São Paulo WorldGolf aerodynamics

Golf Ball Aerodynamics

Slow-motion wind tunnel · dimples · drag crisis · Magnus lift
Smoke · speed |V|/V∞0 → 1.8 (126 m/s) Skin · pressure Cp−1.2 → +1 separation point 1/2000× · t = 0.000 ms · contact · compressing

Drag crisis · Cd vs Re

No-spin curves. The golf ball's crisis comes at Re ≈ 5×10⁴ (≈ 18 m/s); a smooth sphere's at ≈ 3.3×10⁵ (≈ 116 m/s), faster than any golfer can hit. Dots: current speed; rings: with spin.

Magnus lift · Cl vs S

S = ωr/V. Dimpled: Cl = 0.47 S0.42 (fitted to tour flights). Smooth: estimated at half. Shaded: beyond the fitted range.

Flight · same launch, smooth vs dimpled

What you're seeing

Each shot starts at the instant the clubface strikes: the ball flattens by several millimetres and springs back over about half a millisecond (close to a second on screen), leaving with its speed and backspin. From then on the camera rides with the ball, so the air flows left to right past it as in a wind tunnel, slowed two thousand times. At the strike the smoke hangs still; the ball's start sets it moving and the wake grows from nothing. Lines of smoke are released from a rake on the left; each line is a streakline, the path a puff of smoke takes, coloured by the local air speed.

The lesson: where the air lets go

Air slides around the front of the ball, speeds up over the shoulders, and then has to slow down and climb back to high pressure on the back. A thin boundary layer next to the surface does not have the energy to do that for long. Where it gives up, it separates (white dots), leaving a low-pressure wake that pulls the ball backwards. That is most of a golf ball's drag.

Smooth ball: its boundary layer stays laminar, orderly but feeble, and separates at about 82° from the front. The wake is wider than the ball and the drag coefficient is about 0.48.

Dimpled ball: the dimples trip the boundary layer into turbulence. Turbulent mixing pulls fast outer air down to the wall, so it hangs on until about 113°. The wake is roughly half as wide and Cd falls to about 0.22 (0.24 with driver spin): the drag crisis, moved from Re ≈ 3×10⁵ down to ≈ 5×10⁴, inside the speeds golfers actually hit.

Backspin

A ball struck with backspin turns so its top moves with the air and its bottom against it. The flow stays attached longer over the top and lets go sooner underneath, so the wake is thrown downward. Pushing air down pushes the ball up: Magnus lift, the force arrow pointing up, often bigger than the ball's weight off the tee.

The close-up

The circle magnifies the boundary layer at one point (click the ball's edge to move it; the default, 100°, is between the two separation angles). It shows the velocity profile: laminar (Pohlhausen) or turbulent (1/7-power), and, past separation, the reversed flow next to the wall. On the dimpled ball, little eddies trapped in each dimple and shed from its rim keep stirring the layer.

How this is modelled

model physics models or measured fits   estimate our estimate, plausible but not from a single source   illustrative drawn to explain, not computed

