What Water Hammer Actually Is in a Steam Line

Why this matters

Water hammer in a domestic water line is a nuisance that loosens fittings. Water hammer in a steam line has parted flanges, launched valve bonnets and killed people, and the reason is not that steam pipes are weaker. It is that a steam line contains two phases with a volume ratio in the hundreds to one, so it can generate forces that have no equivalent in a water pipe and that no arrestor on a supply-house shelf is built to absorb. A tech who carries the plumbing model of hammer into a boiler room will reach for the wrong fix, and the wrong fix here is the one that removes the warning and leaves the event.

If it is hammering now, this comes first

A steam line that is banging repeatedly is telling you a joint is being loaded past its rating on every cycle. Get everyone out of the plane of the flanges, threaded fittings and the outboard side of every elbow, because that is where a failure discharges, and shut the line down before anyone diagnoses anything. Close the supply slowly, standing to the side of the valve and out of the line of the bonnet, and let the drips run. Then keep the standoff. A line hammers on cooldown as well, for the collapse reason set out below, so the plane of the flanges and the outboard side of the elbows stay clear until the line is cold and drained, not merely until it goes quiet. Fast closure adds a second transient to a line that is already producing them.

Do not trace a steam leak by hand or by feel at any point in this work. Steam at usable pressure is invisible for the first stretch out of a leak and it cuts. Look for the plume against a dark background from a distance, and if a leak location genuinely has to be pinpointed on a live line, that is an ultrasonic or thermal job done from outside the jet, not a walk with an open palm.

What it is not

Most of the useful content here is exclusions, because every one of them is a repair somebody has already sold.

It is not high pressure. Pressure is the driving force in one of the two mechanisms below, but a system hammering at 2 psig is common and often more violent than the same building at 15 psig. Lowering the pressure does not reliably stop it and sometimes makes it worse, for a reason given later in this article.

It is not a loose pipe. A loose pipe makes the noise louder, because the pressure event becomes a movement event and the pipe strikes the structure. Tightening the hangers removes the noise and leaves the pressure transient intact, which converts an audible warning into a silent load on the fittings. That is the single most damaging thing on this list and it gets done routinely because it works, in the sense that the complaint stops.

It is not the plumbing hammer that arrestors are for. The domestic version is a valve closing on a moving column of cold water, and the sibling articles in this library on water hammer control and plumbing noise cover it properly. An arrestor or an air chamber does not belong in a steam line: an air chamber is a dead leg that fills with condensate within days and then becomes a slug source of its own, a bladder-type arrestor is not built for saturation temperature, and neither one is positioned to absorb an event that originates in the middle of a pipe run rather than at a valve.

It is not thermal expansion. Expansion ticking and creaking rises and falls over minutes as metal moves against a guide or a hanger, and it happens on a line that is warming smoothly. Hammer is a single sharp report, or a burst of them, tied to an event.

It is not automatically a failed trap. A trap failed closed is often the enabler, because it lets condensate accumulate where the mechanism needs it, but replacing the trap on a line whose real problem is a sag or a reverse grade fixes nothing and buys the shop a callback.

Mechanism one: condensate driven as a slug

Steam moving down a main drags on any condensate lying in the bottom of the bore. At low velocity the water simply runs along the floor of the pipe. Raise the velocity and the drag lifts the surface into waves; raise it further, or let the depth build, and a wave touches the crown and seals the bore. At that instant the water stops being a film and becomes a piston with the full steam pressure behind it, and it accelerates until it reaches something that will not move: an elbow, a reducer, a closed valve, the tube sheet of a heat exchanger.

The pressure produced when a moving liquid column is stopped is given by the Joukowsky relation, and for cold water in a rigid steel pipe with a stop faster than the pipe's own pressure-wave return time it works out to roughly 60 psi for every foot per second of velocity change. Hold on to the conditions attached to that number: cold water, full bore, rigid pipe, effectively instantaneous stop. At 250 F the liquid is about six percent less dense, so the constant overstates slightly, and a real slug in a steam main neither fills the bore uniformly nor stops instantly, because it compresses the steam ahead of it and breaks up on impact. Use the figure as a bound on the event, never as a prediction of it.

Mechanism two: sudden condensation collapse

This is the one with no plumbing equivalent. When steam contacts water that is below saturation temperature, or a pipe wall that is below saturation temperature, it condenses, and a pound of steam that occupied about 26.8 cubic feet at atmospheric pressure becomes about 0.0167 cubic feet of water. That is a volume ratio of about 1,600 to 1 at 0 psig. The space the steam was occupying does not fill from somewhere else at the speed the condensation happens, so it becomes a near-vacuum, and the liquid on either side is driven into it by the full absolute pressure of the line.

