Principal investigator: Kevin Turner
University: University of Pennsylvania
Industry partner: TE Connectivity Corp.
Recently, a new set of lower cost metal additive manufacturing (AM) systems that consist of separate low-temperature printing and high-temperature sintering process have been developed and commercialized. Separating the high-temperature processes needed to bond and consolidate the metal powder from the printing step allows the overall cost of metal AM to be reduced. In the second step of the process in which the binder is removed and the powder consolidates, there is a significant volume change of the part (~20%). Thus, the geometry of the part must be designed and initially printed with dimensions that are substantially larger than the final part. TE Connectivity, the industry partner on this PA Manufacturing Initiative proposal, and others who have sought to adopt these printing techniques for production have noted that current CAD/software tools available for designing parts to account for this shrinkage fail to give adequate results when tight tolerances (e.g. mm-scale and less) are required on the final part and that costly iteration is often required to produce parts within spec. The objective of this PA Manufacturing grant is to develop data-driven models that are informed by both experimental measurements and process physics (e.g. mechanics, heat transfer, sintering mechanisms) for designing additively manufactured metal parts that undergo significant shrinkage in the fabrication process. This objective will be realized through research by a PA manufacturing fellow, a PhD student at the University of Pennsylvania, who will be jointly mentored on the is project by the PI and a scientist at TE Connectivity. The fellow will develop data-driven models that incorporate input from physics-based simulations of the sintering process and experiments. The fellow will develop the computational models and design the experiments, while the industry partner will obtain the experimental data (print and characterize) needed to train and validate the models.