the steam's partial pressure as well as at the total pressure
• Published thermal conductivity data for air, water and carbon steel at service temperature, for the film resistance comparison
• See related: articles in this library on what an air vent is doing in a steam system, on why a condensate line is a chemistry problem too, and on the water treatment and deaeration coverage that owns feedwater gas removal