CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   CFX (http://www.cfd-online.com/Forums/cfx/)
-   -   Rarefied gas (http://www.cfd-online.com/Forums/cfx/91459-rarefied-gas.html)

 thais_b August 11, 2011 12:00

Rarefied gas

Hi!

I'm modelling a reentry vehicle in CFX and I'm working with low density gas. I'm facing a problem with overflow in all my simulations: there are few peaks of pressure near the geometry (a blunted body) and 3 steps later, the simulation crashs. I read somewhere that Navier-Stokes does not work anymore for these conditions. Does someone have experience with this kind of simulation?

 ghorrocks August 11, 2011 19:16

CFX should work fine until the gas can no longer be considered a continuum or until the gas disassociates. But for rarefied gasses you will probably need to use a different EOS.

But to start off I would just use an ideal gas at the temperature and pressure you are using, and once that is working reliably I would swap to a more accurate EOS.

 thais_b August 11, 2011 21:54

Thank you for you answer Glenn.
Do you have any idea when the continuum equation fails? I'm working with density of 10^-5 kg/m3 and I was suposing that this value could be the reason that causes overflow.

 ghorrocks August 11, 2011 22:00

http://en.wikipedia.org/wiki/Knudsen_number

 thais_b August 12, 2011 17:20

Thank you Glenn. I'm gonna check this.

 thais_b August 15, 2011 09:28

Hi again!
My Knudsen number is about 0.004 (M=10 and Re=4000). My conditions are calculated based on air ideal gas. So the solver should work fine but in the fifth step, the solver always returns this error:

Parallel run: Received message from slave
-----------------------------------------
Slave partition : 3
Slave routine : ErrAction
Master location : RCVBUF,MSGTAG=1012
Message label : 001100279
Message follows below - :

+--------------------------------------------------------------------+
| ERROR #001100279 has occurred in subroutine ErrAction. |
| Message: |
| Floating point exception: Overflow

Any idea?
Thank you!

 ghorrocks August 15, 2011 18:57

Your Knudsen number looks OK, so CFX should be good to model this.

You are diverging numerically. Need to improve stability - smaller timesteps, double precision numerics, better mesh quality, better initial condition.

 All times are GMT -4. The time now is 05:16.