The problem consists of two cylinders, one at high pressure and the other at lower pressure connected by a tube (L= 2m). The flow will be initially started by a valve upstream.

At t = 0 sec the valve will be opened and flow will start. The flow will compressible under such high pressure difference and will be choked.Shock wave is also expected which will move down stream.

The pressure of upstream vessel will decrease and that of receiver vessel will increase, until the flow will no longer be choked and ultimately the flow will cease to occur.

i want to model it and calculate the transient time of the process.

i am familier with fluent 6.0

how can i specfy (the varying presures) of both upstream and down stream reserviors?

and which condition to apply for the initial inteface between high and low pressure.?

why dont u try not using any pressure based boundaries at all. Just model it as two reserviors interconnected by the pipe, each having a different initital condition in terms of the pressure.

cheers nandu

Hi Jehanzeb!

You use fluid boundary condition for the inside of the two vessles and the pipe. Then you just patch a initial pressure in on of the tanks up to the point where the valve is located. In the other vessle you patch initial pressure here.

/Mattias

Thanx guys it was a great help.

i liked the approach right away, but i foresee convergence problems.

should i first run it for steady state solution? does steady state soln make any sense in this case

i was wondering would k-epslon or k-omega turbulence models be good? (very highly turbulent flow)

shouldnt i solve for invisid flow first.

jehanzeb

It depents on what information you are seeking. I got the impression that you wanted a transient simulation to get an idea on the coarse of events in the two vessles. If you run a steady state simulation the only thing you will see is the final result, when equlibrium has set. k-e turbulence model is recomended.

/Mattias

i want to see the transient effects in the tube (L=2 m) and the time to reach equilibrium (which theoretically is 6 sec.)

another problem is that there are lot of computaional time involved

my grid size is = 0.005m step size = 0.0001 time steps = 60000 iterations / time step = 1000

is there some way i can make it more efficient

and also y velocity does not converge even after 1000 iterations for that i reduced the y- vel residues from 0.001 to 0.05

k-epslon RNG

residue y vel= 0.05

pressure diff = p2/p1 = 10