Turbulence & aerofoils

[richochet]

For Trebuchet03:- Thank you for your turbulence thread start of 11/10/2007 - terrific stuff!.

Later on you mentioned looking into de Havilland slots - jumping in on this I envisaged an aerofoil with a fairly blunt leading edge, at a moderate angle of attack and airflow velocity high enough for flow separation to be imminent. Because of the angle of attack, the incipient separation point on the upper surface would be further to the rear of the aerofoil than the one on the lower surface. Now, open a channel, a slot, from just downstream of that lower "about to be" separation point to just behind the corresponding point on the upper surface. Even if there is not much difference yet developed between the airflow pressures above and below the leading edge, the angle of attack will assist this small pressure difference in establishing a flow, upwards and backwards through that slot.

This flow, in pulling air into the slot from below, also pulls the general passing flow in towards the underside of the wing, thus preventing separation there and maintaining lift. On the upper surface, leaving the slot this airflow fills in the developing low pressure volume behind the incipient separation point. The air travelling from the leading edge over the upper surface has accelerated according to Bernoulli's principle, as usual, and is ready to separate. It seems likely that turbulence would develop between this separated flow and the upper surface; this turbulence would be more general than that of a turbulent boundary layer, but filling this space with the air leaving the slot and which accelerates because of the lower pressure prevents such turbulence from developing. Stall is defined in some differing ways "...airflow over the upper surface is broken", "...when the wake becomes turbulent". The common factor is an angle of attack greater than for steady lift. At a high angle of attack`the wake is, strictly, to the rear of the upper surface and the difference in those quotes defining stall disappears. Filling that turbulentwake zone with airflow from the slot will restore laminar flow and delay the stall condition, as well as reducing its increased drag.

Having pretty much satisfied myself of the principles, the next step had to be looking for references to de Havilland slots, but with little success, until a de Havilland site referred to "Handley Page" slots - that other British aircraft manufacturer, of Hampden and Halifax bombers and a number of very large flying boats of the 1930's with individual names beginning with "H". Then, NACA (predecessor to NASA) tech. note No. 423, dated July, 1932 was found; entitled "Effect of length of Handley Page tip slots on the Lateral-Stability
Factor, Damping in Roll" by Weick and Wensinger.

The first sentence of the Introduction "Handley Page type wing-tip slots have been found to give improved lateral stability at angles of attack above that corresponding to the stall of a plain wing.", does not tell us actually that Handley Page slots do anything to delay stall. The tech. note is actually the full report of tests carried out in a wind tunnel at the Aeronautics Laboratory at Langley Field, Virginia on a model of a Clark Y wing modified by removing a portion of the upper leading edge and remounting it in front of the remaining
leading edge and below it by about 35% of the wing depth. This portion is called up as "Auxiliary airfoil or slat" in Fig. 1 which illustrates the wing cross-section. The test purpose is quoted "...to determine the best length or position of the inboard end of the slot for good stability."

It appears that the proposal by Trebuchet03 to look into this matter may have been pre-empted by those people who have mounted undertrays or belly-pans to their cars to smooth the airflow under them and have then, in order to avoid generating uplift pressure there, have directed radiator discharge air rearwards and upwards through slots in the bonnet into the low pressure zone to the rear of the "leading edge" of the bonnet. Also, Lotus Elise and Exige appear to have air outlets for this air, even before fitting optional undertray, but these are approximately semi-circular in plan with the most forward part of the curve approximately midway between the front of the car and the base of the windscreen. This would not seem to provide discharge into a typical low-pressure zone. While these models have smallish frontal area because of their low height, their drag co-efficients are quoted as between 0.4 and 0.44 prior to fitting "Aero" options, but are not mentioned for
the cars as fitted with these options, unfortunately.

