Learning Forum

[mhchoudary]

Dear all,

I am facing convergence issues during my simulation of Oil (liquid) and water (water-vapour) mixing at supercritical conditions.

I am using Peng-Robinson EOS for individual substance and mixture. The operating pressure is 300bar and inlet/initial temperature is 420°C.

Oil is composed of one specie A and supercritical water is composed of two species (water-vapour and oil component A).

Water (gas phase) at supercritical condition is injected into the container half filled with Oil. The injected water has only water-vapour as component. As water flows into the half-filled Oil container, mass transfer takes place i.e. specie A from Oil is transfered to Water. Hence, when water (in gas form) leaves from the outlet, it contains water-vapour and specie A. This is a multiphase problem and I am using VOF method.

The problem is that the simulation does not converge no matter what i do. What could be the reason? Is it even possible to simulate multiphase problem with cubic EOS since i just read on fluent user guide that cubic EOS are not available for two-phase region? but at the same time it also states that cubic EOS are compatible with multiphase Euler methods?

I appreciate any guidance or help to overcome this issue.

Many thanks.

 

 

 

 

[Amine Ben Hadj Ali]

You can use Real Gas Models with Two-Phase flows. I think the confusion lies in the definition of two-phase domne where CEOS might fail due to the Mawell Correction. Moreover CEOS or even NIST models alones are not able to describe a state under the saturation dome.

How are you modeling the mass transfer?

Supericritcal flows are always challenging and the usage of Real Gas Models can lead to increased numerical stiffness. 

[mhchoudary]

Many thanks for your reply Dr. Amine. To simplify the problem I did not activate the mass transfer. For mass transfer I will/am use UDF for species mass transfer at the interface based on Fick's law.

The problem is as soon as i decrease the operating pressure to, say, 100bar, everything works fine and simulation converges. 

I am using Peng-Robinson EOS for Oil phase which is basically still liquid. Is it correct? when i select Peng-Robinson from drop down list for density and provide molecular weight (723 kg/kmol) and click change/create, a lot of properties (critical temperature, critical pressure, criticl volume Acentric factor etc) are caluclated automaticaly by the ansys. I am relying on these values! Are these correct? For viscosity, I am selcting "Kinetic-theory".

[Amine Ben Hadj Ali]

I do not know if your oil is liquid or not as I do not know the material. Nevertheless I usually have large doubts on modeling liquids with CEOS.

You need to provide the real critical properties for your oil-fluid: if you do not know that you can stop your runs!

[mhchoudary]

I have tried both: using PR-EOS for Oil and providing manually properties for Oil. The properties which i provide manually are correct.

However, water-vapour is assigned Peng-Robinson. The properties are for individual components are corerct.

The problem is, i think, in the mixture. When i use 'Volume-weighted-mixing' in mixture-templete, the solution converges.

However, as soon as I use PR or any other real gas model in mixture-template, residulas especially "continuity" blow up after say 1500 iteration (depending on the timestep). It looks like that as soon as supercritical water enters into the reactor and mix with the Oil, simulation starts diverging.

I am intializing reactor half filled with supercritical water and half filled with Oil. Supercritical water inlet is at bottom-half which is filled with oil. 

While diverging i get many kind of errors/warnings: Usually it starts with "turbulent viscosity limited to viscosity ratio of 1.000000e+05" and then "reverse flow occured in xx faces of oulet", and "temperature is below the spinodal point in xx cells on zone x". 

I have tried everything, from decreasing time step to increasing mesh size etc. but nothing has worked so far. Any help or hint how to get rid of this issue is much appreciated.