generative design results in static stress simulation

[Anonymous]

Hi,

I have tried to use generative design to design a bracket. I set the safety factor as 2 in the generative design setups and Fusion returned a reasonable result. Then I promoted that to the design space and checked again with static stress analysis. However, it shows that the minimum safety factor dropped to 1.2. I did not change the design at all during this process. Thus, I am wondering what could possibly happen that caused the drop of safety factor? Is it because generative design and static stress analysis use different algorithm so that they calculate the safety factor differently? 

Thank you!

[I_Forge_KC]

Can you share some photos showing where the lowest FoS is located?

The interface between the TSplines geometry and the BRep geometry can occasionally contain a weird riser that pops up. You may also have various extraneous "blobs" around that create risers out of nothing. Just a thought.

[Anonymous]

Hi,

Here are some photos. The first one is the CAD body promoted from generative design. The second and third are the static stress simulation result. The min FoS is at the inner edge of the hollow cylinder where the inner surface is set as a fixed constraint. 

[Ben-Weiss]

Hi @Anonymous!

It looks to me like you're likely seeing a difference in the way the simulation is being handled between generative design and the static stress simulation tool. A few additional details about this behavior were covered in this post.

Hope that helps!

Ben

[Anonymous]

Hi @Ben-Weiss ,

Thank you that helps me a lot! I also wonder which safety factor of the bracket is likely to be more accurate, the setup in generative design (I set it as 2) or the static stress simulation result (1.25)? 

[Ben-Weiss]

Hi @Anonymous ,

The factor of safety calculated as part of the static stress simulation on the output design is likely to be closer to reality.

Ben

[Anonymous]

Hi @Ben-Weiss ,

That is very interesting. Does that mean we cannot fully trust the generative design results yet? For example, do we always need to run an additional static stress simulation to verify the generative design outcome, and make adjustments to the outcome according to the static stress simulation result?

Thank you.

[Ben-Weiss · Accepted solution]

Hi @Anonymous,

I would always recommend running an independent static stress simulation on any design you intend to bring into the real world where factor of safety is important for several reasons:

• The design changes slightly during the conversion to b-rep. This can cause the mechanical performance to change.
• A different simulation strategy is employed inside of generative than in the static stress. Normally, the two line up pretty well, but occasionally you see differences (such as the one you point out in your original post).
• Any additional edits you make to your design after exporting need to be validated to ensure they do not weaken the part (even adding material can reduce the factor of safety in some situations). The intent with generative design is to provide you with a design you can further edit and refine as your requirements evolve or to incorporate aspects of the problem statement that cannot be included in the generative setup (aesthetics, fatigue, etc.)

It's not that the generative design result is particularly untrustworthy so much as all mechanical simulation has a degree of uncertainty. It's important to check and validate, preferably with a different simulation strategy, before relying on the results in any real-world application. If a part is going to be used in any context in which the cost of failure is at all significant (either in dollars or in risk of injury), I would encourage you to work with someone trained in finite element analysis and engage in mechanical testing of the manufactured part to validate intended performance.

Ben

[Anonymous]

Thank you very much for your explanation! It is very helpful!