CAD and Machine Learning Possibilities
As we consider the impact of material usage on climate change, socially responsible engineers need to create products that use less material without sacrificing performance. Parallel to this, the rising capability of machine learning has allowed it to be a powerful tool to produce effienct structural designs across a variety of objectives. Here is an exploration of a CAD data set by the GE Jet Engine Bracket design competition. It reflects diverse, realistic and complex designs produced by contestants. The performace of these design across a variety of structural and material efficiency metrics has been quantified by Whalen, Beyene and Mueller in order to further the use of machine learning in application-focused contexts.
Start Exploring
Carousel View provides a less technical visualization of the geometric data. Hovering over a braket image gives you info on the creator of the design. Clicking on a braket, takes you to the submission page for the design where you can download the geometry directly.
Data View allows the user to dig into strucutural simulation results and see how specific brackets perform on interactive plots. Hovering over a data point gives you a preview of the design it corresponds to while clicking the point loads it in the 3D viewer that users can manipulate with their mouse. Users can also click on the legend to filter the bracket designs by category and use the drop down features on both the x and y axis to construct custom exploratory plots.