Transient or steady state CFD problem

A subforum specific to the development of the OpenFoam-based workbenches ( Cfd and CfdOF )

Moderator: oliveroxtoby

User avatar
Posts: 256
Joined: Fri Dec 23, 2016 9:43 am
Location: South Africa

Re: Transient or steady state CFD problem

Postby oliveroxtoby » Thu Jan 10, 2019 4:07 pm

julieng wrote:
Tue Jan 01, 2019 8:18 pm

Can you have a look on the model please (the model is attached in a previous named flow pass a cylinder Re 05). The model cannot be more simple but I have always a very high value of the error for pressure even for very low inlet velocity.

Best regards

Happy new year
I've been looking into this, and it's actually quite interesting. In pure PISO mode, the residual goes to zero, but with the settings we are using it doesn't behave the way I expected. I think this is due to some relaxation-factor dependence in the formulation, and is probably nothing to worry about. If you were to plot a physical quantity like force on the cylinder, you should find it converging. The residual is not all that useful in a transient simulation, and is of more value to determine convergence within a particular timestep if one is running larger timesteps.
Posts: 1394
Joined: Sat May 20, 2017 12:06 pm
Location: Germany

Re: Transient or steady state CFD problem

Postby thschrader » Sat Jan 12, 2019 9:21 pm

For me this discussion about the residuals in cfd is very complicated.
The reason is:
I work as a structural engineer, doing statics. When doing the statics
for a house/bridge/tower... I have to trust the results of the FEM-static-calculation.
Mostly I use 1D-beam elements. And this works. No crash..
For the cylinder-problem of julieng there exists a lot of experimental/numerical data.
You can calibrate the cfd-model. But what to do when I have a completely
new structure?
How do you solve a new cfd-problem without using the wind-tunnel
for calibrating your model?
My english could be better...
Posts: 40
Joined: Sun Nov 25, 2018 8:57 pm

Re: Transient or steady state CFD problem

Postby julieng » Mon Jan 14, 2019 3:08 pm


Maybe the Reynolds number is too small and the case needs a Stoke traitment. In comsol it is advised for Re<1.
I ask you in the same topic if it is possible to plot forces acting on the cylinder over time directly in paraview. Or maybe it is possible to add in the case some post processing command with openfoam blue-core.

Best regards