Abstract
Material extrusion of cordierite ceramics faces challenges in fabricating dense, high-strength components due to non-linear parameter coupling. To address this, a two-stage framework based on Latin hypercube sampling is proposed. The first stage employs planar dimensional deviation of green bodies for rapid, non-destructive screening to filter impractical parameters prior to sintering. The second stage optimizes flexural strength of sintered bodies under dimensional constraints, enabling a decoupled enhancement of mechanical performance. Sensitivity analysis reveals that layer height (41.8%) and nozzle temperature (39.0%) govern dimensional deviation, whereas layer height is the dominant factor (60.7%) governing flexural strength, which is further validated by the evolution of microstructural defects. Under optimized conditions, the green bodies achieve a planar dimensional deviation of <0.25%, while the sintered samples reach a flexural strength of 196.8 MPa after hot isostatic pressing. To validate the proposed framework, a 148-mm lightweight mirror is successfully fabricated, exhibiting near-zero thermal expansion (−0.45 to 0.05 × 10−6/K, 0–50 °C). This study provides a useful insight into the development of a data-driven fused deposition modeling strategy for the fabrication of high-performance ceramic components.
| Original language | English |
|---|---|
| Article number | 115993 |
| Pages (from-to) | 1-13 |
| Number of pages | 13 |
| Journal | Materials and Design |
| Volume | 265 |
| Early online date | 13 Apr 2026 |
| DOIs | |
| Publication status | Published - May 2026 |
Keywords
- Cordierite ceramic
- Material extrusion
- Low thermal expansion
- Artificial neural network
- Latin hypercube sampling
Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver