CFD Online Discussion Forums

CFD Online Discussion Forums (
-   CFX (
-   -   LES (

mefpz April 8, 2004 04:58

I am practising the LES using CFX5.6 first time. Now I got the problem. For a given transient problem, it can be runed using k-w model perfectly. However, When I just shifted the model to LES, the solution process don't go ahead at the very beganing. Can anybody help me? Thanks!

The error message is:

ERROR #001100279 has occurred in subroutine ErrAction. Message: Stopped in routine MEMERR

Meri April 8, 2004 05:58

Looks like you do not have enough memory, the LES simulations solves extra equations and consequently more memory is needed.

mefpz April 8, 2004 09:02

I don't think so. I just tried to do a case, so the question is very simple and the elements is about15662 and my memery is 526M. Do you have any other advices? Thanks!

Meri April 8, 2004 10:29

I will need more info on exactly what you are doing.Try creating your *.def file gain, making sure all the initial conditions are set properly, also the time steps must be rather small 1e-4s,otherwise it will not get going. Please report back on your success or not.

mefpz April 8, 2004 21:13

I tried a refined resolution with more than 2 millions elements and 1e-4 timestep on the enhanced computer with 2CPU and 3G memmery, but it still doesn't work. Do the initial condition and the inlet boundary condition need any special treatment, say the incoming velocity at the inlet boundary must be fluctuation and the velocity field in the domain must be a fluctuaion field? Thanks a lot!

Bob April 9, 2004 05:44

Why don'tyou try a very simple geometry, say a duct. Try and get this going rather than trying something with loads of elements. Use a vey simple set up, then start adding to this say the boundary conditions of your full model. Once happy with this, then move to the actual model. I am correct in understanding that the final model was run using K w turbulence model ? Can you add your LES def file setup to this site so we can take a look at it ? Bob

James May 18, 2004 21:08

If the LES crashed from the begining its probably a setup thing I would say.

But anyway, you can't realistically expect to run and LES with 15,000 cells, even if they were hexahedral, that is too small.

Also, LES actually solves LESS equations that kw. LES is a laminar flow calculation for any motion larger than the cell size, and it uses zero equation turblence model for any motion smaller than the grid size. You can't say that the timestep should be 0.0001s, it totally depends on the problem. But it needs a very small timestep because you have to resolve the motion of all structures larger than the cell size, in time and space.

LES models need lots of care taken in the setup, they also need lots of understanding about the actual problem being solved and well-thought-out, accurate and realistic boundary conditions are essential. Remember that the kw model is solving for the mean flow condition, LES is solving for many many instantaneous flow conditions. These are two totally different things. If there is no natural disturbance in the flow, like a blockage, you will need to introduce disturbance through boundary or intial conditions.

All times are GMT -4. The time now is 18:44.