  • modelCd(Re): a subcritical sphere curve blended into a post-crisis curve in log Re; smooth crisis at 3.3×10⁵ with minimum ≈ 0.07 (Achenbach 1972); golf ball crisis at 5×10⁴, then ≈ 0.22, nearly flat (Bearman & Harvey 1976; Choi, Jeon & Choi 2006). The curve shapes are smooth fits to those data, not the data themselves; the golf ball's post-crisis value (0.221) is fitted to measured flights (below).
  • modelSpin: Cl = 0.47 S0.42 and Cd += 0.26 S, the power-law form of Smits & Smith (1994) with coefficients fitted to TrackMan PGA Tour average flights of eight clubs (launch S ≈ 0.08–0.46; beyond it is extrapolated, flagged in amber). They agree with wind-tunnel data (Cl ≈ 0.18, 0.24, 0.28 at S = 0.1, 0.2, 0.3). The earlier 0.54 S0.4 with 0.18 S put every apex 5–10 yd too high. estimate smooth-ball lift = half of that: real smooth spheres can even lift the wrong way near their crisis (Kim et al. 2014).
  • modelSeparation angle 82° laminar, 113° dimpled-turbulent, 120° smooth-supercritical (Achenbach 1972; Choi et al. 2006), switched by the same crisis curve. estimate the top/bottom shift with spin (up to ±20°, saturating with S) and the wake width (from the separation line's angle).
  • modelThe flow is a 2-D cross-section: free stream + doublet (the ball) + bound circulation sized from Cl, plus a discrete-vortex wake. Vortex blobs are released at the imposed separation points at the rate the boundary layer feeds vorticity (½Ue², reduced ×0.6 as is customary), with mirror images inside the ball so no air goes through it, and every blob moves with the velocity all the others induce (Clements 1973). The shedding, the wake width, its downward tilt with spin and the Kármán-street rhythm (Strouhal ≈ 0.2) come out of that model, not out of an animation.
  • modelSurface Cp: attached sin² shape to a pressure minimum, recovery, then constant base pressure; shaped after Achenbach's measurements and tuned so its pressure-drag integral matches the no-spin Cd above (0.47 vs 0.48 smooth, 0.221 vs 0.221 dimpled).
  • estimateBoundary layer thicknesses use flat-plate correlations along the surface (δ ≈ 5x/√Rex laminar, 0.37x/Rex0.2 turbulent). In the main view the layer is drawn about 5× thicker than life so you can see it slow the smoke.
  • illustrative small-scale jitter in the wake and turbulent layer, the eddies in the close-up, and the dimple pattern on the ball. A 2-D model cannot resolve real dimples or 3-D sphere wakes, which form helical vortex loops rather than a flat street.
  • modelTrajectory: point mass, gravity + drag + Magnus lift (RK4), Cd and Cl recomputed every step from the current speed and spin; no wind, sea level. estimate spin decays with a 25 s time constant (the fit is insensitive to it). Against TrackMan tour averages for eight clubs (carry · apex · landing angle): driver 263 yd vs 275 · 35 vs 32 · 40° vs 37°; 7-iron 172 vs 172 · 31 vs 32 · 47° vs 50°; PW 130 vs 136 · 32 vs 29 · 53° vs 52°. Every club carries within 5%, peaks within 4 yd and lands within 6°; validate.mjs re-runs these checks.
  • modelImpact: a 2-D rigid-body oblique collision of the ball with the club head. The head has its mass, moment of inertia and centre of gravity behind the face, so a strike above the CG turns the head and takes spin off (the vertical gear effect). Normal restitution en 0.83 (driver, the rules limit) or 0.78 (irons). Tangential restitution et = 0.33: the squashed ball springs back and overspins (Cross 2005); the classic "rolls off the face" model (et = 0, Penner 2003) under-spins irons by 30%. There is a Coulomb friction limit (μ = 0.4). With et fitted once, iron spin comes out within ±10% of tour data from 5-iron to wedge. The driver spins about 20% too much and gives 4% less ball speed than TrackMan. The club delivery (speed, attack, dynamic loft) is solved so the ball leaves with the slider values. estimate contact time 0.45 ms at 50 m/s (Hertz scaling v−1/5), the force following the deflection; head properties (driver 203 g, CG 38 mm deep, struck 12 mm above it; 7-iron 265 g).

About the physics · sources

  • E. Achenbach (1972), Experiments on the flow past spheres at very high Reynolds numbers, J. Fluid Mech. 54, 565–575.
  • P. W. Bearman & J. K. Harvey (1976), Golf ball aerodynamics, Aeronautical Quarterly 27, 112–122.
  • J. Choi, W.-P. Jeon & H. Choi (2006), Mechanism of drag reduction by dimples on a sphere, Physics of Fluids 18, 041702.
  • A. J. Smits & D. R. Smith (1994), A new aerodynamic model of a golf ball in flight, Science and Golf II, 340–347.
  • J. Kim, H. Choi, H. Park & J. Y. Yoo (2014), Inverse Magnus effect on a rotating sphere: when and why, J. Fluid Mech. 754, R2.
  • R. R. Clements (1973), An inviscid model of two-dimensional vortex shedding, J. Fluid Mech. 57, 321–336.
  • R. D. Mehta (1985), Aerodynamics of sports balls, Annu. Rev. Fluid Mech. 17, 151–189.
  • A. R. Penner (2003), The physics of golf, Rep. Prog. Phys. 66, 131–171.
  • R. Cross (2005), Bounce of a spinning ball near normal incidence, Am. J. Phys. 73, 914–920; and R. Cross (2002), Grip-slip behavior of a bouncing ball, Am. J. Phys. 70, 1093–1102.
  • TrackMan PGA Tour averages (club and ball speed, attack angle, launch, spin, apex, landing angle, carry), as published by TrackMan.
  • H. Schlichting & K. Gersten, Boundary-Layer Theory (Pohlhausen profiles, 1/7-power law).

Coefficients are as commonly quoted from these works and smoothed for display; treat the curves as representative of a typical modern ball, not of any one ball. Ball: D = 42.67 mm, m = 45.93 g; air ν = 1.5×10⁻⁵ m²/s.