[viking]

Hi there,

i am using cfx 14.5 and an icem mesh.
the simulation case is as follows:

filled cylinder with helium, valve closed. Valve blocks flow from helium filled cylinder into 'vacuum' pipe (means real low density and pabs like 10 Pa).

then the valve opens (conditional connection interface plus using moving mesh for the large, 16 mm, opening distance).

the problem is now: in initial conditions there is no flow in the domains, e.g. subsonic conditions. but if the valve opens, we will get high flow rates like mach 3 (in helium mach# is about 1000 m/s). and this seems to be transsonic flow.

So what can we do to have this transition from 0mach to 3mach flow?
is it wrong to use the transsonic model from the beginning?

kind regrads
sebastian

[ghorrocks]

Yes, you should use a setup suitable for transsonic flow from the start.

When the valve cracks open you are going to need tiny time steps. I mean really small. And all the other numerical stability things will also help - double precision, good mesh quality and a good initial condition.

[viking]

dear ghorrocks,

thank you for your reply. for the timestep 10^-8 s seems to work. in my preliminary simulations using isothermal conditions, a great wavefront travels through my pipe.

now i have an additional question regarding the initial conditions:
what is about the temperature in my "vacuum" domain?
as we know, a real vacuum w/o any fluid or atoms has no temperature.
but for temperature change due to pressure driven density change, i want to take the temperature into account.

as i know from mr. gay-lussac:
lets say 2 boxes are thermally insulated to their ambience and are connected with a pipe (also a valve in the pipe). one box is filled with a fluid (p1,T1) and the other box is evacuated (p = 0, no Temp.).
if we open the valve, the fluid will distribute homogenous in both boxes and we will get a common pressure p2 at the end.
the temperature in both boxes will have the initial temperature from the fluid T1.

So, is it wrong to set up the same initial temperatur in all domains?

[ghorrocks]

You have to have a velocity, temperature and pressure at all locations in the simulation at all times. CFX is a Navier Stokes equation solver and the Navier Stokes equations do not apply in high vacuums.

There is no such thing as a perfect vacuum anyway. There is going to be some molecules whizzing around. Even in the vacuum of space there is a few molecules up there. As long as the Knudsen number is <<1 then you are fine to use CFX.

http://www.cfd-online.com/Wiki/Knudsen_number
http://en.wikipedia.org/wiki/Knudsen_number