[shivasluzz]

How to do multiphase analysis of gear pump?

Immersed solid method doesnot support multiphase simulation...

By searching in this forum, i found that by 'Junction box routine' it is possible do the that..

Can anybody please elaborate the method or steps involved?

thanks

[ghorrocks]

Gear pumps are really easy with immersed solids, and really hard any other way. So this is going to be challenging....

One way is to use moving mesh with dynamic remeshing. There are some tutorial examples of this around, I am not sure where.

A second approach uses the small blessing with sort of analysis is that as the motion is periodic (each tooth is the same). So you might be able to generate meshes at various points of engagement of a single tooth (say 10 meshes) and use moving mesh to get the motion between the keyframe meshes. This way the simulation becomes a series of simulations over the tooth motion and does not require dynamic remeshing.

[shivasluzz]

Thanks for the reply ghorrocks :-)..
But please tell me how to do that?
Here are my observations..
1. In an Ansys tutorial, it is given that 'a series of mesh files can be read through junction box routine but it need special purpose meshing software since the node numbers should not change in all the mesh files'..
my questions regarding this are
can it be done in icem cfd?
how to read a series a files using junction box routine

2. I came pressure relief valve ansys tutorial and in that mesh at different time intervals are selected using multiconfiguration option.. if it can be done then why we need to do junction box routine?

[ghorrocks]

Sorry I cannot give you detailed instructions of how to do this, I am a busy man.

You do not need to do this with a junction box routine. The standard initial conditions interpolation should be OK for it, then you script up a whole bunch of short CFX runs where each run is a new mesh, with an initial condition interpolated from the previous run. You might also be able to do this as a multiconfiguration run.

This method has nothing to do with the software which generated the mesh. You can use ICEM, ANSYS Meshing or anything else.

The dynamic meshing approach will require linking to the meshing software which probably means ICEM.

[shivasluzz]

Ok.. thanks ghorrocks...

[shivasluzz]

i explored both the methods (dynamic remeshing and reading different mesh files) and here is my observation and doubts:
1. Dynamic remeshing:
a. This option seems to be simple if there is no contact between the teeth
b. how much important is to model the contact? is it safe to ignore?
c. all dynamic remeshing thread in this forum and in other websites gives refers to 'www.edr.no/blog'. But sadly this link is no longer available :-( can somebody please give an alternate link
d. while recording icem-cfd replay file, upto which operation we have to record? record till generation of mesh? or record till saving of mesh file?
e. i believe some more lines should be added with icem-cfd replay file (other than what is recorded automatically) for interacting with cfx solver manager. can somebody show an sample icem cfd replay file for dynamic remeshing?

2. Reading Mulitple mesh files:
a. as told by ghorrocks, we can easily model using mesh for only one pitch and we can reuse those file for full rotation
b. junction box routine is not required. we can use multipleconfiguration and connect all using configuration control.
c. After the one pitch is completed, i think we have to link the last analysis with the first analysis mesh file using configuration control.
d. since solution from one configuration is transfered to another configuration by node number, i hope we need to renumber nodes at the end of one pitch, to make full rotation of gears instead of oscillation for one pitch
e. Renumbering nodes option is available in Execution control of configuration.

3. And the 3rd option is using Workbench parameteric option. Its very simple and everything is almost automated. We need to only parameterize the gear rotation angle and vary the angle through the workbench parametric analysis. Even the mesh generation is automatic. But it takes too much time to complete one time step and to start another.

Expecting expert advice. Thanks :-)

[ghorrocks]

Dynamic remeshing can handle tooth contact. But it does make it a bit harder. Whether tooth contact can be ignored is up to the details of your simulation - only you can assess that. Note that squeeze film effects mean that it is actually quite hard to really get contact, often it just squeezes down to a very thin film.

If I was doing this I would use the pre-generated meshes and a big batch file to do a long series of runs. interpolating the results onto the new mesh as initial conditions. But you can do it however you like.

[shivasluzz]

thanks ghorrocks...
squeeze film effects are really thoughtful.. will study about that..
and i dont understand your interpolation method
can you please give me a sample icem replay file for dynamic remeshing?
i mean what extra command should be added to the recorded replay file to use it in cfxpre?

[ghorrocks]

The interpolation method I describe is just the initial condition interpolation stuff in CFX.

I don't have ICEM so cannot help you there.

[shivasluzz]

ok..anyway thanks for the help ghorrocks.. i will find out and post here..

[shivasluzz]

Check this link
https://www.sharcnet.ca/Software/Ans...remeshing.html

it explains the process of creating Master replay file in icem-cfd for the purpose dynamic remeshing in cfx..
but at the end it tells that through icem-cfd replay method, only translational mesh motion is possible and not rotational

[shivasluzz]