[lhcamilo]

Hi there,

I have been running a few simulations of a piston moving through a fluid from rest and coming to a full stop. The piston consists of a solid cylinder/disk moving through the fluid

For that I have been using the oversetPimpleFOAM, where a "guest" mesh moves through a coarse mesh. Both grids are unstructured and created with cfMesh.

I have been using a variable timestep while keeping the CFL constant at about 0.4.

The max Reynolds number is around 8000 so I am aware that the flow is turbulent, though I am pretty sure these pressure spikes are outliers and probably due to some numerical issue.

I have observed that increasing the host grid coarseness also increases magnitude of the spikes.

Does anyone know what might be causing these discontinuities? Would a structured grid help?

btw, I have run these simulations with both Openfoam 1912 and 2012 and it does not seem to make any difference.


Thanks in advance for any help

[AndreasPe]

I remember reading something about pressure spikes concerning the overset method, but with the foam-extend implementation. Maybe this presentation helps:

https://foam-extend.fsb.hr/wp-conten...ningSlides.pdf

[random_ran]

Originally Posted by lhcamilo

Hi there,

I have been running a few simulations of a piston moving through a fluid from rest and coming to a full stop. The piston consists of a solid cylinder/disk moving through the fluid

For that I have been using the oversetPimpleFOAM, where a "guest" mesh moves through a coarse mesh. Both grids are unstructured and created with cfMesh.

I have been using a variable timestep while keeping the CFL constant at about 0.4.

The max Reynolds number is around 8000 so I am aware that the flow is turbulent, though I am pretty sure these pressure spikes are outliers and probably due to some numerical issue.

I have observed that increasing the host grid coarseness also increases magnitude of the spikes.

Does anyone know what might be causing these discontinuities? Would a structured grid help?

btw, I have run these simulations with both Openfoam 1912 and 2012 and it does not seem to make any difference.


Thanks in advance for any help

It is interesting to see such a unrealistic result. It seems that it is caused by the numerical aspects instead of a presentation of the real physics.

How do you obtain the data? Is this a 3D model? Perhaps a spatial average could ease up a little bit

[lhcamilo]

@Andreas, thanks for your comment. Indeed the issues seem to be related.



@Ran it is a 3D model. It is a disk moving through a quiescent fluid/air with a predefined kinematics.



The graphs are obtained with a (point) probe, and we tried averaging, but the spikes seem to be global rather than local.


Despite the velocity data being in good agreement with the hot wire experimental data we have, though if we calculate the acceleration, the spikes also seem to be present.