[Nauseouslyquick]

Hello everyone,

I use polyflow for 2D newtonian-flow past a circular cylinder problem,and i have two problems. First of all,i can capture vortex street at about ~Re 73 instead of Re 47. However, with the same mesh, i can obtain vortex street at Re 52 by using Fluent.

My domain is 10D inlet, 20D outlet,20D Height. velocity inlet, no slip on the walls and cylinder, zero pressure at outlet.

My second problem is that when i change height of the domain as 5D ( like 2d pipe flow over a sphere case) , i cannot visualize the vortex shedding even at re 300 (delay of the vortex is acceptable, but this much aint though) .In my other cases, even though i ve captured shedding at about 5 to 10 sec; i couldnt see anything in 60 sec with 5D height.

No matter what i do like mesh assymetry for disturbance, time step reduction, enhancing process time; i couldnt figure out the problem.

Any suggestions, solutions and thoughts would be appreciated.
Thanks

[FMDenaro]

Using Fluent, Polyflow or other softwares means nothing if you do not consider discretization and method. Artificial viscosity is one of the main issue in such softwares and you have to minimize it.

What type of BC are you considering?

[Nauseouslyquick]

Hi,

I have not full of knowledge about the subject(discretization_thats why im using package software) but Polyflow runs based on FEM method rather than FVM.

For the integration it uses Implicit Euler, and my convergence criterion is 10^-3.Btw, I mentioned my flow BCs in the former post.
I couldnt modify or change artificial viscosity in both polyflow and fluent (i think its related to truncation error), also i ve checked CFL condition and mesh independency.
Do you mind if you explain how can i reduce artificial viscosity?

Im sorry if i wander off, thank you for your reply sir...

[FMDenaro]

Upwind schemes generate artificial viscosity... However, I suggest to check the solution by performing a grid refinement study