[robo]

Hi all,
I'm fairly new to openFoam and am still trying to get my bearings with its turbulence capabilities. I am investigating some turbulent vortex shedding cases, and am currently stuck on my validation.

I'm running this case for a cylinder at Re = 5e6 & M = 0.2 using the menter SST model in openFoam 2.2.1, using a near wall treatment. Whenever I attempt a run, what appears to be some sort of velocity fluctuation appears at the front of the cylinder, propagates upstream, spread throughout the domain, and eventually the simulation crashes. I've attached two pictures to demonstrate what I mean; the first is as the fluctuations begin to propagate, the second is slightly before a crash (the velocity scale has been modified to exaggerate the fluctuations).

I believe these to be numerical errors, however I'm unsure what's causing them, as I'm not very familiar with openfoam's solver yet. My solution scheme is using the simplest (i think...) options in all cases, ie gauss linear, and backward time stepping. The time step is such that the cfl is around 0.2 in the smallest cell right up until the crash occurs. The turbulent inlet is set with 1% intensity and a viscosity ratio of 1. Boundary conditions are fixed value for k & omega at the wall, k=0 and omega=10*6*nu/(beta1*ds).

I'm hoping that someone has seen similiar behaviour before or can recommend some avenues to explore, as I'm getting pretty frustrated by this.
Thanks in advance
Robo

[flyingd]

Hi,
I am very intersted your case. Could you give your 0 file to me. I can help you check your defination for intinal conditions.

If you like, we can discuss together.

Originally Posted by robo

Hi all,
I'm fairly new to openFoam and am still trying to get my bearings with its turbulence capabilities. I am investigating some turbulent vortex shedding cases, and am currently stuck on my validation.

I'm running this case for a cylinder at Re = 5e6 & M = 0.2 using the menter SST model in openFoam 2.2.1, using a near wall treatment. Whenever I attempt a run, what appears to be some sort of velocity fluctuation appears at the front of the cylinder, propagates upstream, spread throughout the domain, and eventually the simulation crashes. I've attached two pictures to demonstrate what I mean; the first is as the fluctuations begin to propagate, the second is slightly before a crash (the velocity scale has been modified to exaggerate the fluctuations).

I believe these to be numerical errors, however I'm unsure what's causing them, as I'm not very familiar with openfoam's solver yet. My solution scheme is using the simplest (i think...) options in all cases, ie gauss linear, and backward time stepping. The time step is such that the cfl is around 0.2 in the smallest cell right up until the crash occurs. The turbulent inlet is set with 1% intensity and a viscosity ratio of 1. Boundary conditions are fixed value for k & omega at the wall, k=0 and omega=10*6*nu/(beta1*ds).

I'm hoping that someone has seen similiar behaviour before or can recommend some avenues to explore, as I'm getting pretty frustrated by this.
Thanks in advance
Robo

[robo]

Hey flyingd,
Thanks for your offer, but I have resolved the issue since I posted. Using the gauss linear scheme on the convection term was causing the oscillations to appear, as that scheme is unbounded. Switching to upwind or limitedLinearV remove the oscillations, although those two schemes give different answers for the shedding behaviour (upwind gives alternating vortices, while limitedLinearV shows symmetric shedding, which I was not expecting...)