Observations of damage development from compression-after-impact experiments using ex situ micro-focus computed tomography D.J. Bull1*, S.M. Spearing1, I. Sinclair1 1 Faculty of Engineering and the Environment, University of Southampton, Highfield, Southampton, SO17 1BJ, UK *Corresponding author: daniel.bull@soton.ac.uk Keywords: A. Carbon fibres, B. Delamination, C. Damage mechanics, D. Non- destructive testing, X-ray computed tomography The development of damage mechanisms leading up to compression-after-impact (CAI) failure is studied in particle-toughened and untoughened systems. Microfocus computed tomography (µCT) enabled non-destructive monitoring of the internal damage development in three-dimensions (3D) by taking scans after impact, after an application of near failure compression loads and after coupon failure. In combination with µCT work, mechanical CAI testing and ultrasonic C-scans were conducted to determine the effect of the projected damage area on residual CAI strength and to complement the observations made from µCT scans. The important role of the undamaged “cone” of material immediately under the impact site for out-of-plane sublaminate stability is identified. The implication of delamination growth into this region is discussed. It was found that where particle-toughened systems suppressed delamination growth into this region, greater residual CAI strength was maintained on a like-for-like projected damage area. 1. Introduction It is widely accepted that low velocity impact leads to damage in carbon fibre materials which has a direct effect on the residual compressive strength. It is reported that the loss in residual compressive strength scales with the size of the projected damage area (a representation of the extent of delaminations) [1-11]. This conclusion is based on conventional methods of measuring the damage area, typically using ultrasonic C-scans. Whilst this correlation based on projected damage area is widely accepted, the controlling mechanisms leading to catastrophic failure are still debated and not well understood. Such lack of understanding may be attributed to the complex and sudden nature of compression-after-impact (CAI) failure making it difficult to identify the critical mechanisms contributing to the loss of compressive strength. Whilst it is generally agreed that delaminations lead to a reduction in compression strength after impact, due to the formation of sublaminates with a reduced flexural stiffness which results in the earlier onset of buckling [6, 12], it is the sequence of events that is debated. Some studies report that sublaminate buckling leads to a sudden growth of damage extending laterally, leading to a sudden failure of the coupon [2, 13, 14]. Another reported mechanism is that the buckled sublaminates lead to a load redistribution resulting in compressive fibre fracture [4, 15]. One study using X-ray radiography to monitor damage growth at incremental compressive loads suggests that sublaminate buckling leads to a combination of bending and compressive loads in the remaining undelaminated regions. In this model the final failure is believed to occur when these stresses exceed the maximum compressive stress in the 0° plies [16]. To understand damage mechanisms that can develop and contribute towards critical failure, this study uses novel µCT experiments, scanning coupons at various loading stages: after impact, after application of near failure in-plane compressive loads and after compressive failure. This technique allows for a non-destructive three-dimensional evaluation of the damage development within the same coupon [17] and is carried out in combination with conventional ultrasonic C-scan and CAI experiments. The aim of this study is to understand better the damage mechanisms leading up to compressive failure and any additional contribution towards damage tolerance in particle-toughened systems beyond the extent of the impact damage area. Such understanding will aid development of damage tolerant material systems and ensure finite element models for CAI tests are capturing the correct failure mechanisms in order to better predict critical failure loads. 2. Materials and methods 2.1 Materials Five proprietary unidirectional carbon fibre prepreg systems were used in this study. These consisted of four systems with thermoplastic particles introduced into the resin, and one untoughened system without particles. All systems used the same intermediate modulus carbon fibres, with the same fibre-to-matrix ratio by weight, and the same thermoset base resin. Variations in particle sizes and particle chemistries were used in the particle-toughened systems, with the same ratio of particles to resin by weight making up the matrix. The material systems are ranked in order of impact damage resistance as measured by ultrasonic C-scan projected damage area, with the untoughened (UT) system being least damage resistant, and the particle-toughened systems (T1, T2, T3 and T4) listed in order from lowest to highest damage resistance. Test coupons were manufactured in accordance with ASTM D7136M standards. These consisted of 100 x 150 mm, 24 ply coupons using a [45, 0 -45, 90]3S layup. Coupon thickness was approximately 4.6 mm +/- 0.2 mm across all five material systems. 