Experimental observations of fatigue damage in cross-ply laminates using carbon/epoxy ultra-thin plies

Experimental observations of fatigue damage in cross-ply laminates using carbon-epoxy ultra-thin plies Serafín Sánchez-Carmona*, Elena Correa, Alberto Barroso, Federico París Grupo de Elasticidad y Resistencia de Materiales, Departamento de Mecánica de Medios Continuos y Teoría de Estructuras, Escuela Técnica Superior de Ingeniería, Universidad de Sevilla, Camino de los Descubrimientos, 41092, Sevilla, España * Phone number: +34662200786; Email: [email protected] ABSTRACT The damage mechanisms of cross-ply laminates under fatigue loading have been deeply studied during last years. In the case of quasi-static loads the so-called scale effect has recently acquired renewed importance after the appearance of ultra-thin plies. In this work, the experimental observation of the onset of transverse damage and its progression in two different cross-ply laminates under tension-tension cyclic loading is performed. First, a [04/903/04] laminate made of plies with conventional thickness (150g/m2) is analysed. After looking for the expected fatigue damage mechanisms in this case, a [04/90/04] laminate with a single ultra-thin 90⁰ ply (30g/m2) is studied. All specimens have been carefully examined by means of an optical microscope, detecting the expected damage mechanisms for [04/903/04] and non-conventional types of damages for [04/90/04]. The change of fatigue damage mechanisms for cross-ply laminates involving an ultra-thin 90⁰ ply is microscopically detailed, meaning the evidence of the delay of transverse damage onset in the 90º layer under cyclic loading. This evidence shows the existence of a scale effect concerning the appearance of the fatigue damage mechanisms. Keywords: cross-ply laminate; fatigue damage mechanisms; transverse cracking; ultrathin ply; scale effect 2 1. Introduction The use of composite materials in aeronautical industry has been remarkably developed during the last years due to the necessity of more lightweight aircrafts in order to both manage the transport of a larger number of passengers and reduce the quantity of fuel. It is important to know the service life of composites to reach the most possible reliable design, hence, damage and failure phenomena which affect composite materials must be identified. An important type of load closely linked with composite failure is fatigue, so it is needed to comprehend how cyclic loading impairs this type of material, causing its degradation and even its final failure. In this way, fatigue damage mechanisms in carbon fibre reinforced polymers (CFRP) have been deeply studied during last years although, nowadays, still remain as an open research topic. Hahn and Lorenzo [1], Berthelot [2], Harris [3] and Bathias [4], among others, show a comprehensive review on the dependence of the appearance of fatigue damage mechanisms on the material features, such as the plies thicknesses and the stacking sequence, and, on the testing conditions such as the stress ratio, the applied stress level and the applied loading direction. There are noteworthy explanations to describe how fatigue failure is due to the accumulation of damage mechanisms, such as fibre/matrix interface debonding, matrix cracking in off-axis plies, delaminations, splitting in on-axis plies (which describes matrix cracking parallel to the fibre direction) and fibre breakage. One of the most studied configurations is the cross-ply laminates, [0m/90n]s, as can be deduced from the investigations of several research groups (Lafarie-Frenot and coworkers [5-9], Takeda and co-workers [10-11] , El Mahi et al. [12], Yokozeki et al. [13], Tohgo et al. [14], Taheri-Behrooz et al. [15], Hosoi et al. [16], Amacher et al. [17], Shen et al. [18], Carraro et al. [19-20] and Corbetta et al. [21]), although multidirectional 3 laminates, specially quasi-isotropic laminates, have also been analysed profoundly ([22]– [28]). The main failure mechanisms for cross-ply laminates under fatigue loading are illustrated in Figure 1. Transverse cracking, delaminations and fibres fracture are the main fatigue mechanisms which occur in cross-ply laminates, appearing mainly in this order, although longitudinal cracking may also appear combined with fibre breakage. Even under certain conditions, fibre peeling is referred to as another failure mechanism causing the final failure (Tohgo et al. [29]) and which could be due to the combination of longitudinal cracking in the 0º plies and fibres fracture. This mechanism appears on the sample edges and evolves through the complete specimen width while the number of cycles increases, also leading to the final fracture of the laminates. The first and the most relevant damage mechanism during the fatigue life of a cross-ply laminate is transverse cracking. This mechanism appears and evolves up to reaching a distinctive saturation level, which describes its characteristic damage state (CDS), a physical situation well defined in [30-31] and widely manifested in the previous studies from the different research groups. CDS is a unique and distinct situation for a given laminate in which each off-axis layer suffers damage up to reach a saturation level. CDS mainly depends on the laminate properties, including its stacking sequence, and it is the stable damage state which is formed prior to catastrophic failure. According to the observation of cross-ply laminates behaviour, no more additional transverse cracks (TCs) appear in the 90º layers when CDS has been reached. In the first stages of the appearance of transverse damage when conventional thickness plies are involved (greater than 150µm), only delamination traces are induced at the tip of transverse cracking close to the 0º/90º interfaces. Approximately at the same time as a laminate reaches its CDS, full delaminations between 0º and 90º layers start to appear, 4 increasing considerably afterwards and thus implying a detrimental effect on the laminate mechanical performance. Finally, fibre breakage occurs in 0º layers up to reaching the final failure of the specimens, i.e., reaching their fatigue life. During the last decades a special feature of the behaviour of composite materials, the scale effect, has attracted the attention of the scientific community. This phenomenon has always been related to the layer thicknesses reduction which has implied changes in the behaviour of the composite material. Many authors have attempted to provide an explanation for this phenomenon, from Parvizi et al. [32] and Flaggs and Kural [33] up to París et al. [34-35], among many others. In these recent publications [34-35], the scale effect has been physically explained based on the quasi-static damage mechanisms involved in failure, showing how the first damage occurring in the weakest lamina is less detrimental when decreasing its thickness and finding that the damage mechanism also changes as the thickness decreases. Furthermore, based on an energetic approach, the appearance of a progressive mechanism of damage in the case of ultra-thin plies versus an explosive mechanism of damage for thick plies is explained. According to this phenomenon but in connection with cyclic loads, some authors, such as Lafarie-Frenot and co-workers [5-9], have expounded that thinner 90º plies in crossply composites improve their fatigue performance under certain conditions, naming this as size effect of the 90º layer (using it as an alternative term for the scale effect). This improvement has mainly been analysed with conventional ply grammages, i.e., greater than 120 g/m2. Note that the term size effect can be also found in literature referring to a completely different phenomenon such as that accounting for the size of the coupons (see for instance, Zhu et al [36] for metals).To the best authors’ knowledge all the experimental studies related to the scale effect and composites laminates involving 5 ultra-thin plies (UTPs), plies with grammages below 120 g/m2, have been developed under quasi-static testing and only two studies, Sihn et al. [37] and Kötter et al.[38], have provided information about the scale effect under fatigue testing. In reference [37], undamaged and open-hole samples of CFRP unidirectional and quasi-isotropic laminates have been analysed, whereas in reference [38] undamaged and low-energy impacted samples of CFRP quasi-isotropic laminates were tested. Both experimental works, the former by means of X-ray and the latter using computed tomography, have shown that fatigue performances improve as thinner layers are used. Particularly, important improvements are achieved in [38] under tension-compression fatigue loading (R=-0.5, R being defined as the minimum applied stress divided by the maximum applied stress), matrix cracking and delaminations being significantly supressed. In addition to this, impact-induced delaminations do not spread in thin-ply layers under cyclic loading with R=-0.5 ([38]); although reported for both studies, [37-38], the projected delaminations areas were larger after impact. Nevertheless, these two studies, [37-38], do not deal with the detail observation of damage mechanisms occurring at the micromechanical level and the damages