  1. steam pocket trapped over subcooled water

     water .........(  steam pocket )......... water

  2. pocket condenses, volume falls by roughly
     1600 to 1 at atmospheric pressure

     water .........(  near vacuum  )......... water

  3. both columns accelerate inward and collide

     water .................>|<................ water
                          impact

The important part is that the accelerating pressure is absolute, not gauge. On a heating main at 0 psig the columns are being pushed together by 14.7 psi, and the void they are closing is large. On a 100 psig main they are pushed by 114.7 psi, but a pound of steam at that pressure occupies only about 3.9 cubic feet against 0.0177 for the water, a ratio of roughly 220 to 1, so the void is far smaller for the same mass condensed. One term grows with pressure and the other shrinks, which is why low-pressure heating systems are notorious for violent hammer and why dropping system pressure is not a reliable cure. Neither term alone predicts severity, and that is the practical reason this event is eliminated by removing its precondition rather than calculated and designed around.

Why the collapse is the more dangerous of the two

A slug impact is bounded by how fast the slug was moving. A collapse is bounded by nothing the field can see. The columns have a pressure difference of the full absolute line pressure across a void of unknown length, they accelerate over that length, and the velocity at collision is whatever the geometry gave them. Two features make it worse in practice. It happens where the pipe is coldest, which is the equipment end and the far end of a main, so the impact lands on tube sheets, control valves and coil headers rather than on straight pipe. And it needs subcooled condensate, which means it is most likely exactly where a system has been shut down, throttled back or allowed to flood, all of which are normal operating states rather than faults.

The everyday version is a heating coil or converter whose modulating valve has throttled the shell pressure below the return system's back pressure. Condensate stops leaving, floods the tubes, cools below saturation, and incoming steam collapses on the pool every time the valve opens. The sibling article on finding where hammer is being generated works that case in full.

Worked example: putting a number on one sag

A 4 in Schedule 40 main runs at 15 psig, so around 250 F, at a design velocity in the range dry mains are usually sized to, roughly 80 to 130 ft/s. A 12 ft section between two hangers has settled and holds condensate about half the bore deep after a weekend shutdown.

The bore of 4 in Schedule 40 is about 4.026 in, so the flow area is about 0.088 square feet. Twelve feet at half depth is about 0.53 cubic feet, and water at 250 F runs about 58.8 lb per cubic foot, so the pool weighs about 31 lb.

On Monday start-up, steam is admitted and that pool is picked up. Suppose the drag brings the sealed slug to 20 ft/s, which is well below the steam velocity around it. Applying the 60 psi per ft/s figure with its cold-water, full-bore, instantaneous-stop conditions gives an upper bound around 1,200 psi at the first elbow, against a line pressure of 15 psig. That is roughly eighty times the operating pressure, arriving in a few milliseconds, on a fitting that was selected for the operating pressure with a class rating nowhere near four figures.

The real event is smaller than the bound, because the slug is not a rigid column and the steam ahead of it cushions to some degree. It does not need to reach the bound. The failure mode is not a burst on the first bang; it is a threaded joint that works loose over a heating season and then weeps live steam into an occupied space, or a flange whose gasket is crushed unevenly and lets go on a cold morning six months later, at which point nobody connects it to a noise from the previous winter.

Notice what the arithmetic does not depend on: the pump, the boiler, the trap model, the burner. It depends on 12 ft of pipe not draining. Everything upstream can be correct.

Reading one bang for its mechanism

You cannot tell the two apart by loudness, and location is the sibling article's subject. The tell is the trigger and the temperature.

• A bang that happens on admission of steam to a cold or drained line, once or a few times, then quietens as the line comes up to temperature, is slug propulsion. Condensate that was lying in the pipe is being cleared.
• A bang that happens repeatedly during steady operation, especially at low load, and gets worse the more the load falls, is collapse. Something is holding subcooled water where steam can reach it, and the condition is being recreated on every cycle.
• A line that hammers while it is cooling down or after shutdown is collapse as well, and this one is often missed because nobody is present to hear it.

Where the two overlap, treat it as collapse until proven otherwise, because that is the branch where the load has no ceiling you can compute.

References

• ASME Steam Tables, or an equivalent saturated-steam table, for specific volumes and saturation temperatures at your operating pressures
• ASME B31.1 Power Piping, in the edition your jurisdiction or your contract adopts, which binds the piping installation, for drainage and support requirements on steam mains
• 29 CFR 1910.147, control of hazardous energy, for isolating a steam line and its stored pressure before work
• See related: articles in this library on water hammer control and plumbing noise for the cold-water valve-closure case, and on finding where water hammer is being generated