2.2 Mechanical testing Materials were subjected to impact loading using a drop tower in accordance with ASTM D7136M standards at nominal incident energy levels of 25, 30, 40 and 50 J. Results from these experiments can be found in a previous study [18]. The height of the drop determined the impact energy. These were repeated three times on each material system. The velocity of impact was measured and the actual impact energy recorded, this was typically less than the target impact energy (although always within 10%) due to energy lost by friction at the interface with the guide rails of the drop tower. After impact, ultrasonic C-scan was performed to measure the projected damage area. To measure the residual compressive strength, each post-impacted coupon was subjected to CAI tests on a universal testing machine according to ASTM D7137M standards. Coupons were placed in an anti-buckling rig and loaded at a cross-head displacement rate of 1.25 mm per minute. 2.3 Compression-after-impact ex situ µCT procedure To monitor damage progression at near failure compressive loads, ex situ experiments were carried out. For this work, coupons were impacted at 25 J for the UT system and 30 J for T1, T3 and T4 systems. A lower incident impact energy was chosen for the UT system to enable the majority of damage to be within the field of view of the µCT scan. To increase throughput and maximise coverage of the field of view, pairs of coupons were stacked together when scanned. Two local regions were scanned on each specimen, one directly in the vicinity of the impact site and the other at the lateral edge of the projected damage area, determined earlier from an ultrasonic C-scan. Local volumes were obtained by positioning the region of interest over the axis of rotation. The voxel size of the scan was 14 µm, limited by the clearance between the X-ray target and coupon stack. Scan settings were: 115 kV, 100 µA, one second exposure, two frames per projection and 1301 projections over 360° of rotation. A Molybdenum target with no filtering was used. Due to time constraints on accessing the imaging facilities, settings were selected for moderate scan times (~45 minutes) rather than an optimum image, and the T2 system was not scanned. For the ex situ scanning procedure, coupons were µCT scanned after impact and following application of incremental load steps near the critical failure load. The load levels reached in each load step are normalised to the failure load and are reported in Table 1. The first load was applied to two standard deviations below the mean failure load based on previous mechanical testing results. Subsequent loads were incremented at approximately 2 kN load steps or until damage was detected audibly. Across the four systems, the load step immediately prior to failure is referred to as a ‘near failure’ load in this work. In the UT and T1 system, the ‘near failure’ load was slightly higher than the measured final failure load: whilst this may be attributable to sub-critical damage growth, the effect was substantially less than one standard deviation observed in the conventional CAI tests. 3. Results and discussion 3.1 Mechanical testing A plot of normalised failure stress against impact energy is shown in Figure 1. Normalisation of failure stress was calculated by dividing the failure stress against the largest failure stress in the data set. Across all systems tested, there is an approximately linear reduction of in-plane compressive failure stress as the impact energy increased. There are clear differences in damage tolerance of the four material systems tested, correlating with the impact damage resistance of each system. As expected, the UT system exhibits the least damage tolerant properties, T1, T2 and T3 showed intermediate levels and the T4 system had the highest. There was some scatter in the compressive failure stress across all the material systems at different impact energies. The effects of scatter made it difficult to predict the exact failure load during interrupted ex situ µCT tests which had the aim of achieving damage observations at loads near to failure. For this reason, the lower bound of failure to two standard deviations, as determined from these tests was used on the first loading cycle. 