densities evolution during fatigue load. The encouraging results about the scale effect and the appearance of UTPs have motivated the planning of the present experimental work, which consists in the observation of the fatigue damage mechanisms appearing in cross-ply laminates with the 90º layer of both conventional and ultra-thin plies. It is worth highlighting that this investigation does not pursue a massive experimental testing program, instead the aim has been to microscopically evaluate the fatigue damage mechanisms in the most exhaustive way. The objective of the present work is to explore in detail which are the differences in the generation of the mechanisms of damage, if existent, that could lead 6 to an explanation of the scale effect under tension-tension (T-T) fatigue loading in cross-ply laminates. 2. Experimental programmes In the search for an explanation of the scale effect under tensile quasi-static loading, París et al [34] performed an experimental programme to detect the appearance of the first damage in the 90º layer for different cross-ply laminates, covering from conventional thicknesses of the 90º layer to ultra-thin ones. In the present work, in an attempt to follow a similar procedure, two different cross-ply laminates, [04/903/04] and [04/90/04], have been selected to compare the onset and progress of damages under cyclic loading. In what follows, [04/903/04] (with 3 plies of 150g/m2 prepreg in the 90º layer), is referred to as CP-450, and [04/90/04] (with a ply of 30g/m2 prepreg in the 90º layer) as CP-30. 0º plies are exactly the same in both laminates, using 150g/m2 prepreg. With the aim of investigating how scale effect alters the damage mechanisms under fatigue loading, it is important to know to which stress the 90º layer is subjected in each laminate for the force levels selected for fatigue testing. The maximum stress of the 90º layer for each laminate (𝜎22,𝑚𝑎𝑥 𝐶𝑃−450, 𝜎22,𝑚𝑎𝑥 𝐶𝑃−30 ) at each force level has been calculated using the Classical Lamination Theory (CLT). Analytical calculations both for CP-450 and CP-30 have been carried out using the thermo-elastic properties obtained experimentally from the quasi-static characterisation of the 150g/m2 and 30g/m2 prepregs, respectively [39]. Furthermore, it is interesting to point out that thermal stresses have also been taken into consideration during 𝜎22,𝑚𝑎𝑥 calculations. The thermal decrease after the curing process of composite laminates has an influence on their residual stress state, appearing at 3 different levels. First, the whole 90° layer suffers a thermal residual stress due to the different thermal expansion coefficients of each layer at mesoscopic level as it is 7 theoretically expounded, e.g., in Shokrieh M.M. [40]; these stresses can be calculated using the Classical Laminate Theory. Second, associated with the samples free-edge there exists a thermal stress along the thickness as Pagano et al. [41] showed. Third, the fibres and the matrix also have different thermal expansion coefficients so thermal residual stresses at micromechanical level arise as Correa et al. [42] detailed. Taking these influences into account the residual thermal contribution has been partially discarded for the 𝜎22,𝑚𝑎𝑥 selected for the discussion in the results section. The reasons for not considering the whole contribution of the thermal stresses have been the previous authors’ experiences, Barroso et al. [43], and a new study in which the thermal stresses relaxation has been experimentally analysed using the same materials than the present work in two different antisymmetric laminates [04/904] (one fully manufactured with 100g/m2 prepreg and the other one with 30g/m2 prepreg). The noticeable difference between the two different prepregs is seen in Figure 2, in which b1) shows the curvature radius of the laminate, R (cm), used to measure the thermal stress relaxation. In Figure 3, this curvature radius versus the time after the curing process (h) is charted, showing the considerable relaxation, which is more marked for the ultra-thin prepreg CP-30. Based on these results, the thermal stresses considered have only been the 40% of those calculated by means of CLT. 2.1. Specimens’ preparation Both cross-ply laminates have been manufactured using carbon/epoxy prepregs TP402/T700S from North Thin Ply Technology. Whereas all 0º plies have been manufactured with the 150g/m2 prepreg for both cross-ply laminates and it has also been employed for the 90º plies in CP-450, the 30g/m2 prepreg has been used for the 90º ply in CP-30. 