Unphysical solidification
 

I am running two phase simulations of sand in water. Clearly the accumulation of sand has a different time scale than the features of the fluid flow (flow direction, turbulence). So my question is if there is a way to solidify, i.e set a high PRPS value or high viscosity value in each cell reaching a specified volume fraction of sand or specified density.

This could be done after a restart when the "steady state" fluid flow is reached, and the development of a sand dune is mainly dependent on the supply of sand. Setting a new surface based on the density distribution and then restart to get a new flow pattern would speed up the computations of a process which takes several hours in real-time.

Re: Unphysical solidification
 

Dear Matos

Very nice and complicated problem. I am not specialist in nucleation but I will try to give my oppinion.

If you use PRPS, calculating density and viscosity from it, you have two choises (as far as I know) IMAT<100 and IMAT>100. In the former case you are considering the second phase (the sand) as a fluid and therefore you will have a contribution to first phase dynamic as a mixture. The later will give you a solid into the cell with the slip boundary condition as default.

The former option seems valid to me as far as you haven't solidification, ie. sand suspension maybe near a nucleation point, but it could be unappropriated when the sand grains (or conglomerates) become fixed to the substrate.

In such state (after some critical fixed mass) the solid substrate will change its form inside the cell and you should correct the cell volume occupied by solid and fluid (sand suspension) into the eqs. of motion.

PHOENICS has a tool (PARSOL) to consider those situations, but I don't know if it is accessible by way of GROUND codding in order to change it during flow calculation.

I imagine the fluid flow near the substrate modulates the nucleation process (stochastically driven) in a non stationary way, but (as I am not specialist on this), if the flow pattern can be considered as locally stationary, your idea (second paragraph) could be coupled to PARSOL technique to consider arbitrary solid surfaces inside the fluid domain.

Another way is to work with porosities (VPOR) to consider the substrate evolution inside the cell.

Regards

Kike

Re: Unphysical solidification
 

Thanks Kike,

Then to a practical question: Is there a way to access the PHI file with the help of PIL after a restart (I don't have a recompilable version). I want to test all computational cells and set porosity low for all cells with a high content of particles (fluid 2) if one of the neighbouring cells has low velocity (to ensure a continous surface).

Regards

Thomas

Re: Unphysical solidification
 

In answer to your specific question - here at Flowsolve we have tackled this sort of problem in slightly different contexts, e.g. freezing of a flowing liquid, and electrostatic deposition.

We used IPSA for the two-phase simulation, and made the viscosity a function of the solid volume fraction R2 so that for R2>0.95 (say), the viscosity ramps up very steeply.

This viscosity function was defined in Ground.

The approach worked very well.

I am not clear however whether there are any specific characteristics of sand deposition which make the problem more complex, i.e. where you refer to different characteristic time scales. More thought may be required here?

Re: Unphysical solidification
 

Dear Thomas

I am sorry to answer you so late. I think you can't do what you want with a non-recompilable version.

If you can't recompile your exe, you can check PHI file and locate the "zones" where you have high content of particles (I assume you can control this).

Then you can insert some patches (into a new Q1) which cover those regions to change the porosity for a constant different value.

Nevertheless, if you want to fix the porosity value for each cell as a function of solid concentration you will need to insert GROUND code and recompile.

Regards

Kike