CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   ANSYS Meshing & Geometry (https://www.cfd-online.com/Forums/ansys-meshing/)
-   -   Meshing of two airfoils (https://www.cfd-online.com/Forums/ansys-meshing/84958-meshing-two-airfoils.html)

Malohm February 13, 2011 13:12

Meshing of two airfoils
 
Hey,

I have to solve the following problem: Given are two airfoils right behind each other. The incidence angle of both the first and the second airfoil is zero degrees. I'm wondering what the most efficient way to mesh this geometry would be. I haven't found any papers dealing with this problem.

Thanks for you help,
Marko

http://dl.dropbox.com/u/20358584/Gambit_ImageFile3.jpg

PSYMN February 14, 2011 17:45

High lift...
 
4 Attachment(s)
I recommend 2D Hexa Blocking... And actually these sorts of things are done quite regularly...

See this configuration from last years AIAA workshop... (actually the workshop was done in 3D, but you get the idea).

DoHander February 14, 2011 18:25

Quote:

Originally Posted by Malohm (Post 295048)
Hey,

I have to solve the following problem: Given are two airfoils right behind each other. The incidence angle of both the first and the second airfoil is zero degrees. I'm wondering what the most efficient way to mesh this geometry would be. I haven't found any papers dealing with this problem.

Thanks for you help,
Marko

I suspect you are using Gambit, a good strategy will be to split your flow domain in few simpler subdomains that can be meshed with MAP. Using a block structured mesh will give you a clean solution and faster convergence. You can use as guide the images posted by PSYMN, however the 2D Hexa Blocking mesher is available only in ICEM-CFD included with Fluent 12 and up.

If you want a fast solution just used one of the unstructured meshing algorithms from Gambit.

Do

PSYMN February 15, 2011 13:40

I agree...

If you are already using gambit, just forget about mapped for now and use a nice tri mesh with Gambit's sizing function... It won't be as accurate, but it will be pretty good and you could have your solution the same day. The structured mesh takes more time up front, and so is more worth doing if you plan to do a study of a series of designs (the initial effort will be divided out over the series and could actually save you time) and are very concerned about accuracy or rapid convergence.

If you are starting from zero (no Gambit experience was my earlier assumption), then I would suggest that you would be better off learning ICEM CFD (for mapped quads) or ANSYS Meshing (if you want unstructured triangles).

Malohm February 15, 2011 14:06

ok, thanks for all your advices! I think I'll switch to ICEM to try the hexa meshing.

And thanks for the pictures :)

Malohm February 25, 2011 06:48

1 Attachment(s)
Ok, this is a result after my first trials with ICEM. What do you think?

The next step would be to add an additional wall to study the wing in ground effect for this configuration.

PSYMN February 25, 2011 11:27

It looks pretty good. Perhaps you would have saved some mesh if you had used a CGrid. (if you did, I can't tell from this pic).

For your ground plane, just start with the previous model, add a horizontal split and associate it with the new ground curve. Then delete (or change material on) the blocks below the split.

Malohm February 25, 2011 12:34

1 Attachment(s)
Actually I did :)

Thanks, I'll run some calculations and then add the ground plane.

Malohm February 26, 2011 08:42

2 Attachment(s)
When trying to import the mesh into fluent I get this error message. Any ideas what could be wrong?

Details are:


Ans.Fluent.Cortex.CortexCommandFailedException: An error occurred in FLUENT when executing command (wb-read-case "C:\Users\Marko\Uni\STS\Mesh\free1.msh" 'cas)
An error occurred in FLUENT when executing command (wb-read-case "C:\Users\Marko\Uni\STS\Mesh\free1.msh" 'cas)
at Ans.Fluent.Cortex.CortexCommunicator.SendMessage(S tring msg, String msgKey, String errorMessage, Boolean waitForReply)
at Ans.Fluent.Cortex.CortexCommunicator.SendMessage(S tring msg, String msgKey)
at Ans.Fluent.FluentCommunicator.MenuLoad(String cmd)
at Ans.Fluent.Data.SetupData.ReadMeshFile(ICommandCon text context)
at Ans.Fluent.Data.SetupData.ReadMeshAndModelInfo(Com mandContext context)
at Ans.Fluent.Data.SetupData.LaunchFluentAndRead(Comm andContext context, Boolean batchMode, Boolean loadSolution, DataContainerReference fromContainer, Boolean internallyStarted)
at Ans.Fluent.Commands.EditCommand.Execute(CommandCon text context)
at Ans.Core.Commands.Concurrency.CommandWorkUnit.exec uteInContext(CommandContext subContext, IExecutionEngineCallback tracer)
at Ans.Core.Commands.Concurrency.BaseWorkUnit.doExecu te(IExecutionEngineCallback executionEngine, CommandContext subContext)
at Ans.Core.Commands.Concurrency.BaseWorkUnit.Execute (IExecutionEngineCallback executionEngine, Boolean dontCatchExceptions)
--- Ans.Core.Commands.CommandFailedException: An error occurred in FLUENT when executing command (wb-read-case "C:\Users\Marko\Uni\STS\Mesh\free1.msh" 'cas)
An error occurred in FLUENT when executing command (wb-read-case "C:\Users\Marko\Uni\STS\Mesh\free1.msh" 'cas)
Command: Fluent.Edit(Container="Setup")
at Ans.Core.Commands.CommandAsyncResult.Wait(Int32 milliSecondsTimeout, Boolean exitContext)
at Ans.Core.Commands.CommandAsyncResult.Wait()
at Ans.Fluent.Commands.EditCommand.InvokeAndWait(ICom mandContext context, DataContainerReference Container, Boolean Interactive)
at Ans.Fluent.Gui.GuiUtilities.GetLauncherSettingsAnd Edit(GuiOperationContext operationContext, DataContainerReference cref, LauncherEditMode editMode)
at Ans.Fluent.Gui.OpenInFluentGui.Invoke(GuiOperation Context operationContext)
at Ans.UI.UIManager.<>c__DisplayClass9.<InvokeOperati on>b__8()
at Ans.UI.UIManager.InvokeOperationCore(String pseudoname, OperationDelegate callback, Boolean allowOSMessages)

PSYMN February 26, 2011 16:47

Mesh not case...
 
I think you read it in as a case file instead of a mesh file...

A case file has a lot more info that is missing...

Malohm February 26, 2011 18:37

Ok it works. I used the repair function in Icem. Thanks for your help.

Malohm March 12, 2011 09:35

This might be a stupid question, but I'm not sure yet whether my grid is structured or unstructured. I use quadrilateral elements and following this tutorial I converted it the pre-mesh to an unstructured mesh to export it.

So, I think after importing my mesh to fluent it is unstructured, isn't it? I'm using spalart allmaras to calculate my solutions. I choose "Gree-Gauss Node Based" as solution method. So I guess a finite volume method is applied, am I right? Diffusion terms are discretized with second-order central differencing scheme.

So just to be sure, I use an unstructured grid with finite volume method. Or does Fluent treat my mesh as structured anyways?

Edit: worked through the theory of finite volume methods. The only remaining question: Even though I convert my mesh to an unstructured mesh, is it still structured since it fulfills all criterias for a structured mesh?

PSYMN March 12, 2011 22:26

Unstructured Solver...
 
Structured vs unstructured is more about how the data is stored than how it looks... ICEM CFD hexa can generate the same mesh and output it as either an unstructured or a multi-block structured mesh.

Fluent (since version 4) has been an unstructured solver. But this doesn't mean there aren't numerical advantages to having a nicely ordered unstructured mesh...

RGRUIZ March 13, 2011 20:54

2D_airfoil meshing
 
2 Attachment(s)
hI SIMON i wrote you before asking for some meshing..i have the same problem that malohm..i am using gambit for meshing and i need a y+<=1 because i am trying to obtain the transition point on an airfoil we design for an asme competition..i a mechanical engineer student from venezuela..

i have seen your youtube videos, so i can learn a little bit about icem, i have a question on video #3 where you fast forwarded the video, can you tell me what did you do there??

this is what i did on gambit but i think i can get better results using icem

RGRUIZ March 13, 2011 20:58

I am using SST transition model from fluent 12 but it doesn't converge the results...i think i have serious problems with my mesh so, i need to improve it..

I have check my boundaries conditions and they are good, i can send you if you want..

I need this improve so i can work on my thesis..i choose this model to investigate and i am stock on the mesh...

PSYMN March 13, 2011 21:27

Your Gambit mesh looks pretty good from here... what time in the ICEM video are you talking about? I haven't watched them since I made them, so you need to be a bit more specific with your question.

I should note that my work is getting much busier lately so I am scaling back on CFD Online... I am also going on vacation in a couple days...

RGRUIZ March 13, 2011 21:33

Hi simon

Thanks for your answer..it is on 1:05 from video #3..

RGRUIZ March 13, 2011 21:53

Hi mister Simon As you request..this the link


http://www.mediafire.com/?v1b5bq5q7yfc4xk

http://www.mediafire.com/file/v1b5bq5q7yfc4xk/Ruben.rar


All times are GMT -4. The time now is 15:45.