15. Wildcard week

Assignment:
Design and produce something with a digital process (incorporating computer-aided design and manufacturing) not covered in another assignment, documenting the requirements that your assignment meets, and including everything necessary to reproduce it.

Planning

It was difficult for me to decide what to do this week, and that worked against me because, without realizing it, I spent a lot of time exploring different options and couldn't make up my mind. At first, I had planned to work on composites. I read about the different materials available in the lab, how to work with them, but ultimately, I didn't like the idea. Then, I started exploring artificial intelligence. I wanted to use AI to design and manufacture some mechanical component for my final project. I read the article "How Can Large Language Models Help Humans in Design And Manufacturing?" which talks about using ChatGPT for 3D design. I also dabbled a bit in SolidWorks. During a research on applying AI in Fusion 360, I came across several videos on Topology Optimization and Generative Design. It seemed like an interesting area, and I was intrigued because it was a completely new concept for me. So, I finally decided to experiment with this option by designing something simple: a guitar hanger that I could 3D print and attach to the furniture I designed during the Computer Controlled Machining week.

Design - Topology Optimization

While Generative Design and Topology Optimization aren't strictly AI in the traditional sense, this functionality relies on advanced algorithms that iteratively explore and evaluate different design configurations to find optimal solutions. In my case, I will first explore what Topology Optimization entails. My design should support a weight of approximately 4 kg, using PLA as the printing material. Taking these parameters into account, along with the surface that will bear this weight, the software modifies the design by generating a topology that will support the weight while optimizing material usage.

For the design, I followed the steps explained in this video tutorial. My original design was as follows:



This was the result after applying the topology optimization:



This was the result after applying the topology optimization with different parameters, such as the surface that will support the weight and the preserve region:

Design - Generative Design

Next, I decided to experiment with the Generative Design function and evaluate the generated designs. To do this, I followed the steps shown in this video tutorial.

After following the steps mentioned in this video tutorial, several designs were generated according to the parameters configured in the process. I started with the following design so that Fusion could generate the functional designs.



After analyzing the various generated designs, I decided to use the following design for 3D printing.

Printing and Testing

Topology optimization performs calculations based on an existing design to determine the parts of the object that are necessary to support stress according to the weight and manufacturing material configurations. In this way, this function can economize material usage by reducing parts that are functionally unnecessary. However, the generative design function uses more complex iterative algorithms to create a new construction and design that didn't exist before, performing real-time calculations as it builds according to the configured parameters. The configuration basically involves determining the function the design will fulfill (in my case, to support a weight of approximately 40 N), delimiting the volume where the software will build, the direction and orientation of the construction (x, -x, y, -y, z, -z), assigning the location and direction of the weight, the construction material, among others. In this way, unlike topology optimization, generative design generates multiple novel design options. As I mentioned before, I decided to print one of the results obtained with generative design because I liked them more.

Downloads