3.1 Failure modes and global behavior
Detailed information of the failure patterns and typical load–displacement responses of three groups of specimens with different W/D, E/D and D/t obtained from experiments is illustrated in Fig. 5. For Group 1, distinct fracture surface roughly perpendicular to the loading direction can be observed for type A with the W/D of 2. Moreover, a hardly visible hole bearing deformation accompanying with flat surface around the hole illustrates almost no bearing failure occurred. Besides, no cracking is seen on the end face of the plate, which excludes shearing out and cleavage failure. Thus, net tension failure is diagnosed for type A. In respect of type B, except for the fracture surface along the width direction of the plate, an ovality resulting from a larger hole deformation and a small amount of matrix fragments and fractured fibers near the edge of the hole are able to be observed, which indicates a combination of net tension and slight bearing failure occurred. As regards type C, massively local destructions adjacent to the hole and a serious hole deformation induced by bearing can be obviously observed. The local damage includes matrix crushing, transverse fibers bending, longitudinal fibers micro-buckling and a piece of material out-of-plane bulging. In addition, a fracture surface approximately along the center line of the plate and starting from the end face and extending into the plate can be found. It follows that remarkable bearing resulting in secondary cleavage is the final failure mode. For Group 2, representative shearing out failure mode characterized by evident two fracture surfaces roughly parallel the loading direction, materials pushed-out from the end face and a small deformation can be observed in type D. When E/D increases to 2.5 for type E, compared to type D, a smaller hole deformation occurred and a cleavage rather than shearing surface almost from the end face to the hole edge is apparent, which derives that cleavage led to the final failure. With the E/D increases to 3.5, the largest hole deformation and most severe local damage appeared for type C and, consequently, as mentioned above, full enough bearing failure happened. As to Group 3, hole deformations in the same extent and similar bearing failure modes leading to secondary cleavage can be found, which reveals D/t has little effect on the failure mode.
The load–displacement response of type A present a simple linear elastic shape signifying almost no damage occurred until the ultimate load. Then the load-bearing capacity dropped sharply, which manifests typical brittle failure characteristics. Except for type A, the curves for other types can be seemed as a three-phase variation as depicted on the curves of type C. Phase I: After initial contact leading to assembly clearance elimination accompanied by a short nonlinear form, the curves of all specimens are nearly linear until a knee point and no damage occurred. Similar stiffness can be obtained at this stage for all specimens since it was primarily dominated by the elastic hole deformation and dependent on the local bearing between the pin and hole contact surface, while hardly influenced by the geometry. Phase II: Stiffness decrease was followed with the plastic deformation of hole and composites initial damage occurred at the knee point where stiffness changed abruptly. In most samples, before stiffness and load recovery, evident load decline forming the first peak can be observed. Compared to the first phase, the curves ascend at a smaller slope accompanied with intermittent fluctuations indicating material damage occurring. It can be found that, with the increasing of W/D, E/D and D/t, the magnitude of the slope fluctuations augmented. This change can be mainly explained by that the single net tension, shearing out or cleavage failure mode show a stable load rising for type A, type D or type E, whereas bearing or multiple failure modes exhibit labile load variation for larger geometries. Phase III: The load gradually increased up to peak then suddenly drop, which reveals the joints lost the load carrying capability. It is can be seen respectively from the three groups of test results that the larger the geometry is, the larger ultimate load and longer the second phase can be obtained. For group 3, similar bearing failure modes and global shapes of the curves can be found, which reveals D/t has little effect on the failure mode. However, it is worth noting that, as the increase of D/t, the slope of curves in phase II and critical load rose obviously. This is mainly ascribed to the fact: the damaged material corresponding to the same degree of hole deformation increased. Furthermore, after the bearing initial failure, stiffness recovery observed is also related to the out-of-plane expansion of the damaged materials near the pin. With the increase of D/t, more damaged materials quickly filled the gap between the composite plate and the fixture plates, which is known to lead to significant increase in bearing strength for laminates .