Trebuchet03 then put a question regarding the tripping of boundary layers into turbulence. This had to do with the clinging of such turbulent layers to a surface where laminar flows would not so cling. Take a look at NACA Report 1247, "Characteristics of Turbulence in a Boundary Layer with Zero Pressure Gradient" - P. S. Klebanoff. Apparently published by the National Bureau of Standards, Washington, on May 8, 1953, superceding NACA Tech. Note 3178,
although even numbered pages are headed Report 1247 - NACA. I had downloaded a copy some time ago but had been daunted by the intensity of the mathematics. There is no doubt that its conclusions are of significance. The May 1953 date seems suspect because one of its quoted references is dated 3rd December 1953 and TN 3178 also is given an later date, 1954.


NACA Report 1247 describes wind tunnel tests on a 12 feet long 1/4 inch plate mounted parallel to the flow. With airspeed at 50 feet per second, tripping the boundary layer into turbulence and artificially thickening it to assist instrumentation, had a result - 3 inches of boundary layer thickness was achieved at the working station 10 feet 6 inches downstream from the leading edge. The tripping and thickening was done by a 2 foot length of No. 16 floor sanding paper at the leading edge. Proving and checking this method is
recorded in NACA Report 1110 of 1952, archival quote - "Report gives an account of an investigation conducted to determine the feasibility of artificially thickening a turbulent boundary layer on a flat plate. A description is given of several methods used to thicken artificially the boundary layer. It is shown that it is possible to do substantial thickening and obtain a fully developed turbulent boundary layer, which is free from any distortions introduced by the thickening process, and, as such, is a suitable medium for
fundamental research."

It is to be hoped that this information will still be of use to Trebuchet03. Tripping a boundary layer into turbulence is not likely to require as much of a device (2 feet of sandpaper) as described above - artificial thickening would probably not last long outside of the carefully contrived environment of the wind tunnel and is not needed in order to form a smaller wake.

[GasSavers_RoadWarrior]

The advantage of slats or slots is to increase lift at high angles of attack and delay the stall point by delaying separation...

Although delaying separation sounds good, we don't really want to be moving the center of pressure back or increasing induced drag by tilting the lift vector backwards beyond vertical.

[richochet]

Originally Posted by RoadWarrior

The advantage of slats or slots is to increase lift at high angles of attack and delay the stall point by delaying separation...

Although delaying separation sounds good, we don't really want to be moving the center of pressure back or increasing induced drag by tilting the lift vector backwards beyond vertical.

In stall, the centre of pressure probably moves to where we don't want it anyway; some aircraft can be recovered from stall easily, others suffer sideslip, hence the 1932 NACA investigation into roll stability with slots. With swept wings, locating the centre of pressure is more due to their geometry.

Surely the lift vector is always vertical - up? In the wind tunnel, the resultant (of lift and drag) vector will tilt backwards. Out in the real world, in level flight the resultant vector must be straight ahead, from vectors of lift, - up, weight, - down, thrust, - ahead and drag, - backwards. On the road, downforce more use than lift.

[richochet]

Originally Posted by richochet

Out in the real world, in level flight the resultant vector must be straight ahead, from vectors of lift, - up, weight, - down, thrust, - ahead and drag, - backwards. On the road, downforce more use than lift.

Oh, h*ll, that wasn't well said. Reckon it's probably true enough, but clumsy! Look at it this way - the ahead thrust vector (force) needs to balance the backwards drag vector (force) at any particular speed, for a steady velocity, so the ahead velocity vector still has a positive figure, dammit it's moving forward without acceleration - but if the uplift exceeds the gravity downforce then the ahead velocity vector angles a bit upwards, and the aircraft climbs at a steady speed. Oh boy, I'm gonna leave it at that for the moment.

[richochet]

Yes, well, this thread could still have some relevance to road vehicles. There are a lot of them out there with rear spoilers - these are upside-down aerofoils, often with fairly high angle of attack. Even if not too close to stall, and achieving useful downforce, this can mean low pressure drag where it's not wanted, perhaps turbulence adding to drag behind vehicle. Any answers?