Resources · Solutions Blog · Explicit Dynamics / Physics & Curiosity

Why a Whip Cracks: The Tip Breaks the Sound Barrier

Rand Simulation — Applications Engineering AI
Explicit transient dynamics · Ansys LS-DYNA · 7 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.

The crack of a bullwhip is not leather slapping leather. It is a miniature sonic boom: the tip of the whip briefly outruns sound itself, and the crack is the shock wave it sheds. High-speed photography settled that decades ago — real whip tips have been measured near Mach 2. What we wanted to see was the machine that does it. So we built a two-meter whip in Ansys LS-DYNA as a chain of 800 articulated links — each link a tiny truncated cone, thick at the handle, whisper-thin at the tip — cast it back over the hand, and threw it the way a person actually throws it: an arm stroke with a 160-degree wrist flick laid over it. The chain did what a whip does. It carried the twist of the handle down its length, amplified it link by link as the cord got lighter, and snapped the tip through a supersonic hairpin at 570–594 m/s — Mach 1.66–1.73 — before the hand had even finished moving. Then it did it again, a second crack as the loop ran out of whip over the top. No motor, no spring, no trick: an ordinary human throw, funneled into ever-lighter cord.

The finest-resolution solve (1,600 links, 1.25 mm each), rendered from the solved chain as a variable-radius tube and colored by local material speed — cord moving faster than sound ignites in teal. The whole event is under a fifth of a second, shown in solver slow-motion: the whip starts cast back over the hand, the arm strokes forward while the wrist sweeps up and over, and the tip fires through two hairpins — one at the bottom of its dive while the hand is still moving, one over the top when the rolling loop runs out of whip. The big readout is a peak-hold of the solver’s finely-sampled tip trace (each flick lasts under a millisecond — faster than the animation’s frames); the bar shows the instantaneous tip speed. Tube radius is exaggerated 4× for visibility. The geometry is a generic, self-authored bullwhip profile, not any maker’s design.

A whip is a chain that remembers your wrist

Lay a bullwhip out straight and it is nothing but a taper: a handle about 25 mm across, a long thong thinning to 8 mm, a thinner fall, and a final wisp of cord — the cracker — barely 2 mm across. Ours is 2.0 m long and weighs 319 grams, and along that length the mass per unit of length collapses by a factor of about 156. We modeled it literally as what it is: a chain of 800 short links (1,600 in the refinement run), each one a tiny truncated cone matching the local diameter, joined end to end at joints that carry tension and bending — so whatever the handle does, the chain has to relay it, link by link, all the way to the tip. That relay is the whole machine. A wave traveling toward the tip keeps its momentum and energy while the links carrying them get lighter and lighter, so the motion has to get faster and faster.

And this time we threw it with a wrist. Earlier drafts of this study dragged the handle forward in a straight line, and the whip answered with an honest but tame wave — the model never quite cracked. A real crack is thrown: the arm strokes and the wrist turns the handle over, and that rotation is what races down the chain. So the drive is now a composite of both, each held to human numbers: the grip accelerates through a 600 mm stroke to 10 m/s (inside the published 5–12 m/s range), while the handle sweeps 160 degrees up and over — peaking near 4,200 deg/s, between a hard wrist snap and a pitcher’s arm — and everything arrests together in 4 ms. The prescribed motion is exact: each handle node follows the analytic arm-plus-rotation path, and the solved handle-tip speed matches the hand calculation to five digits (22.281 m/s), a receipt that the drive we described is the drive the solver ran. The solve is planar, gravity is on, and there is deliberately no mass scaling: this study’s product is a speed, and mass scaling fabricates kinetic energy. Before trusting anything we gated the rig: a free-fall check landed within 0.003% of the exact answer, and a uniform cord under known tension carried a transverse pulse at 59 m/s against a 50 m/s string-theory value — a +19% gap that is the bending stiffness of a 6 mm rod adding to the string speed, exactly the kind of honest discrepancy you want a gate to surface rather than hide.

Segment rotation at nine stations along the whip over time, showing the handle's 160-degree sweep amplifying to over 900 degrees at the tip — The receipt that the chain relays the wrist: local segment rotation at nine stations from handle to tip. The handle sweeps 166 degrees (the prescribed flick plus a whisker of elastic overshoot); each station down the chain turns further and later than the one before, and by the tip the same twist has compounded into more than **900 degrees of rotation — the handle’s motion amplified 5.4×** by the time it reaches the lightest links. The two cracks fire where the darkest trace goes vertical.

