Embedded Files
High-Precision Rapid Prototyping Framework for Complex Structures
Team Members: Sina Rastegar (Lead), Mohammad Abu-Mualla, Xinnian Wang, Muhammad Areeb, Jida Huang, Jun Wang
Supervisor: Jida Huang
Key achievements
Eliminates Mesh-Based Representation
Supports Multiple 3D Printing Technologies
Automated MATLAB-Based Slicing
Resolution-Independent Fabrication
High Computational Efficiency Material Versatility
Future Development Directions
Figure: Building large-scale structure by part to illustrate the scalability of the presented pipeline. (a) parts printed separately. (b) parts assembled. (c) scaling up the design to illustrate the scalability of the proposed method with different. Model’s sizes.
Hard Skills
Additive Manufacturing & 3D Printing
Computational Geometry & Function Representation (FRep)
MATLAB & Algorithm Development
Finite Element Analysis (FEA)
Computer-Aided Design (CAD)
G-code Generation & Toolpath Planning
Data Optimization & Memory Management
Numerical Methods & Mathematical Modeling
Soft Skills
Problem-Solving & Critical Thinking
Collaboration & Teamwork
Attention to Detail
Time Management
Technical Communication
Adaptability & Continuous Learning
Project overview
This project successfully developed a mesh-free slicing pipeline utilizing function representation (FRep) to enhance the efficiency and scalability of high-resolution additive manufacturing (AM). Traditional slicing methods relying on polygonal mesh representations were found to be computationally intensive and memory-demanding, making them unsuitable for intricate multi-scale material designs. By eliminating the need for STL-based representations, this project introduced a novel approach that significantly reduced memory usage and computational overhead.
The implemented slicing pipeline employed zero-level set functions to implicitly define solid region contours, making the process independent of design or manufacturing resolution. Unlike conventional methods that store large binary images or require point-by-point classification, this approach streamlined slicing while maintaining high geometric accuracy. As a result, the computational efficiency of AM processes was greatly improved, enabling the fabrication of highly detailed structures with minimal processing constraints.
To validate the method, several complex hierarchical structures were successfully fabricated using an 8K Stereolithography (SLA) printer. The project also addressed the geometry frustration problem that occurs when connecting distinct cellular structures in functionally graded materials, ensuring smooth transitions between micro-scale unit cells. The results demonstrated that the proposed slicing framework effectively bridges the gap between computer-aided design (CAD) and high-resolution manufacturing, offering a scalable solution for future advancements in AM technology.
Figure: Developed stl-free slicing pipeline using zero level-set values to formulate solid region contours as implicit functions.
Figure: Defining various topological characteristics: a) the relative volumetric error is defined as the relative difference between the CAD model and printed model volume; b) the average cusp height.
Femur Bone Implant
Two topological characteristics of the printed structure are considered here.
The first one is the relative volumetric error. As illustrated in Figure (a), this characteristic is the relative error between the volumes of the printed structure and the CAD model.
The second topological characteristic is the average cusp height of sliced layers, Figure (b).
Figure: Printed samples of F1 CAD model using SLA printer: a) uniform slicing with the maximum number of layers (Nl = 600); b) uniform slicing with the minimum number of layers (Nl = 150); c)adaptive slicing optimizing volumetric error with 300 layers.
Discussion
In this figure the CAD model of an architected material is printed using the SLA printer for three scenarios. The printed model with the finest thickness layer, figure (a) took the longest to print ( 2hrs); the roughest printed model, figure (b) took the shortest to print (35min); and the adaptive sliced printed model, figure (c) took almost one hour to print. While the adaptive sliced model was printed twice faster than the finest model, they are visually identical, and the surface error of the adaptive sliced model is not noticeable.
Google Sites
Report abuse