8 These laminates have been manufactured by means of a hand lay-up in a clean room and cured in an autoclave using a vacuum bag. Once cured, glass-epoxy tabs have been bonded to both panel sides in the gripping zones and, afterwards, the panels have been cut by means of a water refrigerated diamond disc saw as recommended in ASTM D3039 [44]. The specimen dimensions are shown in Figure 4, with Ls=180mm, w=25mm and Lt=55mm. All coupons have been sanded and polished on their lateral free edge to be able to observe the damage by means of an optical microscope (ex-situ inspections). Additionally, a replica system in the same lateral polished edge has been used in which the replica sample can be also observed in the optical microscope (in-situ inspections), combining them with ex-situ inspections. The sanding and polishing process is almost the most significant step during these experimental programmes because of the importance of detecting precisely the first damage that occurs in 90º layer, in which inter-fibre/matrix cracks (debonds) can be involved, and thus dirtiness or certain marks could cast doubt on the experimental observations of the sample free edge after cyclic loading. With this issue in mind, a tailor-made device for the sanding and polishing process (Figure 5) has been developed. This device assures that the specimen is kept in vertical position whereas the user can side-to-side move it manually (green arrow in Figure 5.a), applying certain pressure in both sample ends (red circles in Figure 5.a) while the sanding and polishing disc is rotating (blue arrow in Figure 5.a). 2.2. Tests performance Before fatigue testing, five specimens from each laminate have been quasi-statically tested, obtaining their ultimate tensile force (Fu) and the apparent elastic modulus (EL) 9 (Table 1). These tests have been carried out in an Instron 4482 electromechanical testing machine, controlling the displacement at a speed of 1mm/min. Dynamic tests have been carried out in an Instron 8802 hydraulic testing machine, controlling the load by means of a sinusoidal function with a force ratio R=0.1 (R being defined as the minimum applied force divided by the maximum applied force), using combined frequencies of 1 and 3 Hz. The different testing frequencies have been used to accelerate the number of cycles applied after 104 cycles, using a thermocouple to ensure that the specimen temperature increase during cyclic loading has been low and therefore, undesirable changes of the material behaviour have been avoided. Different force levels have been selected to perform fatigue testing, 0.6, 0.4, 0.3, 0.2 and 0.16 of their ultimate tensile force (Fu). Table 2 shows 𝜎22,𝑚𝑎𝑥 for each cross-ply laminate at each force level selected, first without considering thermal stresses and second taking the 40% of thermal stresses into account. Whereas the highest force level has been selected as a scenario in which transverse damage is going to be relevant from the first cycle, at least for CP-450 as it is known from literature, the lowest force level has been chosen as the situation in which the 90º layer has a stress level below its transverse strength, which is around 30-35 MPa for the material used in the present work [39], in order to look at a more progressive appearance of transverse damage. 3. Results: Development of fatigue damage The experimental results of each cross-ply laminate have been analysed through the damage density (ρD) versus the number of cycles, defining ρD as the number of damages divided by a gauge length of 50 mm. 3.1. Typology of observed damages 16 It is worth pointing out that a damage mechanism change has been found for each force level tested when the 90º layer thickness has decreased. First, the main transverse damage in the 90º layer for CP-30 is GTD instead of TC as occurs for CP-450, as it is detailed using the energetic approach in París et al [34-35]: when the first ID can appear, CP-450 has the sufficient energy to directly produce a complete TC, whereas CP-30 has not enough energy to produce a complete transverse damage, mainly showing IDs or I-CDs up to become GTDs without reaching the whole 90º layer. And, second, a relevant delay of transverse damage appearance has been found when the 90º layer has an ultra-thin thickness. Whereas CP-450 shows the onset of conventional transverse cracks (TCs) and their subsequent density increases along its free edge (as it is well reported in literature and it has been obtained in this work), CP-30 shows an evolution of different forms of transverse damages up to evolving to generalised transverse damages (GTDs). In other words, the onset and progress of transverse damage in