SubModel in OptiStruct - Global/Local Zooming

Really cool example, thanks for sharing!

Just a tip for anyone that will explore this feature (it's on your pdf file but I've had trouble in the past). If you have more than one Load Step you need to create one SPC load collector for each load step that will be analyzed with this approach.

Updating:

In HM/HW 2020.1 the SPC(M) option is already available and it is mcuh easier to create the submodel setup, with no need of editing the solver deck outside of HyperMesh. (see the updated attachment)

Thanks you for sharing this.

Example and illustrations are very helpful to understand the solution.

Thank you for this example.

I don't understand what I supposed to do at the beginning. The part of the model should be remeshed and rest of the model should be deleted? Or what exactly?

I don't understand what I supposed to do at the beginning. The part of the model should be remeshed and rest of the model should be deleted? Or what exactly?

1) run your global model, and generate your H3D file

2) create the mesh only of the part that you want to study in detail (delete all the rest)

3) create the appropriate SPCM and IMPORT cards

4) run your submodel

Hope this clarifies.

Wondering if its supported for Inertia relief instead of mapping displacement I can take driving as stresses Like other solver ?

i didn't get your use case. For submodel approach, you should have the displacements coming from the previous run in order to map it. You want to use Initial Stress and Inertia Relief for that? Maybe it works, but i don't know what is your use case.

My use case is where I have a large frame structure which I am simulating in the field scenario subjected various loads on the structure. For this I am doing inertia relief analysis. I am interested in doing the submodelling for local structure for quick iteration for mapping the stresses from global model.

Is the Option Initial stress works for Linear models in Optistruct. If yes I can connect with you by creating a support case on altair and explain my work flow

i believe initial stress works only with NL cases.

Submodel should work in your case, as it will map the global effect (even with IR) for the local model though its ingterface nodes.

But in IR the displacements are normalized and no significance unlike static . Is there any document where I can refer for IR submodel ?

i'm not sure if your statement is true. the relative displacement that generates stresses in the body are still there, right? IR compensates the rigid body motion with an equivalent acceleration/force to make sure the body doesn't fly away, but still the general behavior is there.

I have the POC model which I tried which I can share. But that is in Nastran Format . Not in Optistruct format .

This should work as well for INREL condition. Look here a test. Shell beam with central force under bending and Inertia Relief. Displacements are mapped from the global model to the local model at the boundaries, and stresses match well.

This second test applies only a force at the beam end, similar to a cantilever beam,

but with no SPC. Then from global, I took the displacements and mapped into the local submodel. No force applied in the submodel, only enforced displacements coming from the global model h3d file.

Running your NAstran model in OptiStruct should be pretty straightforward.

just zooming in to give an idea of mesh refinement used for the submodel vs global. It is not necessary to go so refined, obsiously, this model was just a sample file. But it is clear that the submodel will capture well the boundary displacements and stresses from the global model.

Why are we calling SPC in Submodel ? and it says its constrained here in all six dof ? Please let me know

SPC with M flag is the key for mapping displacements from the imported global H3D file into the submodel.

Submodeling here is based on mapping the global displacements into enforced displacements into the submodel, thus enforcing the same loading deformations.

Hello Adriano

Thanks for answering my Questions. Submodels are definitely are reducing the time and for quick design changes its helpful and also for locally refines the mesh and mapping the global boundaries to predict accuracy state

I have one last question . I took the Beam problem , solved as is the Global model. Instead of taking your submodel I took the part of the global model and solved for submodel as shown here. I also attached files for reference. Why the stress difference is observed when the model has same mesh in both models and even though interpolation displacement didn't change . I understand stress is more sensitive to mesh than displacement

Why the stress difference is observed in Global model though Submodel is dervied from global model in this case ?

i've opened your results and just adjusted the legend to be the exact same on both global and submodel. I've set maximum and mininum at the legends.

Submodel usually presents a higher local maximum value at the boundaries, related to mapping. So usually I would recommend ignoring stresses at the ends (SPC'd).

I'd say that althought they don't match identically (maybe even because we're not using rotations here) it is pretty decent.

thank you Adriano. Yup with same legend sync the stress pattern matches. I was surprised when the numbers are differing ~ 3% still they are good away from boundaries. thanks a lot for sharing insight wrt to optistruct on this query

@Adriano_Koga Please let me know ASSIGN ,OP2,1000,'Globl.op2' works too instead of H3DRES ???

is there any syntax for OP2.

In OptiStruct, the ASSIGN,OP2,... statement does it support importing results for submodeling ?