I am doing a 3d simulation of a mixing problem where a sonic jet is injected into supersonic crossflow. I am using second order extrapolation boundary condition at the exit. But I am getting negative densities at the exit boundary condition. Any suggesations ? Amol

Is the wall normal velocity into the domain or out? Is the wall normal velocity subsonic or supersonic?

What is your intent for the boundary? Do you want a nonreflecting boundary, a force free boundary, or maybe one where some quantities are fixed?

I guess the earlier discription was very short. so this:

A a sonic (Mach 1) jet is injected into a constant area duct. The crossflow (flowing across the duct) is of Mach 2. I am having trouble at the exit boundary where flow leaves the computational domain. Right now I am doing a inviscid mode simulation and hence the velocity normal to exit boundary is supersonic. But in practise, the flow at the exit will be a supersonic boundary layer having both subsonic and supersonic layers. My intent for the boundary is to have a smaller computational domain. THanks for the reply Runge_Kutta (cool name).

For outflow boundaries one should specify this number of boundary conditions:

Euler Subsonic: 1

Euler Supersonic: 0

Navier-Stokes Subsonic: N_dim + 2 + (N_species - 1)

Navier Stokes Supersonic: N_dim + 1 + (N_species - 1)

where N_dim is the number of dimensions in your simulations and N_species is the number of species in for which you are solving a species equation. Take each boundary grid point on a point by point basis and classify it as inflow/outflow, subsonic/supersonic. Next, be careful about Euler boundaries in a Navier-Stokes calculation (things like inviscid slip walls). Right now, you are not permitted to impose any BCs for an Euler supersonic outflow. That will change when you add diffusive terms.

I am using a commercial code for the simulation and I have to use the boundary conditions avilable with software. The list is at : http://www.aerosft.com/Gasp/Manual/reference/x1885.html

The most relevant boundary condition I found was extrapolation from within the cell. Thanks a lot for the help.

First, you and GASP aren't talking to each other. It sounds like GASP isn't interested in Euler flows. Maybe you should presume that all is Navier-Stokes and proceed accordingly.

Navier Stokes - Supersonic outflow :

All wall-normal components of the viscous stress tensor = 0. (n dot tau = 0)

The wall normal heat flux is zero (n dot q = 0)

All wall normal species diffusion velocities are zero.

Navier Stokes - Subsonic outflow :

Both tangential wall-normal components of the viscous stress tensor = 0. (n dot tau dot t1 = n dot tau dot t2 = 0) where t1 and t2 are two tangential unit vectors to the surface.

The wall normal heat flux is zero (n dot q = 0)

All wall normal species diffusion velocities are zero.

The last BC is usually specified based on nonreflection criteria.

Thanks a lot.

Hi Everybody

I am looking for experimental data on bed to wall heat transfer in bubbling fluidized beds. Could anyone cite some references.

rahel