CP-30 seems to be more progressive, starting as isolated debonds (IDs) and/or isolated-continued debonds (I-CDs) up to becoming GTDs, without evolving to a full TC as it could be expected. This second relevant point is a direct consequence of the physical explanation of the scale effect previously provided by París et al [34-35]. It is important to highlight two relevant facts: only 10 GTDs have been detected for CP- 30 at 0.2Fu appearing between 105 and 106 cycles, and few GTDs have kinked from 6·104 to 105 cycles when the GTD level started to saturate, keeping almost constant up to 106 cycles. It is worth pointing out that I-CDs have been detected for CP-450 at 0.2Fu, although their evolution to TCs has been less progressive and much faster than for CP-30. This conclusion is deduced from both comparisons, in terms of either force or stress level. 17 Neither CP-450 nor CP-30 at the lowest force level tested have shown any type of damage up to either 104 or 6·104 cycles, respectively. This fact seems to corroborate the non-consideration of the whole residual thermal stresses during CLT calculations. Once the 𝜎22,𝑚𝑎𝑥 reaches the material transverse strength, YT, the 90º layer must show some transverse damage. In this specific case, the samples tested at 𝜎22,𝑚𝑎𝑥 ≈YT have not presented any transverse damage after a high number of load cycles. Hence, this absence of transverse damage would be incoherent if this 𝜎22,𝑚𝑎𝑥 actually present were even higher considering the total thermal stresses. 4.2. Considerations on longitudinal damages Focusing on the longitudinal damage in 90º layer (LD), it is important to highlight that the number of LDs is remarkable in relation to the number of GTDs; nevertheless, LDs seem not to be induced by transverse damage so its profound analysis will be performed separately in an ongoing publication. Finally, delaminations are clearly delayed, or seem to be even supressed, when the 90º layer has an ultra-thin thickness. Whereas full delaminations appear in a considerable number for CP-450 after TCs reach CDS, just a few shorts DLs (that indicate the appearance of future full delaminations) have been detected for CP-30 (even at the highest force level tested when its GTD level starts to saturate). 5. Conclusions Specimens from two carbon/epoxy cross-ply laminates: [04/903/04] (named as CP-450 and manufactured from 150 g/m2 prepreg) and [04/90/04] (named as CP-30 and manufactured from a 30g/m2 prepreg for the 90º layer and a 150g/m2 for the 0º layers) have been tested under T-T cyclic loading at different force levels (0.6, 0.4, 0.3, 0.2 and 0.16 of their ultimate tensile force, Fu). The study has focused on the exhaustive experimental microscopic observations of the fatigue damage appearing in each 18 laminate, mainly centring on the differences encountered in the damages detected in the 90º layer as the thickness decreases. The most highlighting fact is the noticeable difference of fatigue damage appearance detected between CP-450 and CP-30. First, the most relevant result is the inexistence of a full conventional TC for CP-30 at any of the force levels tested, GTD being the greatest transverse damage observed. And second, the appearance of transverse damage is more progressive when the 90º layer thickness decreases (as has been detailed with the appearance of IDs and I-CDs evolving to GTDs for CP-30), a fact related to the delay encountered for the transverse damages, observing even the suppression of delaminations. The onset of the comparable transverse damage (TCs for CP-450 vs GTDs for CP-30) seems to slightly delay when the 90º layers stress is the same. Nevertheless, the appearance of the first transverse damage in each laminate, I-CD, is produced approximately at the same number of cycles when the 90º layer has the same stress level, as it could be deduced from París et al. [34-35]. The results obtained have led to conclude that transverse damage has substantially decreased for CP-30 in comparison with CP-450 and, consequently, the next expected damage (delaminations) changes its pattern too. Therefore, the present results show that a scale effect, concerning the appearance of the damage mechanisms in cross-ply laminate under fatigue loading, exists. The effect identified in the actual investigation, after observing mainly the first damage mechanism which appear, can be understood with the physically based explanation about the phenomenon called scale effect, profoundly studied under quasistatic-loading as it is expounded in [34-35]. More experimental tests involving intermediate thicknesses and analysing the subsequent damage