CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   3D panel code (http://www.cfd-online.com/Forums/main/15668-3d-panel-code.html)

yaaw September 3, 2008 11:45

3D panel code
 
hi,can any one help me in finding 3D panel method code(in matlab)to calculate aerodynamic coeffecient for complete aircraft,thanks

LBRS September 4, 2008 03:58

Re: 3D panel code
 
I don't know any CFD code in MatLab, but some exists in Fortran, but probably not freeware. Otherwise there is JavaFoil on the web, in Java but it is only 2D

Ahmed September 4, 2008 04:58

Re: 3D panel code
 
The book "Computational Fluid Dynamics for Engineers: From Panel to Navier-Stokes Methods with Computer Programs" comes with a CD, check what is there

yaaw September 4, 2008 05:27

Re: 3D panel code
 
thanks for your help,many codes written in matlab for 3D,i want any one

yaaw September 4, 2008 05:29

Re: 3D panel code
 
thank you,it not a matter,if not free i will buy

gocarts September 4, 2008 09:22

Re: 3D panel code
 
Does it have to be a MATLAB panel method?

If not try the 3D Panel Flow add-on for SymLab.

Full Disclosure: I represent Symscape the developer of SymLab.

Praveen. C September 4, 2008 10:30

Re: 3D panel code
 
No idea about matlab code, there is one in fortran here

http://web.mit.edu/drela/Public/web/avl/

Paolo Lampitella September 5, 2008 09:28

Re: 3D panel code
 
In my knowledge does not exists such a code wrtitten in matlab. Or at least has not public availability. Maybe someone has written its own code but i don't think so because matlab would strongly slow down the solution process and this is not feasible today for any panel code.

For wings only a steady code is available from the Prof. Davenport at Virginia Tech:

http://www.aoe.vt.edu/~devenpor/aoe5104/

Also several similar codes (some of which are in matlab) are available from the Prof. Mason

http://www.aoe.vt.edu/people/faculty.php?fac_id=whmason

But, the most simply available and complete 3D Panel Code is PanAir (obviously in fortran) from:

http://www.pdas.com/

for a very little fee. Obviously you will not get any support for any of theese codes.

Charles September 5, 2008 17:59

Re: 3D panel code
 
Just a word of warning here. Perhaps because panel codes generally represent an earlier era in computational aerodynamics, a misconception exists that they are a cheap and easy option. In experienced hands, the right 3D panel code is an amazingly fast design-level analysis tool. However, in particular because you have to specify wakes, and keep track of topology and connectivity issues, these codes are far from simple to use correctly. You really do need to know what you are doing if you want to use them. For example, I take exception to the solution described in http://www.symscape.com/examples/panel/dlr_f4 . That is just wrong!


Peter Attar September 5, 2008 23:01

Re: 3D panel code
 
lol..I especially like the fact that they chose the angle of attack of the simulation to best match the experiment...

gocarts September 6, 2008 09:24

Re: 3D panel code
 
What angle of attack do you suggest?

Given that the experiment didn't quote an angle, it just provided the overall lift and drag.

Point noted - you do need to know the limitations of panel methods to get best value from them: http://www.symscape.com/blog/why_use_panel_method

I presume anyone looking for a panel method is willing to accept its limitations.

Peter Attar September 6, 2008 10:38

Re: 3D panel code
 
I suggest if you want to compare to experimental data then you should pick a case where you know the relevant important parameters...

gocarts September 6, 2008 11:16

Re: 3D panel code
 
"I suggest if you want to compare to experimental data then you should pick a case where you know the relevant important parameters..."

I agree, I would prefer a better defined case. I was after a relatively complex 3D configuration, with easily accessible geometry (public domain iges, step) and fully subsonic - the DLR F-4 was the best I could find.

I'd welcome suggestions for other cases that fit these requirements.

I ran a simulation for a 2D multi-element airfoil (NLR 7301) that you might find more interesting: http://www.symscape.com/examples/panel/nlr7301


Peter Attar September 6, 2008 11:25

Re: 3D panel code
 
I'm not doubting the ability of the code...I am sure it is fine. I just think it is not good practice to compare to experiment when you don't even have the angle of attack that the experiment was run at.

Charles September 7, 2008 04:45

Re: 3D panel code
 
The thing with this particular test case is that it was used for the drag prediction workshop, where the objective obviously was to compare the abilities of various codes to calculate drag accurately. As we all know, induced drag scales with the square of the lift, so I think the decision was made to allow the participants to run their submissions at the angle of attack that was required to reach the target lift coefficient, thus effectively normalizing everybody's submissions. It's an unusual approach, but I believe that it was done this way to compensate partly for any inaccuracy in the tunnel angle of attack measurement, due to deflection of the model, balance an support. In fact, the supplied geometry was already partially modified to account for the flexing of the wing, which highlights the organisers' concern about deflections.

Now back to gocarts' solution on the web page .... what I don't like is the treatment of the inboard edge of the wake. This is one of the trickiest areas with lifting panel codes, and there is no simple solution. However, arbitrarily terminating the wake well short of the wing root just can't be right, as it effectively means that there can't be any lift produced at the wing root. We know that is not right. I've seen two approaches used - one is to run the leading edge of the wake down the side of the fuselage, joining up with the plane of symmetry at the back. PMARC works like this. This obviously does not work well with long fuselages, or ones that don't taper to the back. The other approach is to extend the wake into the body, joining up there, the so-called "extra strip" method, which I think is used in DacVine. This method seems to produce a very realistic distribution of lift.


gocarts September 7, 2008 11:32

Re: 3D panel code
 
As you point out - not extending the wake into the wing-body root is an issue. This reflects a limitation in our current force-free wake model - which doesn't limit collisions between wakes and geometry.

Also as you said earlier Panel Methods aren't a general solution for aerodynamic analysis, especially given their constraints in terms of physics and wake specification. With this in mind we are currently working on an OpenFOAM integration for more general aerodynamic analysis: http://www.symscape.com/newsletters/june-2008

With such a toolbox of techniques I think the DLR-F4 in the drag prediction workshop would make a good case study. I imagine you could use the panel method (even with its limited wake model) to get an estimate of the incidence that best matches the pressure coefficient profiles and then use the full RANS (OpenFOAM) CFD solver to perform the drag prediction simulation. I believe the overall turnaround time would be significantly faster than just using a RANS CFD solver in trial and error mode to determine the incidence.


All times are GMT -4. The time now is 15:15.