5 To show how the extent of the projected damage area correlates with a reduction in failure load, these data are plotted in Figure 2. The projected damage area based on ultrasonic C-scan measurements is representative of the scale of delaminations; detailed µCT measurements of impact induced delaminations shows the through-thickness delamination distribution to be fairly uniform across the though-thickness. The projected damage area was normalised by dividing the particular damage area for a particular impacted specimen by the largest damage area in the set of data (all materials, all impact energies). Comparing the failure stress against the projected damage area, it is clear there is a correlation between the two parameters, allowing for the scatter in the data. Within the same systems, an increase in projected damage area clearly results in a loss in residual compressive strength. It is interesting to note however, that there are variations in the gradients and overall values of the CAI strength-damage area relationships between material systems. As such it may be conjectured that several damage modes or toughening mechanisms may contribute to the damage tolerance. Compared to the UT system, the T2 and T3 systems share a similar gradient across the range studied. However, for a given damage area, the T2 and T3 systems showed a higher compressive failure stress, in the order of ~30% more than the UT system. There was also a higher damage tolerance exhibited by the T1 system over the UT system, particularly at lower impact damage areas. In the most damage resistant, T4 system, the slope is significantly steeper than the other systems tested. Whilst T4 exhibits good CAI damage tolerance when considered in terms of impact energy, when plotting CAI load against damage area, failure loads are in fact quite similar to T1, T2 and T3 in the regime where they overlap with the T4 data. It is clear that the CAI failure stress decreases considerably more rapidly with damage area in the T4 system: extrapolating the T4 and UT results indicates that area-for-area, the T4 material may in fact be worse than the UT material at intermediate to high damage areas. As such, T4’s CAI engineering performance may be identified as strongly impact damage resistance- driven, whilst T2 and T3 demonstrate greater damage tolerance for a given impact damage area. Overall it is clear that factors other than simple delamination area- controlled buckling contribute to residual CAI strength. 3.2 Compression-after-impact µCT damage development observations 3.2.1 Delamination development µCT cross-sections at the impact site after impact and after application of a near failure compressive load are shown in Figure 3 for the T1 and T3 material systems. Common to the UT, T1 and T3 systems, an increase in residual crack-opening displacement two to four plies from the back face (i.e. opposite to the impact site) was observed as highlighted in (i). This increase in crack-opening displacement was a consequence of the applied compressive load. In combination with an increase in crack-opening, delamination growth was clearly detected, propagating into the undamaged cone beneath the impact site, e.g. (ii). Figure 4 shows a 3D segmentation of two modes of delaminations: (i) “central” where delaminations are confined between two matrix cracks of the same orientation and (ii) “45° segments” where delaminations grow away from the impact site, constrained at the interface between two matrix cracks 45° apart. The 45° delamination segments form a “spiral staircase” of delaminations through the thickness of the material; this is illustrated by the example shown in Figure 5 representing segmented delaminations at four ply interfaces with each colour representing a delamination at a particular ply interface. The combination of the 45° delamination segments forms a near circular pattern surrounding an undamaged “cone” of material immediately below the impact site. The projection of the near circular patterns represents the damage area typically obtained through ultrasonic C-scan methods. Across all systems studied, no detectable delamination growth was observed on the 45° delamination segments during loading up to the near failure loads. Delamination growth was only observed to occur on central delaminations within the undamaged cone region defined by the impact. To quantify this “inward” delamination growth, the total length across central delaminations was measured after impact and after application of near-failure loads. The arrows in Figure 4 (i) show how the total length of these central delaminations was measured. This was achieved by taking the edge to edge delamination distance along a centreline parallel to the orientation of the matrix cracks that confine the delamination. In cases where an undamaged region exists between the central delaminations, the sum of the two lengths was taken as shown by the arrows (i) in the unloaded case. Should the delamination propagate through the undamaged region then a single measurement was taken as shown by the arrow in (i) for the near failure case. Central delamination measurements are plotted in Figure 6 representing the normalised central delamination length against the ply interface for impact and near failure loads. Delamination extents were normalised by dividing the measured length by