Adobe/Purdue tool gives structural strength to 3D printed objects

Depending on the material they’re made from, objects created using 3D printing can have a common flaw: They are fragile and often fall apart or lose their shape over time. Although CAD software, combined with high precision printers, will allow designers to create and fabricate previously un-manufacturable parts, that isn’t always a good thing. The printed fabrications often fail at points of high stress.

“You can go online, create something using a 3D printer and pay $300, only to find that it isn’t strong enough to survive shipping and arrives in more than one piece,” said Radomir Mech, senior research manager from Adobe’s Advanced Technology Labs.

To address this problem, researchers at Purdue University and Adobe’s Advanced Technology Labs have jointly developed a program that automatically imparts strength to objects before they are printed.

“It runs a structural analysis, finds the problematic part and then automatically picks one of the three possible solutions,” said Bedrich Benes, an associate professor of computer graphics at Purdue University.

Coded by former Purdue doctoral student, Ondrej Stava, the software application automatically strengthens objects either by increasing the thickness of key structural elements; adding struts; or reducing stress on structural elements by hollowing out overweight elements.

“We not only make the objects structurally better, but we also make them much more inexpensive,” Mech said. “We have demonstrated a weight and cost savings of 80 percent.”

The tool also automatically identifies “grip positions” where a person is likely to grasp the object. A lightweight structural analysis solver analyzes the object using a mesh-based simulation, but requires less computing power than traditional FEA tools.

The research team presented their findings at SIGGRAPH but say that more testing is necessary before the software is ready for wider release. Future research may focus on better understanding how structural strength is influenced by the layered nature of 3-D-printed objects. The researchers may also expand their algorithms to include printed models that have moving parts.
www.purdue.edu