[leonardoasfora]

Hi,

I'm having some problems on specifying an ASM simulation. I followed the guidelines that one could find in ANSYS CFX-Solver Modeling Guide about this topic, but im not convinced that i'm doing this right. It says that in order to set up a ASM simulation, you first need to create a material for a variable composition mixture. I'm insterested in water-air flow, so I created a mixture of liquid water and air ideal gas, which I called MaterialMixture. The mixture properties were set up as ideal mixture.

In the Fluid's and Particle Definitions on the Domain's tab, I inserted only a item named Mixture, related to MaterialMixture. On Fluid Model's tab, the dispersed phase (air) was set to Algebraic Slip with Schiller Neumman model (mass fraction of 0.015), and the continuous phase (water) was set to Constraint.. The boundary conditions was pressure (10atm) at the Outlet and mass flow rate (0,6 kg/s) at the inlet.

I than proceeded to the simulation, which consists of water-air flow in a convergent nozzle. Convergence was nicely achieved (10^-5 RMS residuals). However, i noticed that the mass fraction of gas varied along the axis of the flow, what shouldn't happen since im not modelling mass tranfer between phases. I think i might be doing something wrong when setting up the ASM simulation, but I cant find the problem. Could someone help me? I really need to make this work.

Thanks in advance.

[ghorrocks]

What are the imbalances doing? If you are having problems with conservation but the residuals are well converged you might need to add imbalances to the convergence criteria.

[leonardoasfora]

Thanks ghorrocks.

I actually already used imbalances as a convergence criteria setting the conservation target to 0.01. However, gas mass fraction from the inlet to the outlet varied about 50% (0.015 at the inlet and 0.010 at the outlet).

This situation is very weird indeed! It got me thinking why an ASM should be set as variable composition mixture (VCM). By definition, in a VCM mass fractions of each component are allowed to vary, and their somation are constraint to unit. But ASM models not necessarily have to allow it, as in the case i'm trying to study.

I dont know what's going on. I tried to find something in the tutorials but there's nothing there. I appreciate any help one could give me.

[ghorrocks]

You need to determine whether what you are seeing is a result of numerics or the physical models you have selected. By numerics I mean your simulation is not yet fully converged. To check your numerics are OK, do another simulation with the convergence tolerances event tighter than you currently have (residuals to 1e-6, imbalances to 0.001). If the result does not change the issue is from your physics selections, if the result changes then your simulation is not fully converged.

[leonardoasfora]

Thanks ghorrocks.

I actually realized what the "problem" was. I misinterpreted the definition of mass fraction by CFX. I thought it was the contribution of gas mass flow rate to the total mixture flow rate (something like "quality"). However, the definition of mass fraction done by cfx is simply the division between the phase density and the mixture density, which of course should vary if there's a compressible phase.

[leonardoasfora]

I am actually talking nonsense. Forget what I said earlier. The problem still persists.

As I said before, the results didnt changed when i used tigther convengence criteria, which means that there's some problem with the physical model i'm trying to simulate. I've tried many things but nothing worked. I must be doing something wrong but i cant see what. The solver solves a mass fraction equation, and the results is that it varies throught the domain. However, this comes with the definifition of variable composition mixture . Is there anyway to "discard" this equation?

Thanks

[ghorrocks]

Please post an image of what you are getting, an image of your mesh and your output file.

Also: What happens if you run a full eularian multiphase model rather than the simple ASM one?

[leonardoasfora]

Thanks ghorrocks.

Here's the mesh i'm using. I think it's okay:



Aspecto ratio: Min 1.16/ Max 8.34/ Average: 2.38
Skewness: Min: 10^-5/ Max: 0.76/ Average: 0.18

I used a full eulerian two fluid model (only drag force included with the same bubble diameter and boundary conditions) and I found the same behavior: gas mass fraction varies along the domain. The residuals criteria was 10^(-6) for all equations and conservation target of 0.001. Here are the output flies:

TwoFluid.zip

Drift.zip

Here's a comparative between gas mass fraction averaged in cross section for both model along the x axis.



I'm starting to think that the problem is actually me...maybe i'm misinterpreting the definition of mass fraction...Maybe it's not supposed to be constant . I made some tests and mass fractions were only constant with negligible slippage between phases (homogeneous model). Is this case, volume fractions can be directly calculated using volumetric flow rates, and gas mass fractions turns out to be equivalent to the ratio between gas mass flow rate and the bulk mass flow rate, which of course is constant if there's no mass transfer.

If there's slippage between phases, volume fractions are not directly related to volumetric flow rates, and therefore the equivalence to mass flow rates is lost. I thought it should remain constant because somehow it's not intuitive to me that there's a definition of mass fraction that varies even if there's no mass transfer between phases. But what do I know...

[ghorrocks]

Can you show a cross section of the mass fraction?

Can you explain why you say your comment in post #5 is wrong?

[leonardoasfora]

I think you are reffering to this cross section:



I was looking for the definition of mass fraction and i found the following in ANSYS CFX-Solver Theory Guide:

(1) mass fraction = (component density/mixture density)

If this was the case, it would actually vary throughout the domain if there's a compressible phase in the mixture. However, in the context of ASM, if you follow the model derivation you'll see that in fact:

(2) mass fraction = (phase density/mixture density) * volume fraction

And as I said, i thought it should remain constant.

I checked the results and mass fraction values given in CFX-post obey equation (2).

[CFXILMENAU]

Originally Posted by ghorrocks

What are the imbalances doing? If you are having problems with conservation but the residuals are well converged you might need to add imbalances to the convergence criteria.

Hi.

I have seen all the replies and then I have found that this is a water-air multiphase simulation. Why is the setting up a multicomponent simulation with a variable composition mixture?

Best regards

[Opaque]

The Algebraic Slip Model can be implemented either way, i.e. as a multicomponent material, or multiphase flow with pure materials.

ANSYS CFX used the multicomponent approach early on, and it has not been modified since.