CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   OpenFOAM (http://www.cfd-online.com/Forums/openfoam/)
-   -   channelFoam for LES of Flow over a Wavy Wall (http://www.cfd-online.com/Forums/openfoam/98208-channelfoam-les-flow-over-wavy-wall.html)

sunliming March 6, 2012 01:50

channelFoam for LES of Flow over a Wavy Wall
 
Hi Foamers,
I adopted channelFoam to simulate flow over a wavy wall which is sinusoidal shaped,
seeing that merely the geometrical configurations are different from tutorial channel395.

----------------------------------------------------------------------------------------
Sizes of the domain are:
(Lx,Ly,Lz)=(4Lambda,Lambda,2Lambda),whereLambda=1 is wave length;
X:streamwise;
Y:vertical or channel depth direction;
Z:spanwise.
Refer to Case 76 of http://cfd.mace.manchester.ac.uk/ercoftac/ for details!

----------------------------------------------------------------------------------------
Grids:192 X 96 X 96;
Vertcial expansion ratio 1.04 with y+ equals about 0.8;
Bulk velocity Um=1m/s;
Re=Um*Lambda/nu=6760;

----------------------------------------------------------------------------------------
Boundary Conditions:
All are the same as channel395 but the topWall,here I removed the topWall exposing
the channel to the atmosphere,making it an open channel .Hence I set BC of slip for both
U and p at the topWall,i.e.,
topWall patch;
bottomWall: Wall;
inlet,outlet,frontWall backWall: cyclic;

----------------------------------------------------------------------------------------
Sub-grid model:homogenousdynSmagorinsky

----------------------------------------------------------------------------------------
Unluckily,I encountered several problems:

*First, after about 4 flow-times through the channel I got some results,and I'd like to get
a field average in spanwise direction Z ONLY,but the postChannel application exerts averaging
throuth all over the field including streamwise direction which is undesired in my simulation,
cause my homogenous direction is ONLY Z! So how can I revise the postChannel code and recompile
to meet my command? Or are there already some existing codes that can help?

*Second,after about 7 flow-times,the calculation exploded due to the gradually growing Courant
number which started at a mean value of 0.2(maximum 0.6) and ended up to an amazing value of several
thousands!Meanwhile, the pressure gradient ‘gradP’is dropping all the way from 0.00886 to below
0.0001,yielding friction velocity u_tau dropped form 0.09 to 0.01!
I'v no idea how this happened!So I wanna know how is the gradP calculated in the straight
channel,is it a mean value of the whole field which no longer suit my wavy case?Can I still use
this channleFoam application for my simulation od should I change for another application e.g.
pisoFoam?

----------------------------------------------------------------------------------------
Any advise is appreciated!
Best Wishes!
sunliming
Tongji University
Shanghai, China
Email:sunliming59045@hotmail.com

sunliming March 6, 2012 02:07

1 Attachment(s)
Attached is the geometrical configuration of the open channel copied from Hudson's paper,but the coordinate system should be adjusted.

sunliming June 6, 2012 23:35

moving wavy channel
 
1 Attachment(s)
moving wavy channel

sunliming June 6, 2012 23:57

moving wavy channel
 
Hello all,
I am doing a LES simulation of flow over a flexible wall undergoing a streamwise travelling wave motion, the lower boundary of the domain is shown in the picture below. The elevation of the lower wall is a function of both time and space, i.e. y_wall=amplitude*sin[k(x-ct)], where c is the phase speed and k is frequency, which means every point of the lower boundary oscillates up and down its own equilibrium position with different phase angle.
So how can I deal with such kind of boundary motion and corresponding dynamic mesh? Can I use codeFixedValue BC? How?
Thank you in advance!
Beat regards!
Sunliming
Tongji University,
Shanghai, China,
Email: sunliming59045@hotmail.com
http://www.cfd-online.com/Forums/att...g-boundary.jpg

sam1364 May 1, 2013 15:59

How did you fix Re for this flow geometry?


Quote:

Originally Posted by sunliming (Post 365167)
Hello all,
I am doing a LES simulation of flow over a flexible wall undergoing a streamwise travelling wave motion, the lower boundary of the domain is shown in the picture below. The elevation of the lower wall is a function of both time and space, i.e. y_wall=amplitude*sin[k(x-ct)], where c is the phase speed and k is frequency, which means every point of the lower boundary oscillates up and down its own equilibrium position with different phase angle.
So how can I deal with such kind of boundary motion and corresponding dynamic mesh? Can I use codeFixedValue BC? How?
Thank you in advance!
Beat regards!
Sunliming
Tongji University,
Shanghai, China,
Email: sunliming59045@hotmail.com
http://www.cfd-online.com/Forums/att...g-boundary.jpg



All times are GMT -4. The time now is 21:47.