Shape optimisation vs. shape generation

[david.montgomery]

I have a simple problem where I am loading a thick circular plate on the top surface and supporting it on the bottom. In Fusion, shape optimisation for minimum mass gives a type of hollowed out vaulted structure as might be expected but the shape generation always returns a solid and not 'hollowed out' part.

Is this normal for the shape generator function or am I doing something wrong?

For the shape generation problem I have defined no-build volumes, preserved volumes and original shape volumes. The original shape is solid and not hollow.

[krunoslav_knezic]

As far as I understand, the optimization tool will always optimized given geometry regarding defined condition.

This means that you as designer have made some design, and now you have give it to software to optimize.

In generative design, you give software to design best possible solution in defined condition, meaning that software will find best possible design with given constrains.

That bean said it's normal to have two different outcome, because first is optimization of given shape, and second results from generative design is optimized shape.

Best way could be to use both, starting with generative design, and then run FEM analysis or even optimization to create perfect shape.

Look at this video, it might give you more insight.

[I_Forge_KC]

It's hard to make an apples to apples comparison without seeing the setups in more detail.

From a quick glance, I see:

- The Generative Design process has a directional constraint built in which causes a different shape entirely. Essentially, there is little or no undercuts unless they can be supported by your overhang angle.

- It looks like the preserve geometry is basically two disks separated by some distance. These preserve geos will not change or reduce in volume during the generation phase.

- On the Fusion Topology Optimization you have your LPC slider set to hide the absolutely critical geometry. What you're showing isn't a realistic solution to the problem (as defined).

Can you share more about the setup in Generative Design?

[david.montgomery]

Thank you for the reply and link.

I have run this analysis many times in generative design and none of the outcomes have created internal voids.

The shape optimisation produces a shape with voids (which  is the better generic solution).

I was just wondering if this will always be the case and if I had done something wrong/missed  

[david.montgomery]

Thanks for the post.

I was trying to set up the same problem with the two methodologies.

I understand about the slider, I just wanted a screen cap that shows the voids.

In the generative design I have reserved the top and bottom disk volumes to a depth of 2mm. The 8mm gap in between is where I want the generative design to suggest an optimal structural shape. I have defined a reserved annular cylindrical area outside the disk diameter for 'obstacle' geometry  and a disk that contains the reserved areas and the volume in between them as original part. 

The disk should be supported on the lower surface and a load applied across the top surface.

I am not sure about your comment on undercuts. There are two buttons for 'unrestricted' and 'additive manufacturing'. The latter gives options for defining undercut angles.

I am thinking that Gererative design is producing external topology variant options and will not provide solutions of hollowed out  sections. Or am I missing something?