CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   HELP! (http://www.cfd-online.com/Forums/main/404-help.html)

 a CFD student December 19, 1998 05:16

HELP!

I am a begineer in CFD. Currently, I am trying to use the SIMPLE's method to do the calculation of a flow past a sphere. Could somebody kindly tell me where I can get(download) the corresponding free codes written in Fortran? Another problem I am considering to solve is like this: a flow passes a sphere while the computational domain is a hexahedron with slip boundary. How can I handle this problem? It seems I have to use two different coordinate(Cartesian and spherical) system to do the calculation, am I right? Is there any mesh generator which is straightforward to use for this problem? By the way, I only have some knowledge and experience on FD and FV, mainly on FD. Some people told me FEM is more appropriate for this problem but I would like to use the method I am more familiar with. Thanks a lot in advance!

 x December 21, 1998 10:02

Re: HELP!

xx

 John C. Chien December 21, 1998 13:00

Re: HELP!

I don't know where to find a SIMPLE code which can handle the general coordinate transformation. If can visulize the sphere as if it is a cube, then you can use H-type mesh to cover the flow field. In the case of O-type mesh, you can transform the outer boundary into a cube ( or rectangular shape. This type could be a better choice, if you are mainly interested in the flow near the surface of the sphere. The adaptive unstructured mesh would be ideal for the whole flow field. Regardless of what you used, as long as you have enough mesh points to cover the high gradient flow region, the selectionof mesh type is not a critical issue. Especially, if you are computing unsteady wake flow, your mesh also need to follow it ( or use a very fine mesh).

 Duane Baker December 21, 1998 18:57

Re: HELP!

Hi " CFD student ",

Here is a start for you. There are some excellent source codes which accompany a text written by J. Ferziger and M. Peric' entitled "Computational Methods for Fluid Dynamics". The codes are FORTRAN and are educational/experimental and incredibly well written and documented...this is to say that they are not optimized for computational efficiency, memory requrements, etc. but are logical and understandable and permit a quick learning curve for modification!!!

The code that you probably want to start with is either THE 2GL/2GT code. They are only 2-d but have a grid builder for 2-d structured non-orthogonal curvelinear grids which can handle o-grids. Then you want to look at the 3D code for cartesian grids. It is then possible to make 3D structured grids (even block structured grids) from a series of 2D grids. The 3D code then needs to be modified for the non-orthogonal and transient(if required) aspects...and then...voilla..you have what you want!

Just a roughg estimate is that this is the order of months worth of coding, debugging, etc. assuming a pretty good knowledge of the basics of CFD! If this is not the case a good place to start is to work through the text and the simpler examples.

If you are still interested after realistically looking at the work required..I may be able to help sith some of the aspects...such as the 2D to 3D grids and a few other things which I have already coded up!! Email me.

Also, anybody else who has done some modifications to these codes, it would be usefull to pool them together into a library or something!!! What do people think about that???

The following table summarizes a few of the codes:

CODE DIM TRANS COORDS ORTHO COLOC. LAM/TR DISC ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ ^^^^^^^^^^ 2DGL |2D |SS/UNS |CURV |NON |YES |LAM |FV 2DGT |2D |SS/UNS |CURV |NON |YES |TURB |FV 3DC |3D |SS |CARTES |ORTHO |YES |LAM |FV

You can get them by: ftp ftp.springer.de cd pub/technik/peric in "readme":

2DGL -> This subdirectory contains the new codes for solving the laminar two-dimensional Navier-Stokes equations (steady or unsteady) using Finite-Volume method and non-orthogonal structured grids with a colocated arrangement of variables. H-, O- and C-grids can be used, with multiple grid levels. Two subdirectories exist:

SG -> This sub-directory contains two single-grid versions of the code: in one the grid-nonorthogonality and non-smoothnes are specially accounted for when solving the pressure-correction equation and when computing the gradients, and in the other version not.

MG -> This sub-directory contains the version of the above code with multigrid acceleration of outer iterations. More information is available in the README file in this directory.

2DGT -> This subdirectory contains the new codes for solving the RANS-equations using Finite-Volume method and non-orthogonal structured grids with a colocated arrangement of variables. H-, O- and C-grids can be used, with multiple grid levels. Two subdirectories exist:

KEPS -> This sub-directory contains the version of the above mentioned code with the k-epsilon model of turbulence (for high Reynolds numbers) with wall functions.

KOMEGA -> This sub-directory contains the version of the above mentioned code with the k-omega model of turbulence (for high Reynolds numbers) with wall functions. More information is available in the README file in this directory.

Hope to hear from anybody interested!!! Good luck.....................................Duane

P.S. Please put your real name on a posting not just a CFD student!

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