[demm10]

Dear forum,

I faced wit a problem of modeling CO2 compressor.
Inlet total parameters are: P0* = 8 MPa, T0* = 310 K.

The problem is that inlet total pressure and temperature is in supercritical region, but inlet static parameters are in liquid region (also close to saturation line). So working process is actually one phase.

I've made RGP fluid file and use it in CFX as a homogeneous binary mixture.

When I'm trying to solve this case, the solver fails with error:

| Newton's method failed to converge while computing the variables |
| listed below. In each case, the solver continued with the variable |
| field as it was on the final iteration. |
| |
| If this situation persists, you might try decreasing the Newton |
| iteration underrelaxation factor. This can be changed by setting |
| one of the following parameters for your mixture: |
| |
| Temperature : "Constitutive Relation Under Relaxation" |
| Pressure : "Newton Pressure Under Relaxation" |


May be someone have any experience with such cases?
I would be very thankful for any help

[ghorrocks]

This sort of problem is very common with complex material models. You have to start by making sure the simulation works well with a simple material model like ideal gas. Then as you work towards the full material model you need to look at issues like double precision numerics, smaller time steps, better mesh quality and a better initial condition to resolve the problem.

[alirezame]

Hello,

Do you use non-equilibrium model? It would be better if you can get the result with a single phase (supercritical CO2 gas). so you do not need to use homogeneous binary mixture. if you want to get non-equilibrium condensation model in 2 phase, then you have to write some CEL code to force the solver to use the liquid table. also metastable region and spinodal curves should be defined in your RGP table.

[demm10]

Originally Posted by ghorrocks

This sort of problem is very common with complex material models. You have to start by making sure the simulation works well with a simple material model like ideal gas. Then as you work towards the full material model you need to look at issues like double precision numerics, smaller time steps, better mesh quality and a better initial condition to resolve the problem.

Originally Posted by alirezame

Hello,

Do you use non-equilibrium model? It would be better if you can get the result with a single phase (supercritical CO2 gas). so you do not need to use homogeneous binary mixture. if you want to get non-equilibrium condensation model in 2 phase, then you have to write some CEL code to force the solver to use the liquid table. also metastable region and spinodal curves should be defined in your RGP table.

Thanks ghorrocks and alirezame.

Yes, this is really very complex problem.
I tried to simulate with Ideal Gas as a working fluid and get quite good convergence.
I think the problem is in another thing. As I mentioned earlier inlet total parameters are in supercritical region while static parameters are in liquid region. The most complex thing here is that this phase change comes without 2 phase (without condensation or evaporation - see attached picture).
Hence, solver understand that there should be liquid - gas phase change, but cannot understand how can this be possible without 2 phase region (it is only my suggestion).

May be you have some ideas how to simulate in this region correctly?

[aghagoli]

Hello
Could you solve the problem? please let me know how you could solve this.

[evcelica]

If the pressure everywhere is above the critical pressure, you can use just single phase vapor for the simulation. The .rgp file should include data in the vapor portion of the file for the fluid above the critical pressure, even though temperature is below the critical temperature.
At least this is my experience with helium. You never really have a phase change it seems (at least not an abrupt liquid to gas phase with different properties), as you never cross the saturation curve. Look in the CFX Solver modeling guide. Figure 12.5. It looks as if the fluid is labeled as "supercritical" when either the temperature OR pressure is above the critical temp/press. Normally we say temperature AND pressure must be above the critical point to be a supercritical fluid, but that is purely our nomenclature. The CFX diagram makes more sense to me than normal phase diagrams with respect to defining fluid properties, and where a real phase change would occur. Above the critical temp OR Pressure, there is no abrupt phase change, just a change in nomenclature of what we would call "subcooled liquid above the critical pressure", and what we would call a "supercritical fluid". But the fluid properties are continuous and smooth across this nomenclature change.

Hope this makes sense? It took me a while to come to this realization.

See next post for Figure 12.5:

[evcelica]

Figure 12.5 of CFX Solver modeling Guide.

[aghagoli]

Thanks, evcelica
My question is: when we want to define Supercritical RGP table, we need to define the saturate properties for each table for instance, for enthalpy or speed of sound, etc. So, we should define the saturated vapor line (quality=1) or we should define the saturated liquid line (Q=0).

[evcelica]

I have always used an RGP generator, not made my own table, so I'm not 100% sure all what is in there. I could send an example table of what my generator makes. I use "NISTtoRGP", I'm sure you can search in this forum for "NISTtoRGP" for my other posts which describe where to get it and how to use it.