Shape Optimization Final Geometry Help

Shape Optimization Final Geometry Help

aruiz40DNGK7
Contributor Contributor
372 Views
1 Reply
Message 1 of 2

Shape Optimization Final Geometry Help

aruiz40DNGK7
Contributor
Contributor

Hello,

 

I've been trying to complete a final geometry for a shape optimization simulation for a GE Jet Bracket. What I would like to know is what's the best way to remove all the material that's unnecessary once the design is promoted to the design workspace. I've been trying to do sketches and then cutting, but that leaves me with a mess since it is a shape I'm not used to work on and the result is not smooth. I'm quite new with shape simulations and removing the material so hopefully some of you can have some ideas that make the job easier while sketching and cutting.

 

I provided the file. The simulations are also there, just for you to consider, study 4 is the one I selected and want for the final optimized geometry.

0 Likes
Accepted solutions (1)
373 Views
1 Reply
Reply (1)
Message 2 of 2

henderh
Community Manager
Community Manager
Accepted solution

Hi @aruiz40DNGK7,

 

Thank you for providing the study. Although I'm not able to help much on how to replicate the optimized mesh shape using the starting shape in the Design workspace, I can offer an alternative approach that doesn't involve the complicated modeling step. However, I'd like to comment on the simulation study setup first.

 

  1. If I understand correctly, the 2013 GrabCad challenge (https://grabcad.com/challenges/ge-jet-engine-bracket-challenge) requires that the single mass reduced shape will be able to withstand four different load condition scenarios. The Shape Optimization study type does support multiple load cases. Instead of defining a different study for each scenario, I propose using a single study that has the four different load conditions defined in four unique load cases. More on this later.
    GE bracket challenge setup.png
  2. The model orientation is different in your model by a +7 degrees rotation about the Z direction compared to the challenge:
    rotated +7 deg about Z.png
  3. The clevis holes aren't aligned to the global coordinate system, yet the moment axis will be aligned to global, rather than the hole's axes when defining the load magnitude and direction using the Vector direction type (other direction types will align to hole axes):
    Moment vectors not aligned with hole axes.png

  4. The use of a moment load for case 4. Under the hood moments are converted to nodal forces,  which will not only push on the cylindrical face, it can also pull on the face. In my opinion, pulling on the face wouldn't be as realistic of the physics as much as using a Bearing load might be, since the Bearing load type can only push, and will not pull on the cylindrical face. Also, the bearing load will apply a parabolic load profile instead of equally distributed as Force does. Using the equation moment = force x perpendicular distance, you can back out the load value and direction that needs to be applied to each hole.
  5. About the Shape Optimization study type supporting multiple load cases, which it does in a special way. What this means is it takes into account all defined load case scenarios during the solve, and 'combines them' to arrive at a single shape result. This is different than how multiple load cases are handled in a linear static stress study type, where each load case is considered independently of the others. Moreover, if we define a different study for each scenario, the shape that is strong enough to withstand load case 1 may not be able to withstand load case 4.

Another consideration is the result itself from Shape Optimization. Using the criteria settings defined as the problem's constraints, the solver determines the best shape that both reduces mass and maximizes the stiffness from the results of critical load paths. What might not be obvious is the shape will be the same if you have a 1 Newton force, or a 1000 Newton force applied since the load paths themselves are not changing. However, if there are multiple loads defined in a single load case, the values of the load magnitudes relative to each other does make an impact on the result.

 

The alternative is to use Fusion's Generative Design, which does take the load magnitudes into account, and the outcome shapes are automatically created as Breps with associative design history:

 

GD Outcome - GE bracket challenge.png


GD Explore GE bracket challenge.png

 

A quick start tutorial for setting up a Generative Design study like the GE bracket challenge is also available: https://help.autodesk.com/view/fusion360/ENU/?guid=GD-TUT-GE-BRACKET-QS

 

Hope this helps! Please let us know if you have any additional questions, comments or concerns.

 



Hugh Henderson
QA Engineer (Fusion Simulation)
0 Likes