Researchers İlke Demir, Daniel G. Aliaga, and Bedrich Benes tackle one of the most popular topics in 3D printing today: optimization. While the many benefits of digital fabrication are oft discussed—from greater affordability, improved speed in production, and the ability to create and re-design without a middleman—challenges continue to arise due to continual innovation. Ever on the search for perfection, users are continually seeking ways to predict mechanical properties, decrease defects, and monitor additive manufacturing systems.
In this study, the authors focus on reducing the amount of material used, reducing print times, and refining accuracy. Detailing the efforts of their research in ‘Near-convex decomposition and layering for efficient 3D printing,’ we learn more about their ‘divide-and-conquer approach,’ featuring automatic decomposition and configuration of an input object into print-ready components.
“3D printers have both limitations and advantages depending on the coherency between the printer features and the model geometry,” explained the authors. “Instead of relying only on improvements of the 3D printing technology, we provide a solution that optimizes the model in order to maximize that coherence by segmenting the model into easily printable components.”
They noted 15% improvement of quality, 49.4% savings in material, and 50.3% reduction in printing.
The sample for this study is a polygonal model. Decomposition included separating the beginning clusters into an ‘optimal’ set of components. In the next step they were prepared for printing in a configuration phase, saving time as in most other cases labor is extended as the print bed must be moved down, or the printhead must be moved up. Production is also more efficient as parts are printed at once. In evaluating properties, the researchers examined:
- Volumetric approximation
- Number of components
- Amount of support material
- Faster print time
- High quality resulting from less angular surfaces
The algorithm consists of subspace creation and segmentation. A set of similarly shaped clusters (triangles) is defined, and then clusters are ‘iteratively merged and split’ for balance.
“During each iteration of this step, we compare cluster-by-cluster, mark similar clusters, and merge-split at the end of each iteration, until convergence. We also highlight that our method uses the same threshold parameter values for all models,” explain the authors.
For improved printing, components must possess:
- Surface angles
- Sizes and Numbers
Of the 20 samples applied to the framework in this study, some were manually modeled, and some were acquired commercially. Complexity averaged 23.9K, with the new method suitable for both solid and shell forms. Pre-processing time for segmentation and configuration was around 15 minutes for a medium complexity model.
Printed examples were compared with the initial and segmented models, ‘with better approximated surfaces, and multi-color support.’ Real models were also examined in their initial form, after supports were removed, and before and after assembly.
“… our approach prevents wasting material, and provides higher fidelity objects, with multi-material support. Note that, even if the approximated surface is highly curved, our decomposition finds segments that connect well, even after printing with accumulated printing errors.”
The authors did note, however, that the printed model did not ‘approximate’ the original—although the segmented model did. Upon superimposing printed versions in wireframe, they were able to show that improved approximations can be achieved—using the same printer.
“The coloring in the point cloud version indicates that our algorithm decreased the overall error more than 35% based on the Hausdorff distance of sampled surface points. We have not evaluated based on a measurement of the real printed models, because parameters contributing to this surface error is more constrained in simulation,” concluded the researchers.
“Our results show that the framework can reduce print time by up to 65% (fused deposition modeling, or FDM) and 36% (stereolithography, or SLA) on average and diminish material consumption by up to 35% (FDM) and 10% (SLA) on consumer printers, while also providing more accurate objects.”
What do you think of this news? Let us know your thoughts! Join the discussion of this and other 3D printing topics at 3DPrintBoard.com.[Source / Images: ‘Near-convex decomposition and layering for efficient 3D printing’]
You May Also Like
3D Printing Webinar and Event Roundup: September 12, 2021
Buckle your seatbelts, it’s going to be a busy week of webinars and events, both virtual and in-person! RAPID + TCT and FABTECH will both be held in-person this week...
Sixth Bioprinting Acquisition in One Year from Cellink Parent Company BICO
Pioneering bioprinting firm Cellink, now part of a larger company rebranded as BICO (short for bioconvergence), has already been making quite a name for itself and is preparing to capture...
Complete Tumor 3D Printed to Facilitate Faster Treatment Prediction
There are more than 120 different types of brain tumors, many of which are cancerous, but the deadliest, and sadly most common, is the aggressive, fast-growing glioblastoma multiforme (GBM), a...
3D Printing Webinar and Event Roundup: August 15th, 2021
From convincing your professor they need a 3D printer and the future of static mixers to biomaterials and bioprinting, we’ve got another week of webinars and events to tell you...
View our broad assortment of in house and third party products.