NACA Sections

Rick, I invite you to please read the papers enclosed so far, A&D book included before going any further with this story about the Mach number. It truly embarasses me to read these things.

 

Peter
You may be interested in a more like-for-like comparison of a NACA0012 foil operating at Re#2.88e6 and Mach 0.16. I have not checked if I have modelled all the test parameters. The JavaFoil data is only for smooth surface so it should be compared with the appropriate test curve.

I should add that there was a fault in the test set up as well in the earlier run where the angle measurement could have an 0.5 degree error. I think that is why the session 3 data is included as well.

Rick

 

If Rick is referring to compressibility effects in Abbott & Doenhoff, then he's correct. See here:
http://groups.yahoo.com/group/xfoil/message/5000
http://groups.yahoo.com/group/xfoil/message/3133

Summary for those that don't feel like chasing the links:
Many "low speed" tunnel tests actually have locally supersonic flow at the leading edge at high alphas due to the intense pressure spike, at least for airfoils with small leading edge radii. This can cause enormous variations in computed CLmax when the tunnel speed is varied between Mach=0.10 to Mach=0.20 say. Normally such a Mach number change is considered irrelevant, but in some cases it can often explain the large variation in measured CLmax between different tests.

Related factoid:
A typical jetliner landing at Mach=0.2 can easily have locally-supersonic flow over slat during the flare.

 

Mark
I guess one could take a hint from the paper that they were making an effort to get a foil into the same Mach number as the real heli blade they wanted to develop. I could not imagine any reason for going to this effort if it made no difference to the result.

The pressure is not being clipped as you get with pointier foils or higher Mach on the rounded nose. I have attached the Mach curve.

I have also attached the boundary layers for the two cases. I have not made any effort to understand the reason but there is substantial difference due to the inclusion of the Mach number.

I wonder what Xfoil produces for this situation? (I have never taken the time to get to know it but Tom Speer has suggested I should)

Rick W

 

Some time ago, I recall seeing a basic program (Excel) for calculating displacement, weight, etc. for a bulb.

Does anyone have the link?

Just a simple amateur...

 

- deleted the link for Meebles, the software was only for model boats. sorry. -
However, if you are good in math, you can quite easily calculate the surface and the volume of a generic body-of-revolution bulb via the Guldino (or Pappus') Theorem: http://mathworld.wolfram.com/PappussCentroidTheorem.html
Once you know the volume, the calculation of the weight is straightforward.
If you do not feel sufficiently good in math to try it out, just let me know. I'll help you out.

 

Hello mr. Drela,
Well, didn't look to me like that was Rick's intention, but if it was, then he wasn't good at explaining his thought. Here is the phrase:

A high Mach number (which in wind-tunnel testing is a free-stream Mach number, M_inf, not a local Mach number at some point on the body surface - unless explicited differently) is anything up to hypersonic speeds. M_inf=0.16 is not a high, but a low subsonic free-stream Mach number.

That said, I reckon that local compressibility effects are important, especially at high angles of attack. In fact, for many foils it is not rare to see the Cp peak to -12 or more at high AoA's - which means a local Mach number of around 0.6 when M_inf is, say, 0.15.

But, if I have not understood it wrongly, both Javafoil and Xfoil should be equipped with appropriate compressibility correction methods, be it Prandtl-Glauert or Von Karman-Tsien or yet another one.
(If that is not the case, I think I will faint right here in front of the PC because I am extensively using an Xfoil-based software for a current project of mine... )

And, back to the case of NACA0012, I have analysed it with QFLR5 (which is the Xfoil engine plus a very nice GUI, so the output should be identical to Xfoil's) and have found out that the agreement of Cl and Cd with experimental data (both A&D and Gregory&O'Reilly) is really good, when Ncrit=6 is used for turbulent transition prediction. The Cm is well-predicted for AoA up to 3-4° and then blows off.
All in all, definitely better than Javafoil results seen before - particularily when Cl-Cd curves are compared. Damn, if Xfoil only had the possibilty to deal with multiple lifting surfaces...

I enclose the resulting graph, for M=0.00 and M=0.16 (basically the same curves, except for the Cm at high AoA). It is good all the way to the stall, correctly predicted at 16° AoA.

 

I would like to learn more about this discrepancy of wind-tunnel data at high AoA's due to local compressibility effects, at low free-stream mach numbers.
Could you please point me to some research papers which address this issue?

 

Where did the 16 degrees come from?

How do you know it is "correct"?


Rick W

 

I agree that there isn't much effect on the 0012, which has a fairly large LE radius. But it's much more pronounced on thinner "laminar" 6-series sections that might appear attractive for smaller boat keels.
See the two PDF plots.
At M=0.16, the effect is noticable for the N64-010, and VERY noticable for the N66-010.
The effect on the NACA 0009 is similar to that of the N64-010.

 

Ok, I see what you intend. Could you please help me understand beter the physical explanation of said effect of local compressibility on Cl,max and on Cl-Cd curve? Is it just because the absolute value of Cp increases on the suction side, as indicated by Prandtl-Glauert (or another similar) correction factor, thus creating a more adverse pressure gradient downwinds, or are there some other mechanisms involved?

 

I don't know of any papers that specifically examine this effect. But I'm not sure what any such paper would prove. I'm only trying to make this point:
When comparing experimental and computed CLmax values, it might be important to run the calculations at the same Mach number as the experiment, even if the wind tunnel test was "low speed". My two previous PDFs clearly prove that.

The physical reasons for the effect are clear. As the local M in the Cp spike approaches 1.0 it does two things:
1) It is "self-reinforcing", and increases the Cp spike even higher than the incompressible value. This is mostly captured with a Prandtl-Glauert correction, and somewhat more accurately with a Karman-Tsien correction.
2) It reduces the BL's resistance to adverse dp/dx, primarily through a temperature and corresponding density gradient across the BL thickness which appears at significant local Mach number. This is captured by any compressible boundary layer formulation.

Both of these effects promote earlier separation and stall, and the physical reasons are 100% understood. There is no mystery here.

XFOIL captures both 1) and 2).
It appears that Javafoil captures 1), and I don't know about 2).

 

Perfect, very clear explanation. Thank you.