Species transport in multiphase flow
 

Hello,

I need to simulate the distillation process without combustion (mixing of liquid and gas phases) in which light components of liquid transfer to gas and heavy components of gas transfer to liquid. I tried to simulate this process with following options:

-Eulerain multiphase model, Species transport, Henry's law species mass transfer
phase-1: mixture 1 (hexane-liquid+benzene-liquid)
phase-2: mixture 2 (hexane-vapour+benzene-vapour)

This model requires that the sum of species mass fractions equal to 1 in whole domain whereas I need that this condition was satisfied only in phases (i.e. mxiture 1/ mixture 2 is located only in phase-1/phase-2).

Please explain what is my mistake and how solve this problem.

Best Regards.

I'm not sure if I understand you correctly; but it isn't strange that the total mass fraction of all components, over all phases, should add up to 1 right?

Dear Cees Haringa,

Thanks for the reply. I meant that total mass fraction of mixture-1 components in whole domain should be equal to 1 and the same condition for mixture-2. I want that components of the mixtures flowing in the respective phases: liquid components - phase-1, vapour components - phase-2. And use interphase species mass transfer.

Using the above-described model vapour components are in the liquid phase, liquid components are in the gas phase.

Could you be a bit more specific about what you did?

Correct me if I'm wrong anywhere:

- set up the basic case (turbulence model, energy eq., blabla)
- Enable species transport. Make 2 mixtures:

• M_gas: benzene (g) + hexane (g)
• M_liq: benzene (l) + hexane (l)

go to phases > interaction > mass transfer and set up 2 mass transfer mechanisms:

• benzene (g) > benzene (l)
• hexane (g) > hexane (l)

Choose the appropriate mass transfer mechanism (so condensation-evaporation, or make a UDF for your specific rule)


run the simulation.


Am I right with that? If so, I still cannot see where the problem is. There is no reason why total mass should not be conserved, nor why the species mass fractions in their respective phases should be erroneous... So please give a bit more background on your setup

- laminar flow, energy equation, no gravity

- Eulerian multiphase with Species transport (Diffusion Energy Source):

• Primary Phase - M_gas: benzene (g) + hexane (g)
  M_gas:
  • Density: incompressible-ideal-gas
  • Cp: mixing-law
  • Thermal cond: mass-weighted-mixing-law
  • Viscosity: mass-weighted-mxing-law
  • Mass diff: kinetic-theory
• Secondary Phase - M_liq: benzene (l) + hexane (l)
  M_liq:
  • Density: incompressible-ideal-gas
  • Cp: mixing-law
  • Thermal cond: mass-weighted-mixing-law
  • Viscosity: mass-weighted-mixing-law
  • Mass diff: kinetic-theory
• Surface tension force modeling
• Now I consider a simple problem without mass transfer mechanisms

- BC (see pic):
inlet_bot:

• gas: velocity-0.1m/s, vf gas - 1, hexane (g) mf - 0.65
• liq: velocity-0m/s, vf liq - 0, hexane (l) mf - 0

inlet_top:

• gas: velocity-0m/s, vf gas - 0, hexane (g) mf - 0
• liq: velocity-0.05m/s, vf liq - 1, hexane (l) mf - 0.35

outflow
-Solution method: PC-Simple
-Solution initialization:

• velocity - 0
• hexane(g) - 0.65
• liq vf - 0
• hexane(l) - 0

Thanks for the details!

Ok, I can see some potential problems now.

1) you use inc. ideal gas for both liquid and vapor phase, which means the phase density will be equal between the phases. I assume this is not what you want. Instead, use volume or mass weighed mixing law for the liquid phase density (or for simplicity, set it to constant for the time being)

2) why do you model surface tension force? is it important in part of the problem (just curious)

3) If there is no mass transfer, what is going to happen? I guess your domain will be 2/3 filled with gas, with 65% hexane and 35% benzene, and the 1/3th with liquid (since your volumetric flowrate is half that of the gas) with 35% hexane and 65% benzene. Is that what happens, or is that where it goes wrong?

Also why do you have 2 inlets and one outlet? A simple distillation setup should have 1 inlet (gas/liquid feed) and 2 outlets right?

Dear Cees Haringa,

Thank you for advices!

1) I agree with this.
2) In the future I plan to calculate the distillation model where there is a layer of liquid free surface.
3) I have same above-described problem.
4) About the outlets, this test case to determine the correct options for distillation process. In my main model will be two outlets.

I think I've tried everything and with more confidence come to the conclusion that Ansys does not calculate such problems (2 phases+2 mixture). On this topic I have not found any tutorial.
And notice I set hexane(l) to 0 in Solution initialization. It leads to that benzene(l) is equal 1 in all domain. Although initially the liquid is not present in the domain (Solution initialization: liq vf=0).

I don't know how to solve my problem, probably the only way is to try to identify the transport of species through UDS. Although I'm not sure how to work the conservation equations for UDS species.

I attached pictures of result (small calc time) where "species 0" is hexane(l), "species 1" - benzene(l).

I think the main issue is on how the Mass fraction is defined in Fluent. If you take a look at the units of the mass fraction (30.4 Alphabetical Listing of Field Variables and Their Definitions) is on Kg of species "i" per Kg of the mixture. The mass fraction is calculated based on the volume of the mixture (and not on the volume of phase-2, which is what you want). This is why the contours of volume fraction of phase-2 and mass fraction of species "i" look so different, and it looks like the species are gaining mass.

What I've seen in the Combustion tutorials is that they have one fluid zone (one volume mixture). They use inlet-velocity where you can put the mass fraction, and volume fraction. Then, they impose that all the cells within the domain have the same volume fraction and mass fraction for that whole fluid zone. I think that this helps to conserve the mass.

Were you able to solve the problem or deploy a workaround?