Abstract
As a thermoplastic composite with superior sustainability and mechanical performance, carbon fiber reinforced polyetheretherketone (CF/PEEK) has become the preferred material for fabricating prototype components in next-generation aircraft applications. However, drilling-induced delamination damage is prone to occur due to the material's anisotropy and lamination characteristics, severely compromising the service reliability of CF/PEEK components. This study focuses on the entrance and exit delamination damage in drilling CF/PEEK. Specifically, drilling experiments under various parameters are conducted to examine the distribution characteristics and formation mechanisms of delamination. Subsequently, considering thrust force, torque, and drilling temperature, a small-sample sensitivity analysis framework based on a hyperparameter-optimized support vector regression (SVR) model and the Sobol method is developed to identify sensitivity factors influencing delamination. The results indicate that delamination primarily occurs within the fiber cutting angle range of 90°-180°, where fiber bending fractures lead to an increase in subsurface damage. The hole entrance primarily exhibits mode Ⅲ delamination caused by intralaminar tearing, whereas the hole exit primarily exhibits mode III delamination, accompanied by mode I delamination due to interlaminar debonding. Furthermore, torque is identified as the most sensitive factor influencing both entrance and exit delamination. Compared to entrance delamination, the exit delamination is more susceptible to interactions among multiple factors.
| Original language | English |
|---|---|
| Article number | 114663 |
| Journal | Thin-Walled Structures |
| Volume | 224 |
| DOIs | |
| State | Published - May 2026 |
Keywords
- CF/PEEK
- Delamination damage
- Distribution characteristics
- Drilling
- Formation mechanisms
- Sensitivity factors
Fingerprint
Dive into the research topics of 'Investigation on entrance and exit delamination damage in drilling CF/PEEK via experimental and machine learning approaches'. Together they form a unique fingerprint.Cite this
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver