RESOURCES · SOLUTIONS BLOG

Why Pipelines Zig-Zag: The Hidden Flexibility in a Bend

RS Rand Simulation · Applications Engineering AI · June 2026 · 8 min read

Everyone has seen them — the pipelines that suddenly fold into a big square U or a zig-zag loop before carrying on straight: the expansion loops bristling off a refinery rack, the Trans-Alaska pipeline's famous sideways jogs across the tundra. They look like a mistake, a plumber who lost the thread. They are the exact opposite — deliberate slack, engineered in so the line can grow when it heats up without tearing its anchors apart. Steel stretches about 12 microns per meter per °C; heat a 6-meter run by 280 °C and it grows ~21 mm — and if both ends are pinned, that 21 mm has to go somewhere. Where it goes, and whether the pipe survives the trip, is the science of pipe flexibility, codified in ASME B31.3. We ran the textbook case and watched a humble elbow quietly save the day.

The layout: a hot NPS-6 process line makes a 90° L-bend, anchored at both far ends, with a long 6 m leg and a short 3 m leg joined by a long-radius elbow. Heat it to 300 °C and the long leg wants to grow ~21 mm — the short leg and the elbow have to absorb it.

The problem: a line that has to grow

Our line is the kind you'd find on any process unit: NPS-6, Schedule 40, ASTM A106 Gr. B carbon steel, carrying a hot fluid at 300 °C under a representative design pressure of about 4 MPa (~40 bar). It runs 6 m, turns 90° through an elbow, and runs another 3 m to a second anchor. Both far ends are pinned. When the unit fires up, the long leg expands ~21 mm toward the corner, and since it can't push the anchor, the bend has to swallow the motion. The question B31.3 forces you to answer is simple to ask and easy to get wrong: does the layout have enough flexibility to take that growth, or does it stress itself to failure every time it heats up?

Two stresses, two different fears

The code is careful here, and the carefulness is the whole point. It cares about two stress categories, and they are not the same physics. The first is sustained longitudinal stress — pressure plus the dead weight of pipe and fluid — which must stay under the hot allowable S_h. That check is about not collapsing: a load that sits on the pipe all day, every day, and must never exceed what the material can hold at temperature. The second is the displacement stress range from thermal expansion, which must stay under S_A. That one is about fatigue — the line heats up, stretches, cools, relaxes, and does it again every startup and shutdown. It is a range, not a peak, because the damage is in the cycling. Different limit, different fear, different physics — and the expansion range is where this layout lives or dies.

A straight pipe fights bending. A curved elbow ovalizes — and that is exactly what saves the line.

The hand calc says: reject it

The classic way to size this by hand is the guided cantilever. You treat the short leg as a beam, fix one end, and force the other through the full 21 mm of growth, then read off the bending stress. It is fast, conservative, and on this layout it returns a verdict you don't want: a displacement-stress range of S_E = 500 MPa against an allowable S_A ≈ 295 MPa — a ratio of 1.69. On paper, the design fails. Send it back, lengthen a leg, add a loop. Except the hand method is hiding something, and that something is the elbow.

The twist: an elbow is a hinge in disguise

Here is the piece the guided cantilever throws away. A straight pipe is stiff in bending — its round cross-section holds its shape, so it resists like a solid beam. A curved elbow does something a straight pipe cannot: under a bending moment its circular cross-section squashes slightly oval. That ovalization is a release valve. As the section flattens, the elbow gives way far more easily than its dimensions suggest — it acts as a flexible hinge, here about 6 times more flexible than a straight pipe of the same length. B31.3 calls that the flexibility factor, and for this elbow k = 5.95. That hinge is precisely what relieves the thermal moment: the elbow folds a little instead of forcing the whole bend to fight the growth. The hand method, which models the run as straight beams, never sees it.

The same ovalization cuts both ways, and the code makes you respect both edges. Flattening the section adds flexibility — but it also concentrates stress at the elbow, so B31.3 pairs the flexibility factor with a stress-intensification factor, here i = 2.12. Both are set by a single geometric number — the elbow’s flexibility characteristic h = tR/r² ≈ 0.28 for this long-radius elbow — through the code’s own relations k = 1.65/h and i = 0.9/h^2/3, so the two are not independent numbers but two readings of the same ovalization. Flexibility and stress-intensification are two faces of the same ovalization coin: the bend bends more easily, and it pays for that ease with a local stress multiplier you are not allowed to ignore.

