Generative Mini Coffee Table and Basket Hanger

This week in Computational Fabrication, we were tasked to create an object using Autodesk's generative design or topological optimization algorithms.

I wanted to try out both so I made two different designs. The first one was an idea I found online which is a mini coffee table for a singular mug. The second one was an experiment with topological optimization for a hanger to hold small stuffed animals on a hook in my apartment - inspired by this peppa pig that currently gazes over the living room in my apartment.

Generative design is the process where the user defines a set of structural constraints, forces, manufacturing specifications, target attributes and other parameters to create a series of outcomes for the user can choose desired results.

Topological optimization is given a fully defined geometry and some parameters, the program will try to create a new mesh that either maximizes stiffness or minimizes material.

• 3D printer
• Autodesk fusion 360

For the first version, I first made a basic cylinder with a hole in it wide enough to fit a regular coffee mug. I then made a pentagon and replicated it using the circular pattern tool to generate all 5 of the table legs.

Then I selected the geometries to preserve and put a force downwards of 1.25 lb force. 1.25lb was selected because according to this site, the average weight for ceramic mugs are .55-.88lb and one cup of water is .5lb so a good approximation of the sum would be 1.25lb force.

I then ran the generative design study from Fusion 360 and got some simple but smooth results.

An issue I encountered when printing was that all my models imported to be 1/10 of the size I made them in Fusion 360 and were rotated to the side. I scaled them and flipped them back in Cura.

Outcomes printed in this study:

• outcome 1

I realized from these results that it probably wasn't necessary to print an entire solid cylinder so I converted it into a ring and beveled the sides. To ensure that something would get generated to hold the cup in the center I placed a small sphere with half the force constraint. So the ring and the point had .625lb force each.

But this created a result that obviously would not be able to hold a cup in a stable position so I did not try printing and made some modifications.

Outcomes printed in this study:

• None

I then replicated the small sphere 20 times to different randomly generated points within the ring and then ran the study again. I kept the center one so in total there are 21 spheres. I kept .625 lb force for the ring but split the force evenly between the sphere .625/21 = 0.02976lb force and ran the generator.

This resulted in some alien looking designs with points sticking out. I printed one of the outcomes for this and the print had a lot of areas where the plastic did not stick/failed but resulting structure can still hold a mug fairly well. Here I also experimented with different materials which caused some interesting results.

Outcomes printed in this study:

• outcome 6

To avoid the weird dot-like alien structure I added a obstacle cylinder geometry below the spheres so that the generative algorithm would not make any structures in that area.

This created a more neuron looking design that is able to sturdily hold up the mug and also print fairly well. I tested with some different mugs and ones that have a smaller radius weren't as balanced(probably because the spheres were not distributed evenly).

Outcome 5 of this design actually failed but I thought that model was also interesting so I printed it as well.

Outcomes printed in this study:

• outcome 3
• outcome 5
• outcome 10

For the basket hanger, since I wanted to try using the topological optimization program, I designed a simple basket with three poles connecting to a cylinder for the hang. I used the Loft tool to connect the different shaped sketches together.

At first, I tried static stress optimization on this model but got errors that the model was "overengineered." This was because I had accidently left other bodies in the study. I switched over to shape optimization as suggested from the Fusion 360 dialogue box.

Another problem I encountered was under defining preserve geometries to perform the optimization on. As you can see from the third picture, the "optimized model" would actually be floating in space.

I changed the geometries, but then had errors when trying to specify the target mass to be 30%. Because of the way the model was made, Fusion 360 kept notifying me that the minimum mass possible was 52%. I changed the parameter to accommodate but was a bit disappointed in the resulting model. It did not look that much different from the original. I believe I might over constrained it.

I decided to go back and do generative design on the basket hanger to see difference in geometry.

I changed the base to be a ring instead of a square and changed some other constraints thus comparison might be inaccurate. However, from what I see, the generative design clearly produced a much smoother outline than the shape optimization did.

When printing, I chose to minimize the support material. The blue model clearly had a much harder time trying to print everything together but eventually was able to connect together.