CFD Online Logo CFD Online URL
www.cfd-online.com
[Sponsors]
Home > Forums > Main CFD Forum

Comments on moving grid algorithm

Register Blogs Members List Search Today's Posts Mark Forums Read

Reply
 
LinkBack Thread Tools Display Modes
Old   August 2, 2005, 06:56
Default Comments on moving grid algorithm
  #1
quarkz
Guest
 
Posts: n/a
Hi,

I intend to code a FVM NS solver for an airfoil which moves in 2d, and maybe followed by a wing moving in 3d. I already have a working 2d NS solver in cartesian coordinates

There have been different methods proposed for this type of problems. These includes grid regeneration, immersed boundary methods, least square meshless methods etc.

Can someone comments on these methods in terms of simplicity, accuracy, reliability?

Thanks alot!

  Reply With Quote

Old   August 2, 2005, 09:57
Default Re: Comments on moving grid algorithm
  #2
A.S.
Guest
 
Posts: n/a
the algorithm is quite simple read paper by Leonard (1979) probably AIAA journal, or refer any publication on Geometric Conservation laws.I had coded it for turbomachinery application (sliding mesh), equation are quite straight forward. Your code is in Fortran or C/C++, if in fortran I can help.

I would personally prefer remeshing.

Good Luck

A.S.
  Reply With Quote

Old   August 2, 2005, 16:55
Default Re: Comments on moving grid algorithm
  #3
Mani
Guest
 
Posts: n/a
If you are looking for optimum performance and ease of implementation, it really depends on your application. For example: If your single 2D airfoil is not deforming (just translating and rotating), it makes no sense to remesh or deform the grid. In that case the grid simply moves rigidly with the airfoil (simulated by grid point velocities).

For more complicated geometries (multiple airfoils, cascades) and for 3D motion and deformation, you should think about the order of magnitude of your displacements. What is it that you want to study? Flutter? Limit cycle oscillations? Forced response? Flapping wings? Store separation? All of the above? Which method is best for you will depend on that. For example: For most flutter computations you can focus on very small amplitudes, since you are only performing a stability analysis. In that case you can pretty much apply any of the methods you mentioned, but for highest efficiency you could simulate the motion by a perturbation boundary condition, without ever changing the mesh. On the other hand, if you are looking at relatively large limit cycle oscillations, you'll want to do some remeshing and/or grid deformation.

In essence: Methods that are most efficient and easy to implement will be only applicable to certain problems, while the more robust and general methods are usually less efficient. Be clear about your objective before you make a choice. Many people tend to shoot for some brute force method by default, which is often unnecessary and very inefficient. I don't know of any book on this particular subject, but a survey of research papers will give you a quick overlook.

If you're a student/researcher at NUS and have the chance to talk to someone at the Temasek Labs I would suggest that you do that...
  Reply With Quote

Old   August 4, 2005, 01:13
Default Re: Comments on moving grid algorithm
  #4
quarkz
Guest
 
Posts: n/a
Thanks a lot A.S. and Mani

A.S.: I'm programming in fortran so I may need your help later on ;-)

I am specifically interested in simulating flapping wings which means large amplitudes. Initial goal should be in 2d and if possible, extend to 3d.

I believe airfoil or wings are not too complex geometries. Hence I'm wondering if I should attempt using immersed boundary method (IBM), which sounds attractive or just go on with the traditional grid movement/regeneration.

Correct me if I'm wrong, but it seems that IBM's algorithm is more difficult to program.

I am also concerned about speed and accuracy.

Hope to select a most appropriate approach. Please give your comments.

Thanks alot
  Reply With Quote

Old   August 4, 2005, 02:28
Default Re: Comments on moving grid algorithm fo A.S.
  #5
Takasaka
Guest
 
Posts: n/a
Hi A.S. Would u please give me the complete reference of the Paper by Leonard(1979). Is it possible to incorporate the moving grid algorith in case of staggereg grid in cartesian coordinates. In my problem finer grid is needed at certain position in the domain and other portion of the domain (less significant) can be calculated on coarse grid. Would u please comment.
  Reply With Quote

Old   August 4, 2005, 02:42
Default Re: Comments on moving grid algorithm
  #6
A.S.
Guest
 
Posts: n/a
Hi,

Visit Prof. Shy Webpage at Ufl's (ufl.edu) website.

http://aemes.mae.ufl.edu/~cfdweb/cgi...ex=1&altmenu=1

His group is working on aeroelasticity, using coupled FEM-CFD solver, hope it will be more helpfull.

Good Luck

A.S.
  Reply With Quote

Old   August 4, 2005, 02:57
Default Re: Comments on moving grid algorithm fo A.S.
  #7
A.S.
Guest
 
Posts: n/a
Hi Takasaka,

Most of my coding is in coupled solver, am very poor in segragated algorithm. So cannot comment more.

I hope Peric's book has section on moving grids.I have mailed you paper on incompressible deforming/moving grid. The only difference is that in deforming grid volume will also change with time (remeshing), where as in rotating domains like turbomachine volume will remain same for all the time-steps.

As far as grid is concerned, I had worked with variable grid size only thing is sliding interface interpolation becomes tedious. Hope it helps.

Also visit Prof. Shy's web page at UFL, for further deatils.

http://aemes.mae.ufl.edu/~cfdweb/cgi...ex=1&altmenu=1

Hope this would be helpfull.

Thanks

A.S.
  Reply With Quote

Old   August 4, 2005, 14:43
Default Re: Comments on moving grid algorithm
  #8
Mani
Guest
 
Posts: n/a
>I am specifically interested in simulating flapping >wings which means large amplitudes. Initial goal should >be in 2d and if possible, extend to 3d.

As long as it's 2D I wouldn't worry about either grid deformation or immersed boundaries. Simply have the grid move with the airfoil. That's not only the easiest way but also the most accurate and most efficient! Only as you go to 3D, with deforming wings, you have to think about your options.

The immersed boundary does seem attractive because potentially applicable to a wide variety of problems that may be very hard to tackle otherwise. Just think of complicated geometries such as heart valves. I would agree with you that an implementation of the ibm will probably be more difficult than a grid deformation routine, but that really depends on where you start from and what your background is. Do you already use cartesian grids with a gridless treatment of boundary conditions, or do you have experience with that? Only if you are already familiar with some of the concepts of immersed boundaries, it would be justified to go for this approach, because your simple geometry doesn't really justify it. If your concern is speed and accuracy, I'd have to say the same: go for a simple and fast grid motion technique (algebraic), keeping the grid boundary attached to your structural boundary.

But actually: Why not do both? Start with grid deformation which is quite easy to do, and then implement the immersed boundary method as another option in your code. It would be good to see a comparison of the methods for flapping foils/wings.
  Reply With Quote

Reply

Thread Tools
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are On
Pingbacks are On
Refbacks are On


Similar Threads
Thread Thread Starter Forum Replies Last Post
who have the education codes for moving grid? aiya Main CFD Forum 0 August 14, 2006 09:53
moving grid with domain size change Hu Main CFD Forum 5 May 13, 2004 21:14
GRID TO GRID INTERPOLATION in FLUENT calogero FLUENT 3 June 4, 2003 08:32
Loading transient data of moving grid leo CD-adapco 3 April 9, 2003 01:35
moving grid for helicopters jy Main CFD Forum 1 January 31, 2000 12:49


All times are GMT -4. The time now is 01:01.