Flick, wave, snap — twice

The tip’s own trajectory tells the story. For the first 100 ms it loiters near where the cast left it, drifting while the arm strokes and the whip straightens. Then the wrist wave arrives. The tip is hauled into a sweeping dive — most of a meter of drop in about ten milliseconds — and at 121.3 ms, three milliseconds before the hand has even stopped, it snaps through a hairpin at 594 m/s on the baseline chain and 570 m/s on the twice-refined one: Mach 1.66–1.73. That is the crack. But the throw is not done: the arrest dumps the arm’s remaining momentum into the fold, the whip unrolls skyward, and as the rolling loop runs out of whip over the top the tip turns over a second time — 452–462 m/s, Mach 1.32–1.35, at 145 ms. Two cracks from one throw, which any whip handler will tell you is exactly what a good overhead crack sounds like when it double-taps. Across both bursts the tip spends roughly half a millisecond supersonic — violent brevity being precisely why a crack sounds like a crack and not a whoosh.

Map of the whip tip's solved trajectory in the vertical plane, colored by tip speed, with the first supersonic hairpin zoomed — The motion of the end of a whip, mapped: the solved tip trajectory over the whole event, colored by tip speed. The tip loiters through the stroke (pale tangle, center), dives as the wrist wave arrives, and fires through the supersonic hairpin of crack 1 (inset) while the hand is still moving; it then rides the unrolling whip up and over the top into the slower, second hairpin of crack 2. Time ticks show the pacing: a hundred milliseconds of loiter, then both cracks inside the next 45.

The result: a human throw — 10 m/s arm stroke plus a 160° wrist flick — drives the tip of a 2 m tapered whip through two supersonic cracks: 570–594 m/s (Mach 1.66–1.73) and 452–462 m/s (Mach 1.32–1.35), the peak agreeing to 4% across a 2× mesh refinement and every number taken only where the run’s energy accounting closes to within 3%. The identical throw into a uniform whip of the same length and total mass peaks at 86 m/s — seven times slower and never within shouting distance of Mach 1. The taper is the engine; the wrist is the ignition.

Tip speed against time for the tapered whip at two mesh resolutions and for the uniform control whip, with Mach 1 marked and both cracks annotated — The tip’s speed history from the solver’s own nodal velocity output, sampled every 0.02 ms. Crack 1 fires at 121.3 ms — inside the wrist’s sweep, before the hand arrests — and crack 2 at 145 ms as the loop runs out of whip; between and after them the light end rings audibly on the plot. The two mesh resolutions agree on the story, the timing (to 0.02 ms), and the first peak to 4%. Each trace is drawn only while its energy accounting closes to within 3%; the uniform-whip control (gray) never gets out of second gear on its way to a late, gentle 86 m/s.

The taper is the engine, and the control proves it

The claim “the wave accelerates because the whip gets lighter” deserves a control, not a narration. So we ran the same cast and the same arm-plus-wrist throw into a whip of the same length and the same 319 g total mass, but with the taper removed — a uniform 13.6 mm cord. Same hand, same stroke, same wrist, same arrest, same gravity, same material, same starting fold. Its tip manages 86 m/s, a factor of seven below the tapered whip’s crack 1, and it never comes near the barrier. Put the two side by side and the mechanism is not subtle: one of these is a momentum funnel, the other is a rope.

The causal receipt, played together in solver slow-motion: identical cast and throw into the tapered whip (top) and the mass-matched uniform whip (bottom), both colored by local speed on the same scale — supersonic cord ignites in teal. Both feel the same wrist; only the tapered chain compounds it. Line thickness tracks the local diameter.

Space-time map of transverse speed along the whip, showing the wrist wave amplifying as it approaches the tip and both cracks — Every node’s transverse speed, plotted over arc position and time: reading upward is reading toward the tip. The wrist wave crosses mid-thong carrying about 16 m/s of material speed and arrives at the cracker carrying about 86 — five-fold amplification earned purely by running into lighter links — then crack 1 fires at the top edge before the hand-arrest line, and crack 2’s chevron follows at 145 ms. This is the tapered-cord amplification that whip theory predicts: the wave doesn’t coast, it compounds.

An honest band, and honest wreckage

First, the good news the gates delivered. Halving the link length from 2.5 mm to 1.25 mm moved the headline peak by 4.0% — 594 to 570 m/s — inside our planned ≤5% mesh-independence gate, with the crack timing agreeing to two hundredths of a millisecond. Earlier, gentler versions of this model failed that gate; the wrist-driven crack is better behaved because the flick is bigger and earlier, and the supersonic verdict no longer hangs on which mesh you believe. We still publish the band, not a favorite: Mach 1.66–1.73.

