Wind Tunnel -- initial testing speed?

[Bispen]

Hi!

I would like to submit a model glider to a CFD wind tunnel so I can determine what would be its optimum glide ratio and forward speed at this incidence.
I am not supposed to know in advance the range of speed this glider can fly (between stall and structural damage), so the speed I will initially set in the CFD software is arbitrary.
From documents I have read, the speed is given by the target Re number which has to be kept constant during the whole measurements at different conditions. Some reports mention something between 1.75e5 to 3e5.
How to choose then? And what about if I go beyond, won't it respect anymore the nature of the flow in reality?
If anyone could bring some clarifications and a recommendation for this test, it would be fantastic.
Thanks

 

[bigfooted]

Can you elaborate on what kind of CFD simulation you want to perform? Do you want to do 3D simulations of glider for a number of different Reynolds numbers/velocities? What kind of information do you want to get out of such simulations?
I also do not understand your reference to the 'documents you have read'. What kind of documents? How do they determine the conditions in your cfd simulation? What kind of 'conditions'?

CFD simulations, even with today's computers, are expensive. Are you talking about doing a simulation of a 2D wing profile or something similar or a complete 3d model of a glider? The first simulation you can run on a pc, for the second you need a workstation or a cluster. Depending on what you want to achieve, it might be impossible to do so with your resources.

 

[Bispen]

Hi bigfooted,

Thanks for your participation.
Our simulation is 3D one in OpenFoam, RANS wSST with wall function, maybe Low-Re if necessary. I have all the necessary hardware to compute this. My concern is not about the CFD environment but the physical and mathematical approach.

The purpose is some performances analysis : to determine 1/ the optimum glide angle, 2/ the corresponding forward speed for this specific angle of attack/incidence, 3/ the stall speed if possible

I know the weight, the altitude but playing around with mg,L,D,Cl,Cd,attack angle and incidence angle and all the curves found in the books I have (polar, drag polar, Cl/alpha, Cd/alpha, Cl/Cd, h/V...) I am still ending up with an unknown variable (either attack angle or speed). Lift varies according to alpha and speed. Same for total drag. And all is dependent from incidence angle.

I suspect an iterative approach might be the only solution. I thought it would have been simpler.

http://www.rjplancaster.net/documents/PrinciplesOfGliderFlight-LiftDragPerformance.pdf
http://www.free-online-private-pilot-ground-school.com/Aerodynamics_in_flight.html
http://www.langleyflyingschool.com/Pages/CPGS+4+Aerodynamics+and+Theory+of+Flight+Part+1.html
http://www.skybrary.aero/index.php/Drag
http://www.pilotfriend.com/training/flight_training/aero/drag.htm
https://www.grc.nasa.gov/WWW/K-12/airplane/bga.html

 

[bigfooted]

If you know nothing about the performance of the wing then your only option is to compute $Cl(\alpha,Re)$ and $Cd(\alpha,Re)$ for a range of angles and Reynolds numbers.
If your wing shape is close to an existing naca profile, you could use a table lookup to get an estimate of Cl/Cd, or you can use a program like xfoil for 2D wing profiles. If your wing shape is pretty simple (single profile, non-swept) you can also do 2D simulations in openfoam first. You can then limit the number of 3D simulations, as they will be quite expensive. My guess would be around 10 Million cells for a medium 3D mesh and between 1-2 days of computing time on a recent 32 core machine for openfoam. But you also need to do a mesh refinement study, and maybe even a validation study where you compare the simulation results with measurement results if that hasn't been done yet. Especially when you are interested in the stall angle you will need that. You might need to switch to an unsteady solver with appropriate turbulence model for high angles of attack to get the accuracy you need.

Here is an openfoam study of a 2d airfoil to get an idea of the accuracy you can expect (not that for a 2D airfoil they already need 0.5M cells to get mesh independent results. so 50 cells in the depth-direction already gives you 25M cells):
http://www.lr.tudelft.nl/fileadmin/Faculteit/LR/Organisatie/Afdelingen_en_Leerstoelen/Afdeling_C_O/Graduate_School_AE_01/PhD_Poster_Day_2013/Zhang__Y..pdf

The sites you refer to are extremely basic. A very good (but still quite basic) introduction is J.D. Anderson's 'Introduction to Flight'
https://www.amazon.com/dp/0073380245/?tag=pfamazon01-20
You can get them for a good price at the second hand market.

 

[Bispen]

bigfooted,

The following link illustrates an even more complete experiment, closer to what we are looking for
www.standardcirrus.org/ThomasHansen/StdCirrusSimulation-1.pdf

I said, maybe we will figure out the stall speed - reason is as you mentioned time computation. We usually use DES but yes, simulation takes ages. Good for educational cases but not business oriented.

Dealing with Cl, Cd, incidence and weight is OK. But as angle of attack comes into the equation, it complicates things. But as it seems that for different Re numbers (i.e. different speeds), there are few changes in Cl or Cd, I can take anyone in the estimated middle range as starting point, build my Cl/Cd over alpha curves and determine the angle for the largest ratio. Then knowing the weight I can determine the corresponding incidence - I suppose.

As for Anderson, I already have this book. I found in it all the equations behind the scene but again nothing about this equilibrium state I am investigating for and the corresponding experimental approach.
https://www.amazon.com/dp/0073398101/?tag=pfamazon01-20

Thanks for your post