Generative Design Fail

[tommywengXCEDU]

Hey y'all,

I've been having issues creating a frame for an aircraft project with the generative design option. Specifically, every time I try to solve the design, the progress bar would stagnate at 9% and then report "Fusion encountered a problem and failed to complete the outcome generation." The Pre-check tool shows no issues either and I've already turned the generation resolution to the coarsest option. 

I've put the CAD link below in case anyone wants to take a stab at it. The floating elements are "preserve geometry" components, while the outer shell is the "obstacle geometry". 

https://a360.co/4mL1U5F

[CGBenner]

@tommywengXCEDU 

Hello, and welcome!  I've moved your post to this Design forum where more of the experts might be able to take a look at it.  Good luck.

[henderh]

Hi @tommywengXCEDU,

I noticed two issues that if remedied, could enable generating the desired outcomes.

1. The setup isn't statically stable with only a single vertex with a Fixed constraint. To the solver, it has unbalanced forces, and the bodies will spin about the vertex fixed in 3D space. Since in real-life there is no ground from which to abstract a fixed constraint reference, it requires a solver option to take this into account.
   • 
2. The thin shell fuselage body is tricky obstacle geometry for the solver, and the likely reason which is causing it to be ignored. BTW, very thin preserve geometry creates an impossible design space to solve, if thickness is less than the size of a Voxel mesh element (and therefore cannot be created small enough to represent the geometry in the solver).
   • 

What I'd try for 1) is to:

• Go to the Study Settings > Enable 'Remove rigid body modes' option > Delete / Suppress the Fixed constraint. This tells the solver to calculate, then include artificial body loads. These loads are equal in magnitude, and opposite in direction with respect to the sum of the applied loads, to create a balanced / statically stable load case that can be solved.
  <nerd alert!> Under the hood, the balancing forces are included by adding those global linear and angular body load accelerations to the boundary conditions matrix, before solving the system of simultaneous equations. At least that's how it works in the Finite Element Method / Autodesk NASTRAN solver used in the Fusion Simulation workspace. I presume the general algorithm is the same (or similar) employed in our Generative solver.
  •  
  • If this doesn't work, fix an entire face to eliminate the open Degrees of Freedom in the study setup to allow you to tweak other knobs as necessary, once the iterations can proceed beyond the initial few.

For 2) it will make it much easier on the solver if the fuselage thickness could be increased dramatically. I tried using the Thicken command, but could only increase it by 1mm. Perhaps you can try using it as a tool body, to cut out the negative shape from a dummy solid box body. The geometry should no longer be too considered too thin with the 'mold-like construction body' to act as better suited chunky body obstacle.

Hope this helps, and best of luck! Please let us know if you have any additional questions, comments, or suggestions.

Best regards,

[tommywengXCEDU]

Gotcha, thank you!