Hey there,
It’s Robin from CFD Engine & I’ve been puzzling my way through the OpenFOAM v10 release notes this week.
I don’t use the Foundation version, so this is like having a peek over the garden fence to check that the neighbours haven’t built a pool while I wasn’t looking.
Good news (for me) – they haven’t 👏
This release is dominated by one big development and a host of smaller ones. I say “small”, but what I really mean is that they don’t touch my workflow, your mileage may vary.
v10 may well contain a killer feature for you, especially if you’re into trickier physics. There’s a long list of developments around multiphase & reacting flows, combustion & chemistry, plus dynamic meshing. But very little in the way of mesh generation & post-processing.
If you’re currently using AMI or ACMI, you’ll definitely want to pay attention 👇
The big one
The headline development appears to be a new method for handing non-conformal interfaces –non-conformal coupling– which replaces the Arbitrary Mesh Interfaces (AMIs) we’ve had for a decade.
In case you’re not familiar, non-conformal interfaces are a way for disconnected, adjacent meshes to talk to each other. They’re commonly used to allow regions to slide past each other without having to be remeshed (a spinning propeller or a rotor-stator array for example) stationary models are fine though too.
The non-conformal interface is the way we deal with flow passing from one mesh region to another when the faces at their boundary no longer match up. The previous method was called an Arbitrary Mesh Interface (AMI plus its cyclic sibling, ACMI).
This new version is the same basic idea, but better – i.e. conservative, second-order accurate in space, with simpler boundary conditions & easier parallel handling.
It looks like a good step forward but it has killed ACMIs (they’re no longer in v10) and signed the death warrant for AMIs (they’ll be gone by v11) 😬
The relevant tutorials have been updated to help you migrate.
I’m not sure how I’d feel about this if I had years of AMI-based results in my archive. On the one hand, this should improve future simulations, but it would pretty-much invalidate my archive (or I could just stick with v9 forever) 🤔
Bits & Bobs
I’m struggling to find anything as significant as that change, but here are a few other things from the full release notes that that caught my eye (not sure if they’re big news or not):
Multiphase Bits
- A new coalescence & breakup model for bubbly flows has been added to
multiPhaseEulerFoam. - Actuation discs can now be used in multiphase models.
Chemistry Bits
- There’s a new load balancer that will rebalance the mesh based on the chemistry being solved (as opposed to just rebalancing based on the cell count).
- Chemistry models have switched to mass fractions from concentrations to improve calculations where expansion is significant.
- There were a number of changes to ISAT models which might be of interest.
Miscellaneous
- A new redistribution library (Zoltan from Sandia Labs) has been added. Looks like it has some unique features but…you have to download & build it yourself & it only works in parallel (i.e. you can’t use it for decomposition, only redistribution).
bouyantSimpleFoam&bouyantPimpleFoamhave merged intobouyantFoam.- We can now add (coupled & uncoupled) particle clouds to lots more solvers
fluent3DMeshtoFoamhas been updated to Fluent R2021
What did I miss?
Like I said, even if I used the Foundation version, most of this would pass me by. I don’t have a true appreciation of the significance of the developments, so what have I missed?
Is there something in v10 that will be a big deal in your CFD? Drop me a note, I’m interested to hear your perspective.
As the very least, you should check out the release notes (whether you use the Foundation version or not) there’s a lot to digest.
Finally, kudos & thanks to the OpenFOAM Foundation team (& their financial supporters) for giving us another excellent option in the free & open-source CFD space 🙏 a huge effort with only 8 contributors 👏
Until next week, stay safe,
