Hello,

I tried to use SIMPLE-algorithm (3D laminar flow), it works fine with Re=1 ( mu=1 ), but when I use higher Reinolds numbers, the code becomes unstable.

Is SIMPLE appropriate for high Relnolds number flows? Re=1000-100000 ( mu = 0.001-0.00001 )

1) Do you have introduce any artificial viscosity (upwind scheme) ? It is mandatory for higher Reynolds nnumber flow.

2) Is your relaxation parameter for the p' equation correctly set-up (around 0.5)?

3) Do you use an implicit or explicit scheme?

1) Do you have introduce any artificial viscosity (upwind scheme) ? It is mandatory for higher Reynolds nnumber flow. I use QUICK scheme (2nd order of accuracy) and I have not introduced artificial viscosity. If I do so, will my solution be correct and suitable? (Air viscosity is about 10e-5.) And will I obtain a backward flow behind my body (it is a cube) with artificial viscosity? As I realize, artificial viscosity means decreasing Reinolds number. Is that right?

2) Is your relaxation parameter for the p' equation correctly set-up (around 0.5)? The relaxation parameter helps a little (up to Re=10), but with Re=100 my code is still unstable. I also use inertial relaxation, but it does not help.

3) Do you use an implicit or explicit scheme? I consider a steady-state flow.

1) Artificial dissipation is implicity introduced by using any upwinding (such as the QUICK scheme). There was a discussion on artificial dissipation some times ago on this site. For your Reynolds numbers I think the QUICK scheme should be ok (I never used this scheme. This advice is based on some readings I have done. Some people on this board could give better advices on this subject.) Or it can be explicitly introduced (I personnaly think it not a good idea).

2) Is "inertial relaxation" a false time step? (i.e using a time step even with permanent flow).

3) I had similar problems with permanent flow and high Rayleigh numbers. I had to introduce a false time step in order to get a converged solution.

4) Is it your own code? Did you validate your code on some well known problems (flow between two infinit planes, lid-driven flow, backward facing step,etc.. ) There could be a bug in your code.

Good luck.

(1). What kind of 3-D problem you are trying to solve? You will have to describe it more clearly first. (2). For Reynolds number=1, how was the Reynolds number defined? and How many grid points or cells were used in the computational domain? Is it 1x1x1, or 10x10x10, or 100X100X100? (3). Assuming that your code is all right, then you can change the mesh size first to see how it affect the solution. (4). If your Reynolds number=1 case is all right, then your Reynolds number=100000 case will probably need 100000 times the mesh size to get a solution. The fact is, Reynolds number changes the solution behavior, and in general it makes the boundary layer thinner. So, in order to capture the correct solution, you will need more mesh point in the right place. (5). This is only one area which you can do some experiment easily.

"You will need more mesh points in the right place" means that mesh needs be refined near the wall, is my understanding right? Can we just use wall functions to avoid the cost of refinement?

Thank you!

(1). If the flow is nice and clean, yes, normally you can use wall function to avoid expensive calculation near the wall. (2). If you have separated flow situation, no, the wall function is not designed for such condition.

Dr. John C Chien

For a submerged jet of a newtonian incompressible fluid having low Prandtl (~0.1), impinging on a wall, how should one model the flow near the wall for flow and energy solution. Pls. advice with ref. of books/journals.

(1). I don't know.