[huangxianbei]

Hi, all:
As I have post some threads in our forum, I implemented the dynamic Vreman model in OF. While, for the first version, the program can run as good as dynamic Smagorinsky. The results are similar,however, the problem is that the nuSgs file contains only the boundaries set as in the initial condition.

Here is the version-1
DVM-1.tar.gz

While, I rewrite the code like this
DVM-3.tar.gz
The nuSgs field can be output as usual, but the calculation breaks down in the second time step

Could anyone give me some help about this problem? I have been worked on it for two days, problem still exists

Any help is appreciated

Xianbei

[huangxianbei]

here is the equations used in this model

[huangxianbei]

well, I think I get some idea about how.

First, I rewrote the code again( once again!!)

Here is the final version
DVM-final.tar.gz

First problem I faced: diverge after one time step

reason: unstable nuSgs values in the field. As can be seen from
http://www.cfd-online.com/Forums/ope...agorinsky.html
The dynamicSmagorinsky model is bounded to avoid negtive nuSgs. This is a good way to avoid divergence in dynamic model. Thus, I rewrote this part

nuSgs_ = max(-0.5*nu()*cV(alpha),-nu());
    nuSgs_.correctBoundaryConditions();

This significantly prevent divergence. Also , this is correct because in most cases, the backward energy transfer is negligible.Now the nuSgs field can be output. Also, in order to further improve the steadiness of the calculation, fvc::average is also used before the volume average.

Second problem: why the nuSgs field contains only part of the values. Not all the cells' value in the domain are output.

Reason: I think it's due to the zero value of nuSgs. Only the non-zero values are output in the nuSgs file.

Although problem seems been solved, still, most of the nuSgs value may be 0 as the calculation runs for several steps. I'll test it with a much longer time running.

If anyone is interested in this model, I'll be glad one can test this model in other cases.

PS: my case is the centrifugal pump impeller