The Rupture Disk: A Weak Point Built to Burst and Save the Tank

The job: the last line before the tank tears

Every pressure vessel is built to a maximum allowable working pressure (MAWP). Push past it — a blocked outlet, a runaway exotherm, an external fire boiling the contents, a slug of liquid thermally expanding in a trapped line, a water-hammer spike — and the steel yields, then tears, releasing the stored energy of a compressed fluid all at once. The rupture disk is the calibrated weak spot that fails first, on purpose, at or just above MAWP. It converts an uncontrolled vessel rupture into a directed release to a flare, scrubber, or safe area.

Crucially, it is passive: no spring, no pilot, no power, no signal. It sits sealed and inert for years and must perform exactly one action, once, perfectly. That austerity is its strength — there is almost nothing that can fail closed.

Mechanism: two ways to build a controlled failure

Two geometries dominate, and they fail by opposite physics.

Forward-acting (tension-loaded). The domed foil faces the process on its concave side, so pressure stretches it. Treat the dome as a thin spherical shell of radius R and thickness t; the membrane stress is

σ = P·R / (2t)

and it bursts when σ reaches the material's ultimate tensile strength σ_u. Laser-etched score lines — deliberate stress concentrations — steer the tear so it opens in clean petals instead of firing a metal fragment downstream.

Reverse-buckling (compression-loaded). The dome is flipped so its convex side faces the pressure, which now compresses it. At a critical pressure the shallow shell snaps through — an oil-can inversion — and knife edges or cross-scores shear it open. Snap-through is an elastic instability set by geometry and stiffness,

P_cr ∝ E·(t/R)²

not by reaching σ_u, so it repeats far more tightly and shrugs off pressure cycling. That is why reverse-buckling disks may run at up to ~90% of their burst pressure, versus ~70% for forward-acting.

The numbers: a worked example and the tolerances

Worked example. A forward-acting disk with t = 0.1 mm, dome radius R = 20 mm, and σ_u ≈ 550 MPa bursts near

P_b ≈ 2·t·σ_u / R = 2 × (1×10⁻⁴ m) × (550 MPa) / (0.02 m) ≈ 5.5 MPa ≈ 55 bar

The scaling is transparent: halve the foil thickness and the burst pressure halves; double the diameter and it halves again. Makers hit a target set point by trimming exactly these three levers — then burst-testing samples from every production lot to confirm it.

The performance figures a designer cares about: burst tolerance is typically ±5% of the stamped value (above ~2.8 bar / 40 psig); opening time is on the order of 1 ms for reverse-buckling; and the burst pressure is temperature-dependent, because both σ_u and E fall as metal heats. A disk is therefore stamped with a burst pressure at a coincident temperature and derated a few percent per 100 °C. Off-the-shelf disks span a fraction of a bar (thin graphite for storage tanks) to well over 1000 bar (heavy Inconel for research autoclaves).

Why it fires fast — and only once

Why milliseconds? Because burst is an instability, not a gentle yield. Past the peak load the membrane can carry no more, so once a crack initiates at a score line it runs at a large fraction of the material's Rayleigh (surface-wave) speed — kilometres per second — and the compressed fluid behind it does the rest of the work peeling the petals back. A reverse dome's snap-through is likewise dynamic, inverting in under a millisecond. This speed is the whole point: it outruns a deflagration's pressure rise and the sluggish lift of a mechanical valve.

The same physics dictates the cost. The disk is non-reclosing: having burst, it cannot reseat — the system stays open until someone fits a new disk. And a poorly designed disk can fragment, throwing metal shards downstream. Scored and reverse-buckling designs are certified non-fragmenting, which is mandatory when the disk sits at the inlet of a relief valve, so debris can never jam the valve open or shut.

In real engineering practice

Rupture disks are code devices. In the U.S. they fall under ASME BPVC Section VIII, UG-127 and UG-134; internationally under ISO 4126-2; and relief loads are sized per API 520/521. Every disk is stamped with its burst pressure at coincident temperature, lot number, manufacturing range, minimum net flow area, and flow characteristic (a discharge coefficient of ≈0.62, or a certified flow-resistance factor K_R).

• Paired with a relief valve. A disk installed upstream of a PRV isolates the valve's seat from corrosive or fouling fluid; a telltale pressure gauge between them flags a pinholed disk. ASME applies a 0.9 combination-capacity factor when the pair isn't jointly certified.
• Materials. 316L, nickel, Monel, Inconel, Hastelloy, tantalum, even gold and graphite; foils run 0.05–0.5 mm.
• Where they live. Chemical reactors, pipelines, rail and road tankers, fire-suppression cylinders, aerosol cans (a micro rupture disk under the crimp), spacecraft propellant tanks, and dust-explosion vent panels under NFPA 68.

Failure modes, limits, and a common misconception

The limitations follow directly from the mechanism. The disk is a one-shot device that vents the entire inventory and cannot be tested in place — proof-testing destroys it, so confidence comes from lot sampling. It is vulnerable to fatigue if pressure cycles near the set point (the classic reason to prefer reverse-buckling), to corrosion pinholes, to mechanical mishandling, and to vacuum or backpressure that can invert a forward-acting disk — hence vacuum-support rings. Running a forward disk too close to its burst pressure shortens its life rather than adding safety.

Common misconception. A rupture disk is not a crude weak spot that lets go 'somewhere around' a pressure. It is a precision, lot-certified component that stays bubble-tight — zero leakage — right up to a set point held to ±5%, then fails in a millisecond. And counter-intuitively, the most accurate modern disks are loaded in compression, with the dome deliberately facing away from the pressure they protect against.

Rupture disk vs. spring-loaded relief valve — two philosophies of overpressure protection.

Attribute	Rupture disk (bursting disc)	Spring-loaded relief valve (PRV)
Reclosing?	No — one-shot, must be replaced	Yes — reseats when pressure drops
Opening time	≈1 ms (reverse-buckling)	Tens of ms to ~100+ ms
Leak before set point	Bubble-tight, zero leakage	Can simmer / weep near set point
Set-point accuracy	±5% of stamped burst	≈±3%, drifts with fouling
Dirty / corrosive / slurry service	Excellent — no moving parts	Poor — seat fouls and sticks
Opening area	Full bore, all at once	Modulated, partial lift
Relative cost	Low	High
Field testability	No — proof test is destructive	Yes — lift-test in place