the largest measured length for that material system. Ply interface numbers are labelled in order of distance from the impact side of the coupon with ply 1/2 representing the interface closest to the impact side. From the plot, it is clear central delamination growth is occurring into the undamaged cone in the UT, T1 and T3 systems at near failure loads. What is also interesting is the initiation of new central delamination sites that were captured at six interfaces in the UT system and two on the T1 system. These delaminations were observed initiating from pre-existing matrix cracks. The T1 system had central delaminations at nine ply interfaces after impact, this resulted in more impact-induced damage growing into the undamaged cone in comparison to the other systems, see Figure 6. Delamination growth can be explained by considering the out-of-plane buckling of the sublaminates created by compressive loading. Such out-of-plane buckling of the sublaminates provides the driving force to propagate delaminations into the undamaged cone. The out-of-plane deflection during post-impact compression has also been observed in other studies through surface profilometry [2, 11, 12, 19]. µCT cross-sections of the T4 system are shown in Figure 7. In the T4 system which has the greatest damage resistance, no delamination growth into the cone was observed at the near-failure load. However in the scan taken after CAI failure, delaminations were observed in this undamaged region indicated at (i). It is likely that the load step prior to CAI failure was not sufficiently high in this case to create delamination growth into this region. In comparison, the other systems were within 0.4% or exceeded the failure load in the load step prior to failure, whilst the T4 load step prior to failure was 1.3% below the failure load. The other observation with the T4 system is the significant permanent out-of-plane deformation caused by the impact event highlighted in (ii) creating an indentation and locally bowed plies. An out-of-plane deformation of ~0.3 mm was measured at the midplane directly beneath the impact site. This was similar to the T1 system. On the UT and T3 systems, the permanent out-of-plane deformation beneath the impact site at the mid-plane was approximately half, i.e. ~ 0.15 mm. The effective lateral extent of this out-of-plane deformation is 15 mm for all systems tested. 9 3.2.2 Fibre fracture development Pre-existing 0° fibre fracture in some of the particle-toughened system was found to grow laterally across the ply at near-failure loads, as shown circled in Figure 8 (a). Similar observations have been made previously [20]. A 3D segmentation of this fibre fracture (b) is shown in red, and can be seen to propagate across a load bearing 0° ply, which has delaminations at its interfaces with neighbouring ±45° plies, shown in blue and yellow. In systems where no 0° pre-existing fibre fracture was present after impact, there was no detectable fibre fracture at near failure loads. It is conceivable that pre- existing failed fibres led to a redistribution of load to the neighbouring 0° load bearing fibres, leading to a growth of fibre fracture laterally across the ply. Such fibre fracture in the load bearing 0° plies will inevitably contribute towards a reduction in compressive failure load. The current observations indicate this is most important for the T4 system which exhibited a greatly reduced projected impact damage area, but sustained significantly more fibre fracture during impact in comparison to the other systems. An interesting point to consider is the degree to which CAI load performance in these materials is a convolution of, or competition between, load bearing fibre fracture (and associated growth during loading) and loss of constraint from delamination. Independent of which case applies, the increased incidence of this additional damage mechanism appears to provide a simple explanation of greater CAI load sensitivity of the T4 system when compared on an equivalent delamination area basis in Figure 2. 3.3 Sequence of events leading to compressive failure In the UT, T1 and T3 material systems, central delamination growth was observed propagating into the undamaged cone beneath the impact site. The progress of failure at the impact site is shown for the UT system in Figure 9 at three stages: after impact, after application of a near-failure compressive load and after the coupon failed. Key features contributing to critical buckling of the sublaminates in this region may be linked to: (i) representing the undamaged cone, (ii) showing delamination growth into this undamaged cone, and (iii) showing critical sublaminate buckling. The importance of the undamaged cone is simply illustrated schematically in Figure 10 where in (a) it offers support to the sublaminates at the centre resulting in a shorter unsupported length ‘L’ of the sublaminates. As the compressive load is increased, out- of-plane deflection of