An API for L-PBF process parameter recommendation
Building a data-driven starting point for additive process development, against the "dark art" culture of trial-and-error.
I’ve always pushed back against the “dark art” of additive culture that emphasises intuition and gut feel. While important, I think that a solid data-driven approach, combined with process knowledge and experience, is the only way to truly industrialise areas like AM.
I’ve been building a small side project, an API for L-PBF process parameter recommendation.
The core tool is a DoE starting point: give it a target density, your material, and your printer model — get back a recommended process window to start your investigation from. This is useful when starting out designing experiments.
It also works the other way; feed in a parameter set, get a ballpark density estimate back. This is good for screening out obviously bad combinations before committing to a build.
My models are trained on a public dataset from Barrionuevo et al., extended with material properties pulled from NIST and manufacturer datasheets. That extra data is what makes the process window recommendation possible. The models account for why different materials behave differently, not just that they do.
If you know of any other public L-PBF datasets, I’d love to include them.