[s.amirzadeh]

Hi,anyone can explain how to define a vacuum domain in cfx? the vacuum domain is between 2 surfaces which transfer heat only by radiation.
i know there is another topic as this one but there is not a clarifing answer,
if anyone know something please share it.

[ghorrocks]

CFX cannot model a pure vacuum. You are going to have to approximate it. So instead define the vacuum domain as a solid domain with a material property with a very low thermal conductivity, and maybe a very low density (you could use the properties of air at the pressure of your vacuum chamber). If the conductivity is low enough that no significant heat transfer occurs it will work.

[s.amirzadeh]

i have defined a material with these propeties:
state:gas
molar density :10e-5
density :10e-5
thermal conductivity:10e-5
specific heat coefficient :10e-5
viscosity :10e5
but my residuals become weird,take a look at them,any ideas?
111.jpg

222.jpg

[ghorrocks]

If you make the values too small then the numerics will go unstable - like you see. You need to make them small enough that it is effectively no heat transfer but large enough the numerics are stable.

I would recommend you use a solid domain using the properties of air at the pressure in your vacuum chamber.

[s.amirzadeh]

thanks for the tip,but how can i define solid domain with air properties?please tell me step by step,and solid domain won't effect the radiation?its solid anyway!
i couldn't find a place to put a desired pressure at the domain,can you explain with details?
and introduce a reference for air properties at a vacuum pressure.

[ghorrocks]

Take an existing solid material and edit it with the properties of air. There are only going to be a few properties to change, probably only density, thermal conductivity, specific heat. And if you want it transparent for radiation then edit the radiation properties to make it transparent.

[JuPa]

In reality a perfect vacuum does not exist (except for perhaps in space).

In reality vacuum pressures are usually the order of 0.1 ~ 100 mbar - pressures which CFX can work with.

There may be no need to model this as a solid domain if you know the true pressure, and only if you don't assume that P = 0 bar!

[ghorrocks]

Yes, you can model this as the real fluid but then you will be solving the NS equations in that domain and the flow will probably do all sorts of weird things - but won't affect the heat transfer one jot. That is why I recommended to model this as a solid domain with the heat transfer properties of air - it gives it a realistic set up numbers to use, but it only solves the heat equation and does not waste CPU effort on the NS equations.

[JuPa]

Originally Posted by ghorrocks

Yes, you can model this as the real fluid but then you will be solving the NS equations in that domain and the flow will probably do all sorts of weird things - but won't affect the heat transfer one jot. That is why I recommended to model this as a solid domain with the heat transfer properties of air - it gives it a realistic set up numbers to use, but it only solves the heat equation and does not waste CPU effort on the NS equations.

The true solution of this problem will be radiative effects, with some conduction and some convection. If does s.amirzadeh model this as a real fluid, then you can gauge how important the conductive and convective effects are in relation to radiation (arguably you can do this without CFD). I agree with your assertion that it means you solve the N-S equations.

However a much simpler way (imo) to model this is to just set everything up as normal, using real properties and the correct pressure, and then in expert parameters turn of fluid modelling and run the simulation. That way if you then later decide you do want to model the fluid, at least you have a decent initial guess. I don't think the solid domain approach you described will give an ideal initial guess to the same problem with conduction & convective effects are also considered (because you would then need to change material props from solid to fluid).