[modest_may]

There are many types of boundary conditions for outlet in CFX. However, it seems to be impossible to impose both velocity and static pressure simultaneously. Here is a problem, at the outlet B.C., the pressure and velocity are coupled:

Velocity u*Ay + miu*Velocity u. Gradient y - Ax*Ay=0 (Eq.1)
Velocity v*Ay + miu*Velocity v. Gradeint y - Ay*Ay
& + (Velcoity•A) - 0.5*Velcoity*Velcoity = Pressure+Q (Eq.2)
Velocity w*Ay + miu*Velocity w. Gradient y - Az*Ay=0 (Eq.3)
where A=(Ax,Ay,Az) is a predefined vector (it is defined using initialized profile data) and Q is also a given value.

Now, the outlet boundary condition needs to be specified based on Eqs.(1-3). However, the unknown variables, including the pressure and three components of the Velocity vector, are more than the available equations by one.

Before I present the question, it might be worth mentioning that during the solver run, the pressure, velocity and velocity gradient can be get using the "USER_GETVAR" routine. So the questions are:

1. since the available equations are more than the unknown variables, can I define the pressure arbitrarily and then calculate and specify the three components of the velocity at the outlet boundary condition (see the figure attached, where pout=constant)?

2. As pointed out by the CFX modeling guide, USER_GETVAR cannot be used to change the value of the solution field.
If the static pressure is specified at the outlet and the pressure value is calculated using Eq.(2), then after every iteration step, is there any ways to correct the solved Velocity u and Velocity w using Eq.(1) and Eq.(3) respectively? This involves to manually change the outlet velocity field.

PS: Eqs.(1-3) are derived formulas, they may have no physical meanings and can be regarded as constraints. So the simulation problem is not described here.

Looking forward to any advice and comment. Thanks in advance. :-).

[ghorrocks]

Why do want to do this? In every case I have seen the person who wants to set a simultaneous velocity and pressure boundary condition actually wants to do something else.

And the fact that simultaneous setting of velocity and pressure at a boundary is mathematically ill-posed in most situations is a give-away that this is something you probably don't want to do.

[modest_may]

Thank you for the reply.

In fact, this is the adjoint outlet boundary equation for an optimization problem. A similar description is presented in this thread, but is for OpenFoam.

adjoint method- uniformity

As mentioned before, this condition may have no physical meaning, I just wonder if CFX can realize those two techniques whatever the final results are. .

Apart from the boundary condition, is it possible to modify the velocity filed after every iteration? I just found the velocity is stored in the directory of '/FLOW/BOUNDCON/ZN1/BCP3/VARIABLES/VEL_FL1', while i have no idea to do that.

[ghorrocks]

You have access to change just about everything in CFX through user fortran, but low level stuff like this is very tricky to implement and not well documented. I have not this type of fortran stuff for many years.

[modest_may]

Many routines are not documented, this makes the User Fortran mysterious and elusive.

[modest_may]

Originally Posted by ghorrocks

You have access to change just about everything in CFX through user fortran, but low level stuff like this is very tricky to implement and not well documented. I have not this type of fortran stuff for many years.

Can you tell me what Routines are appropriate for this situation?