How to Support a Run That Has to Move

Why this matters

A run that changes temperature is going to change length. That part is not negotiable and no amount of hardware stops it. What you get to decide is where the movement happens and what it pushes against on the way. Get that wrong and the run still moves, but it moves by tearing a fitting, dragging a hanger sideways, buckling into a ceiling grid, or walking out of a piece of equipment over three winters.

The number that makes people take this seriously is the force. A fully restrained straight steel section pushes on its anchors with a force that does not depend on its length at all: a short one pushes as hard as a long one. The worked section below comes out around 24,000 pounds on a 2-inch line. Nobody picks a clamp that holds twelve tons by accident.

Before you work on a run that is hot or pressurized

Relieve pressure and drain before opening any joint on a heating or hot-water run, and let it cool. Hot water above its atmospheric boiling point flashes to steam the instant a joint cracks open, so the point of release is not a place to stand and a partly-backed-off flange bolt is not a pressure-relief method. Isolate and lock the source under 29 CFR 1910.147, which is the correct standard here because this is mechanical and stored thermal energy rather than an electrical utilization circuit.

Contact with a 180 F surface burns immediately. Measure with a non-contact instrument or a probe on a handle, not a hand and a rule.

If you have to cut or weld, that is hot work: a permit, the area cleared of combustibles, and a fire watch during and after, under 29 CFR 1910.252(a), with NFPA 51B applying in the edition your authority having jurisdiction or your insurer has adopted, which is how it reaches you. Heating galvanized pipe releases zinc oxide fume, which is an inhalation route and needs local exhaust ventilation or respiratory protection under a program meeting 29 CFR 1910.134. A welding glove does nothing about fume.

And if the run is insulated with thermal system insulation in a building constructed no later than 1980, that insulation is presumed asbestos-containing under 29 CFR 1926.1101 for construction work, with general-industry duties at 29 CFR 1910.1001. Do not pull it back to find the pipe.

Step 1: establish the real temperature range, not the operating temperature

Movement is driven by the difference between the extremes the run will actually see and the temperature it was installed at. Write down three numbers: installation temperature, maximum, minimum.

A heating line installed at 60 F, operating at 180 F, sitting in an unheated space that reaches 40 F in winter has a 120 F rise in one direction and a 20 F drop in the other, for a total travel range of 140 F. Sizing anything from the 120 F alone leaves 20 F of the 140 F range unaccounted for and, more importantly, gets the direction wrong at every guide, because the cold position is not the installed position.

Skip this step and everything downstream is computed against the wrong number.

Step 2: divide the run into sections between anchors

An anchor is a point you have decided must not move. Everything between two anchors is one section, and one section's movement is entirely its own problem.

You are choosing anchors, not discovering them. If you do not choose, the run chooses for you: the stiffest connection in the section becomes the anchor, and it is usually a piece of equipment, a branch tee, or a wall penetration that was grouted solid. Those make terrible anchors and they will tell you so.

Step 3: compute each section's free movement

Length change equals the material's expansion coefficient times the length times the temperature change. Near room temperature, steel runs about 6.5 parts per million per degree Fahrenheit and copper about 9.2. Plastics move several times as much and vary widely by material and by temperature, so take a plastic's coefficient from the manufacturer's own data rather than from any general table.

Free movement is the number the section wants. It is not yet the number anything sees.

Step 4: decide where the movement goes, using one gate

The gate: does this section's own geometry give the movement somewhere to bend, or does it need a device?

A change of direction perpendicular to the movement is a spring. The leg bends, the movement is absorbed, and nothing is overstressed as long as the leg is long enough. Required leg length grows roughly as the square root of the pipe diameter times the movement, which has a useful consequence: doubling the movement only lengthens the required leg by about 40 percent, so a leg that is close is usually worth extending rather than abandoning. The constant in front of that square root depends on the material's modulus and its allowable stress, so the actual required length comes from the piping specification or the engineer's calculation, never from a remembered rule of thumb.

If there is no leg, you install the absorption: an expansion loop, an offset, or a manufactured expansion joint. Each has its own anchor and guide requirements published by whoever makes or designs it, and those requirements are part of the device, not advice about it.

Step 5: guide between the anchor and the absorber

A section under compression that is not guided does not shorten neatly, it bows sideways. Guides constrain the run laterally while letting it slide along its axis, which converts a buckling problem into a sliding problem.

The guides closest to the absorber matter most, because that is where the run is least supported against going sideways and where the manufacturer's own spacing requirements for a loop or joint apply.

Step 6: give every joint in the section the full budget

This is the step that gets skipped, and it is the reason a correctly calculated run still fails at one fitting.

Movement in a section is not shared out among the joints along it. A joint two feet from the anchor barely moves. A joint two feet from the absorber sees essentially the whole section's travel, every cycle, for the life of the system. So the joint types that live near the absorber have to tolerate that travel: what each mechanical joint tolerates has its own article, and this is exactly the question it answers.

