Different displacement results between optimization and analysis

Diogo Ferreira

Hi there,

 

Im looking for some help to understand the differences between the results from the size optimization, in the displacement and failure index, between the result from the analysis and the result from the optimization itself (the ones that you see in Hyperview when you choose to observe the last iteration). Does it make sense to have reasonable values in the last iteration results for the load case, and when i import the' _shufflingX.fem' and run the analysis again, the results are different and do not agree with my boundary conditions?

 

Thanks

 

Diogo

Altair Forum User

Hi @Diogo Ferreira

 

Sorry for a late reply.

 

The reason is OptiStruct uses 90/60% Material density (depending on response used).

 

In order to match the results for zeroth iteration with analysis results, use 100% material using OptiControls>> MATINIT>>1

Nachiket Kadu_22143

Hi Prakash

I was going through similar problem here, i am performing maximization of frequency by increasing volume fraction (taking shield example in tutorial as basis). The optimization results show a lower value of frequency so i kept MATINIT=1. The initial frequency (at 0 iteration) came closed to actual natural frequency but as the analysis proceeds the frequency instead of maximizing tends to lower down at first few frequency and again increases to the almost the initial value of frequency. For your reference i am attaching the .out files kindly compare and suggest your views.

upper bound value of vol_frac=0.4

Original frequency=23Hz

Final Frequency obtained using matinit as 0.6=  23Hz (Mass=154 Kg)

Final Frequency obtained using matinit as 1=  22.9Hz [Initially (at 0 iteration) 167 kg and get reduced to 154 Kg]

 

 

Nachiket Kadu_22143

Hello

Can anyone clarify doubt regarding the above comment?

Altair Forum User

Hi @Nachiket Kadu

 

Sorry for a late reply, 

 

I was travelling and could not able to reply. I will get back to you soon.

Altair Forum User

HI,

 

Both optimizations converged to similar feasible designs - same volume fractions and very close frequencies. With different matinit, the optimizations start with different design volfrac, but not necessarily to converge to different designs.  

Nachiket Kadu_22143

Hi Prakash

Thanks for replying, you can also see that though it is a maximization problem the mass seems to be decreasing upto a great extent. I have one more doubt, in maximization problems, element with element density=1 suggests addition of stiffness to that region?

 

Diogo Ferreira

Hi @Prakash Pagadala 

Thanks a lot by your answer! My point of mine first doubt is should i trust more the hyperview results of size optimization, or more the analysis of the _shuffling.fem of SZTOSH card? Because they give diferente values of composite failure

Altair Forum User

Hi 

You can absolutely trust it.

 

Redesign and validate