Comparison of static condensation and CMS


Hi all!Smiley Happy

 I am currently working on a comparison between the results of a static condensation and CMS for my university but I have not been already suceeding or agreeing with the results. In my opinion, I might have forgotten some basic things in the analysis and I would be really really glad if someone of you could help me.

Now to the problem:

Lets break the model down to the basic: It's a hollow cylinder as a shell model. I implemented RBE2 spiders on both ends, which shall represent the bearings and their midnodes shall represent the masternodes. Now I am planning to do a static and a CB reduction to get an superelement out of it, which I can integrate in Simpack later on.

Static reduction:

First, I tried to do it via the SOL 101 and was confirmed by the line "Static condensation" in the .f06, but later I realized, that this was more a "static static" condensation and not a "static dynamic" condensation. So I read myself into the nastran user guide and got to the conclusion, that I am using sol 103 ( is Sol 103 real eigenvalues the right choice?) for both static and CB reduction. Is SOL 103 Real Eigenvalues the right choice? Because SOL 103 is asking for Generlized DOF's which are not necessary in static condensation.

The main difference lies in the SPOINT and QSET definition, e.g. static condensation does not use them. Am I right until now?

Therefore, I defined the midpoints of both RBE2 elements as my ASET (Constraint type ->fixed boundary dof, all fixed)

Next step was to configure the case control -> Eigenvalue Method (Lanczos) -> Lanczos Data -> Here I did not change anything to the default settings (e.g. Number of Desired Modes 10, Number vectors "", Mass)

Is this the right way to approach the static condensation? Does the model itself has to be constrained somewhere aside of the ASET?

When I'm writing the input file, deleting the QSET and SPOINT lines, and solve the input file after that, the .f06 files states a "static condensation". Unfortunately, I am not able to open the results in NX Smiley Sad


Same procedure as above, just another solver -> SOL 103 Superelement.

Difference lies in the declaration of the generlized DOFs. The midpoints of the RBE2 elements are defined as BNDFIX as well, which should (acc. to quick reference guide), be identical to BSET1, which is required. (constraint type -> fixed boundary degrees of freedom)

Same Lanczos method. But the results are somehow confusing. Different modes and different frequencies.


Can somebody please help me here?



Thanks in advance!Smiley Happy


Kind regards




EDIT: I found another workaround: Via Menu->Insert->DOFSets. DOFSets can be grouped there and afterwards used for ASET, BSET etc. in the solutions dialog. BUT: As I want to automate the workflow, node labels are changing and therefore the DOFset is empty after a mesh update Smiley Sad


Re: Comparison of static condensation and CMS



I've proceeded a little bit now.

Now I am using Sol 101-Superelement for Static condensation. There I am definining the midnodes of the RBE elements as BNDFIX via Fixed boundary DOFs. Additionally I could use a DOFSET and define it as ASET for additionnal accuracy. Am I right?

The problem still remains in reading the results. I guess NX does not know what kind of results I am expecting.

CMS: I use Sol 103-Superelement where I define the midnoes of the RBE elements as BNDFIX (fixed boundary DOFs) as well and choose the number of desired eigenmodes (connected to spoint and qset).


I would like to do a kind of comparison (like it is done in the superlement user guide chapter 9-7) of the eigenfrequencies of baseline, static condensation and CMS model.

The baseline values can be derived by using Sol103-Real Eigenvalues, I think. How can I extract the first eigenvalues of the static condensation?