Step 7: clear the path the movement will take

Walk the section and remove everything that will accidentally become an anchor. A hanger with no travel clearance. Insulation packed tight into a wall sleeve. A pipe passing within a quarter inch of a strut. A second trade's conduit strapped to your run. A drain line teed off and clamped hard to a wall six feet away, which will be dragged sideways by the section it is attached to.

Step 8: set the cold position deliberately

Where the movement is one-directional, the run can be installed offset from its final position so that it arrives centred when hot, which halves the travel each end has to accommodate. That is a specification decision with a stated offset and a stated installation temperature, and it only works if the person installing it knows the number and records the temperature at which they set it.

The gate run against two sections of the same building

Steel heating distribution, installed at 60 F, operating at 180 F, so a 120 F rise. Both sections are 60 feet between structural points.

Section A is straight, ceiling to ceiling, no change of direction anywhere in the 60 feet.

Free movement: 6.5e-6 times 720 inches times 120 F, which is 0.56 inch.

Apply the gate: there is no perpendicular leg, so the geometry absorbs nothing. If both ends are held, the section is fully restrained, and a fully restrained straight member develops a stress equal to its modulus times its coefficient times the temperature change. For steel at roughly 29 million psi, that is 29e6 times 6.5e-6, which is about 188 psi for every degree F, so 120 F gives about 22,600 psi.

Note what that expression does not contain: length. A 6-foot restrained section and a 600-foot restrained section develop the same stress and push with the same force. Force is that stress times the metal cross-sectional area, so for a 2-inch schedule 40 pipe at about 1.07 square inches of metal, the anchors see roughly 24,000 pounds. That figure assumes a uniform temperature change along a straight member, purely elastic behaviour, and no relief from bending or buckling, so it is the upper bound rather than a prediction. It is the right upper bound to design against.

Section A therefore fails the gate. Read the 24,000 lbf correctly, though: that is what the anchors see if you do nothing, so it is the argument for an absorber rather than the anchor specification. Install the absorber and the anchors instead carry that device's published reaction, its spring force, guide friction and, for a bellows, the pressure end thrust the pipe wall no longer carries across the joint. Take that number from whoever makes or designs the absorber. Either way it is an engineered detail rather than a clamp out of the van.

Section B is the same 60 feet with a 12-foot leg turning off at one end and running away perpendicular.

Free movement is the same 0.56 inch, because the length and the temperature change are the same. But now there is a leg to bend. The gate asks whether 12 feet is long enough to absorb 0.56 inch within the allowable stress for that pipe, which is the square-root relationship above with the specification's own constant. On a run of this size it is a comfortable margin rather than a marginal one, and the calculation is quick because the answer is not close.

Section B passes, and what it needs instead is discipline: one anchor at the far end so the movement is aimed into the leg, guides along the straight portion so it slides rather than bows, and nothing clamped rigidly to the leg itself. Put one hard clamp in the middle of that 12-foot leg and you have removed the spring, and Section B becomes Section A with extra steps.

The two sections have identical movement and opposite answers. The variable is geometry, not size and not temperature.

Verifying it, once, with a scribe mark

Compute the expected travel at one guide before the system runs. Then scribe a mark on the pipe against a fixed reference at the guide and record the pipe temperature when you made it. Bring the system up, let it stabilize, and read the offset with a non-contact instrument or a rule held against the fixed reference rather than the hot pipe.

Three outcomes, and each says something different.

• Travel close to the computed figure. The section is doing what you designed. Record it as the baseline, because a future reading that has changed means a support has seized.
• Travel much less than computed. Something is restraining the section. The movement did not vanish; it turned into stress and it is being absorbed by whatever is stiffest, which is usually a fitting or an equipment connection. Find the unintended anchor.
• Travel more than computed, or movement at a point that should not move. An anchor is slipping, which means the anchor is now a guide and the next section along is taking a load it was never given.

Then look at the guides themselves after a season. Bright wear on the sliding surface is the system working. Bright wear on the pipe wall means the guide has no wear surface and is machining your run.

References

• 29 CFR 1910.147 for isolating and controlling stored mechanical and thermal energy before opening a pressurized or hot run.
• 29 CFR 1910.252(a) for welding, cutting and brazing fire prevention including the fire watch; NFPA 51B applies in the edition your authority having jurisdiction or your insurer has adopted, and reaches you through that adoption or your hot-work permit program.
• 29 CFR 1910.134 for respiratory protection where heating galvanized material releases zinc oxide fume; 29 CFR 1926.1101 (construction) and 29 CFR 1910.1001 (general industry) for presumed asbestos-containing thermal system insulation.
• The piping specification or the engineer's calculation, and the expansion joint or loop manufacturer's published anchor and guide requirements, which own the leg lengths, anchor forces and guide spacing.
• See related: What Thermal Movement Does to a Joint; The Mechanical Pipe Joints and What Each One Tolerates; What a Support Has to Do Besides Hold Weight.