View Issue Details
ID | Project | Category | View Status | Date Submitted | Last Update |
---|---|---|---|---|---|
0000816 | OpenFOAM | Bug | public | 2013-04-15 10:40 | 2013-05-23 10:47 |
Reporter | Lieven | Assigned To | henry | ||
Priority | urgent | Severity | major | Reproducibility | always |
Status | closed | Resolution | fixed | ||
Platform | Linux | OS | Ubuntu | OS Version | 12.04 |
Summary | 0000816: Incorrect definition of magS in dynamic LES models | ||||
Description | By definition, the strain rate magnitude |S| is computed as |S|=sqrt(2*S:S) where S = symm(gradU) (see e.g. Pope, 2000, Eq. 13.74). In the dynLagrangian model however, this rate is computed as |S|=mag(S) (line 123 of dynLagrangian.C) which is only |S|=sqrt(S:S). Therefore, a factor sqrt(2.0) needs to be added to this line. Although I'm not very familiar with them, it seems to me that the same is true for the homogeneousDynSmagorinsky model and the dynOneEqEddy model. | ||||
Tags | No tags attached. | ||||
|
|S|=sqrt(2*S:S) is a definition of |S| but not the only way to define it and to many the factor of 2 is odd; why not 4 to compensate for the 1/2 both the Ss? We generally use the definition |S|=sqrt(S:S) unless it causes serious problems with the formulation of the model it is used in. What is important is that the formulation of the model is consistent with the formulation of |S|. Can you show that the implementation of the dynLagrangian model is inconsistent with the definition of |S| used? |
|
Both OF and Pope compute nuSgs as (at least in the case of dynLagrangian): flm/fmm*sqr(Delta)*sqrt(2*S:S) so if the ratio of flm and fmm is independent from the definition of |S|, this would be fine. However, it is not. As only M is proportional to |S|, the fields LM and MM scale with sqrt(2) and 2 when the factor sqrt(2) would be introduced in the definition of |S|. Because flm and fmm are basically time averages of LM and MM, the ratio flm/fmm changes with a factor 1/sqrt(2) when the definition is changed. |
|
In the original paper we used as a reference for this implementation |S| is not defined as sqrt(2*S:S) and the behaviour of the model appears independent of odd the coefficient some use in the definition of magnitude for this particular tensor. |
|
Ok, let me rephrase my issue. The fact that nuSgs is computed as flm/fmm*sqr(Delta)*sqrt(2*S:S) gives me the strong impression that |S| = sqrt(2*S:S) since nut = sqr(cs*Delta)*|S| where sqr(cs) = flm/fmm unless of course an other definition for nut is used in the paper (my apologies, I don't have access to the paper so I can't verify it myself). If not, magS should be calculated according to the same definition. So either the sqrt(2) needs to be removed from nut or added to the definition of magS. |
|
Thanks for the clarification; we need to reformulate the k function for consistency both internal and with the Smagorinsky model. |
|
Thanks for the commenst. The db6ebb050ba8d4db4f4967f4f9225cd067222c77 should solve the inconsistency. |
|
Sorry to say so, but nuSgs is still calculated incorrectly. Instead of nuSgs_ = 2.0*(flm_/fmm_)*sqr(delta())*mag(dev(symm(gradU))) it should be nuSgs_ = (flm_/fmm_)*sqr(delta())*mag(dev(symm(gradU))) (the original expression needed to be devided by sqrt(2), not multiplied by it) |
|
After checking the definition of mag(S) for this particular model we introduced the modification you mentioned. Thanks. The commit is 3e0c93dbdb9406230a187d828058596b8377c82b |
|
For future reference. This is probably related to the bug-report of #255 (And for more discussion: http://www.cfd-online.com/Forums/openfoam-solving/82780-improved-implementation-dynamic-smagorinsky-3.html ) |
Date Modified | Username | Field | Change |
---|---|---|---|
2013-04-15 10:40 | Lieven | New Issue | |
2013-04-15 11:05 | henry | Note Added: 0002115 | |
2013-04-15 11:37 | Lieven | Note Added: 0002116 | |
2013-04-15 11:42 | henry | Note Added: 0002117 | |
2013-04-15 12:55 | Lieven | Note Added: 0002118 | |
2013-04-15 14:16 | henry | Note Added: 0002119 | |
2013-04-15 15:14 |
|
Note Added: 0002120 | |
2013-04-15 15:43 | Lieven | Note Added: 0002121 | |
2013-04-16 14:15 |
|
Note Added: 0002122 | |
2013-05-23 10:16 |
|
Note Added: 0002239 | |
2013-05-23 10:47 | henry | Status | new => closed |
2013-05-23 10:47 | henry | Assigned To | => henry |
2013-05-23 10:47 | henry | Resolution | open => fixed |