open TELEMAC-MASCARET The mathematically superior suite of solvers

Hi there,

Does anyone have any experience or insight as to whether TELEMAC-3D is capable of accurately simulating complex phenomena in the vicinity of a vertical cylinder? I have seen some discussions elsewhere in the forum concerning von Karman vortex streets but I was wondering if it is possible to generate the horseshoe vortices responsible for scour? I believe there was some work on this at the University of Hannover in the past?

If it is possible, I would particularly be interested in any comments on the best options for:

- Turbulence models (vertical and horizontal)
- Time steps (e.g. I understand very small time steps can prevent the development of von Karman vortex streets)
- Mesh / layer resolution in the boundary layer
- Solution schemes

Thanks in advance,

Enda

Hello,

Yes, it has already been done in Hannover university, including the scour. A constant and very small diffusion coefficients on the horizontal is probably the better, and on the vertical I do not know (mixing-length ?). As a matter of fact the k-epsilon model should not work properly, as it is a RANS model, supposed to give you the average flow, so it raises in this case a philosophical problem. Practically k-epsilon could give you a diffusion that kills the eddies (however in practice it does not with the mesh size we use).
An important thing is the numerical diffusion in the advection of velocities, so the method of characteristics will perform well with a Courant number in the order of 1 or even more. If you are starting your work, in the while you will get our version 6.3, with the weak form of characteristics that is even better for von Karman eddies.

With best regards,

Jean-Michel Hervouet

Thanks again, Jean-Michel.

I have run a few tests and counter-intuitively (or so it seems to me), TELEMAC-3D running in hydrostatic mode seems to better capture the horseshoe vortices at the front of the pile compared to the non-hydrostatic mode. I read a paper from work at the University from Hannover and they seemed to find something similar with their numerical modelling work, hypothesizing that this was due to incompatibility with boundary conditions in the non-hydrostatic code. Does anyone have any insight or further references they could share to help me understand this issue?

Best regards,

Enda

Hello,

Well if you look at the test case of non linear waves it does work in non hydrostatic only. Maybe a key point is to have the dynamic pressure taken into account in the propagation step (DYNAMIC PRESSURE IN WAVE EQUATION : YES). Otherwise I have no experience on the horseshoe vortex. For von Karman eddies, it is a 2D phenomenon, and they will not be better in 3D, even with non-hydrostatic option.

With best regards,

Jean-Michel Hervouet

Hello,
I am also interested in this topic, and I have the same question that the hydrostatic version can capture the horseshoe vortex better than the non-hydrostatic version even including the Dynamic pressure in the wave equation.

Is there any further reference for this to clarify why it occurred, and how to deal with it in TELEMAC?

Best regards,
Wenjun