Now the wreckage, because a hairpin this violent has consequences. Six of the seven whip runs eventually blew up numerically at the near-massless cracker — LS-DYNA terminated them on out-of-range velocities — every one of them at or after its own crack. So every number in this post is energy-gated: a peak counts only while the run’s total-energy accounting closes to within 3%, only up to the last recorded energy audit, and never within 2 ms of a blow-up. That gate has teeth, and what it bites is instructive. The raw output of the baseline run “peaks” at 8,300 m/s at the instant of its death — numerical garbage, discarded. The 7 m/s throw died mid-crack, so its clean window ends at 317 m/s while the tip was still accelerating: we report it as a floor, not a peak. The softest whip (E = 100 MPa) never reached a clean crack at all and is excluded outright. What survives the gate behaves like a whip: throw harder, crack harder — 7, 10, and 12 m/s arms give ≥317, 594, and 869 m/s of tip speed. That last number is Mach 2.53 with clean energy closure, solved at one resolution only in a drag-free plane, so we read it as a trend, not a certified speed — but it brackets the shadowgraph measurements of real expert cracks near Mach 2 from above, exactly where a model with no air resistance should sit. Stiffness, meanwhile, tunes the machine: the stiff whip (E = 500 MPa, 405 m/s) resists the tight terminal turnover and undershoots the 300 MPa baseline. A real whip is not just tapered; it is tuned.

Left: peak tip speed against arm speed with the measured real-whip band; right: the study's gates and receipts — Left: where the solved peaks land against Mach 1 and the measured real-whip band. The 7 m/s marker is open with an arrow because its run died mid-throw — the value is a floor. Right: the receipts — the units and wave-speed gates, the twist-relay check, the drive cross-check, the enforced energy-closure rule, the mesh gate the crack finally passed, and the honest accounting of six cracker blow-ups.

The same physics that snaps tow cables

A whip is the recreational version of a serious failure mode. Any long, tension-carrying line whose mass per length drops toward a free end — a snapped tow cable, a broken crane hoist line, a mooring rope parting under load — is a taper-amplifier waiting for a wave, which is why cable snap-back zones are marked on ship decks and why a parted line can be lethal at its free end. And the wrist lesson generalizes: what the free end does depends not just on how hard the anchored end moves, but on how it turns — the rotation you put into a line is relayed and amplified right along with the translation. Explicit transient dynamics is the tool that turns that intuition into numbers: launch the real wave into the real taper and read off what the free end does.

Honest scope. The whip is a generic, self-authored taper (handle/thong/fall/cracker, 25 → 2 mm over 2.0 m), not any maker’s geometry, modeled as a chain of homogenized elastic links — no braid mechanics, no fraying, no failure. The starting shape is an idealized cast: the whip folded back over the hand with the two legs 200 mm apart, a snapshot of the instant after the throw, with no self-contact modeled (the unrolling never requires the legs to touch). The throw is a prescribed, human-bounded handle motion — a 10 m/s arm stroke with a 160° wrist sweep, not arm biomechanics — and the solve is planar (the render is 3-D; the hero’s tube radius is exaggerated 4× for visibility). There is no air in the model: no drag, no acoustics, and therefore no solved sonic boom — the tip-speed-to-shock link is the measured result of Krehl, Engemann & Schwenkel (1998), who photographed the shock and clocked real tips near Mach 2; tapered-whip theory (Goriely & McMillen, 2002) supplies the amplification mechanism, and a drag-free plane should — and does — overshoot the measured band at the hardest throw. The headline peak is a 570–594 m/s band across a 2× mesh refinement (−4.0%, inside our ≤5% gate); the supersonic excursions are sub-millisecond. Every published number is energy-gated — peaks count only inside ±3% total-energy closure, before the last energy record, and 2 ms clear of any blow-up — because six of the seven whip runs error-terminated at the near-massless cracker at or after their crack: their contaminated tails contribute nothing, the 7 m/s run’s peak is reported as a floor, the soft-stiffness run is excluded, and the 12 m/s run’s Mach 2.53 is a single-mesh trend point, not a certified speed. No mass scaling, no global damping; tip speeds are the solver’s own nodal velocities. No sound is synthesized — the crack stays in the physics, not the speakers.

Have a long, light, fast-moving structure — or a cable that could part? The workflow behind this study — Ansys LS-DYNA solving the transient wave mechanics of a tapered line end to end, gated on closed-form checks and anchored to published measurement — is how Rand Simulation turns “everyone knows it cracks” into numbers an engineering team can act on. That’s innovation through insight.