[deepbandivadekar]

Originally Posted by Tobi

Can you give us some description / draft about what your case should demonstrate? I am a bit doubtful about the physics behind. No wall, nothing?

No, there's no wall involved, as the physical wall is a porous surface (so that's basically an internal surface). So basically imagine a cube with top boundary face as inlet for external flow and bottom boundary face as inlet for flow through porous medium. A blockMeshDict file entry defines the porous zone at the bottom. I am trying to see how much difference it makes to the heat flux or the internal surface temperature, if the internal wall were impervious, as against porous in current case.

Does it make some sense? Let me know if you want something specific.

Meanwhile, I tried changing the mesh size from mm to m to cm, it doesn't affect the outcome. Any pressure other than close to 1e05 definitely fails. I wondered about the definition of pressure when it is totalPressure since in Openfoam pressure is actually pressure/rho, but the tutorial cases involving totalPressure surely use values in format 1e5 but not larger. I just changed the underrelaxation for rho to 0.5 and less. Doesn't change outcome.

[Tobi]

Hi,

now you made a key point. Pressure does not matter here (incompressible). The only point that matters is the dp. So you can set it to 1e5 or -10 or 1e54. However, I would expect that you have compression effects here and you should use a compressible solver. What is your max. velocity you will get? Is it higher than Ma > 0.3?

A pictures would be good. As far as I get it.

                                 
channel 1       porous      outlet                  channel 2
----------------                                            ---------------
                  x x x
-> -> ->          x x x                                      <- <- <-
                  x x x
----------------                                            ---------------
                            outlet

While you neglect the channel. Maybe it is better to have it. I do not know how the numerics relate to inlet face and direct porous media. Never worked with porous things.

[deepbandivadekar]

Originally Posted by Tobi

Hi,

now you made a key point. Pressure does not matter here (incompressible). The only point that matters is the dp. So you can set it to 1e5 or -10 or 1e54. However, I would expect that you have compression effects here and you should use a compressible solver. What is your max. velocity you will get? Is it higher than Ma > 0.3?

A pictures would be good. As far as I get it.

                                 
channel 1       porous      outlet                  channel 2
----------------                                            ---------------
                  x x x
-> -> ->          x x x                                      <- <- <-
                  x x x
----------------                                            ---------------
                            outlet

While you neglect the channel. Maybe it is better to have it. I do not know how the numerics relate to inlet face and direct porous media. Never worked with porous things.

Well this is precisely a compressible flow, as the flow is indeed supersonic. And the pressures are high. I thought rhoPorousSimpleFoam was anyway a compressible solver, isn't it so? That's the reason I used it.

Here's the image. All sides are outlets. Top and bottom inlets.

[Tobi]

To be honest, I am sorry. I did not check you solver. I was reading about your comment with the incompressible stuff and thus, I was thinking you are using an incompressible solver.

I wondered about the definition of pressure when it is totalPressure since in Openfoam pressure is actually pressure/rho, but the tutorial cases involving totalPressure surely use values in format 1e5 but not larger. I just changed the underrelaxation for rho to 0.5 and less. Doesn't change outcome.

After re-reading (more carefully) you made a wrong statement. In OpenFOAM, the pressure is not p/rho. This is only the case for incompressible solvers. That was the reason why I thought you are using an incompressible one.

The picture is clear as it is your numerical domain and we can check it out by executing blockMesh. I was interested more about the whole system not the system you are drawing / simulating -> the big picture.

In addition, I was re-reading the whole thread and figured out that there are no information about your set-up or what you want to do. The big picture is missing. We just know you apply a big pressure and everything blows up. So, again, I please ask you to give more information. I refer to this thread from Niels: How to give enough info to get help

Additional information:

• you should think about the pressure at your outlet. I cannot imagine that it is atmospheric pressure, or is it?
• Depending on your system, maybe a transient solver is better at the beginning
• I get a negative pressure value which is not possible
• Furthermore, you should check which density ranges the air could have - is it correct that rho is between 0.4 and 2 ?
• Save each time step and analyze what is going on. I run now 30 iterations but the velocity field looks terrible as well as the density field
• I would suggest you to extend you domain and remove the cells to end up with 20.000 (for the beginning)
• Reduce the relax to 0.1

Good luck.

[deepbandivadekar]

Originally Posted by Tobi

To be honest, I am sorry. I did not check you solver. I was reading about your comment with the incompressible stuff and thus, I was thinking you are using an incompressible solver.
After re-reading (more carefully) you made a wrong statement. In OpenFOAM, the pressure is not p/rho. This is only the case for incompressible solvers. That was the reason why I thought you are using an incompressible one.

Okay! That makes sense, as the units definition at the top of p file confused me, which is different for incompressible solvers. Alright.

Originally Posted by Tobi

The picture is clear as it is your numerical domain and we can check it out by executing blockMesh. I was interested more about the whole system not the system you are drawing / simulating -> the big picture.

In addition, I was re-reading the whole thread and figured out that there are no information about your set-up or what you want to do. The big picture is missing. We just know you apply a big pressure and everything blows up. So, again, I please ask you to give more information. I refer to this thread from Niels: How to give enough info to get help

Additional information:

• you should think about the pressure at your outlet. I cannot imagine that it is atmospheric pressure, or is it?
• Depending on your system, maybe a transient solver is better at the beginning
• I get a negative pressure value which is not possible
• Furthermore, you should check which density ranges the air could have - is it correct that rho is between 0.4 and 2 ?
• Save each time step and analyze what is going on. I run now 30 iterations but the velocity field looks terrible as well as the density field
• I would suggest you to extend you domain and remove the cells to end up with 20.000 (for the beginning)
• Reduce the relax to 0.1

Good luck.

Alright. I'll try to do these changes and see what happens. One quick last question:

Originally Posted by Tobi

• I get a negative pressure value which is not possible
• I run now 30 iterations but the velocity field looks terrible as well as the density field

How do I monitor this (or any other properties, say at a particular location)? By field do you mean an average of all cells ?