CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   LES with ANSYS 14 (http://www.cfd-online.com/Forums/main/109079-les-ansys-14-a.html)

mazdak November 8, 2012 15:51

LES with ANSYS 14
 
hi
i want to solve turbulent channel flow problem with ANSYS 14 and use LES model. the Re number is 6652. I dont have any idea about the grid size.
the length is 0.4 m and the width of channel is 2cm.
as this is the first time that i'm working with LES model, could you please help me and give me some paper or references about the grid size?
thanks

FMDenaro November 8, 2012 16:05

Quote:

Originally Posted by mazdak (Post 391082)
hi
i want to solve turbulent channel flow problem with ANSYS 14 and use LES model. the Re number is 6652. I dont have any idea about the grid size.
the length is 0.4 m and the width of channel is 2cm.
as this is the first time that i'm working with LES model, could you please help me and give me some paper or references about the grid size?
thanks

First, LES is a formulation not a turbulence model... therefore you have to switch your run time-dependent solution until an energetic equilibrium state is reached. Use centred scheme and dynamic modelling.
If you are simulating a periodic channel flow you have also to force your pressure gradient.
As the grid size is concerned, I suggest to consider 3-4 grid points within y+=1 using a stretched grid along the nornal-to-wall direction.
In streamwise and spanwise direction I suggest to use dx+ = 20 -30 and dz+ = 15 -20.
Is your Re number referred to channel height and centerline velocity? Then I suppose your Re_tau is less than 600.
A DNS could be also realized with several million grid, however there are several database you can use

mazdak November 8, 2012 16:30

Thanks for your time. I have written code for turbulent
Channel flow using ke model but I'm not very familiar with
LES. The only thing that I know is that there
Small scale are filtered and modeled while
Larger scales are directly resolved.
Yes Re is based on channel width and center line vel.
Vcenter=5m/s
Channel width: 2 cm
Den.=1.2
Vis=1.85 e -5
It's air.
And channel length is 0.4 m
That it is periodic or not I don't have any idea
So how many points in each direction shod be used?

atmcfd November 16, 2012 21:09

Quote:

Originally Posted by mazdak (Post 391091)
Thanks for your time. I have written code for turbulent
Channel flow using ke model but I'm not very familiar with
LES. The only thing that I know is that there
Small scale are filtered and modeled while
Larger scales are directly resolved.
Yes Re is based on channel width and center line vel.
Vcenter=5m/s
Channel width: 2 cm
Den.=1.2
Vis=1.85 e -5
It's air.
And channel length is 0.4 m
That it is periodic or not I don't have any idea
So how many points in each direction shod be used?


I guess this paper would be of some use to you, assuming you are interested in channel-flow like cases
Frohlich, J., Mellen, C. P., Rodi, W., Temmerman, L., and Leschziner, M. A., 2005. Highly resolved large-eddy simulation of separated flow in a channel with streamwise periodic constrictions. J. Fluid Mech. 526, pp. 19-66.

mazdak November 16, 2012 21:14

Quote:

Originally Posted by atmcfd (Post 392641)
I guess this paper would be of some use to you, assuming you are interested in channel-flow like cases
Frohlich, J., Mellen, C. P., Rodi, W., Temmerman, L., and Leschziner, M. A., 2005. Highly resolved large-eddy simulation of separated flow in a channel with streamwise periodic constrictions. J. Fluid Mech. 526, pp. 19-66.

Thank you, unfortunately I'm not familiar with
LES, and don't know how grid size can affect the results. Also
I don't know what is periodic condition. It means that
After some time steps flow shows similar behavior ?

atmcfd November 16, 2012 21:25

Mazdak,

A periodic condition at the boundaries specifies that the flow is of "repeating" nature over a period of time. i.e. the initial and final flow quantities are same at the locations where periodic b.c.s are applied.You may google this - lot of info available.

As for LES, I'm afraid that you *must* have a good understanding of it if you want to perform any meaningful simulation, even if you are using a commercial software. Grid size is very important in the quality of the simulation, and its effect largely depends upon the grid resolution in various flow regions. eg. you may want the grid to be finer near walls, where you would like to observe boundary layers, eddies , etc. An alternate approach would be to use a somewhat coarser grid, and use wall models instead, which try to model the viscous effects in the region w/o you spending too much computation time on fine grids.But their accuracy might greatly differ.

Finally, I suggest you go through the basics of LES and understand its physical significance.You might want to search for FMDenaro's posts on LES in this forum , since they answer many questions which any beginner would have.

Good luck!

mazdak November 16, 2012 21:39

thank you for your advice. actually i'm using ANSYS 14 for the simulations and i use the same grid size and time step which has been mentioned in the following paper:

http://www.sciencedirect.com/science...07570403001138

x=0.9078 m (259 nodes)
y=0.0667 m (25 nodes)
uniform gird, but i can not get the same results. i do not know what should i do? i have changed the grid size, the time step but i got different results.


All times are GMT -4. The time now is 10:10.