the sublaminates occurs and is linked to delamination growth into the undamaged cone. When delamination growth within the cone allows the delaminations to extend, unbridged across the full near-circular impact damage region, it effectively more than doubles (in a one dimensional sense at least) the unsupported length ‘>2L’ of the sublaminates, (b). This sudden increase in unsupported length significantly reduces the load-carrying capability of the sublaminates resulting in local buckling. The significance of the undamaged cone has been reported in finite element models by Craven et al. [21] which shows that the undamaged cone leads to two smaller local buckles compared to a single larger buckle in a system without an undamaged cone. As a result, the inclusion of an undamaged cone led to an increase in local buckling strain by a factor of approximately two, although the model did not include delamination growth into this region. Considering the occurrence of delamination growth into the undamaged cone, the local buckling strain is expected to be greater than a system modelled without an undamaged cone but lower than the inclusion of an undamaged cone with no delamination growth. The gain in buckling strain by the undamaged cone highlights a clear advantage for systems preventing delaminations propagating into this region during loading in order to maximise the critical failure strain and load. Prevention of delamination propagation is achievable by maximising the quasi-static fracture toughness. In particle-toughened systems, the increase in fracture toughness is achieved by providing traction sites between sublaminates; see Figure 11 (a). Whilst the presence of bridging ligaments was not easily observed from the µCT scans, used in the present study, their presence has been observed using higher resolution SRCT in previous work, see (b) [18, 22]. This process can be seen to suppress the onset of delamination growth into the undamaged cone. By suppressing delamination growth into the undamaged cone, sublaminates maintain a shorter unpinned length thereby increasing stability and load carrying capability. This is consistent with the observation that the T2 and T3 systems maintained ~30% greater failure stress for a given projected damage area compared to the UT system. The marginal improvement in the T1 system is consistent with initial delaminations within the undamaged cone reducing the extent of material available for delamination propagation. In addition to suppressing delamination growth, should the ligaments extend sufficiently far behind the crack tip they may also counter out-of-plane deflection of the sublaminates by ‘tying’ regions together in a similar way to Z-pinning strategies [7, 23]. Use of high resolution facilities, e.g. synchrotron radiation laminography [17] in future work would allow confirmation of the role of particles in restricting delamination and thus determining the CAI failure load. The observation of delamination growth into the undamaged cone and the presence of bridging ligament formation highlights some of the limitations in simplified sublaminate buckling models such as those using a Rayleigh-Ritz solution of circular or elliptical delaminations [21, 24, 25]. Based on this solution, it is found that the buckling 12 load and strain generally follows an inverse square dependency on delamination length as given by Equation (1): (1) where is the buckling strain, is the buckling load and is the delamination length across the sublaminate in the loading direction. Based on this relationship, the buckling strain/load is governed by the size of the delamination length across the sublaminate. This is not so straightforward in this study, the ~30% increase in failure stress for a given damage area in two of the particle-toughened systems compared to the untoughened system is attributed to the observed suppression of delamination propagation through the undamaged cone, and bridging ligament formation creating traction sites between sublaminates. The complexity of the mechanisms observed in this study highlights features that are important to capture and include in models to accurately predict failure beyond a simplified critical delamination size. 4. Conclusions The CAI performance of one untoughened and four particle-toughened carbon fibre composite systems were examined. Consistent with previous work, the loss in CAI strength correlated strongly with an increase in the projected impact damage area for a given material system. Compared to untoughened systems, particle-toughened systems demonstrated up to 30% improvement to failure stress for a given damage area, highlighting that the link between failure stress and the size of the delamination area is not straightforward. Through use of µCT to study CAI damage growth, several observations were made regarding: delamination growth into the undamaged cone 13 immediately under the impact site driven by out-of-plane deflection of the