UC San Diego

Composites are frequently used in aerospace structural applications due to their high strength to weight performance, but due to their layered structure they are vulnerable to transverse impacts. Impact damage in composite laminates often consists of highly interactive damage modes composed of delamination, matrix cracking, and fiber breakage. In order to ensure the safety of composite structures, a variety of non-destructive evaluation (NDE) techniques are used to characterize impact damage. However, procedures for utilizing NDE to create and validate models of residual strength after impact are not yet established due to either limitations in the characterization of impact damage, as in the case of Ultrasonic pulse-echo scanning (UT), or due to the complexity of interpretation of the NDE technique, as in the case of X-ray computed tomography (CT). Improved quantification of damage from CT and UT characterization may lead to improved predictive capabilities for the prediction of structural performance after an impact event.

This work presents a novel automatic damage segmentation procedure for CT scans of impacted composites that converts the complex 3D dataset into simplified damage visualizations and 2D damage maps for each composite layer. The results of this procedure were utilized to create and validate a modeling procedure to improve UT characterization of impact damage, and to validate and generate finite element models of impact damage and residual strength performance. The generated residual strength models were created with varying levels of damage modeling fidelity and it was found that the level of damage modeling needed for accurate failure prediction depends greatly on the structural geometry and the presence of major damage features. This NDE and modeling effort was supported by a series of impact and residual strength experiments for flat and stringer stiffened composite panels. The developed techniques proved capable of characterizing impact damage in a variety of structural configurations and establishing models that incorporate this damage at different levels of complexity.