CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   Main CFD Forum (http://www.cfd-online.com/Forums/main/)
-   -   how to decide free boundary condition (http://www.cfd-online.com/Forums/main/2935-how-decide-free-boundary-condition.html)

 laofan December 22, 2000 03:56

how to decide free boundary condition

I need to calculate a axisymmetry jet sprays to air atomsphere,how to decide free boundary conditions to know its Velocity and Temperature distribution in the air??

 Ghanshyam Singh December 22, 2000 11:55

Re: how to decide free boundary condition

Look at my previous posting on "open B.C." U have to fix the pressure and on out flow, convective B.C. should be used also tengential velocity must be set to zero.

 Ghanshyam Singh December 22, 2000 11:55

Re: how to decide free boundary condition

Look at my previous posting on "open B.C." U have to fix the pressure and on out flow, convective B.C. should be used also tengential velocity must be set to zero.

GS

 Patrick Godon December 22, 2000 13:50

Re: how to decide free boundary condition

In order to avoid reflection of the jet or oscillations emanating from the open boundary, you might want to consider imposing the boundary conditions using the method of Characteristics.

See for example the review paper on non-reflective boundary conditions:

Givoli, 1991, J. Comput. PHys., vol. 94, page 1.

Happy Holiday! Patrick

 kalyan December 22, 2000 16:58

Re: how to decide free boundary condition

There really isn't a clean way of modeling free boundaries in CFD. There are so called non-reflecting boundary conditions which can help if the flow is hyperbolic in space (i.e. supersonic flow) and is going out through the free boundary. If the flow is going out at subsonic speed (or it incompressible, i.e. elliptic in space), there is always a certain approximation in the BCs. All the fixes proposed in this context merely minimize the impact of these (BC related) inaccuracies/approximations on the flow in the interior of the domain. So if you are interested in the flow in certain regions (e.g. near field jet dynamics), the free boundaries should be as far away from these regions as possible.

In jets, you have entrainment. The flow can enter the computational from the outside. This problems persists although to a lesser extent even after the free boundaries are very far away. In such cases, some empiricism is necessary. You can use the approximation entrainment velocities (derived from scaling laws for self-similar regions of the jets, both laminar and turbulent) at the free boundaries. If you jet is highly unsteady, use of these entrainment velocities could be problematic since they are obtained for steady state flows. If the unsteadyiness time scale is very large, then it is perhaps justifiable under a pseudo-steady state approximations.

The codes that are most stable for jet flows are those based on parabolized NS equations. The parabolization, however, is a debatable issue and hence so is the accuracy of these methods.

 John C. Chien December 22, 2000 23:32

Re: how to decide free boundary condition

(1). There is always a standard approach to this problem, or any other CFD problem. (2). This is based on the testing or experimental setup. Somehow, you will have to test it to make sure that your calculation is correct. (3). It is likely that the jet will be inside an enclosure or a large box. In this case, you just model the test geometry directly, with walls, jet inlet, downstream exit. The jet boundary will be formed naturally as part of the solution. (4). Since it is not a good idea to test the jet in an open space, like the empty room where small movement around the room is likely to disturb the test result, a better boundary condition is still a solid wall of the test enclosure or the test cell. In this way, the far field condition can be easily controlled for the test.

 Ghanshyam Singh December 23, 2000 02:49

Re: how to decide free boundary condition

This will not work, specially if you are interested in capturing the entrainment. I will do the following:

In the sides where from entrainment is taking place, set Dv/Dy=0, u=0 and P=0.0. At the exit, again fix the pressure i.e. P=0 and convective B.C. for all other variables i.e. D_Fai/D_t+U(D_fai/D_x) = 0.0 [Where Fai=u or v]. You have to keep these boundary sufficiently far away. But convective BC are pretty good. You need not to have too big computational domain.

GS ~

 David December 23, 2000 09:09

Re: how to decide free boundary condition

You can find the following papers

1. Grinstein F.F ~J.C.P,Vol.114,1994,pp 43-55 2. Thompson,K.W ~J.C.P,Vol.68 ,1987,pp 1-24 3. Thompson,K,W ~J.C.P,Vol.89 ,1990,pp 439-461 4. Poinsot,T.J & LeLe,S.K

~J.C.P,Vol.101,1992,pp 104-129

(Excellent)

 Jim Park December 23, 2000 10:52

Re: how to decide free boundary condition

You say that you have

"axisymmetry jet sprays to air atomsphere"

Is this a liquid into air, or air into air, or another gas into air?

If a liquid, is it coming through a spray head, a nozzle, or some other piece of hardware? Or is it simply introduced as an inflow condition along a piece of the boundary?

"free boundary conditions"

and it appears some of the folks are interpreting this to mean at the 'edge of the jet', others as the 'edge of the computational domain.'

What do you mean?