Pipe Bend Minimum Wall Thickness Calculator (B31.3 ¶304.2.1)

Resolves the after-bend minimum wall t_m at intrados, extrados and centerline, together with the I correction factors that separate them, per ASME B31.3 ¶304.2.1. The intrados governs; how much a given bending procedure actually thins the wall is not predicted here.

Method last updated 2026-07-03 (calculation changelog) · fixture-verified on every build — most recently 2026-09-03.

Built and fixture-verified by Matthew Norris, P.E. — active P.E. licensure in Arizona, California, Kansas, Missouri, North Carolina, Texas.

Bending thins the outside of the bend (extrados) and thickens the inside (intrados), but the pressure stress redistributes the other way — the intrados needs more wall than straight pipe. B31.3 ¶304.2.1 handles this with an I correction on the allowable in the wall-thickness equation, evaluated at both surfaces. This is the calculation an engineer reaches for when specifying induction, hot or cold bends in place of fitted elbows: it sets the minimum finished wall the bend must deliver at each surface, which in turn drives the starting wall ordered for the mother pipe. It is also the check a reviewer runs on a field bend that arrived thinner than expected. The card returns both I factors and the required thickness at intrados, extrados and centerline, so the governing surface — and the margin over straight pipe — is explicit.

Method

With R the bend radius to centerline and D the OD:

I_intrados = (4(R/D) − 1) / (4(R/D) − 2)   I_extrados = (4(R/D) + 1) / (4(R/D) + 2)

t_m(I) = P·D / ( 2( S·E/I + P·y ) ) + A

The I factor works by scaling the allowable stress inside the straight-pipe relation: at the intrados I > 1 divides S·E down, demanding more wall, while at the extrados I < 1 credits the lower hoop stress on the outside of a torus with a thinner requirement. Both factors are pure functions of the radius ratio R/D — a tighter bend pushes them further from 1 in both directions. The calculator computes R/D, evaluates I at each surface, and solves the corrected thickness equation three times: intrados, extrados, and centerline, where I = 1 and the result reproduces the straight-pipe value as a built-in sanity anchor. The allowance A is added to each. The same form is referenced by B31.1 ¶102.4.5 and the B31.4/B31.12 bend rules, so one card serves the process, power and pipeline bend checks with the appropriate user-supplied S, E and y. Because the correction rides on the allowable rather than the geometry, every input keeps the same meaning it has in the straight-pipe calculation — the bend check adds exactly one new piece of information, the radius, and reports exactly what it costs.

Limitations — what this calculator is not

• Computes required after-bend thickness — it does not predict how much your bending procedure actually thins the wall. Thinning depends on the method and the machine; specify a starting wall such that the finished extrados and intrados both meet these minima, using the fabricator's demonstrated thinning, not an assumption.
• R ≤ D/2 is rejected (not a physical centerline bend radius); R/D < 1 triggers a warning to verify the bending procedure, since very tight bends are where both the I factors and the fabrication thinning are most severe at once.
• Induction/hot bends may have ovality and hardness acceptance criteria beyond thickness — out of scope. Passing the wall check here does not accept the bend.
• Outputs are minimum finished-wall requirements including the allowance A. Converting them into a purchased starting wall — bend thinning plus the mill under-tolerance of the mother pipe — is the purchase specification's job, not this card's.
• Internal pressure only: no external-pressure or vacuum evaluation of the bend, and no structural, sustained or displacement loading.
• Flexibility and stress intensification of the bend under moment loading are a separate question — use the SIF calculator and the displacement stress-range check for that side of the design.
• S, E and y are user-supplied from the governing edition, exactly as in the straight-pipe calculation, and the check applies to the bend body — the tangents and end welds are straight-pipe territory.
• Forming quality is not assessed: wrinkling, local flats, and surface defects on the intrados of a tight bend are workmanship and acceptance matters between the specification and the fabricator, entirely outside a thickness calculation.
• One bend, one pressure, one set of factors per run — a piping class covering several bend radii is qualified by its tightest radius, and each distinct R/D deserves its own run rather than an assumed envelope.

Worked example — fixture-verified

NPS 8 (8.625 in OD) bent at R = 12 in (≈1.4D), P = 1,000 psi, S = 20,000 psi, E = 1, y = 0.4, allowance 0.065 in.

Step by step

1. R/D = 12/8.625 = 1.39130 → 4(R/D) = 5.56522.
2. I_intrados = (5.56522−1)/(5.56522−2) = 4.56522/3.56522 = 1.28049; I_extrados = 6.56522/7.56522 = 0.86782.
3. Intrados: t_m = 1,000·8.625 / (2·(20,000/1.28049 + 400)) + 0.065 = 8,625/32,038 + 0.065 = 0.33421 in.
4. Extrados: 8,625/46,893 + 0.065 = 0.24893 in; centerline: 8,625/40,800 + 0.065 = 0.27640 in.

The intrados needs 21% more wall than straight pipe here — the reason tight bends are ordered a schedule heavier than the run.

Sources & citations

• ASME B31.3, Process Piping — ¶304.2.1, pressure design of bends.
• ASME B31.1, Power Piping — ¶102.4.5 (same I-factor form on the Power line card).

Per the source & citation policy, allowable-stress and factor table values are user-supplied. Where a page does reproduce specific ASME data (the B16.5 ratings, the quick-reference tables), it states the source table and conditions inline.

FAQ

Why does the intrados need MORE thickness when bending makes it thicker?

