Is there a way to run a 103 solution using PARTs super elements without super element reduction?

Question

I have a model that is made up of many smaller components. Some use repeated ids for grids and elements. I can combine the components easy enough using BEGIN SUPER, SEBULK and SECONCT for each. Each PART is a single INCLUDE statement with an associated BDF file. This builds the model and it runs. However the frequencies and shapes are vastly different from a renumbered model. Looking through the F06 and F04 it appears that super element reduction is done. I want displacement and other outputs for each node in a given model as well. Is there a way to tell Nastran to forgo reduction? Of course I could be wrong that reduction is occurring...

I use one subcase with SUPER = ALL and a single set of boundary conditions. I only care about mode shapes and frequencies at this stage. Using PARTs positioning and inclusion is much easier for preprocessing than renumbering and relocating each little part. Also the part models are changing often so renumbering/repositioning is a big time sink.

Don.Graff · Accepted Answer

Hi- what you are asking can be accomplished using Modules instead of Part SE... see a webinar on Modules here:

simcompanion.hexagon.com/.../simacademy-webinar--overview-of-the-modules-wm446

They are also documented int he Superelement User Guide at the end.

But, just to be clear, Parts SE can result int he SAME answers as a baseline NON-SE analysis. The most likely reason for getting incorrect results is not doing Component Modal Synthesis in the upstream SE. This is triggered by have a METHOD call out an EIGRL within the PART, as well as providing placeholders to store the component modes. This can be done by defining SPOINTs and assigning them as QSET DOF. It is advisable to over-define the number of placeholders to ensure all component modes, as well as any residual vectors are stored and passed to the residual. Make sure the upper frequency of the upstream SE is at least 3-5 times that desired for accuracy in the assembled downstream SE0 portion.

If the model is large, Part SE will be a more efficient means of solving the overall problem.