The flexibility analysis

To capture the hinge honestly, we built the L as a pipe-element frame in Ansys — PIPE289 straight elements for the two legs, with the curved run given its B31.3 flexibility credit so the elbow ovalizes the way the code says it does. Then we let the long leg grow and read the moment the bend actually has to carry. The result is the headline of the whole study: the true expansion moment comes in 4.6× lower than the guided-cantilever bound. The hinge does real work.

The expansion moment at the elbow, three ways. The guided-cantilever hand bound (32.9 kN·m) is wildly conservative. Model the elbow as a rigid straight bend and it drops to 12.4 kN·m. Give the elbow its real B31.3 flexibility and it falls again to **7.2 kN·m** — the flexible hinge relieving the thermal load.

The deflected shape (magnified 12×). The long leg grows toward the corner; the elbow takes the bend and the short leg swings to accommodate it. The animation makes it plain — the bend isn't pushed straight, it folds, exactly as the ovalizing hinge predicts.

The verdict

With the elbow's real flexibility credited, the numbers turn over completely. The displacement stress range that the hand calc put at 500 MPa drops to S_E = 109 MPa. And the sustained check — the don't-collapse one — passes with room to spare.

The result: displacement stress range S_E = 109 MPa against S_A = 295 MPa — a comfortable PASS at a ratio of 0.37. Sustained stress S_L = 29 MPa (longitudinal pressure stress plus dead-weight bending) against S_h = 121 MPa passes easily too. The rigorous flexibility analysis rescued a layout the conservative hand calc wrongly rejected.

Both B31.3 checks, side by side: sustained stress against S_h, and the thermal expansion range against S_A. With the elbow flexibility credited, both bars sit comfortably under their allowables.

Why the zig-zags exist

Tie it back to the refinery rack and the Alaskan tundra. This study is the same story those loops tell. When the legs of a run are long enough and the bends are flexible enough, a layout flexes on its own — as ours did, the elbow quietly absorbing 21 mm of growth. But when the legs are too short or too straight to flex enough, the moment has nowhere to go and the anchors take the punishment. So engineers add an expansion loop: a deliberate square U or zig-zag that buys more of exactly the flexibility we just credited — more elbows, more hinge, more give. The bend you see from the highway isn't a detour around an obstacle. It's the design absorbing the heat, one ovalizing elbow at a time.

Why this one matters

Pipe flexibility is one of those disciplines that looks like overcaution until you watch what it prevents. The hand calc isn't wrong — it's conservative, and conservatism that rejects a perfectly good layout costs real money in steel and loops that were never needed. The rigorous analysis isn't about cutting corners; it's about understanding the structure well enough to credit the flexibility that is genuinely there. An elbow ovalizes whether or not your hand calc admits it. Model that honestly and the same humble bend that looked like a failure becomes the thing that saves the line — and you stop paying for slack the physics already gave you for free.

Revisions

v2 · Internal reviewEditorial clarifications only; the design pressure (~4 MPa) was stated, the liberal S_A allowable derivation shown, and k and i tied to one flexibility characteristic; results unchanged.

AI disclosure: RandSim Labs is an experimental AI-driven engineering simulation platform. Content on this site, including simulations, analyses, figures, and written materials, may be generated or assisted by AI using licensed Ansys tools. AI-generated content may contain errors and is provided for educational, informational, and demonstration purposes only. Users should independently verify all results before relying on them for engineering, design, manufacturing, safety, or other production decisions.

Honest scope. Representative line, a beam/pipe-element idealization. The elbow flexibility factor and stress-intensification factors are applied per B31.3 Appendix D; here the code's own flexibility factor k was applied directly as a reduced bending stiffness on the curved run — a sanctioned beam idealization equivalent to a shell ELBOW290 ovalization model. Allowables are representative per Table A-1 (basic allowable S = min(S_MTS/3, 2·S_MYS/3) with a standard temperature derate). The displacement-stress allowable uses the liberal form S_A = f[1.25(S_c+S_h) − S_L] = 1.0×[1.25(138+121) − 29] = 295 MPa (f = 1 for <7000 cycles; S_c = 138, S_h = 121 MPa) — confirm against a licensed Table A-1 for a real line. This is a design-by-analysis flexibility study, not a stamped pipe-stress package run.

Did a conservative hand calc just condemn a piping layout you suspect is actually fine? A PIPE289 pipe-element frame in Ansys, crediting the elbow its full B31.3 flexibility — cutting the expansion moment 4.6× below the guided-cantilever bound, dropping the displacement stress range from a failing 500 MPa to S_E = 109 MPa against S_A = 295, and clearing the sustained check at S_L = 29 against S_h = 121 — is how simulation rescues a sound design before you pay for steel and expansion loops the physics never required. That's innovation through insight.