View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0001155 | OpenFOAM | Bug | public | 2014-02-11 19:30 | 2014-02-12 07:28 |
| Reporter | matthias | Assigned To | henry | ||
| Priority | normal | Severity | minor | Reproducibility | always |
| Status | closed | Resolution | fixed | ||
| Platform | Linux | OS | Other | OS Version | (please specify) |
| Summary | 0001155: LES laplace filter issue | ||||
| Description | Hi OpenFOAM Team, I think the laplace LES filter isn't correct defined in OF. According to the original definition of differential filter class (Germano 1986a) the filtered field is obtained by subtracting the contribution of high wave numbers from the unfiltered field and not by adding them. In the current implementation, this leads to a kind of "high" pass filter but I suppose the original intention was to create a low pass filter. Of course it is possible to specify negative coeffs what would equal the subtraction operation but for better understanding one should keep the coeffs positive and change the sign of the laplace term. | ||||
| Tags | No tags attached. | ||||
|
|
The current implementation effectively averages the cell value with the neighbours in a manner which is more consistent with the geometry of the cells than the simpleFilter. From our test it is not clear if this is actually beneficial and we do not use generally use this filter. What tests have you done to show the benefit of changing the sign of the Laplacian? It is not clear to me that this will provide a suitable filter. |
|
|
Actually I'm doing some LES calculations using explicit filtering methods and in my previous post I didn't compare it with spatial top hat filter! I mentioned only that the current implementation of the differential filter not the same as proposed in the original work of Germano. Please find attached the paper where you can see the derivation of this filter. According to the definition there should be a minus in front of the laplacian. (eq. 8) |
|
|
|
|
|
laplaceFilter in not an implementation of the differential filter proposed by Germano, it is an alternative to the simpleFilter to perform local averaging. Have you compared results from the two approaches? |
|
|
You are right, I did some tests with a prescribed velocity field and the laplace filter acts like a low pass filter. When using the other approach the filtered field is sharpened. However I'm a little bit confused now. I always thought that the differential filter (proposed by Germano) should do the same thing ... perform a local average like a low pass filter and not enhance the structures. In my application I use the filtering for scale separation as uDelta = U - filter(U). So uDelta should contain contributions from the smallest resolved scales. However, the difference between both approaches was extremely small in the resulting uDelta field, that's why I didn't see any discrepancies before. Nevertheless it is obviously not a bug in the laplace filter implementation, so this thread can be closed. |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2014-02-11 19:30 | matthias | New Issue | |
| 2014-02-11 20:35 | henry | Note Added: 0002813 | |
| 2014-02-11 20:49 | matthias | Note Added: 0002814 | |
| 2014-02-11 20:49 | matthias | File Added: 1.865649.pdf | |
| 2014-02-11 20:50 | matthias | Note Edited: 0002814 | |
| 2014-02-11 22:05 | henry | Note Added: 0002815 | |
| 2014-02-12 00:57 | matthias | Note Added: 0002816 | |
| 2014-02-12 07:28 | henry | Status | new => closed |
| 2014-02-12 07:28 | henry | Assigned To | => henry |
| 2014-02-12 07:28 | henry | Resolution | open => fixed |
