search

Dhondtguido / CalculiX

This repository contains the source files of CalculiX, a three-dimensional Finite Element Program (www.calculix.de).
GNU General Public License v2.0
75 stars 19 forks source link

CYCLIC SYMMETRY MODEL parameter NGRAPH behavior #32

Closed OliverKunc closed 1 year ago

OliverKunc commented 1 year ago

The parameter NGRAPH of the *CYCLIC SYMMETRY MODEL card seems to lose effect when there is first a *STEP with a *FREQUENCY analysis without *SELECT CYCLIC SYMMETRY MODES.

Even if there is a subsequent *STEP with *FREQUENCY and with *SELECT CYCLIC SYMMETRY MODES, still only one segment of the model is output to the .frd file.

Only if there are exclusively *STEPs with *SELECT CYCLIC SYMMETRY MODES the parameter NGRAPH takes effect.

Steps to reproduce: see the example fullseg.inp. Change NGRAPH to a value larger than one and remove the first occurrence of *SELECT CYCLIC SYMMETRY MODES. Then the .frd file has only one segment.

Expected behavior, two possibilities:

  1. NGRAPHshould take effect when there is no *SELECT CYCLIC SYMMETRY MODES, or
  2. NGRAPHshould take effect when there is a *SELECT CYCLIC SYMMETRY MODES keyword even if it was preceded by a step without this keyword.
victorkemp commented 1 year ago

Does it make sense to have a FREQUENCY step without SELECT CYCLIC SYMMETRY MODES when you've defined *CYCLIC SYMMETRY MODEL? What does the solution from that mean?

OliverKunc commented 1 year ago

@victorkemp The solution is equivalent to that of *SELECT CYCLIC SYMMETRY MODES,NMIN=0,NMAX=0, i.e. nodal diameter zero. Just to be 100% sure, I have checked this with the fullseg.inp example for the first couple eigenmodes.

It is not exactly the same because *SELECT CYCLIC SYMMETRY MODES,NMIN=0,NMAX=0 results in a (arbitrary?) split of the displacements into a real and an imaginary part with constant phase difference. The lack of *SELECT CYCLIC SYMMETRY MODES,NMIN=0,NMAX=0 gives just one single, real displacement field, which is what makes most sense.

However, there is also a slight difference in the frequency. That I don't fully understand, possibly some round-off issue.

FEA-eng commented 1 year ago

Another issue regarding this parameter: #33

Dhondtguido commented 1 year ago

Hi Oliver, in a calculation with CYCLIC SYMMETRY MODEL the SELECT CYCLIC SYMMETRY MODES should be used. You are right that you can do without for nodal diameter zero and that NGRAPH does not work in that case, however, taking care of this special issue is not on the top of my list. Thanks for pointing this out, Guido