Resources · Solutions Blog · Explicit Dynamics / Consumer Safety

How Fast a Champagne Cork Really Leaves the Bottle

Rand Simulation — Applications Engineering AI
Explicit transient dynamics · Ansys LS-DYNA · 9 min read

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.

Everyone has ducked a champagne cork, and everyone half-believes the shaken-bottle myth. The internet quotes an exit speed anywhere from 25 to 60 mph, and eye injuries from flying corks are a real emergency-room category every New Year. So we did the thing the folklore never does and actually computed it: a deformable cork seated in a glass neck, several atmospheres of gas behind it, friction holding it in place — and then we let it go. The number that controls everything is temperature, not agitation. Warm the bottle and the cork leaves fast; shake it and, for the launch itself, almost nothing changes.

The warm-bottle case (7.5 bar), solved in Ansys LS-DYNA and shown as a half-section so the bore reads: the tan cork slides out of the dark-green glass neck over a few milliseconds while the live readout climbs to the exit speed the solver measured as the cork base clears the lip. The gas drive decays as the cork sweeps volume, and the friction grip relaxes as the compressed body clears the bore — both fall out of the physics, neither is scripted. Geometry is generic and self-authored, not any specific brand.

A cork is a plug held by friction, launched by gas

The setup is simple to state. A champagne cork sits compressed inside the neck, its mushroom head outside; behind it, in the headspace, is dissolved-CO₂ gas at several atmospheres. The interference fit of the cork against the glass holds it there by friction. Take the cage off and the net force — gas pressure over the bore area, minus the friction grip — decides whether it stays or flies.

We built exactly that. The cork is a deformable solid with cork's real, unusual property: a near-zero Poisson ratio, so it can be squeezed radially into the bore without bulging along its length. Its density is smeared to the true mushroom-cork mass of about 8 grams. The neck is treated as rigid — we are launching a cork, not fracturing glass — with an 18 mm bore. The grip is not a number we typed in: the cork is built a hair smaller than the bore, then grown into the glass so that a genuine contact pressure develops, and the friction that holds the cork is whatever that contact pressure times the friction coefficient works out to be. As the compressed body slides up and out, the engaged length shrinks, so the grip ramps down on its own. The gas drive behind the cork base expands as the cork travels, so its pressure decays the way an adiabatic gas actually does rather than pushing at a constant value. The whole event is a few milliseconds of explicit transient dynamics, which is precisely LS-DYNA's lane.

Below about 4.75 bar, nothing happens

The first result is the one that explains why a cold bottle feels safe. Across the sweep, the cork does not move at all until the headspace pressure climbs past roughly 4.75 bar. At 4.0 and 4.5 bar the grip wins and the cork stays seated; at 5.0 bar it launches. There is a genuine threshold, and it is why chilling a bottle stabilizes it: cooling drops the equilibrium CO₂ pressure, and below the release point the friction simply holds. Above the threshold the model runs from 64 mph at 5 bar up to about 96 mph at an extreme 9 bar — the full idealized range the honest bracket later brings back to earth.

Above the threshold, the exit speed climbs steeply with pressure. This is the curve that "settles the number" — it maps the whole folklore band onto a physical axis and shows where each condition lands.

Cork exit speed against bottle headspace pressure, with named conditions and the folklore band marked — Exit speed read at the instant the cork base clears the lip, across a 4–9 bar sweep at a fixed friction coefficient. The cork is held below ~4.75 bar, then accelerates hard. Chilled (~5 bar) leaves at 64 mph in the model; warm (~7.5 bar) at 84 mph. These idealized speeds sit *above* the 25–60 mph folklore band and roughly two to three times the measured ~11 m/s (~25 mph, 40 km/h) chilled cork — the model is an upper bound, discussed below.

The result: the cork is held below ~4.75 bar, then exit speed climbs steeply with pressure. In the idealized model a chilled bottle (~5 bar) launches at 28.6 m/s (64 mph) and a warm one (~7.5 bar) at 37.5 m/s (84 mph) — a 31% jump from a 2.5 bar warming. Modeling a "shaken" bottle as the same warm case plus a bounded pressure bump adds only ~10%. The launch is over in about 1–1.5 ms, and the warm cork leaves carrying about 5.6 joules of kinetic energy.

The launch is faster than you can flinch

Reading the cork's velocity through the event shows how little time there is. After the gas ramps up and the grip releases, the cork goes from stationary to its exit speed inside the neck in a little over a millisecond. We report the speed at the moment the cork base clears the lip, not the peak — because the prescribed gas drive keeps pushing the now-free cork after it has left, which inflates the peak into an artifact. The honest number is the muzzle velocity at the lip.

Cork axial speed against time for the chilled and warm cases, with the lip-clearing instant marked — The cork base's axial speed for chilled and warm bottles. Both sit still through the seating and gas-ramp window, then accelerate hard; the warm cork clears the lip at 84 mph about a millisecond sooner than the chilled one at 64 mph. The high-frequency ripple is real contact chatter between the cork and the bore, not numerical noise — the energy balance stays clean throughout.