Two separate effects that are easy to conflate. Fabrication thickens the intrados, but hoop pressure stress at the intrados of a torus is inherently higher than in straight pipe — the same pressure is carried by a shorter length of wall on the inside of the curve — so I > 1 raises the required wall there. The Code requires the finished wall to meet tm at each location: usually fabrication thickening covers the intrados, which is why the check matters most for the thinned extrados on tight bends. The two effects also determine what an inspector should measure and where: after bending, the acceptance question at the intrados is whether the thickened wall still exceeds its raised requirement, and at the extrados whether the thinned wall still clears its reduced one — so a UT survey of a bend that only records the minimum reading has thrown away half the check. Record wall at intrados, extrados and neutral axis separately, compare each against its own tm from this card, and the bend is qualified location by location, which is exactly how ¶304.2.1 frames it.

Which surface actually governs the purchase decision?

Each surface is checked against its own requirement: the intrados demands the most wall but fabrication thickens it, while the extrados demands the least and fabrication thins it. In the worked example the intrados needs 0.33421 in and the extrados 0.24893 in — the order is decided by comparing the predicted after-bend wall at each surface with its own tm, and on tight bends the thinned extrados is usually the one that forces a heavier starting wall. A rule of thumb for anticipating the outcome before running numbers: fabrication thinning at the extrados runs on the order of D/(2R+D) as a fraction of the starting wall — heavier as the radius tightens — while the required wall there is the most forgiving of the three locations. On generous radii (5D and up) the effects nearly cancel and the run schedule usually survives; at 3D they compete; at 1.5D the thinning typically wins and forces the heavier starting wall. That is why induction bends on tight radii are quoted with a wall upcharge as a matter of course, and why the check belongs before the purchase order, not after the first UT report.

What bend radius R do I enter?

The bend radius to the pipe centerline, from the bend drawing or the bending machine setup. R ≤ D/2 is rejected as geometrically impossible for a centerline radius, and R/D below 1 draws a warning to verify the bending procedure. Radius drives everything: the worked example's 12 in radius on NPS 8 — about 1.4D — produces an intrados factor of 1.28049, and tightening R pushes both I factors further from 1. Watch the units convention on drawings: bend radii are commonly called out in pipe-diameter multiples — 3D, 5D — and the D in that shorthand is the nominal size, so a '3D' bend on NPS 8 is R = 24 in to the centerline. Enter the resolved inches, not the multiple. The R/D < 1 warning exists because that territory belongs to specialized forming — tight induction bends and formed elbows with their own qualification — where springback, ovality and wall control are procedure-dominated and the simple I-factor picture starts to fray at exactly the radii where its corrections are largest.

How do I pick the starting wall before bending?

From the bending procedure's demonstrated performance. The finished wall at the extrados equals the starting wall minus the procedure's thinning there, and that result must meet the extrados tm; the intrados works the same way with thickening. This card supplies the right-hand side of each comparison — the required minima — while the fabricator's qualification data supplies the left. Ordering one schedule heavier than the run is the traditional outcome of that arithmetic on tight bends. Two procurement details close the loop. Bending specifications express capability as percent thinning at a stated radius — 'max 12% at 5D' — so the arithmetic is: starting wall × (1 − thinning) ≥ extrados tm, solved for the starting wall, then rounded up to a schedule. And ovality rides along with wall in the same qualification: a bend can meet every thickness minimum and still fail on out-of-roundness, which has its own service limits, so accept bends against both numbers from the procedure qualification record. Where the bender's data is a guess, a test bend with a UT grid is cheaper than a rejected production lot.

Does the same check apply to B31.1 power piping?

The same I-factor form is referenced by B31.1 ¶102.4.5, and the Power line carries its own card built on it. Enter the B31.1 allowable stress and factors for the material and temperature, and the geometry side of the calculation — the I factors and the three-location evaluation — is identical to what this page documents. One caution when hopping between the books: carry the whole basis across, not just the equation. The allowable-stress tables differ, B31.1's use of SE as a combined term changes what you enter, and the editions move independently — so a bend qualified on the Process line is not automatically qualified for a power-piping code of record even at identical conditions. The geometry side transfers perfectly; the materials side must be re-entered from the governing book. The Power line's own card exists so that each calculation file cites one code cleanly instead of a hybrid that satisfies neither.

What does the centerline result tell me?

With I = 1 the equation collapses to the straight-pipe form, so the centerline tm is the wall the same pipe would need with no bend at all — 0.27640 in in the worked example, sitting between the extrados and intrados requirements. It is the sanity anchor for the calculation: the intrados result should always exceed it and the extrados result should always undercut it, and the spread between them is a direct read of how much the bend geometry costs. The sanity-anchor role is genuinely useful in review: if a submitted bend calculation shows an intrados requirement below its centerline value, or an extrados above it, the I factors were transposed — the single most common error in hand-done bend checks, and one this ordering test catches without redoing any arithmetic. The spread also prices the geometry decision directly: when intrados-to-extrados requirements span only a few mils, the bend is nearly free and a tighter, cheaper routing radius is on the table; when the spread is wide, the bend is expensive in wall and the routing conversation should hear about it.

Related calculators

• ASME B31.3 Pipe Wall Thickness Calculator — Straight-pipe baseline the centerline case reproduces
• ASME B31.1 Power Piping Wall Thickness Calculator — B31.1 ¶102.4.5 uses this same I-factor form
• Mitered Elbow Allowable Pressure Calculator (ASME B31.3 ¶304.2.3) — Mitered alternative when smooth bends aren't practical
• Stress Intensification & Flexibility Factor Calculator (B31.3 Appendix D) — Same bend geometry, flexibility and SIF side