11-18-2015 08:47 AM - edited 11-18-2015 10:52 AM
Hello Folks!
I got an structure were some parts may allow buckling till a given load ratio but others may not allow it or allow different loading ratios;
In order to correct verify/size/optimize such structures we need to address buckling analysis of them separating the modes regarding each structural part (by filtering the excitation point location: eigen-vector largest displacement) and taken the required actions (reinforcing or lightening) accordingly.
Practical issues:
Perform such task in simple structures is straight-forward, once a simple buckling analysis can answer all the questions even with different 'allowable' eigen-values over different regions.
For complex structures, however, the amount of required number of modes to find all relevant buckling modes for each structure (especially when several subcase are needed) is sometimes tremendously challenging.
In order to better solve this issue we need a way to filter the buckling modes by region prior to write them in results file (sol 105) or to consider them in optimization engine (imagine the optimization solver trying to address a ‘fake’ buckling problem occurring in one region - maybe where buckling is just allowed - by increasing shell thickness in a not related region, where the design variables are provided. It will result in unnecessary heavy or infeasible design).
Others solvers like OptiStruc has the case control parameter EXCLUDE (used to exclude the region we are not interested in), which do exactly what we need. Such control card appears does not exist in NX-Nastran
NX-Nastran do have the MODESELEC case control, but it does not filter results by region, but is just a complementary filter for EIGXX cards, operating on eigen-values ranges and values itself, not in the region where they occur.
The question is: Is there any other way to achieve this in NX-Nastran, by any other means? maybe using DMAP o some seldom used CARDS?
I really appreciate any help.
11-19-2015 10:14 AM
Hello,
Buckling analysis needs to solve eigenmode problem with stiffness and differential stiffness.
I understand you would like to study buckling of a part of the structure neglecting buckling of the rest.
In SAMCEF, you can decide to neglect differential stiffness on a group of elements. In that case, the group must be the "complement" of the group of interrest.