**4. Conclusions**

In this paper, a time-lapse ex-situ helical X-ray CT imaging strategy assisted by staining was used to track the development of fatigue damage under tension-tension fatigue in a non-crimp UD GFRP. In essence, the contrast agen<sup>t</sup> could favour the observation of damage where a penetration path exists, but this could be both the weakness and the strength of this method. The weakness is that a damaged region without connection to the outer surface could not be stained; while the strength is that we are able to identify the connection of different damage modes based on the stained path, even for cases where the full crack path is difficult to be ascertained using X-ray CT. This enables us to experimentally prove the hypothesis on the linking of the different fatigue damage mechanisms and also provide insights into their interaction in 3D. Helical X-ray CT makes it experimentally feasible to follow the fatigue damage evolution over a sufficiently long region in the composite along the UD fibre direction. Overall, four main damage modes were identified,

• off-axis matrix and interface cracking,


Off-axis matrix cracks initiating from the specimen edges were sometimes deflected by stitching threads by debonding. Off-axis cracks between backing fibres were found to be associated with UD fibre fractures evidenced by the penetration path of the contrast agent. Moreover, these UD fibre fractures tend to nucleate and propagate locally in the vicinity of cross-over regions of backing bundles instead of being evenly distributed along the UD fibre direction. In addition, UD fibre fractures also tend to be initiated by the presence of extensive debonding and longitudinal splitting, which were found to develop from debonding of the stitching threads near surface. The isolation of the UD fibre bundle caused by longitudinal splitting potentially makes the composite susceptible to compression and bending loads as well as environmental impact in service. It could be inferred from the results here that further research into the better design of the positioning stitching threads, and backing fibre cross-over regions is required in the future, as well as new approaches to fix the positions of UD fibres. The work presented here (all the X-ray CT datasets are available online [17]) could be of significance to the further improvement of analytical and numerical models to predict the fatigue failure of composite materials.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/1996-1944/11/11/2340/s1, Video S1: A supplementary video accompanies Figure 6.

**Author Contributions:** Conceptualization, Y.W., L.P.M. and P.J.W.; Investigation, Y.W. and G.P.; Writing-Original Draft Preparation, Y.W.; Writing-Review & Editing, L.P.M. and P.J.W.; Funding Acquisition, L.P.M. and P.J.W.

**Funding:** This research was funded by the allianCe for ImagiNg of Energy Materials (CINEMA) project under DSF-grant number 1305-00032B.

**Acknowledgments:** The authors would like to thank LM Wind Power and DTU Wind Energy for sample preparation, and FEI (Thermo Fisher), especially Dirk Laeveren, for technical support and loan of the mk I HeliScan. P.J.W. is grateful to the Engineering and Physical Science Research Council (EPSRC) for funding the Henry Moseley X-ray Imaging Facility (grants EP/F007906, EP/F001452 and EP/I02249X, EP/M010619/1, EP/F028431/1, and EP/M022498/1) and a European Research Council Grant CORREL-CT (No. 695638).

**Conflicts of Interest:** The authors declare no conflict of interest.
