search

sanguinariojoe / aquagpusph

Free CFD software based on SPH and accelerated with OpenCL
GNU General Public License v3.0
34 stars 13 forks source link

Spurious movement of particles in a 3D Stationary Tank #10

Open calderonsj opened 7 years ago

calderonsj commented 7 years ago

It has been found that when carrying out a 3D simulation of a tank with no movement particles disorder massively, inducing undesirable accelerations for some of the particles which end up with them at non-physical positions and hence huge forces appear. This suggests that some variable may tend to zero at a point for some of the particles. The main consequence is a loss of height of the filling level after several seconds of simulation. Boundary integrals methodology at the boundaries is used.

Different values of speed of sound have been tried. Other alternatives, such as 2D simulations and 3D simulations with different resolutions may be tried in order to find the source of this unexpected behaviour.

Another possibility is moving to a semi-analytic formulation instead of the pure numerical formulation used in the code, but I would like to ask for other opinions that might solve the issue.

Please find attached here the case, that has been run for version 3.0.

Thank you

StationaryTank.zip

asoutoiglesias commented 7 years ago

this is the geometry. tanque_chaflanes

sanguinariojoe commented 7 years ago

Thank you for the report! I'll take a look ASAP!

That is probably related with the boundary-integrals formulation.

sanguinariojoe commented 7 years ago

In the way the gradient of the pressure is computed, errors of order 1 can be expected, and therefore heavy problems. Also the viscous term would not help at all.

I'm trying to develop a boundary condition similar to the applied by Bousscase in his thesis

sanguinariojoe commented 7 years ago

@calderonsj How long should you wait to see the artifacts?