mechanisms, particularly focusing on 19 delaminations, need to be performed to completely define the extension of the scale effect under fatigue loading. Acknowledgements This work was supported by the Universidad de Sevilla [grant number VIPPIT-2018- II.2] and Consejería de Economía y Conocimiento Junta de Andalucía [project number PY18-3360] corresponding to the I+D+I Programa Operativo FEDER 2014-2020. The authors would rather appreciate to Dr. Jesús Justo Estebaranz and Dra. María Luisa Velasco López for their advice during the definition of fatigue tests and microscopic observations, respectively, and to Mr. Roberto Prieto García for his collaboration in some experimental tasks. Data availability The raw/processed data required to reproduce these findings cannot be shared at this time as the data also forms part of an ongoing study. References [1] H. T. Hahn and L. Lorenzo, Fatigue Failure Mechanisms in Composite Laminates. International Congress on Fracture (ICF), 1984. doi: 10.1016/b978-1- 4832-8440-8.50041-7. [2] J.-M. Berthelot, “Transverse cracking and delamination in cross-ply glass-fiber and carbon-fiber reinforced plastic laminates: Static and fatigue loading,” Appl Mech Rev, vol. 56, no. 1, pp. 111–147, Jan. 2003, doi: 10.1115/1.1519557. [3] B. Harris, Fatigue in composites. Woodhead Publishing, 2003. doi: 10.1533/9781855738577. [4] C. Bathias, “An engineering point of view about fatigue of polymer matrix composite materials,” Int J Fatigue, vol. 28, no. 10 SPEC. ISS., pp. 1094–1099, 2006, doi: 10.1016/j.ijfatigue.2006.02.008. 20 [5] M. C. Lafarie-Frenot and C. Henaff-Gardin, “Formation and growth of 90° ply fatigue cracks in carbon/epoxy laminates,” Compos Sci Technol, vol. 40, no. 3, pp. 307–324, 1991, doi: 10.1016/0266-3538(91)90087-6. [6] C. Henaff-Gardin, M. C. Lafarie-Frenot, J. Brillaud, and A. El Mahi, “Influence of the stacking sequence on fatigue transverse ply cracking in cross-ply laminates,” Damage Detection in Composite Materials, vol. 1128, pp. 236–255, 1992. [7] M. C. Lafarie-Frenot, C. Henaff-Gardin, and D. Gamby, “Matrix cracking induced by cyclic ply stresses in composite laminates,” Compos Sci Technol, vol. 61, no. 15, pp. 2327–2336, 2001, doi: 10.1016/S0266-3538(01)00125-7. [8] D. Gamby, C. Henaff-Gardin, and M. C. Lafarie-Frenot, “Propagation of edge cracking towards the centre of laminated composite plates subjected to fatigue loading,” Compos Struct, vol. 56, no. 2, pp. 183–190, 2002, doi: 10.1016/S0263- 8223(02)00004-1. [9] F. Foti, M. Gigliotti, Y. Pannier, D. Mellier, and M. C. Lafarie-Frenot, “The effect of the environment on high temperature fatigue of cross-ply C/epoxy laminated composites,” Compos Struct, vol. 202, no. April, pp. 924–934, 2018, doi: 10.1016/j.compstruct.2018.04.065. [10] N. Takeda, S. Ogihara, and A. Kobayashi, “Microscopic fatigue damage progress in CFRP cross-ply laminates,” Composites, vol. 26, no. 12, pp. 859–867, 1995, doi: 10.1016/0010-4361(95)90879-5. [11] S. Kobayashi and N. Takeda, “Experimental and analytical characterization of transverse cracking behavior in carbon/bismaleimide cross-ply laminates under mechanical fatigue loading,” Compos B Eng, vol. 33, no. 6, pp. 471–478, 2002, doi: 10.1016/S1359-8368(02)00028-8. [12] A. El Mahi, J. M. Berthelot, and J. Brillaud, “Stiffness reduction and energy release rate of cross-ply laminates during fatigue tests,” Compos Struct, vol. 30, no. 2, pp. 123–130, 1995, doi: 10.1016/0263-8223(94)00061-1. [13] T. Yokozeki, T. Aoki, and T. Ishikawa, “Fatigue growth of matrix cracks in the transverse direction of CFRP laminates,” Compos Sci Technol, vol. 62, no. 9, pp. 1223–1229, 2002, doi: 10.1016/S0266-3538(02)00068-4. [14] K. Tohgo, S. Nakagawa, and K. Kageyama, “Fatigue behavior of CFRP cross-ply laminates under on-axis and off-axis cyclic loading,” Int J Fatigue, vol. 28, no. 10 SPEC. ISS., pp. 1254–1262, 2006, doi: 10.1016/j.ijfatigue.2006.02.011. [15] F. Taheri-Behrooz, M. M. Shokrieh, and L. B. Lessard, “Residual stiffness in cross-ply laminates subjected to cyclic loading,” Compos Struct, vol. 85, no. 3, pp. 205–212, 2008, doi: 10.1016/j.compstruct.2007.10.025. [16] A. Hosoi, K. Kurihara, N. Sato, and H. Kawada, “Prediction of first transverse crack formation in cross-ply CFRP laminates under fatigue loading,” in 8th Asian- Australasian Conference on Composite Materials 2012, ACCM 2012 - Composites, Kuala Lumpur, Malaysia, 6-8 Nov, 2012, pp. 173–178. 