As 3D printing evolves, researchers have gone beyond mere fabrication processes to developing techniques for optimizing how particular materials can be printed. To that end, researchers at Carnegie Mellon University’s College of Engineering have developed a new approach to optimizing the 3D printing of soft materials. This approach combines expert judgment with an algorithm designed to search parameter combinations relevant for 3D printing, they said.

Images of 3D prints made using a new method developed by researchers at Carnegie Mellon. Their approach combines expertise with an algorithm and applies that to a 3D-printing process to optimize the printing of soft materials. (Image source: Sara Abdollahi, Alexander Davis, John H. Miller, Adam W. Feinberg, Carnegie Mellon)

Calling their method Expert-Guided Optimization (EGO), the technique—designed by a cross-disciplinary team of biomedical, materials, and social scientists—enables optimal printing for high-quality soft materials with a completely new approach, said Sara Abdollahi, a Ph.D. student in biomedical engineering at Carnegie Mellon.

“We developed the EGO strategy after realizing the lack of systematization in 3D printing, especially involving new materials and processes on which little prior information is known,” she told Design News. “We were seeking an approach that would be easy to implement and flexible enough to modify as needed.”

Algorithm Used for Predictions

Studies have shown that experts are good at selecting factors that matter in prediction, but not so good at combining those factors to make an actual prediction, Abdollahi explained. The team used this idea, but applied it in an engineering context to develop a product. “Specifically, the expert was used to inform and start off the search, which was followed through with an algorithmic search to combine the expert selected parameters to make a prediction for the optimum [result],” she explained.

In essence, what the team achieved for the first time was to use a concept from the social and decision sciences to inform an engineering problem, Abdollahi said. “From a practical standpoint, we were able to systematically optimize 3D printing of soft materials without the need for complex physical models, a data training set, or haphazard trials,” she said.

The team published a paper on its work in the journal PLOS One. It demonstrated the EGO method by printing objects made of liquid polydimethylsiloxane (PDMS) elastomer resin, which is often used in wearable sensors and medical devices, with a freeform reversible embedding (FRE) printing method.

Researchers Printed “Calibration Objects”

Abdollahi said the concept can be applied for other 3D-printing processes, such as binder jetting, vat polymerization, and others, as long as the parameter space for each of those processes can be defined.

“Broadly speaking, the method explores random combinations at first, giving an opportunity to gauge different sets of factor-levels selected by the expert,” she explained. “Once a promising combination is found (i.e., a parameter set that produces the fittest print), the approach is to work around this combination iteratively toward improvements.”

Specifically, researchers printed what are called “calibration objects” in 3D printing—that is, a cube and a cylinder, Abdollahi told us. “This is fitting in the context of soft-material 3D printing, using the FRE technique, than is a more recent 3D printing approach on which an extensive library of geometries and sizes is lacking,” she explained. “To get an idea of the sort of prints that can be developed with this tool, a good starting point would be to look at what is possible to create with the technology at hand.”

Standardizing the Approach

Researchers plan to continue their work to explore using other materials, more complex geometries, other search algorithms, and other 3D printing processes or even a different process altogether, Abdollahi said. “A farther vision would be in determining the steps to standardize this approach and see it being applied in industry and real-world context,” she added. “This model has the potential to save time and effort, but also money if we think about the costs that go into the iterative design of products in manufacturing and prevent unnecessary waste products, which can be significant, for example, in 3D printing of metal parts.”

Elizabeth Montalbano is a freelance writer who has written about technology and culture for 20 years. She has lived and worked as a professional journalist in Phoenix, San Francisco, and New York City. In her free time, she enjoys surfing, traveling, music, yoga, and cooking. She currently resides in a village on the southwest coast of Portugal.

Source: MDDI+Qmed