Researchers engineer method to ensure accuracy of additively manufactured parts

4/6/2022

Profs. Chenhui Shao, Bill King, and Sam Tawfick have developed a way to ensure a quality system that monitors and controls part geometric accuracy in industrial-scale additive manufacturing.

Written by

Chenhui Shao
MechSE Associate Professor Chenhui Shao

MechSE professors Sameh Tawfick, Bill King, and Chenhui Shao, along with two recent doctoral students, recently developed a new methodology that ensures the accuracy of additively manufactured parts. Their work was published in the journal Additive Manufacturing.

Additive manufacturing (AM) is increasingly used in industrial-scale production applications. Scalable AM production is often carried out using hierarchical schemes where parts are produced by multiple machines that operate in the same or different factories. Existing quality systems can neither account for the interactions among these hierarchical factors nor predict part geometric accuracy at the feature level. The researchers’ new modeling methodology is able to characterize the hierarchical effects of AM production parameters while taking process physics into account. It also brings a new capability of predicting feature-level geometric accuracy of AM parts.

William King
Professor Bill King

The team validated the methodology using 70 polymer hexagonal lattice parts produced in the real-world production environment of Fast Radius Inc., a startup that was borne out of King’s research – and they have proven their methodology to be applicable in a variety of AM production environments.

Imagine a manufacturer has been using several 3D printers for some time and has collected some historical production data. Now a new printer is being deployed, and its quality needs to be monitored and controlled, but very little information is available about this new printer. Conventional quality systems require a large amount of data, which translates into significant time and costs, to build a reliable quality model.  The new methodology can establish a quality model for this new printer quickly and cost-effectively by transferring knowledge learned from historical production—significantly reducing the production launch time and ensuring the consistency, quality, and accuracy of AM-produced parts.

Sameh Tawfick
Professor Sameh Tawfick

The study, “Hierarchical data models improve the accuracy of feature level predictions for additively manufactured parts,” is a follow-up to published research published in the same journal, from the same team, in which they developed software to improve the accuracy of 3D-printed parts, seeking to reduce cost and waste for companies using AM to mass produce parts in factories.

First author Yuhang Yang is now a research scientist at Meta, and second author Davis McGregor is a senior manufacturing scientist at Fast Radius.

The team acknowledges the assistance of FastRadius as well as the Dynamic Research Enterprise for Multidisciplinary Engineering Sciences (DREMES) at Zhejiang University and the University of Illinois Urbana-Champaign. 


Share this story

This story was published April 6, 2022.