sublaminates, growth of pre-existing 0° fibre fracture, and permanent out-of-plane deformation. Regarding the T4 system, it is known from an unpublished study on this material system that despite being the toughest material system by suppressing delamination growth, it also led to more extensive fibre fracture. This behaviour provided a simple explanation of the greater CAI load-damage area sensitivity for this material system. It is therefore suggested that very tough material systems, may be developed at the cost of being more prone to fibre fracture during impact, which may in turn also significantly contribute to a loss in CAI strength. After application of near-failure compressive loads, no sub-critical delamination growth was observed beyond the envelope defined by the projected damage area caused by the impact event. The importance of the undamaged cone of material under the impact site is to constrain the buckling deformation of the sublaminates. When delamination growth into this region occurs, it connects the surrounding delaminated regions, greatly increasing the unsupported length of the sublaminates and significantly reducing its residual load bearing capability. This mechanism apparently controls the buckling of the sublaminates and provides a mechanistic explanation for the role of toughness in determining the CAI strength for a given damage area. Whilst higher resolution would be required to increase confidence that these micromechanistic effects are indeed occurring, the findings in this work have pointed to a key topic for further study. If, as might be suspected, particle-bridging plays a key role determining the residual in CAI strength; this may be a vital area of improvement for the development of superior damage tolerant materials. It also raises questions regarding the transferability of CAI test data to the practical damage resistance and damage tolerance in composite structures. It is indicated that CAI performance is strongly determined by quite subtle, local effects, such as rate dependences, formation (or not) of an undamaged cone, and the occurrence of fibre fracture. There is scope for work to be conducted to evaluate the robustness of CAI data in determining the in situ damage resistance and damage tolerance of more complicated built up structure subjected to typical in-service damage events. This may be best achieved by modelling, with limited experimental verification. 5. Acknowledgements The authors wish to thanks the µ-VIS computed tomography centre at the University of Southampton for providing the µCT facilities used in this study, particular thanks to Anna Scott for performing the µCT scans. The paper also acknowledges Cytec Engineered Materials Ltd for their sponsorship and supply of materials used in this project. The authors are particularly grateful for the help and support of Dr. Kingsley Ho, the technical point of contact at Cytec. 6. References [1] Cartie DDR, Irving PE. Effect of resin and fibre properties on impact and compression after impact performance of CFRP. Compos Part a-Appl S. 2002;33(4):483-493. [2] Reis L, de Freitas M. Damage growth analysis of low velocity impacted composite panels. Compos Struct. 1997;38(1-4):509-515. [3] Garg AC. Delamination - a damage mode in composite structures. Eng Fract Mech. 1988;29(5):557-584. [4] Wisnom MR. The role of delamination in failure of fibre-reinforced composites. Philos T R Soc A. 2012;370(1965):1850-1870. [5] Davies GAO, Hitchings D, Zhou G. Impact damage and residual strengths of woven fabric glass/polyester laminates. Compos Part a-Appl S. 1996;27(12):1147-1156. [6] Sanchez-Saez S, Barbero E, Zaera R, Navarro C. Compression after impact of thin composite laminates. Composites Science and Technology. 2005;65(13):1911-1919. [7] Zhang X, Hounslow L, Grassi M. Improvement of low-velocity impact and compression-after-impact performance by z-fibre pinning. Compos Sci Technol. 2006;66(15):2785-2794. [8] Ishikawa T, Hayashi Y, Matsushima M. Compression after impact (CAI) properties of Cf/Peek (Apc-2) and conventional Cf-epoxy stiffened panels. Icas Proceedings 1992, Vols 1 and 2. 1992:1225-1234. [9] Recker HG, Altstadt V, Eberle W, Folda T, Gerth D, Heckmann W, et al. Toughened thermosets for damage tolerant carbon-fiber reinforced composites. Int Sampe Tech Conf. 1989;21:283-293. [10] Zhou G, Rivera LA. Investigation on the reduction of in-plane compressive strength in thick preconditioned composite panels. Journal of Composite Materials. 2007;41(16):1961-1994. [11] de Freitas M, Reis L. Failure mechanisms on composite specimens subjected to compression after impact. Composite Structures. 1998;42(4):365-373. [12] Habib FA. A new method for evaluating the residual compression strength of composites after impact. Composite Structures. 2001;53(3):309-316. [13] Chai H, Knauss WG, Babcock CD. Observation of damage growth in compressively loaded laminates. Experimental Mechanics. 