WComp for RADIOSS Optimization

Crashphys

Hi,

 

I'm wondering if it's possible to run an optimization to minimize compliance rather than mass in RADIOSS. Should the objective be to minimize strain energy instead in this case?

 

Also, there seems to be a very limited selection of responses. Is there anything for center of gravity like in OptiStruct?

 

Thanks in advance.

Simon Križnik

Hi,

 

to minimize compliance use minimize internal strain energy response (RTYPE=8).

 

Unfortunately, not all responses are available for Radioss optimization. When the desired response is not directly available from Radioss, it may be calculated using /DRESP2. Or the model could be converted to Optistruct and run explicit dynamic analysis (Radioss simulation in the background).

Unable to find an attachment - read this blog

Crashphys

Hi,

 

I found that the best solution might be to extract interface forces and import them to OptiStruct to do either explicit dynamic or a linear static. I have an output request for the contact forces, however I was wondering how I can get the raw data rather than a graph?

 

Edit: I was also wondering if you could specify: is your recommendation to get the interface loads and run OptiStruct EXPDYN? I thought EXPDYN runs already with the ESLMO method?

 

Thanks

Simon Križnik

Good thinking regarding extracting the contact force- check the procedure in this thread:

https://community.altair.com/community?id=community_question&sys_id=5266003a1b2bd0908017dc61ec4bcbd2

Once the contact forces are extracted from the explicit run they can be used to perform linear static analysis/optimization in Optistruct. Sorry for the confusion: my recommendation to run ESLM using explicit in Optistruct is because I feel running explicit ESLM in Optistruct offers more optimization control, but not all Radioss capabilities are available and vice-versa.

Anetlam

You might check on eBay for these parts too, as there is some people parting out XJ600 Seca II engines and theyre auctioning off the box of engine bolts or something like that.

Crashphys

Hi,

 

I ran the solver with the rigid applied, however I found that the time step is rapidly degenerating. My time step is 3.7366041328229E-05 for the SPH according to the output file, and the model begins with a time step of approx 1e-4, which is definitely not consistent. The time step keeps reducing lower and lower, and by the 5th cycle is triple the initial starting time. I realise I drastically increased the number of elements, however this degeneration seems to happen regardless of element size. Anyone have any suggestions?

 

Thanks in advance.

Unable to find an attachment - read this blog

Simon Križnik

Hi,

 

The timestep reported in the starter out file is reduced by a scale factor of 0.9 by default for computation.

The timestep reported and used in the coarse model is consistent. The contact was edited in the refined model.

 

The most common reason for timestep dropping during computation is:

• element compression: check element formulation and material properties
• deep contact penetrations: increase minimum gap for contact activation

 

Unable to find an attachment - read this blog