Unusual Foil Requirement - Thoughts?

[jehardiman]

Originally Posted by daiquiri

Your requirements lead towards a body of revolution, a submarine-body or a dolphin-body shape, rather than towards a foil. See this paper too: http://cafefoundation.org/v2/pdf_tec...-48131-445.pdf

You need to stress that one should be very careful with that paper. What they were doing was a comparison of drag to low Reynolds (1x10^7) deep ocean drop bodies. Check out the first line of section II

Once the reference Reynolds number Rv is fixed for the
zero incidence, nonseparating flow, the value of the drag
coefficient CD depends only on the shape Y(X) of the axisymmetric
body

Boat sized bodies traveling horizontaly near the surface are rarely in either of those cases. For an analysis of this situation as applied to real ship design see this paper. http://www.dtic.mil/cgi-bin/GetTRDoc?AD=ADA470163

[tspeer]

Originally Posted by PI Design

Hi,

I have an unusual foil brief. I thought I would throw it open to ideas for appropriate sections.

It needs to be highly buoyant (so tending towards a fat section, >25%), low drag and low lift ( for AOA 0-10 degrees). It also has to be very low aspect ratio, no more than 1.0. Reynolds number approximately 1.5e6.

Any thoughts on suitable sections very welcome!

You could use something like XFOIL to design a section specifically for your requirements. I would set Ncrit=1 and trip the flow close to the leading edge so as to simulate a fully turbulent boundary layer. If you get any laminar flow in real life, that would be a bonus.

Even though fully turbulent, you should probably adopt a roof-top pressure distribution at your design lift coefficient to minimize the incipient cavitation speed. Use a short rounded transition segment between the rooftop and the pressure recovery regions so that if you were to get laminar flow, it will only cause a short laminar separation bubble.

The big tradeoff will be how far you can carry the rootop vs how aggressive you dare to be on the pressure recovery. The longer the rooftop, the farther back the maximum thickness will be and the thicker the section can be for a given cavitation speed (height of the rootop pressure distribution), but the more prone the tail will be to separation. Being a low aspect ratio shape, you can expect spanwise flow over a fair amount of the trailing edge, and this may lead to earlier separation than the 2D results would predict, so you need to build in some margin. The pressure recovery will be steepest at the start, and then flattening out, which means either a straight line or concave contours.

A squared off trailing edge will be easier to build and give you a larger thickness ratio. There will be some base drag as a result, but there's a limit to how rapidly you can close the shape down from the maximum thickness and if you try to bring it to a sharp trailing edge it will probably separate much earlier. So you might as well accept that you aren't going to have 100% attached flow and ensure that the flow is attached to the trailing edge. But with a thickness ratio of 25%, you shouldn't need a very thick trailing edge. (Compared to going for a thickness ratio of, say, 40%!)

The shape that BMcF posted looks to be the product of this kind of design approach.

[DCockey]

Originally Posted by tspeer

You could use something like XFOIL to design a section specifically for your requirements. ......

How much difference does the combination of the aspect ratio of no more than 1.0 and "fat section, >25%" make to the pressure and velocity distribution compared to a 2D assumption such as used by XFOIL?

I would assume with the combination of aspect ratio of no more than 1.0 and "fat section, >25%" the "tip" shape would be as important to the drag as the centerline profile.

[tspeer]

I agree that the tip shape would be important, as would the amount of lift. At least the 2D shape would be a starting point.

Once the initial shape was created, the simplest way to get the 3D effects would be with a panel code like CMARC or VSAERO. The boundary layer characteristics along surface streamlines could then be used to guide modifications of the shape.

[daiquiri]

Are you guys aware of any reliable experimental data of very low AR wings or lifting bodies, which could be used as a benchmark for the evaluation of panel codes?

[PI Design]

Thanks very much for all this. Much to take in and digest. Is java foil as suitable as xfoil - it seems to have an easier interface?

[Leo Lazauskas]

Originally Posted by daiquiri

Are you guys aware of any reliable experimental data of very low AR wings or lifting bodies, which could be used as a benchmark for the evaluation of panel codes?

There are some for thin lifting surfaces in a pdf attachment to a previous thread:
http://www.boatdesign.net/forums/att...lat_draft1.pdf

Low Rn effects are likely to introduce great scatter in experiments of thick
lifting surfaces. Looking at the scatter of results for the drag of airfoils, I
imagine it will be even greater for low AR thick bodies.

Good luck!
Leo.

