CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   Please: Looking for a Panel method or vortice lattice for windmill (http://www.cfd-online.com/Forums/main/63166-please-looking-panel-method-vortice-lattice-windmill.html)

be_inspired March 31, 2009 06:03

Please: Looking for a Panel method or vortice lattice for windmill
 
Hi all,

I am doing some studies about windmills and in the current situation, i am looking for a panel method or a vortex lattice method to realize a parametric study of different designs of blades, so rans calculations are not affordable in this stage.
The best option would be a 3d panel method, but i am not be able to find any program.

I have founded several programs for aircrafts, but in my situation a need that the geometry is rotating with a constant velocity.

I don't mind if it is neccesary to pay something, but it would be better if it is free or i can try it firstly.

Thank you for your help

sbaffini March 31, 2009 10:15

If you are a U.S. citizen there are a lot of codes made available for free by the DOE or the NASA.

Another option is http://www.pdas.com/ and you'll take a lot of codes for 300$, but i don't know if PANAIR is suitable for this (it should). Obviously there is no support for nothing as it was for free, but i don't think it's a problem.

Also, a lot of papers from http://ntrs.nasa.gov/search.jsp are available for some of the codes (for free, of course), included PANAIR, so, even if you can't try the codes, you can surely check their capability via the referenced papers.

gocarts March 31, 2009 10:24

Is this what you are looking for?
 
1 Attachment(s)
Here at Symscape we have a 3D Panel Method as part of our SymLab add-ons. It can simulate rigid body motion (including rotation) and you can also specify an angular velocity free stream condition - which for your case is probably easier.

The attached image is for a simple 3 blade turbine with both linear and angular velocity free stream conditions. This example is available as a free download for SymLab.

be_inspired March 31, 2009 12:06

It is a pity but I am not a U.S. citizen. I had seen PMARC and other very interesting codes, but i can not get it.

Thanks for the answers. I am going to try to contact with somebody of pdas to tell me if PANAIR is suitable for my particular case.

In the case of Symscape, i am going to use one geometry to validate it.

If anybody knows about other codes and can help me, I am very interested.

Thank you

A_Lyaskin March 31, 2009 16:32

Check also QPROP:
http://web.mit.edu/drela/Public/web/qprop/

be_inspired April 1, 2009 03:51

I have checked Qprop and it seems to be like a BEM code ( blade element momentum). I need to work with the geometry taking into account the 3d geometry and the 3d effects.

be_inspired April 1, 2009 05:01

Gocarts, I have a question concern to the mesh definition along the span. How can I improve the definition of the mesh ( Where is the mesh?) along the span ( mainly near tip and root) and along the chord?

I need to do some 3d modifications around the tip zone and the accuracy in this zone is important.


Maybe the software can be useful for me, but I need test it with other codes and tunnel results.

gocarts April 1, 2009 09:39

SymLab Panel Flow
 
To control the mesh density, you use the Accuracy Tool in the Physics Tool Palette. If the grading you specify doesn't allow mapped meshes then we automatically switch to triangles. You apply the Accuracy tool directly to geometry edges/faces or volumes. If you haven't already I suggest you review the tutorial "Flow Over an Airfoil."

If you don't have connected geometry (part of your model) in the vicinity of the area you want to resolve, then place a separate edge (source) there and attach the Accuracy tool to it and target the face of your model you want it to influence. So drag the Accuracy tool over to the face of the model and select the Set Target menu item, then pick the source and select the Done menu item. For a detailed explanation of the Accuracy tool try "Accuracy Tool."

The mesh is automatically generated on demand - that is to say when you request values (e.g., pressure coefficient). To view the mesh once values are present select all faces, then in the Properties Panel change the Style property to Flat, and keep the Transparent property unchecked. Try reviewing our tip "View a Mesh."

You are likely already aware of the limitations of panel methods, but remember a panel method assumes incompressible invisicid flow, so it is unlikely to predict viscous tip effects. Accurately representing the geometry of the tip is likely just to slow the simulation without a comparable increase in solution accuracy.

Hope this helps.

Edit:
To create/modify simulations you will be best served by activating the SymLab Professional Add-0n. Without this you can only view a simulation for free in SymLab.

RenardP April 1, 2009 14:52

It will be nice to have a comunity driven 3D panel code coupled with a boundary layer code and transition capabilities (for pure CFD we already have OpenFoam).

I suppose an Open Source development model for this program will work.

What do you think ?

Paul

be_inspired April 1, 2009 15:15

I think it would be a great idea with a lot of people and institutions interested in this program ( to use it, I do not know how start it ). I think a pure CFD is not suitable when you are
in the design stage.
I have some experience with commercial cfd codes and panel codes ( currently I am trying to start with openfoam but it is a bit difficult )

Do you have any idea ?

RenardP April 1, 2009 16:53

Well,

first you need a program specification, these probably can be done here on cfd-online.

1. You will need to write down in a clear an concise form the theory that will support the program. These can also be the work of the comunity.
2. Chose carefully your programming language, for a panel code even an interpreted language will work well. Never use a comercial language (this will drastically limit your coders base even if it looks easy to develop in such a language e.g. Matlab, Mathematica).For a compiled language I would suggest C++, for an interpreted one Python or Ruby.
3. Separate the graphical part from the solver part of your code, this can be easily done by using in the begining a simple solution: write your data to a text file and generate a simple Gnuplot script that will show/save this data in various formats.
4. The solver can also be separated from the surface grid generation part.
5. Create a large base of analytical tests for validating the code. Every capabilities of the code must be validated ( later these examples can be compiled into a nice tutorial).
6. Above all try to find a lot of people that will work on this. Do not use a single genial coder that can live you and your comunity for a well paid job. Always use for coding at least 2-4 people that will manage the main code.
7. Using a scripting language will permit to the average user to easily extend the code. These can be later incorporated in the main code.
8. All the outputs must be done in simple text files or in a clear xml format. Never use binary - you must be able to see/check in a normal text editor the output of your code.
9. Always test the code on Linux (any POSIX system will do like FreeBDS, Mac OS X ...) and Windows (you need a large users base).


All times are GMT -4. The time now is 16:53.