3.2 Damage mechanism
There are, in general, four basic joint failure patterns involving net-tension, shearing out, cleavage, and bearing referred to the composites in this research, although, sometimes combinations of these patterns occurred. The damage behavior in failure of pin-loaded satin woven C/C-SiC composites is complicated, particularly in the bearing failure including multiple damage mechanisms. Thus, further exploration of the occurrence and growth of the failure is obbligato to the damage tolerance evaluation of composite structures. To examine the accumulation and evolution of internal damage in detail during the experiments, microscopic images and AE signals including three features: amplitude, counting and energy of events were applied. The following discussion concentrates on the damage mechanism of representative failure modes in type A, C, D and E.
3.2.1 Net tension failure
Evolution of the amplitude, normalized cumulated counts and normalized cumulated energy of AE events for the type A together with the load-displacement curve and fractographic images were shown in Fig. 6 (a). In the beginning, besides noise seemed to be the influence of friction between the contact surfaces, AE signals are small, corresponding to an approximately linear elastic deformation with the stable mechanical performance of materials until the ultimate load. Accompanied by simple shapes of curves and small hole deformation, almost no damage near the hole and a flat and smooth hole edge can be observed for this type as shown in Fig. 6 (b). At the point of peak load, the energy and the counting increase dramatically, revealing serious damage including the most remarkable longitudinal fiber fracture with the amplitude close to 100dB, fiber/matrix interface debonding with the amplitude of 60-70dB, fiber pulling-out with the amplitude of 70-80dB and matrix cracking with the amplitude below 60dB, which resulting in the net tension failure.
3.2.2 Shearing out failure
As to type D, after the linear phase, the first sharp rising in the AE signal records at the knee point of the load-displacement curve reveals the distinct loss of stiffness and initial damage. Due to the small value of E/D, the stiffness transition is mainly attributed to the large number of matrix shearing cracking of 45-60 dB and weft yarn bending and debonding of 60-80 dB from hole edge to end surface along the shear-out planes, other than local bearing failure of hole edge. This can be also characterized the small hole deformation and relatively flat hole edge as shown in Fig. 7. Subsequently, the previously smooth load-displacement curve is replaced by an obvious but short fluctuation indicating more and more damage occurring, and the energy and event counting increase continuously but with the low amplitude. This is explained as: before the transverse fibers fracture, the main damage of this fluctuation phase is a large number fiber/matrix debonding and weft yarn bending with amplitude of below 80dB. When reaching the ultimate load, drop of load is dominated by transverse fiber fracture with high amplitude beyond 80dB, accompanied by further interface debonding and fiber bending.
3.2.3 Cleavage failure
For type E, the flat and smooth hole edge can be found, which is similar to type D, but the sudden dramatic rising of AE signals corresponding to the initial damage reveals a gentle falling of stiffness. The subsequent fluctuation phase is longer than that of type D and the energy and event counting increase at a roughly uniform speed. This difference can be interpreted by the fact that, under the pressure of the pin, bending deformation occurred on the part of the composite plate from hole edge of to end face as shown in Fig. 8. According to the beam bending theory, the maximum tensile stress appeared at the center of the end face and the maximum compressive stress at the hole edge, leading to the weft yarn tension fracture and bulking to compression fracture. Additionally, the bending is featured by a progressive damage process along the center line of the plate with fiber fracture beyond 80dB and mass of matrix cracking below 60dB, corresponding to the gradual accumulation of the energy and event counting, until the cleavage failure is completed.
