Using the Newton solver for k-epsilon turbulence models

[charlieowen3065]

Check these boxes if you have followed the posting rules.

• Q&A Navier Stokes is the most appropriate category for my question
• I have consulted the posting Guidelines on the Discussions front page
• I have searched the Discussions forum and my question has not been asked before
• I have searched the MOOSE website and the documentation does not answer my question
• I have formatted my post following the posting guidelines (avoid screenshots if possible, triple back quotes before/after pasted text, etc)

Question

Hello!

I am currently attempting to design a RANS model using the k-epsilon formulation you all have implemented recently, and had a question on the matter.

For my specific purposes, it would be useful to use the Newton-solver to simulate my experiment (I am attempting to do some ROM analysis, and algorithms I have written in the past are based on the Newton update). For non-RANS models (just using simple mass, momentum, and heat transfers), the Newton algorithm seemed to be working, but once I added in TKE and TKED variables and kernels, it gave me the error

For iteration type 'nonlinear': solution state 1 was requested in nl0 but only up to state 0 is available.

And it looks like the error is being given by the INSFVTKESourceSink object. Are Newton capabilities not implemented for the k-epsilon model yet? Or is there a work-around that I have not found?

I've seen the SIMPLE algorithm, and I could probably work with it, but it would be nice to work with what I'm familiar with.

Any ideas on where to go with this?

Thanks!

Charlie

Additional information

No response

[GiudGiud]

And it looks like the error is being given by the INSFVTKESourceSink object. Are Newton capabilities not implemented for the k-epsilon model yet?

They are enabled, though we do not leverage them anywhere so you would be a first-adopter

Or is there a work-around that I have not found?

Did you set the parameters in the problem to request the previous solution? Did you follow the tests for TKE + Newton ?

[charlieowen3065]

I did not manually set the parameters in the problem to request the previous solution - I was not aware of that functionality, but I will look into that!

For the tests of TKE + Newton, they all use the Physics block. To build my model, I wanted to start at the basics (I started with just steady-state mass-momentum) and have been building my model up from there, and recently reached the point where RANS was being necessary, and in that, I have not been using the physics block. Does this block work similar to the Modules block, where it is a short-hand for a larger set of input blocks (such that you could completely rebuild it with just blocks in the input file), or is it made deeper in the code?

I was basing my Turbulence models on the channel_ERCOFTAC.i channel-test, and was trying to rework that to use Newton updates.

I am mostly trying to learn how to properly use the Navier-Stokes module right now, so I was avoiding using the larger Module/Physics blocks to have a better handle of what is happening under the hood, and to try and do minor parametric studies to see how much various inputs affect response.

[GiudGiud]

For the tests of TKE + Newton, they all use the Physics block.

once #29177 is merged you ll be able to dump the Physics syntax to regular kernel syntax. It might already work in your case. Try setting (just once) Problem/type=DumpObjectsProblem

Regardless, you can see the parameters of the Problem and executioner block and try to use those

Does this block work similar to the Modules block, where it is a short-hand for a larger set of input blocks (such that you could completely rebuild it with just blocks in the input file), or is it made deeper in the code?

No it's exactly the same as the Modules block, except that there's tons of additional helper and checks that we did not have back when using the Modules block

[charlieowen3065]

Okay! the dump wasn't quite working, but it was a mpicxx error, which I haven't set up yet, so that may be on my end. I'll look into that.

I see the source file, so I'll just parse through that in the meantime - since it does work though, I'm all good!

Thank you!

Charlie