CFD Online Discussion Forums

CFD Online Discussion Forums (
-   Main CFD Forum (
-   -   Calculation with COMSOL Multiphysics (

sega January 16, 2008 10:26

Calculation with COMSOL Multiphysics
Hello World.

I'm trying to compute the following problem with COMSOL.

The geometry is just a square. On the boundarys (right, left and lower) the velocity in both x- and y-direction is given with 0.

On the upper boundary the x-velocity (u) is given with u=100., y-velocity (v) v=0!

This is some kind of popular benchmark-problem for FEM?

As a matter of fact, that all velocities normal to the boundary are given with Dirichlet boundary conditions, there has to be a point with given pressure!

But I dont know how to specify both velocity AND pressure boundary-conditions with COMSOL.

The drop-down-menue just gives me the option to choose either velocity OR pressure boundary.

Does anyone know a solution?

Thanks in advance.

otd January 16, 2008 14:12

Re: Calculation with COMSOL Multiphysics
First I must confess that I'm not familiar with COMSOL. Are you simulating an incompressible fluid? If so, your question is really a fluids question, not a solver question.

For incompressible fluids, the pressure is known only to an additive constant. That is, only the pressure gradient appears in the equations.

Thus if some function p1(x,y) is known for 0 < x < x_o, 0 < y < y_o,

then the function p2(x,y) = p1(x,y) + Constant is also a solution.

The solver should be able to find some solution p(x,y) without specifying a value for pressure. Since the solver doesn't need a bc on pressure for with velocities specified on the boundaries, it doesn't give you that option.

sega January 17, 2008 09:23

Re: Calculation with COMSOL Multiphysics

I had to think about your response. You are right. But I'm not sure if that is the correct answer to that special problem.

Please correct me if I'm wrong. As I have despribed all velocities normal to ALL boundaries are equal to zero - given by Dirichlet-bc's. Therefore there are no Neumann-bc's for this problem!

With no Neumann-bc's there is no absolute pressure, because all boundray-integral over the stress-tensor normal to the boundary are equal to zero.

So I can't compute absolute pressure, without specifying a pressure at some point?

I have recognized that COMSOL does not find a solution for this problem, although it looks rather simple.

I have found a solution with a rather unknown (?) code called SEPRAN. In SEPRAN I was able to specify a pressure for some point, and got the right solution.

I will try to simulate the problem with FLUENT later.

Discussion is opened! :)

Mikhail January 18, 2008 09:49

Re: Calculation with COMSOL Multiphysics

A student of mine had the same problem. I cannot engage the result but it seems that you need to set the pressure in any ONE point of the domain (boundary). It might work.

sega January 20, 2008 15:06

Re: Calculation with COMSOL Multiphysics
Exactly. But COMSOL doesn't give me this option. (As discribed in my first post)

All times are GMT -4. The time now is 22:12.