Scroll through the story below. The gull on the right is a fit to the Flow360 runs, rebuilt in your browser, so at the end you can set the fold and angle of attack yourself and cut through the flow anywhere. The written summary follows it.
Why are birds' wings swept back?
A post in my feed asked exactly that, and answered: because evolution favoured the less draggy design, and your simulation can find it the same way.
There are plenty of CFD posts with beautiful flow fields that never say what question they answer.
"Results viewed in this way take on an air of authority out of proportion to their accuracy."Doug McLean, Understanding Aerodynamics (2012), on colourful CFD pictures; quoted in "Colorful fluid dynamics: say it again, I dare you" (Siemens Simcenter blog)
So here is one we can answer. Is a gull's sweep about drag, or about stability?
A gull is a flying wing with a small tail.
The original post's explanation doesn't smell right to me - sweep on airliners is at speeds markedly higher than how birds fly (caveat below). But if you can change your wing's shape, then you have variable stability - and perhaps that's the better lesson to take from birds/evolution.
Neutral point: the aircraft analogue of aerodynamic center - where the pitching moment doesn't change with lift. If your CG is behind this point, you are statically stable which means less effort to fly in a straight line, but more effort to manoeuvre.
A real gull wing, through a recorded fold.
I built the bird model from Harvey et al.'s measured landmarks on a glaucous-winged gull: twist, dihedral, their gull aerofoil sections and a body. It moves through 87 frames of one recorded fold, from tucked to fully open. Scroll to open the wing. It's fun.
First, does the CFD work at bird scale?
Gulls fly at a Reynolds number near , where the boundary layer separates and reattaches in a laminar bubble. So I first ran a slice of the Eppler 387 aerofoil, which NASA measured at exactly this Reynolds number.
Hover the chart for values. Switch to to see the pressure on the aerofoil.
Then nine gull wings from a wind tunnel.
Harvey et al. 3-D printed nine wing-body models of a second gull, each at a different elbow and wrist angle, and measured their pitching moment. I rebuilt all nine from their published data and ran them at the tunnel's speed and turbulence.
Half-folded, the wing sheds its strongest vortex well inboard of the tip.
Turn the cut from across the flow, where the tip vortex shows as a spot, to along it, where you see the wing's section and the wake it sheds. Slide it over the wing or far behind it. Colour is the vorticity through the cut, the pressure, or the change in speed.
Flow360 was run at 15 fold states. To fill in between them the flow is described in coordinates attached to the wing: span as a fraction of the tip, height above the wing's own surface. A vortex then sits in the same place at every fold, so any fold can be drawn, the same way the surface pressure is.
Fold the hand and the neutral point moves.
Tucked, the neutral point sits behind the shoulder. Fully open, it comes forward to . The centre of gravity barely moves. Scroll to open the wing.
CG from Harvey et al.'s inertia model of the same bird.
Birds have a "choose your own stability" dial.
So sweeping the hand back is how a gull tunes its stability in flight: tucked and swept for a steady glide in gusty air, open and less swept when it wants to manoeuvre. The same trade an aircraft designer makes once, at the drawing board.
Why are birds' wings swept back and bent up? Evolution settled on the shape with the least drag, and an optimiser running your simulations can find it the same way.
Paraphrased; not the author's words.Tucking the hand moves the neutral point 43 mm aft.
Gull geometry, CG and tunnel data: Harvey, Baliga, Wong, Altshuler & Inman, Nature 603, 648 (2022), and Harvey, Baliga, Goates, Hunsaker & Inman, J. R. Soc. Interface 18, 20210132 (2021), CC BY 4.0. Gulls tuning pitch stability by morphing the wing: Harvey, Baliga, Lavoie & Altshuler, J. R. Soc. Interface 16, 20180641 (2019), and, for dynamic stability, Harvey & Inman, PNAS 119, e2204847119 (2022). Eppler 387: McGhee, Walker & Millard, NASA TM-4062 (1988). All CFD: Flow360.
The short version
A common claim is that birds sweep their wings back for the same reason as jets, or to reduce drag. The better explanation is that a gull sweeps its hand back to move its neutral point aft and so set its own pitch stability, as a tailless aircraft does with sweep. Harvey et al. showed in 2019 that gulls tune their pitch stability by morphing the wing; this post is a way of seeing it, in Flow360, through one whole recorded fold.
I ran Flow360 on a real gull wing (glaucous-winged gull 20_0341, from Harvey et al.’s measured landmarks) through one recorded fold, from tucked to fully open, after validating the solver at bird Reynolds numbers. For the wing and body with no tail:
- Tucked, the neutral point sits about 120 mm behind the shoulder: a static margin of about +27 % of the mean chord.
- Fully open, it comes forward to about 77 mm: about +19 %.
- The neutral point travels 43 mm and the centre of gravity 13 mm, so folding and sweeping the hand works as a stability dial: very stable for a steady glide, less stable and quicker to manoeuvre when open.
Validation
- Bird-scale aerofoil. Eppler 387 at against NASA TM-4062. RANS follows the lift curve to within 30 lift counts from 0° to 8°; drag is 47 counts low at 4°, from slow recovery after the laminar bubble. A wall-resolved LES of the same slice is 15 counts above NASA (0.0247 against 0.0232).
- Nine wind-tunnel gull models. Harvey et al. (2021) tested nine wing-body models of a second gull (17_0285). Rebuilt from their published landmarks with their own geometry code (area, span, mean chord and tip twist match their records), Flow360’s neutral point is within 4.4 mm RMS of the tunnel (−3.3 to +7.8 mm); the tunnel’s own runs at two speeds differ by up to 7.1 mm. Lift slope is 11 % below the tunnel, most on the models with the most twisted-down hands.
Method
- 15 fold states, each at = 0°, 4° and 8°, on a mesh refined behind the wing: 45 runs. The neutral point at each fold state comes from the slope of pitching moment against lift, and a Gaussian-process fit gives it across the fold.
- The surface pressure and the 3-D flow field are fitted across fold and (POD plus Gaussian processes) so the page can draw any fold. The flow is described in coordinates attached to the wing, so a vortex sits in the same place at every fold. I checked the fit on two fold states left out of it.
Limits
- Pitch only: no sideslip, so nothing here on yaw or roll.
- The wing and body have no tail; a real tail would add margin.
- The specimen’s measured hand washout separates the flow over the hand at some fold states, so I don’t report lift-to-drag.
- One mesh level for the tunnel models, and only from −4° to +4° there.
- Harvey et al. (2019) found extending the elbow made the wing more stable, the other way round from here. They tested dried wings without a body, with the wrist held open, and took the pitching moment about the shoulder, with no centre of gravity. Their 2021 wing-body tunnel models, which Flow360 matches above, show the neutral point moving forward as the elbow extends, as it does here.
Sources
- Gull geometry, centre of gravity and tunnel data: Harvey, Baliga, Wong, Altshuler and Inman, Nature 603, 648 (2022), and Harvey, Baliga, Goates, Hunsaker and Inman, J. R. Soc. Interface 18, 20210132 (2021), CC BY 4.0; data doi:10.6084/m9.figshare.c.5275490.
- Gulls tuning pitch stability by morphing the wing: Harvey, Baliga, Lavoie and Altshuler, J. R. Soc. Interface 16, 20180641 (2019), doi:10.1098/rsif.2018.0641.
- Gulls controlling dynamic pitch stability by morphing the wing: Harvey and Inman, PNAS 119, e2204847119 (2022), doi:10.1073/pnas.2204847119.
- Eppler 387: McGhee, Walker and Millard, NASA TM-4062 (1988).
- All CFD: Flow360.