21 [17] R. Amacher, J. Cugnoni, J. Botsis, L. Sorensen, W. Smith, and C. Dransfeld, “Thin ply composites: Experimental characterization and modeling of sizeeffects,” Compos Sci Technol, vol. 101, pp. 121–132, 2014, doi: 10.1016/j.compscitech.2014.06.027. [18] H. Shen, W. Yao, W. Qi, and J. Zong, “Experimental investigation on damage evolution in cross-ply laminates subjected to quasi-static and fatigue loading,” Compos B Eng, vol. 120, pp. 10–26, 2017, doi: 10.1016/j.compositesb.2017.02.033. [19] P. A. Carraro, L. Maragoni, and M. Quaresimin, “Prediction of the crack density evolution in multidirectional laminates under fatigue loadings,” Compos Sci Technol, vol. 145, no. October, pp. 24–39, 2017, doi: 10.1016/j.compscitech.2017.03.013. [20] P. A. Carraro, L. Maragoni, and M. Quaresimin, “Characterisation and analysis of transverse crack-induced delamination in cross-ply composite laminates under fatigue loadings,” Int J Fatigue, vol. 129, no. August, p. 105217, 2019, doi: 10.1016/j.ijfatigue.2019.105217. [21] M. Corbetta, C. Sbarufatti, M. Giglio, A. Saxena, and K. Goebel, “A Bayesian framework for fatigue life prediction of composite laminates under co-existing matrix cracks and delamination,” Compos Struct, vol. 187, no. December 2017, pp. 58–70, 2018, doi: 10.1016/j.compstruct.2017.12.035. [22] J. Tong, F. J. Guild, S. L. Ogin, and P. A. Smith, “On matrix crack growth in quasi-isotropic laminates - I. Experimental investigation,” Compos Sci Technol, vol. 57, no. 11, pp. 1527–1535, 1997, doi: 10.1016/S0266-3538(97)00080-8. [23] N. Takeda, S. Kobayashi, S. Ogihara, and A. Kobayashi, “Effects of toughened interlaminar layers on fatigue damage progress in quasi-isotropic CFRP laminates,” Int J Fatigue, vol. 21, no. 3, pp. 235–242, 1999, doi: 10.1016/S0142- 1123(98)00072-3. [24] A. Hosoi, N. Sato, Y. Kusumoto, K. Fujiwara, and H. Kawada, “High-cycle fatigue characteristics of quasi-isotropic CFRP laminates over 108 cycles (Initiation and propagation of delamination considering interaction with transverse cracks),” Int J Fatigue, vol. 32, no. 1, pp. 29–36, 2010, doi: 10.1016/j.ijfatigue.2009.02.028. [25] G. Just, I. Koch, M. Brod, E. Jansen, M. Gude, and R. Rolfes, “Influence of reversed fatigue loading on damage evolution of cross-ply carbon fibre composites,” Materials, vol. 12, no. 7, pp. 1–21, 2019, doi: 10.3390/ma12071153. [26] H. Ben Kahla, J. Varna, and A. Pupurs, “Microcracking in layers of composite laminates in cyclic loading with tensile transverse stress component in layers,” in ICCM20 - 20th International Conferences on Composite Materials, 2015, no. July, pp. 19–24. 22 [27] P. A. Carraro and M. Quaresimin, “Fatigue damage and stiffness evolution in composite laminates: A damage-based framework,” Procedia Eng, vol. 213, no. 2017, pp. 17–24, 2018, doi: 10.1016/j.proeng.2018.02.003. [28] H. Ben Kahla, Z. Ayadi, F. Edgren, A. Pupurs, and J. Varna, “Statistical model for initiation governed intralaminar cracking in composite laminates during tensile quasi-static and cyclic tests,” Int J Fatigue, vol. 116, no. January, pp. 1–12, 2018, doi: 10.1016/j.ijfatigue.2018.05.030. [29] K. Tohgo, S. Nakagawa, and H. Araki, “Damage behavior and life prediction in CFRP cross-ply laminates under fatigue loading,” Nihon Kikai Gakkai Ronbunshu, A Hen/Transactions of the Japan Society of Mechanical Engineers, Part A, vol. 72, no. 7, pp. 1041–1048, 2006, doi: 10.1299/kikaia.72.1041. [30] K. L. Reifsnider and A. Talug, “Analysis of fatigue damage in composite laminates,” Int J Fatigue, vol. 2, no. 1, pp. 3–11, 1980, doi: 10.1016/0142- 1123(80)90022-5. [31] C. Li, F. Ellyin, and A. Wharmby, “A damage meso-mechanical approach to fatigue failure prediction of cross-ply laminate composites,” Int J Fatigue, vol. 24, no. 2–4, pp. 429–435, 2002, doi: 10.1016/S0142-1123(01)00098-6. [32] A. Parvizi, K. W. Garrett, and J. E. Bailey, “Constrained cracking in glass fibrereinforced epoxy cross-ply laminates,” J Mater Sci, vol. 13, no. 1, pp. 195–201, 1978, doi: 10.1007/BF00739291. [33] D. L. Flaggs and M. H. Kural, “Experimental Determination of the In Situ Transverse Lamina Strength in Graphite/Epoxy Laminates,” J Compos Mater, vol. 16, no. 2, pp. 103–116, Mar. 1982, doi: 10.1177/002199838201600203. [34] F. París, M. L. Velasco, and E. Correa, “The scale effect in composites: An explanation physically based on the different mechanisms of damage involved in failure,” Compos Struct, vol. 257, p. 113089, Feb. 2021, doi: 10.1016/j.compstruct.2020.113089. [35] F. París, M. L. Velasco, and E. Correa, “Chapter 9: Modelling fibre–matrix interface debonding and matrix cracking in composite laminates,” Multi-Scale Continuum Mechanics Modelling of Fibre-Reinforced Polymer Composites, pp. 243–274, Jan. 2021, doi: 10.1016/B978-0-12-818984-9.00009-3. [36] S.