[Shoonya]

Hi all,

Could anybody help me in selecting a correct solver. I have a vertical cylindrical annulus region filled with Argon. Its height is 1m and annulus region is 2cm wide. The bottom is at 803K and top at 303K with pressure 100 millibar above than atmospheric pressure. Using Argon properties in this temperature environment the Rayleigh Number is around Ra=10^7. My question is:
Is that correct to use buoayantBossinsqPisoFoam for solving this problem to see the natural convection. Or I shall have to write my own solver?. Also can I change g (gravity) to change the Rayleigh number to view the various temperature/velcoity profiles for different Rayleigh numbers.

thank you

regards
dinesh

[eugene]

The temperature difference is probably too big for the Bousinesq approximation. Try using the compressible solver buoyantPisoFoam instead.

Yes you can change gravity in the file - constant/g

[Shoonya]

Hello eugene
Thank you for your reply and helpful suggestion. I will try with buoyantPisoFoam. A little confusionis is that in the theory of this problem people have used incompressibility and Bousinesq approximation. So, if I use compressible solver, I may not follow the thoery. Could you tell me please how much max. temperature difference can be handled by buoayantBossinsqPisoFoam?
Second, I was trying to ask if I change gravity does it makes sense ? Since in nuclear reactor all the properties of Argon are fixed at those temperatures. So I can not alter Ra No.=g*beta*DetaT*L^3/nu^2)*Pr more. I can change L, but that have to be changed 10 orders of magnitude which is again impractical. So, I thought to change g just for simulation, since Bousinesq approximation involves g.

thanks you again for your valuable time.

dinesh

[eugene]

Check this paper on the Boussinesq approximation: (http://articles.adsabs.harvard.edu/c...MB&classic=YES)

It gives you everything you need to know about Boussinesq. In general the Boussinesq approx is valid if d(rho)/rho0 << 1 ~< 0.1. For an ideal gas with pressure ~constant, this can be expressed as d(rho)/rho0 = (T - T0)/T0.

Where T0 is something like the mean temperature. As you can see this does not leave a lot of room. For T0 = 300, the min max range of T is only 270 - 330 K. Not a lot at all.

The Boussinesq solver should still work, even with very high temperature differences, it just wont be very accurate.

You can certainly change g, as long as your dimensionless numbers all stay the same the results should be equivalent.

[Shoonya]

Thank you very much Eugene....