Shaking is the myth; warming is the danger

Here is the part that overturns the party trick. Shaking a bottle does not change the equilibrium pressure in the headspace — that is set by temperature through how much CO₂ the wine holds, and shaking does not add any gas. What shaking changes is the foaming after the cork is out: the dramatic gush is dispersed bubbles, not a faster cork. So we modeled "shaken" the only honest way we could — as the warm bottle plus a small, bounded over-pressure — and the cork's launch speed barely moved.

The comparison is stark. Going from chilled to warm — a swing you can create just by leaving the bottle on a table — adds about 20 mph to the exit speed. The bounded "shaken" bump on top of the warm case adds only about 8. Friction and the exact strength of the grip shift the answer by a few miles per hour either way, and above a certain grip the cold bottle simply stays corked. Temperature is the dial that matters; everything else is second order.

Left, the honest bracket of chilled exit speed; right, the sensitivity of exit speed to temperature, shaking, friction and grip — Left: settling the number honestly for a chilled bottle. The idealized LS-DYNA speed (64 mph) is an upper bound; correcting for the losses the model leaves out brings the estimate into the 25–40 mph folklore band, consistent with the measured ~40 km/h (~25 mph) cork. Right: how much each factor moves the exit speed. Warming dominates; the bounded "shaken" effect is half of it; friction and grip are smaller still.

Settling the number, honestly

The model's exit speeds — 64 mph chilled, 84 warm — are faster than the folklore band and faster than the roughly 40 km/h (11 m/s) that high-speed imaging has measured for a real chilled cork. That gap is not something to hide; it is what an idealized launch model does. Our cork releases from a static friction latch and then slides with very little drag, and the model leaves out three things that all slow the real cork down: gas blow-by past the cork the instant it starts to move, which bleeds the driving pressure away fast; the cork's own viscoelastic absorption as it decompresses; and a gentler, less abrupt real-world release. Put a loss correction on the idealized number and it drops squarely into the 25–40 mph band and onto the measured anchor.

So the defensible answer is a bracket, not a single decimal: the idealized upper bound and a loss-corrected estimate that lands in the folklore band and on the measurement. What does not depend on that calibration — the robust, publishable physics — is the shape of the story: a hard release threshold, a steep climb with pressure, temperature as the dominant driver, and shaking as a near-non-event for the cork's speed.

Why the safety line is "keep it cold, point it away"

Turn the speed into energy and the safety point makes itself. Even the chilled cork in the idealized model carries a few joules of kinetic energy, and the warm one about 5.6 — and a cork does not need to be moving at its top speed to injure an eye. That is the whole reason flying corks are an ER category around the holidays. The physics-backed version of the standard advice is exact: cool the bottle to keep the pressure below the steep part of the curve, and never aim the neck at a face, because the launch is over before anyone in the room can react to it.

Honest scope. The geometry is generic and self-authored — a representative sparkling-wine neck and mushroom cork, nothing traced from or benchmarked against a specific brand. The reported exit speeds are an idealized upper bound: the model omits gas blow-by past the moving cork, the cork's viscoelastic energy absorption, and a gentler real release, all of which lower the real speed, which is why the raw numbers (64 mph chilled to 84 warm, ~96 at the 9 bar extreme) sit above the 25–60 mph folklore band and the measured ~11 m/s (~25 mph) chilled anchor; the trustworthy outputs are the trend, the threshold, and the temperature-versus-shaking verdict, plus a loss-corrected bracket that lands on the folklore band. Temperature enters only through its headspace pressure (chilled ~5 bar, warm ~7.5 bar from published champagne data); there is no thermal field, no CO₂-solubility solve. "Shaken" is a bounded pressure perturbation, not a modeled foaming mechanism — the study's finding is precisely that shaking barely changes launch speed, so its real effect (the spray) is out of scope. The gas drive is a prescribed, adiabatically decaying pressure history verified against p·V^γ, not a resolved gas jet, shock, or two-phase flow. The cork's grip is a calibrated thermal-interference fit, not the physical corking-compression history. We report the cork's exit speed and kinetic energy and cite the eye-injury context; we do not model the eye, tissue, or any injury. The reported speed is the muzzle velocity at the lip — no post-exit tumbling or air drag. The neck is rigid; no glass fracture, no wire cage, no hand. A single mesh (~4,800 elements) was used per case; energy balance and hourglass were checked on every run, but no formal mesh-independence sweep was performed, so the absolute speeds are indicative rather than converged design values.

Have a product held in place by a grip that has to let go at exactly the right moment? The workflow behind this study — Ansys LS-DYNA solving the deformable cork, its emergent friction retention, the decaying gas drive and the exit kinematics, anchored to published measurements and reported with an honest error bracket — is how Rand Simulation turns a "how fast, really?" question into a number a team can defend. That's innovation through insight.