CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   Main CFD Forum (https://www.cfd-online.com/Forums/main/)
-   -   I need help in setting up grid for cylinder (https://www.cfd-online.com/Forums/main/94834-i-need-help-setting-up-grid-cylinder.html)

 thethirdfarial November 28, 2011 00:58

I need help in setting up grid for cylinder

Hello CFD Online forumers.
I would like to ask about the best way to set up grid for flow around cylinder.
I am currently doing a simulation around cylinder where the flow is in axial direction. The aim is to get the flow to transition into turbulent flow. So far i have yet to get the flow to be turbulent no matter how much I change the parameters. When I change the grid into those of plane wall the program works fine and the flow transition into turbulent without problem so I think there must be a problem in the way the grid is set up. I am using fortran for the simulation. Here are the equations I used to setup the grid:

http://i166.photobucket.com/albums/u...upequation.gif

r is the circle radius, and its made so that the grid gets denser the closer it get to the cylinder surface.
theta in this case is 1.

Here is a sample of the projected grid

http://i166.photobucket.com/albums/u...GridSystem.jpg

The program I am using for this simulation uses LES model with weak compressibility.

So I am quite clueless what's actually happening that the flow won't transition into turbulent flow. Could sin & cos be the problem here?
Can somebody help me?
Thank you

 cfdnewbie November 28, 2011 04:32

Did you check the Reynolds number? Can you (based on Re) estimate the transition length? Try turning off the LES model, and see if that helps. I suspect that your model might be too dissipative and that you need to trip the BL or add some noise to it to get it turbulent....

 thethirdfarial November 28, 2011 05:57

For plane wall the critical boundary layer reynolds number (ReDL) for transition is around 1100 so in my case i used ReDL of 1500 and above.
For the flow, the reynolds number is around 60,000.
In my assignment I need the reynolds number to be as low possible since I'm working on the minimum diameter needed to keep turbulent flow in the boundary layer.

I did insert noise at the start of the calculation and it works fine on plane wall.

I haven't tried taking off LES and use other model so far. Is there any other suggestion?

 qianquan April 6, 2017 20:44

i also met the same problem

Sent from my iPhone using CFD Online Forum mobile app

 arjun April 6, 2017 20:54

Quote:
 Originally Posted by qianquan (Post 643976) i also met the same problem Sent from my iPhone using CFD Online Forum mobile app
Shift your grid just by a fraction say 1.0E-5 m or so.
The issue is that turbulence to start it needs some noise.

 qianquan April 6, 2017 21:26

Dear arjun: thank you for your reply. i did not add any noise, i would have a try . could you help me with my question? those days i did a lot of work on circle cylinder flow simulation with les (sgs : sma - dynamic). My case as follows, 3D domain : 30Dx20DxpiD , D is the diameter of circle cylinder, D=40mm, the mesh around cylinder ,y pluse =1 , the boundary layers are 40, the max grid size is 4mm, in spanwise the grid is 2.5mm, and use structure grid with icem. and then the setting up in fluent : at first , k-e turbulent model applied to calculate a coarse flow filed with steady, when it converged, transfer to les (smag lily dynamic) , slover method ,piso green gaguas ,second order , spatial discretion is bound central differential,this is my setting. the case applied to HPC with 96cores , after long time calculation, the Cd is 0.8 which less than 1.2 (article),the reynolds are 140000, fluid is water, time step is 0.0001, which the max court number is 2, and runs 10 seconds. thank you in advance.

Sent from my iPhone using CFD Online Forum mobile app

 All times are GMT -4. The time now is 18:33.