Screw Symmetry

bobuhito · January 16, 2022, 15:07

Problems with screw symmetry (simplest example is just a tube with interior flow snaked around an infinitely long cylinder) could most easily be solved with a cyclic patch type with translation and rotation both applied, but it seems that cyclic only allows translation or rotation, not both.

Is there any reason that cyclic is so limited? or is there a different way to work these problems without changing the C code?

JulioPieri · January 17, 2022, 15:44

can't you use 180deg symmetry steps so you can use translational-only cyclic?

bobuhito · January 17, 2022, 18:53

Well, for the simplest example of a snaking tube, I could actually simulate two snaking tubes to get the symmetry you are talking about (and you could even fit more snaking tubes and reduce it to, for example, 30 degrees, depending on the radius of the tube relative to the radius of the central cylinder). But, for the general problem, equivalent cross sections with 180-degree or any rotational symmetry might be impossible.

Anyway, I'm mainly wondering why the cyclic type does not allow simultaneous rotation and translation. It seems trivial to stitch the boundaries/matrices like this, but then maybe numerical PISO algorithm has problems converging? I doubt there would be any problem on first thought.

JulioPieri · January 18, 2022, 09:45

It seems to be not possible in OF. Have a look at this post, where they also suggest a workaround, using noOrdering.

Cyclic boundary with a rotational and translational transform

January 16, 2022, 15:07	Screw Symmetry	#1
bobuhito New Member BH Join Date: Dec 2021 Posts: 9 Rep Power: 4	Problems with screw symmetry (simplest example is just a tube with interior flow snaked around an infinitely long cylinder) could most easily be solved with a cyclic patch type with translation and rotation both applied, but it seems that cyclic only allows translation or rotation, not both. Is there any reason that cyclic is so limited? or is there a different way to work these problems without changing the C code?

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Simulation of a chimney with heat in open atmosphere	Jurado	OpenFOAM Running, Solving & CFD	9	December 18, 2020 08:35
rhoSimpleFoam - Newbie Issues	AndyR	OpenFOAM Running, Solving & CFD	6	March 10, 2020 10:28
High nut values in random place and time	krzychu111	OpenFOAM Running, Solving & CFD	0	January 9, 2019 09:42
Possible Bug in pimpleFoam (or createPatch) (or fluent3DMeshToFoam)	cfdonline2mohsen	OpenFOAM	3	October 21, 2013 10:28
[ICEM] Blocking and Symmetry	BrolY	ANSYS Meshing & Geometry	32	August 24, 2012 04:13

January 17, 2022, 15:44		#2
JulioPieri Senior Member Julio Pieri Join Date: Sep 2017 Posts: 107 Rep Power: 9	can't you use 180deg symmetry steps so you can use translational-only cyclic?

January 17, 2022, 18:53		#3
bobuhito New Member BH Join Date: Dec 2021 Posts: 9 Rep Power: 4	Well, for the simplest example of a snaking tube, I could actually simulate two snaking tubes to get the symmetry you are talking about (and you could even fit more snaking tubes and reduce it to, for example, 30 degrees, depending on the radius of the tube relative to the radius of the central cylinder). But, for the general problem, equivalent cross sections with 180-degree or any rotational symmetry might be impossible. Anyway, I'm mainly wondering why the cyclic type does not allow simultaneous rotation and translation. It seems trivial to stitch the boundaries/matrices like this, but then maybe numerical PISO algorithm has problems converging? I doubt there would be any problem on first thought.

January 18, 2022, 09:45		#4
JulioPieri Senior Member Julio Pieri Join Date: Sep 2017 Posts: 107 Rep Power: 9	It seems to be not possible in OF. Have a look at this post, where they also suggest a workaround, using noOrdering. Cyclic boundary with a rotational and translational transform