Radial Pump with Macros

[Lookid]

I have been using freeCAD to prepare impellers and volute for CFD investigation, so what you see is the fluid zone.
The impeller is made step by step using 4 Parts, the volute is made in one Part. Everything with Macros and the parameters can be changed like a normal parametric part.
Everything is fast except at the last step for creating the fluid zone, I need to make Boolean operations to substract the solid blades from the fluid zone, and it can take ~2/3 minutes, because the base of the blade is defined by several layers of 200 points BSplines.

• First, set up the main dimensions of the meridional section

1_maindim.PNG (117.09 KiB) Viewed 3799 times

• Adjust the meridional section

2_meridional_section_adjustment.PNG (190.32 KiB) Viewed 3799 times

• Set the blade angles

3_2DBlade.PNG (206.58 KiB) Viewed 3799 times

• Set Ellipse coefficient for blade leading edge, blade thickness, and some CFD parameters

4_3D_blades_solid.PNG (179.45 KiB) Viewed 3799 times

• Here a full image ready for CFD, with the volute

6_with_volute.PNG (133.23 KiB) Viewed 3799 times

[kisolre]

Nice
Could you share a sample file and some simulation images?

[Lookid]

I can share the file but it has no value since the MACROS are doing the job. But I can't share the MACROS, it's for work.

But I'm redoing the job in my free time with a better approach (for me), where I set up everything in Python and export points-ready to loft- to make the 2D blade. Making only the last step shown here relevant.
This I'll be able to share, but I struggle for some things so far

About simulations, I don't really have some now, but the goal is to automate python-FreeCAD-(SALOME for proper STL creation??)-OpenFOAM to quickly evaluate the design. I'll update here once I manage to do this if interested!

[johnwang]

About simulations, I don't really have some now, but the goal is to automate python-FreeCAD-(SALOME for proper STL creation??)-OpenFOAM to quickly evaluate the design. I'll update here once I manage to do this if interested!

If use openfoam, why not try to use the cfdof workbench inside FreeCAD?

[Lookid]

About simulations, I don't really have some now, but the goal is to automate python-FreeCAD-(SALOME for proper STL creation??)-OpenFOAM to quickly evaluate the design. I'll update here once I manage to do this if interested!

If use openfoam, why not try to use the cfdof workbench inside FreeCAD?

The cfdof workbench doesn't support a lot of things, like rotating regions, cyclic patches, you can't place probes or so as you want, it is just more convenient for me to use OF normally in this case.
This workbench is great for quick testing really, great job. But I doubt it would be used as an accurate CFD tool.

[kisolre]

But I doubt it would be used as an accurate CFD tool.

If no real usage is tried no real tools can be added, tweaked, developed, ... People like you should push the boundaries

[johnwang]

like rotating regions, cyclic patches, you can't place probes or so as you want, it is just more convenient for me to use OF normally in this case.

I was thinking you may need to simulate a part of the wheel (one passage) and the whole volute separately and how to connect them. It will be great to see one of this kind example.

Hope people are interested to expand the capability of the WB.

[johnwang]

If no real usage is tried no real tools can be added, tweaked, developed, ... People like you should push the boundaries

the free spirit

[Lookid]

like rotating regions, cyclic patches, you can't place probes or so as you want, it is just more convenient for me to use OF normally in this case.

I was thinking you may need to simulate a part of the wheel (one passage) and the whole volute separately and how to connect them. It will be great to see one of this kind example.

Hope people are interested to expand the capability of the WB.

It is not a FreeCAD problem at this point but np .
If you make steady-state simulations of a radial pump like this, you have the problem that the solution will depend on the impeller position. It is due to the volute tongue or nose, depending on the languages. It's just where there's the connection to the outlet pipe. This is called the ''Frozen-Rotor'' Approach.
There is a way to do exactly as you said, simulate 1 passage, average circumferentially and transpose the results to the volute, it is called the ''Mixing-Plane'' approach. A version of this is implement in foam-extend, but I never managed to have it work properly.

Otherwise, as you said, you could compute 1 passage, compute the average velocities in the (r, theta, z) reference frame, and use those as inlet boundary conditions for the volute. I am not sure how accurate this would be, and at this point it might be easier to simply compute the whole pump in an unsteady simulation.

With the Workbench, it would be nice to have more functionality yes, or a full GUI for OpenFOAM. But that would be a LOT of work I suppose and the problem is that it is easy to simply deal with OpenFOAM besides the workbench, especially now that you can have OpenFOAM on Windows easily.

What do you think?

[johnwang]

Unsteady simulation is straight foward, just need more cells.

I haven't tried cfdof, but found the elmer one a bit hard to use. Yes, it is easy now in windows for OF, maybe I should give it a go.