CFD Online Discussion Forums

CFD Online Discussion Forums (https://www.cfd-online.com/Forums/)
-   Main CFD Forum (https://www.cfd-online.com/Forums/main/)
-   -   Flow past cylinder (https://www.cfd-online.com/Forums/main/162513-flow-past-cylinder.html)

Tanmay95 November 12, 2015 01:25
Flow past cylinder
 
Hello guys, I am trying to model a wind tunnel experiment of flow past cylinder in fluent. Cylinder dia is 63.5 mm velocity is around 17m/s and I am using standard air properties in fluent database. Currently I am using a transient analysis with standard k-e and first order accuracy in time with a time step of 0.02 sec. I am not getting in vortices in the down flow and my cd and cl are not showing any oscillations after few time steps. I also tried searching for some research paper for high Re flow over cylinder but I can't find a one which helps. I don't know what I am doing wrong. (I have only done a introductory course in CFD so please try to explain in simole words). Thanks[emoji3]

shakabrade November 19, 2015 14:57
You usually have to wait for some time (could be long) for vortex structures to start appearing. First order k-epsilon is very diffusive and is probably ironing the vortex structures. You could try polluting the inlet flow with some velocity profile or try switching to second order. Also, you could try k-omega SST, or even Spalart-Alamaras (1 equation model). k-epsilon is not very good at predicting the flow separation point and that is crucial for this kind of problem

flotus1 November 19, 2015 18:03
  1. Estimate the time scale of the largest vortices using the Strouhal number
  2. Choose a time step size that is at least a factor of 20 smaller as an initial guess
  3. Make sure the mesh is fine enough to resolve the large vortices you expect to see
  4. Use second order accurate time integration
  5. Use a reasonably low turbulence intensity at the inlet
  6. Run the simulation for at least 20 times the period time of the largest vortices

FMDenaro November 20, 2015 10:53
Quote:
Originally Posted by flotus1 (Post 574105)
  1. Estimate the time scale of the largest vortices using the Strouhal number
  2. Choose a time step size that is at least a factor of 20 smaller as an initial guess
  3. Make sure the mesh is fine enough to resolve the large vortices you expect to see
  4. Use second order accurate time integration
  5. Use a reasonably low turbulence intensity at the inlet
  6. Run the simulation for at least 20 times the period time of the largest vortices

Just to add that URANS formulation can somehow useful but if you really need a transient analysis, LES is the preferred approach.


All times are GMT -4. The time now is 02:15.