3.2.4 Full bearing failure
The net tension, shearing out and cleavage refer to situation wherein the load carrying capacity of the joints lost within a short time, while the bearing is widely deemed to be a desirable failure mode since the capacity to transfer load can be maintained longer. The progressive bearing failure can show a warning which significantly improves structure safety, and is recommended by designers. Compared with net tension, shearing out and cleavage, bearing behavior is relatively complicated and characterized by multiple failure mechanisms. For bearing failure, as shown in Fig. 9, low AE data values have been recorded in early phase during the experiment. Subsequently, the significant rising in AE signal records reveals the first loss of linearity and the energy and event counting increase with undulation but at a relatively stable rate, indicating multiple progressive damage occurring. For examining the features of bearing damage better, micrograph images were taken in corresponding failure morphology of different views including hole edge, pin contact surface and middle cut plane as shown in Fig. 10 – Fig. 12 (the white dotted line is the center line of the bolt in Fig. 10 and the green represents the pin in Fig. 12 during the experiments), and the load levels of type C were divided into about 75%, 85% and 100% of the maximum load. Bearing damage initiatively occurred at the hole edge and then propagated unstably in the plate. Due to the brittleness matrix of C/C-SiC composites, partial damaged material was crushed to fragments and fell off. In Fig. 10 (a), The first observation obtained is very localized damage at the contact surface in front of the loaded hole boundary resulting from micro-damage mainly including warp and weft yarns micro-buckling and splitting, matrix crushed and slight fiber/matrix interface debonding at the level of 75%. Thus, high-amplitude AE signals about 80 dB corresponding to the yarns splitting are mixed with middle-amplitude ones of 60-80 dB representing the yarns buckling and interface damage and low-amplitude ones of 45-60 dB showing the matrix damage. Thereinto the fiber bundles buckling and splitting close to the hole primarily accounted for the initial damage at the knee point of load–displacement curve. It is can be also noticed that partial damaged materials near the hole edge were pushed outwards to form bulge as illustrated in Fig. 11 (a). When the load increased to 85% of the maximum load, the damage region developed more extensively in front of the hole boundary and hole deformation augmented, but still extended over a limited area in front of the hole edge, which is visibly found from the middle cut plane view in Fig. 12 (b). Except for the further weft yarns bending to fracture and warp yarns crushed breakage, the fiber/matrix interface debonding became serious, propagating transversely and longitudinally and leading to delamination on the layers outside. Therefore, the counts of middle-amplitude signals of 60-80 dB indicating fiber/matrix and delamination increased. At the level of the maximum load, yarns failure and delamination aggravated and damaged yarns arrangement became disorganized. Particularly, the delamination occurred inside of the plate as shown in Fig. 12 (c). In research  on bearing failure of laminate, the aggravation of fiber buckling and through-thickness shearing action generated angular shear cracks and large-scale delamination through the thickness of the specimen at this level. However, for the composites in this research, fiber damage was more local and the delamination occurred gradually from outside to inside along the thickness of the plate. Actually, weft yarns and warp yarns were forced to break by a combination of compressive and out-of-plane action, so that for satin woven composites with original fiber bundles bending, yarns crushed buckling and debonding tended to appear near the pin compared with the laminate. While due to a long hole deformation, E/D was small enough to meet the secondary cleavage failure mode. Thus, subsequent fiber and matrix fracture under tension and compression, accompanied by a sudden increase in number of high-amplitude signals beyond 80 dB and a load drop, was found to begin at the center of the hole edge and run directly through the end surface, which was similar to the damage mechanisms of type E mentioned above.
The AE amplitude ranges for the matrix cracking, interface debonding fiber pulling-out and fiber fracture are respectively about 45–60 dB, 60-70 dB, 70–80 dB, and 80–100 dB, which are basically consistent with those of the carbon fiber/epoxy composite laminates by Liu et al.  and the self-reinforced polyethylene composites by Zhuang et al.  as shown in Fig. 13. The main difference may lie in fibrous preform woven form. In addition, Bohse  used release rate of AE events energy and Groot et al.  used AE peak frequency to monitor damage mechanisms. Although AE parameters selected to apply in the investigation are different, the relative signals ranges corresponding to the damage mechanisms are generally consistent. However, comprehensive understanding of the damage mechanisms may require further in-depth insight.