Learning Forum

[sardar.svn]

Setup:
Steady | Eulerian-Eulerian (2 phases) | 2nd phase Granular  | Turbulence Multiphase model per-phase | Initialized, then patching of 2nd phase at tank bottom

Simulation objective:
Seeing the steady state of the granular phase stired into the domain

Procedure through which problem arises:
(1) Initialize ==> (2) Patch granular phase at tank buttom at near-max volume fraction ==> (3) Verify total mass granular phase by an expression ==> (4) Start Calculations all the way to very good monitored convergence ==> (5) back to the expression to see the mass has unphysically declined (to some 10% of the initially patched amount, and apparently replaced by the first phase)

I can't get my head around why this is happening while I do remember that my sim was actually working in the very first day.giving results very much similar to the transient simulation that I had in my rocky fluent two-way coupling simulations.
 Background attempts to solve the problem?
1. Did some transient iterations, the mass remains conserved ==> switched to steady, and the mass decline started.
2. Double-checked my Bc's.

1. Do you confirm that such setup (especially Eulerian+Per-phase+steady+packed bed) is actually solvable in the first place without losing mass of the granular phase? In other words do I remember it correctly?

2. How can I solve this and what is actually causing this problem?

Thank you

[Rob]

If convergence isn't good you'll likely lose mass. Is the expected end result a fully suspended system or do you expect a fair level of settling?  Transient typically doesn't see this effect. 

[sardar.svn]

Thank you for answering to my question.

Decline in the overall solid mass is happening no matter the initial volume fraction of granular phase (hence, whether it's a full or partial suspension), or whether I'm using the mixture model or Eulerian. Once again note that the change in the total mass it's not even by one or two percent- like I said in my original post it can be, and usually is, around 80%.

[Rob]

And how is the convergence? What is the initial amount of solid relative to the liquid phase?

[sardar.svn]

 

 

 

Convergence is monitored down to 1e-4 For continuity and other ones which is very rare in Eulerian simulations and I think this is because of the excellent level mesh quality. But aside from residual-wise convergence, I have very stable torque on agitator and also very stable volumetric integral of (epsilon*density) of phase one, which is power per unit volume.

As for init.ial solid volume fraction like I it ranges from 1% to 60%. The image is from my 60% converged simulation that I can’t do a second time unfortunately.

Rob, can you please verify if my setting is actually feasible, namely, steady+Eulerian+Per-phase+packed bed, and therefore, I’m not recollecting it wrong? My screenshot doesn’t show any timestamp below the image and that already means that the simulation was carried out in steady state, right? Do you confirm?

 

 

 

[Rob]

Why are you setting "packed" in a mixing tank? That's for a static system and isn't widely used. Did you monitor solids mass with iteration? 

A lack of time stamp typically means it's steady, but there are ways to mess with the titles so I can't be certain. 

[sardar.svn]

The packed state is actually the initial condition of the system because I have seen difference in results between this case and the uniformly distributed initial state of particles.

Thank you for your confirmation. Since I haven't messed with the titles nor do I even know how to, for that matter, I can be certain that it must have been a steady state but it's very strange to see that I cannot reproduce the same simulation again.

One important thing that is that the very same setup actually works for transient simulation without mass reduction, but when I go steady state, the mass starts declining.

[Rob]

Except that "packed" in the Eulerian model means something, so isn't what you want here. And please note my earlier comment that transient doesn't typically suffer from the mass loss as you're resolving for time. 

[sardar.svn]

I think you mean the specefic "packed bed" in the Phases subtab of Eulerian window. No, I usually do not check that box, as I guess that keeps the phase in place or sth along those lines.