1983;23(3):329-337. [14] Yan H, Oskay C, Krishnan A, Xu LR. Compression-after-impact response of woven fiber-reinforced composites. Compos Sci Technol. 2010;70(14):2128-2136. [15] Soutis C, Curtis PT. Prediction of the post-impact compressive strength of CFRP laminated composites. Composites Science and Technology. 1996;56(6):677-684. [16] Pavier MJ, Clarke MP. Experimental-techniques for the investigation of the effects of impact damage on carbon-fiber composites. Composites Science and Technology. 1995;55(2):157-169. [17] Bull DJ, Helfen L, Sinclair I, Spearing SM, Baumbach T. A comparison of multi- scale 3D X-ray tomographic inspection techniques for assessing carbon fibre composite impact damage. Composites Science and Technology. 2013;75:55-61. [18] Bull DJ, Scott AE, Spearing SM, Sinclair I. The influence of toughening-particles in CFRPs on low velocity impact damage resistance performance. Composites Part A: Applied Science and Manufacturing. 2014;58(0):47-55. [19] Ishikawa T, Sugimoto S, Matsushima M, Hayashi Y. Some experimental findings in compression-after-impact (CAI) tests of CF/PEEK (APC-2) and conventional CF/epoxy flat plates. Composites Science and Technology. 1995;55(4):349-363. [20] Aymerich F, Priolo P. Characterization of fracture modes in stitched and unstitched cross-ply laminates subjected to low-velocity impact and compression after impact loading. Int J Impact Eng. 2008;35(7):591-608. [21] Craven R, Iannucci L, Olsson R. Delamination buckling: A finite element study with realistic delamination shapes, multiple delaminations and fibre fracture cracks. Composites Part A: Applied Science and Manufacturing. 2010;41(5):684-692. [22] Bull DJ, Spearing SM, Sinclair I, Helfen L. Three-dimensional assessment of low velocity impact damage in particle toughened composite laminates using micro-focus X-ray computed tomography and synchrotron radiation laminography. Composites Part A: Applied Science and Manufacturing. 2013;52(0):62-69. [23] Mouritz AP. Review of z-pinned composite laminates. Compos Part a-Appl S. 2007;38(12):2383-2397. [24] Shivakumar KN, Whitcomb JD. Buckling of a sublaminate in a quasi-Isotropic composite laminate. Journal of composite materials. 1985;19(1):2-18. [25] Chai H. Buckling and post-buckling behavior of elliptical plates: part I—analysis. Journal of Applied Mechanics. 1990;57(4):981-988. Figure 1: Graph showing normalised compressive failure stress vs. impact energy for the five systems tested. Figure 2: Graph showing normalised CAI failure stress vs. impact damage area for the five systems tested. Figure 3: µCT cross-sections of T1 and T3 material systems showing: (i) increase in crack-opening and (ii) delamination growth into the undamaged cone after application of a near failure load. White arrow indicates location of impact and side arrows indicate loading direction. Figure 4: T3 material system showing (i) growth of ‘central’ delamination between the third and fourth ply interface into the undamaged cone. (ii) represents a 45° delamination segment, there was no detectable delamination growth of these delamination segments. Figure 5: 45° delamination segments Figure 6: Measurement of total ‘central’ delamination lengths representing damage in the undamaged cone after impact and at near failure loads through the thickness of the test coupons. Figure 7: T4 material system, µCT cross-section at the impact region after impact, after application of near failure load and after failure. (i) represents delamination growth into the undamaged cone detected after failure and (ii) represents a ~0.3 mm out-of-plane deformation caused by the impact. Figure 8: T1 material system, µCT cross-section (a) showing load bearing 0° fibre fracture growth propagating longitudinally off a pre-existing region of fibre fracture after application of near failure compression load. This occurred on the sixth ply from the impact side. 3D segmentation of this fibre fracture is shown in red in (b) with neighbouring delaminations shown in yellow and blue representing the sixth and seventh ply respectively. Figure 9: µCT cross-section at the impact site, red arrow indicating impact location, and white arrows indicating CAI loading direction. (i) undamaged cone, (ii) delamination growth into the undamaged cone and (iii) sublaminate buckling. Figure 10: Schematic showing (a) the unsupported length of the sublaminate ‘L’ and delamination growth into the impact cone and (b) more than doubling of the unsupported length due to delamination growth. Figure 11: (a) schematic illustrating the effects of bridging ligaments on the sublaminates and (b) a high resolution SRCT image of the T4 system showing the presence of these ligaments within the delaminated region. Table 1: Sequence of compressive load steps applied for each of the material systems tested. Step 1 Step 2 Step 3 Step 4 Step 5 Step 6 UT 87% 94% 101% 102% 103% 100% (Failed) T1 82% 95% 101% 102% 107% 100% (Failed) T3 94% 99% 102% 100% 100% (Failed) T4 94% 98% 97% 99% 100% (Failed)