Marius Comper RO

Henri Coandă, Romanian engineer · patent 1934–1936 · measurements 1956–1999

The Coandă effect: at what angle a jet of air leaves a cylinder

A thin jet of air blown tangentially along a cylinder does not go straight on: it clings to the wall and, if it is fast enough, wraps more than halfway round before it breaks away. Choose the jet's thickness, the cylinder's radius and the speed, and the drawing shows where it leaves, according to published measurements.

Fekete (1963), measuredNewman's formula (1961)Shakouchi & Onohara (1989)Kadosch (1956)your jet

What the jet does, and why

A fast jet leaving a slot entrains the surrounding air. If there is a wall on one side, the air there cannot easily be replaced, the pressure drops, and the jet is pushed towards the wall and clings to it. On a curved wall, the jet follows the curve of the surface.

Henri Coandă, a Romanian engineer born in 1886 in Bucharest, patented the use of this effect. The US application of 19 April 1935, for a "Device for deflecting a stream of elastic fluid projected into an elastic fluid", claimed priority from a French application of 8 October 1934; US patent 2,052,869 was granted on 1 September 1936. Its drawings show a free jet, an inclined flap beside the slot and a curved guide that turns the jet right round; the patent gives no measured angles.

How far the jet stays attached to the cylinder

B. G. Newman published in 1961 measurements of the separation angle of an air jet directed tangentially along a cylinder. For thin jets (thickness 0.7% to 5.3% of the radius) his formula gives an angle between 225° and 242°. In 1963 G. I. Fekete repeated the measurements under Newman's guidance at McGill University in Montreal, on Newman's cylinder and on a new, precisely built one. On a rough cylinder the jet separated earlier.

Fekete found that for thin jets speed matters most: on the new cylinder, at a Reynolds number of 2,000 the jet separated at about 85°, at 10,000 at about 165°, and from about 30,000 the angle hardly grew any more: about 223° on Newman's cylinder and 208° on the new one. On Newman's cylinder the jet thickness (0.3–3.7% of the radius) had no visible effect; on the new one (0.7–5.3%) it had a small one. At high speeds, on Newman's cylinder, the angles from Newman's formula differ from Fekete's measured plateau by less than 10%; on the new cylinder the formula gives angles 17–34° higher than those Fekete measured.

Neuendorf and Wygnanski published new measurements in 1999, with separation around 230°. The angle also depends on the point it is measured from: on the new cylinder, at high speed, Fekete found about 210° counted from the slot and about 225° from the jet's virtual origin.

Why the measurements disagree

For thicker jets, different rigs give different results. Shakouchi and Onohara (1989) found 143–160° for thicknesses of 5–15% of the radius at about 80 m/s, angles calculated from the distance over which the jet stayed attached. In the 1956 measurements, made by Jean Liermann under Marcel Kadosch, a jet a quarter of the radius thick separated after 60°, but the curved wall covered only a 90° arc. When the jet was thicker than half the radius it was barely deflected: only the layer next to the wall separated locally, after about 20°.

So the drawing shows the measured angle only where comparable measurements exist. Beyond that the curve is an extension, and the chart marks the values published for the other setups as separate points.

The spoon under the tap is not the Coandă effect

The usual demonstration, the back of a spoon held under a running tap, is not the Coandă effect: the water flows through air, not through surrounding water it could drag along. Caroline Lubert (2011) suggests calling it the "teapot effect", as when tea runs down the underside of the spout, where how well the liquid wets the surface also matters.

The 1910 aircraft

In October 1910, at the Paris aeronautical show, Coandă exhibited an aircraft without a propeller, with a compressor driven by a piston engine. No contemporary report of a flight has been found: Frank H. Winter, a former curator at the Smithsonian's National Air and Space Museum, found none in the Paris newspapers and aviation journals he checked. Romania's early aviation pioneers, Vuia, Vlaicu and Coandă, have a page of their own.

Share this page

The link keeps the thickness, radius and speed you chose.

Sources and calculation

The angle is counted from the slot, round the cylinder. Up to a thickness / radius ratio of 0.053, the angle is read from the mean curve of Fekete's measurements (1963, new cylinder, four thicknesses from 0.0074 to 0.0531 of the radius, Reynolds numbers up to about 75,000), to within about 5–10°. The Reynolds number is a speed-based estimate (speed × √(thickness × radius / 2) / kinematic viscosity of air) of the one Fekete defined from the slot pressure. For thicker jets the curve is extended with Newman's formula, scaled to join Fekete's curve, and marked as unmeasured. The drawn particles follow the computed angle: they are not a flow simulation.

  1. G. I. Fekete, Coanda flow of a two-dimensional wall jet on the outside of a circular cylinder, Report 63-11, Mechanical Engineering Research Laboratories, McGill University, Montreal, 1963. mcgill.scholaris.ca Fig. 37 (Newman's cylinder, with Newman's points) and Fig. 38 (new cylinder): separation angle against Reynolds number, thickness/radius 0.0033–0.053.
  2. B. G. Newman, “The deflexion of plane jets by adjacent boundaries – Coanda effect”, in Lachmann (ed.), Boundary Layer and Flow Control, 1961; formula reproduced in M. Petrolo et al., Advances in Aircraft and Spacecraft Science 4 (2017), eq. (2). techno-press.org θ = 245° − 391° · (b/a) / (1 + 9/8 · b/a).
  3. R. Neuendorf, I. Wygnanski, “On a turbulent wall jet flowing over a circular cylinder”, Journal of Fluid Mechanics 381 (1999). doi.org Separation around 230° (abstract).
  4. T. Shakouchi, M. Onohara, Transactions of the JSME B 55 (1989), no. 511, Fig. 15. jstage.jst.go.jp 143–160° at thicknesses of 5–15% of the radius.
  5. M. Kadosch, “The curved wall effect”, 2nd Cranfield Fluidics Conference, 1967. marcel-kadosch.eu 60° at thickness/radius 0.25 (wall ending at 90°); local turning of about 20° above 0.5.
  6. H. Coandă, US patent 2,052,869 (1936). patents.google.com French priority 8 October 1934; granted 1 September 1936.
  7. C. Lubert, “On some recent applications of the Coanda effect”, International Journal of Acoustics and Vibration 16 (2011), p. 144. iiav.org The spoon under a tap is not the Coandă effect; the "teapot effect".
  8. F. H. Winter, “Coanda's claim”, Air & Space, Smithsonian. smithsonianmag.com No contemporary report of a 1910 flight.