CFD Online Discussion Forums

CFD Online Discussion Forums (
-   Main CFD Forum (
-   -   NS solver for turbomachnery (

Nima Amani Fard May 13, 2000 07:35

NS solver for turbomachnery
I am preparing for Navier - stockes solver to solve the flow equations in amiddle stage of an axial transonic compressor to investigate the rotating stall instability problem.for doing so,I have to solve 2-D cascade problem in stage(rotor+stator) with at least 8 to 10 blades in each row.the difficulties are as follows:

1 - mesh generation for a stage with multiblades in each row

probabley a multiblock structured mesh system must be

used.please send me some informations about best way of making a structured mesh system for the mentioned problem.

2 - Is are any available software for the mentioed problem

or about some smilar problems?

thank you very much best regards Nima Amani Fard

Sergei Chernyshenko May 13, 2000 11:50

Re: NS solver for turbomachnery
Hi, Nima,

I work on rotating stall in an axial compressor. Our project is to be finished in the end of July. My approach is somewhat different but, obviously, there should be much in common. For example, my mesh generator may suit you. However, the codes developed belong to the Manchester University and I am not sure that I can share them. This depends mostly on what exactly you are (a student? a researcher in a private company? etc., whom your code, when completed, will belong to and certainly on many other things, and this is not for me to decide).

About three years ago when I started to work on this project I made a search but did not find a public-domain mesh generator suitable for multi-row topology. May be there is one now.

There are codes which do it, even in 3D case, in the sense that calculations show a rotating-stall-like behaviour, but they are proprietary and, what is important, their applicability to real cases may be questioned. Go to a library and look through the ASME Journal of Turbomachinery, this is a necessary reading anyway.

May be it is worth to keep in touch. Mail me if you think so:

Rgds, Sergei

CONSULTANT May 14, 2000 01:30

Re: NS solver for turbomachnery

You can generate a multi-block structured grid all by yourself, easily.

Take for instance, two airfoils one on top of each other staggered both in longitudinal and lateral directions. You have a freestream region upstream of the airfoils and downstream region behind the airfoils which you want to include in grid generation.

1. Take the LINE formed by upstream, top surface and downstrean of the BOTTOM airfoil as your bottom boundary for your grid (say j=1)

2. Take the LINE formed by upstream, bottom surface and downstrean of the TOP airfoil as your top boundary for your grid (say j=jmax)

3. the vertical lines forming inlet boundary (i=1) and exit boundary (i=imax).

You can use transfinite interpolation to generate an intial grid with this configuration by choosing appropriate DELTA_X[= (imax-i)/number of grid lines in x-direction)] and DELTA_Y[= (jmax-j)/number of grid lines in y-direction)].

By stacking one on top of each other you can get a multi-block structured grid for this configuration.

Depending upon the solver you are using, you can then impose orthogonality of grid lines at each vertex if necessary. (this is bit involved and may take some time)


When you assemble each block, make sure that the grid lines are continuous and theres an one-to-one correspondence between boundaries.

If you ask me for a better way, I would suggest that generating a triangular unstructured grid is the BEST option. You can write your own grid generation program in 3-months !

Good Luck my dear friend,


John C. Chien May 14, 2000 12:51

Re: NS solver for turbomachnery, check out Joe Thompson's grid generation book online
(1). Try the Joe Thompson's book online. The book includes 2-D mesh generation Fortran codes using various methods. The best place to start. (2). You can download the book from the Internet site, listed in the Resources/ book section.

Cfd May 14, 2000 22:41

Re: NS solver for turbomachnery, check out Joe Thompson's grid generation book online
I agree with CONSULTANT that a structured block mesh for a staggered cascade is NOT the best choice. Fancy grid generation techniques, such as those in Thompson's book, are hard to use and quite inflexible, in addition to requiring significant computational resources. If your code permits it, you can use unstructured meshes with arbitrary matching between the adjacent blades (most present day commercial finite volume can easily handle these kind of grids), or perhaps tetrahedral meshes (although personally, I feel tet meshes are less reliable in the solution phase, though easier to generate). This also results in better cell quality (internal angles).

All times are GMT -4. The time now is 06:50.