[daiquiri]

Thanks Leo, I'll take a deeper look at that paper. From the first glance (and from some of my previous tries) it looks that without a LE suction model, panel codes do not perform well for this kind of geometries. Is that correct?

[Leo Lazauskas]

Originally Posted by daiquiri

Thanks Leo, I'll take a deeper look at that paper. From the first glance (and from some of my previous tries) it looks that without a LE suction model, panel codes do not perform well for this kind of geometries. Is that correct?

That is correct. Results might be better with a non-linear version, e.g. as
developed by Bollay or Gersten, but they introduce other difficulties to do
with the assumed location of the wake. Adding thickness is not going to
make the problem much easier, even if it does get rid of some of the
difficulties arising from the sharp edges.

I guess it all depends on the range of AoA one is interested in. And, of
course, whether the edges are straight or curved. As I have emphasised
before, Vortex Lattice Methods do very well for straight leading edges and
trailing edges, but they can give complete rubbish for planforms with curved
LE and TE.

[DCockey]

No conclusions should be made about the applicability or accuracy of panel codes such as VSAERO and CMARC (which Tom Speer suggested above) based on Leo's paper about the solution of the Lifting Surface Intergral Equation compared to experimental results. VSAERO and CMARC are fundamentally different than the methods discussed by Leo.

The mathematical formulation used by VSAERO, CMARC and similar codes has the normal velocity boundary condition applied "exactly" at the actual surface of the airfoil, wing, body, etc. The surface is modeled by a network of panels on the actual surface.

In contrast to obtain the Lifting Surface Intergral Equation the normal velocity boundary condition is approximated on the xy plane, not the actual airfoil surface, using a "thin airfoil" assumption. This results in the square root singularity at the leading edge which in turn leads to discussion of "leading edge suction". "Panel codes" used to solve the Lifting Surface Integral Equation approximate the airfoil with a planar distribution of panels which approximate the planview of the airfoil but not the thickness.

[Leo Lazauskas]

Originally Posted by DCockey

No conclusions should be made about the applicability or accuracy of panel codes such as VSAERO and CMARC (which Tom Speer suggested above) based on Leo's paper about the solution of the Lifting Surface Intergral Equation compared to experimental results. VSAERO and CMARC are fundamentally different than the methods discussed by Leo.

The mathematical formulation used by VSAERO, CMARC and similar codes has the normal velocity boundary condition applied "exactly" at the actual surface of the airfoil, wing, body, etc. The surface is modeled by a network of panels on the actual surface.

In contrast to obtain the Lifting Surface Intergral Equation the normal velocity boundary condition is approximated on the xy plane, not the actual airfoil surface, using a "thin airfoil" assumption. This results in the square root singularity at the leading edge which in turn leads to discussion of "leading edge suction". "Panel codes" used to solve the Lifting Surface Integral Equation approximate the airfoil with a planar distribution of panels which approximate the planview of the airfoil but not the thickness.

Quite right, but they are the only reliable experimental results I have.
In fact, I don't think of experiments being very reliable until they have been
obtained in more than one wind tunnel. But that's just a personal bias.

[Leo Lazauskas]

Originally Posted by daiquiri

Thanks Leo, I'll take a deeper look at that paper. From the first glance (and from some of my previous tries) it looks that without a LE suction model, panel codes do not perform well for this kind of geometries. Is that correct?

Just an additional comment...
The section on circular planform wings has some experimental results that Mikko
also used for his CFD comparisons. Think of these wings as "coins" (i.e. with
squared-off edges) at AoA. They are low AR, are easy to panelise, and a panel
method should be able to handle them quite easily. But there are some very
difficult issues with flow separation, and how to handle the Kutta condition, if
one actually believes in Kutta condition, of course

[daiquiri]

Originally Posted by Leo Lazauskas

if one actually believes in Kutta condition, of course

Ok, is it necessary for me to ask the next question?

[Leo Lazauskas]

Originally Posted by daiquiri

Ok, is it necessary for me to ask the next question?

As a mathematician, I accept the Kutta condition as a means of picking out
one solution from infinite possibilities. Engineers can suit themselves
Or am I answering the wrong question?

[daiquiri]

Originally Posted by Leo Lazauskas

As a mathematician, I accept the Kutta condition as a means of picking out
one solution from infinite possibilities. Engineers can suit themselves
Or am I answering the wrong question?

No no, that was the question... I was affraid that Kutta condition is no longer valid, which would probably be the trigger for a radical change in my lifestyle. I'd probably start growing grapes, go fishing and play guitar on the streets to make my living.