Velocity depending on time in periodic channel tutorial

Submitted by caalvarez on Fri, 02/23/2018 - 00:21

Hello, I am using the periodic channel tutorial. I want to modify the fluid velocity to make it time depending. In the fvOption file, I need to adjust it to set that the flow velocity starts to act after a specific time. Currently, the flow has a velocity at time = 0 s. I want to fix another start time, for example at t = 1 s. I appreciate any suggestions.

alice's picture

alice | Fri, 02/23/2018 - 08:57

Hello caalvarez,

a very straight forward way would be to run the calculation for one second with zero velocity, trigger the writing of restart files (tutorial: cfdemSolverPiso/ErgunTestMPI_restart) and restart the simulation with changed fvOptions file from there...

Best,

Alice

caalvarez | Sun, 05/13/2018 - 08:54

Dear Alice,
Could you please help with some suggestions about how to eliminate the periodic conditions in periodic channel tutorial? I have been modified all files in folder 0 and the BlockMesh file, but I guess that I also need to change/modify the fvOptions file. So, would be there other files to be changed? Thanks.
Best,
carlos

caalvarez | Wed, 04/11/2018 - 23:37

Hi Alice,
Thanks a lot for your response. I had not noted that you responded me. I was waiting for a reply to my email address. Anyway, I would like to know a little more in details how to use restart files. I have already made some modifications to my case, based on tutorial: cfdemSolverPiso/ErgunTestMPI_restart, but I do know which is the moment when I have to change the velocity at fvOptions files. I start my simulation with a velocity close to zero, to let settle the particles, but after 1 second I want to impose a velocity different to zero. I'll really appreciate any help.
Best,
Carlos

caalvarez | Tue, 05/15/2018 - 22:23

Dear users,
I need to modify the periodic channel to a non-periodic case. Could anyone help me? I'll truly appreciate any suggestion.
Best,
Carlos

alice's picture

alice | Wed, 05/16/2018 - 15:38

Hi Carlos,

for a non-period channel flow please remove the fvOptions file from the system folder and change the inlet and outlet faces from cyclic to patch. The apply the desired inflow velocity at the inlet in combination with zeroGradient velocity at the outlet. For the pressure you can then use zeroGradient at the inlet and a fixed value at the outlet. (In the ErgunTestMPI we do something similar, only in z-direction instead of x).

Hope this helps!

Best,

Alice

caalvarez | Fri, 05/18/2018 - 01:33

Hi Alice,
Thanks so much. This information really helped me.
Best,
Carlos

caalvarez | Tue, 06/12/2018 - 01:56

Hi Alice,
I am trying to implement a RANS model to this case. The laminar simulation have run well, but when I try to add turbulence (using a RAS model) the simulation collapsed. The Courant number increases to very large values. My actual time step for CFD is 0.0001 s (in the original case it is 0.001 s); and the ratio between CFD and Liggghts time steps is 100. The dx in the streamwise direction is 0.002 m. The fluid is water. Could you please recommend me some details that I should have to review?
Best.
Carlos

caalvarez | Thu, 06/14/2018 - 15:20

Dear users,
I need to implement a turbulence model in the periodic channel case. I am facing some problems using a RANS model. The Courant number increases to very large values. Could anyone help me? I'll truly appreciate any suggestions.
Best,
Carlos

caalvarez | Wed, 06/20/2018 - 22:51

Hi Users,
I am modeling the formation of sand patterns in a rectangular cross-section flume, my base case was the periodic channel tutorial. I did some laminar simulations, i.e., without a turbulence model, and I got good results when they were compared to my experimental results, but my experiments were in a turbulence regime (between 11000 and 17000 Re numbers). So, my next step was to implement a turbulence model. Until now, I have employed kappa-epsilon and kappa-omega models. In all cases, even with y^+ of the center of the first cell ~ 2, the turbulence simulations produced "strange" results. After a certain time that my simulation starts, the particles start moving as "solid body" in the streamwise direction, without rotation as it would be expected. I have read a right amount of information (in both forum and the user manual) on how to implement the turbulence, and I am confident that my initial setup accomplishes with the requirements for a good simulation (at least believe that)
I had two gif animations, but I couldn't upload them. I truly appreciate if anybody could give some help to explain, why the laminar case works better than the turbulent one.
Best,
Carlos