-P. Zhu, Y. Ai, D. Liao, J. A. F. O. Correia, A. M. P. De Jesus, and Q. Wang, “Recent advances on size effect in metal fatigue under defects: a review,” Int J Fract, vol. 234, no. 1, pp. 21–43, 2022, doi: 10.1007/s10704-021-00526-x. [37] S. Sihn, R. Y. Kim, K. Kawabe, and S. W. Tsai, “Experimental studies of thin-ply laminated composites,” Compos Sci Technol, vol. 67, no. 6, pp. 996–1008, 2007, doi: 10.1016/j.compscitech.2006.06.008. [38] B. Kötter, J. Endres, J. Körbelin, F. Bittner, H. J. Endres, and B. Fiedler, “Fatigue and fatigue after impact behaviour of Thin- and Thick-Ply composites observed by computed tomography,” Composites Part C: Open Access, vol. 5, no. March, 2021, doi: 10.1016/j.jcomc.2021.100139. 23 [39] S. Sánchez-Carmona, M. L. Velasco, A. Barroso, and E. Correa, “Thermomechanical characterisation data of 30 g/m2 and 150 g/m2 cured unidirectional carbon/epoxy tape prepreg TP 402/T700S,” Data Brief, vol. Submitted. [40] M. M. Shokrieh, “Residual Stresses in Composite Materials,” in Residual Stresses in Composite Materials (Second Edition), Second Edition., M. M. Shokrieh, Ed. Woodhead Publishing, 2021, pp. xi–xii. doi: https://doi.org/10.1016/B978-0-12- 818817-0.09990-X. [41] N. J. Pagano, G. A. Schoeppner, R. Kim, and F. L. Abrams, “Steady-state cracking and edge effects in thermo-mechanical transverse cracking of cross-ply laminates,” Compos Sci Technol, vol. 58, no. 11, pp. 1811–1825, Nov. 1998, doi: 10.1016/S0266-3538(98)00047-5. [42] E. Correa, V. Mantič, and F. París, “Effect of thermal residual stresses on matrix failure under transverse tension at micromechanical level: A numerical and experimental analysis,” Compos Sci Technol, vol. 71, pp. 622–629, 2011, doi: doi:10.1016/j.compscitech.2010.12.027. [43] A. Barroso, D. Vicentini, F. París, and V. Mantič, “Thermal Stresses in Composite Joints,” in Encyclopedia of Thermal Stresses, R. B. Hetnarski, Ed. Dordrecht: Springer Netherlands, 2014, pp. 5333–5344. doi: 10.1007/978-94-007-2739- 7_419. [44] “ASTM D3039/D3039M: Standard Test Method for Tensile Properties of Polymer Matrix Composite Materials,” Annual Book of ASTM Standards, pp. 1– 13, 2014, doi: 10.1520/D3039. [45] J. A. Glud, J. M. Dulieu-Barton, O. T. Thomsen, and L. C. T. Overgaard, “Automated counting of off-axis tunnelling cracks using digital image processing,” Compos Sci Technol, vol. 125, pp. 80–89, 2016, doi: 10.1016/j.compscitech.2016.01.019. 24 Figures Figure 1. Main damage mechanisms in cross-ply laminates under fatigue loading. a1) a2) b1) b2) Figure 2. Measurement of the thermal stress relaxation using the curvature radius (R) for [04/904] laminate manufactured with (a) 100 g/m2 and (b) 30 g/m2. Figures a1) and b1) just after curing process, and a2) and b2) after 1030 h. 25 Figure 3. Curvature radius versus time after the curing process for [04/904] fully manufactured with 100 g/m2 and 30 g/m2. Figure 4. Scheme of specimen and its dimensions. 32 Figure 13. DT, DL and GTD damage density (ρD) versus the number of cycles at 0.6Fu (CP-30). Figure 14. ID, I-CD and GTD damage densitiy (ρD) versus the number of cycles at 0.3Fu for the specimens tested (CP-30) 33 Figure 15. DT, DL and GTD damage density (ρD) versus the number of cycles at 0.3Fu for the specimens tested (CP-30). Figure 16. ID, I-CD and GTD damage density (ρD) versus the number of cycles at 0.2Fu (CP-30). 34 Figure 17. DT, DL and GTD damage density (ρD) versus the number of cycles at 0.2Fu (CP-30). Figure 18. GTD damage density (ρGTD) versus the number of cycles for 0.6Fu, 0.3Fu and 0.2Fu (CP30). 35 Figure 19. Comparison between the TC (CP-450) and GTD (CP-30) scaled damage densities (ρ*) at two tested force levels. Figure 20. Comparison between the TC (CP-450) and GTD (CP-30) scaled damage densities (ρ*) at the same 90º layer stress level. 36 Tables Table 1. Ultimate tensile force (Fu) and apparent elastic modulus (EL) for CP-450 and CP-30 laminates. CP-450 CP-30 Average SD CV Average SD CV Fu (kN) 67.54 ±3.15 4.67 % 64.02 ±4.30 6.71% EL (GPa) 79.42 ±2.13 2.68% 95.21 ±5.55 5.83% Table 2. Maximum transverse stress of the 90º layer (σ22,max) for CP-450 and CP-30 laminates at each force level tested. 𝝈𝟐𝟐,𝒎𝒂𝒙 𝑪𝑷−𝟒𝟓𝟎 (MPa) 𝝈𝟐𝟐,𝒎𝒂𝒙 𝑪𝑷−𝟑𝟎 (MPa) 0.6 Fu [86.53, 96.51] [79.38, 87.80] 0.4 Fu [57.68, 67.67] - 0.3 Fu [43.26, 53.25] [39.69, 48.11] 0.2 Fu - [26.46, 34.88] 0.16Fu [23.07, 33.06] - Table 3. Damages detected in the CP-30 sample at 0.6Fu after each increment of cyclic load (N). N ID I-CD GTD LD DT DL Damages 1.0E+00 0 9 1 0 0 0 10 5.0E+00 0 12 8 0 0 0 20 2.0E+01 35 36 15 6 0 0 92 5.0E+01 48 52 38 13 9 0 160 3.0E+02 89 148 89 18 14 0 358 1.0E+03 161 187 147 26 57 0 578 3.0E+03 186 188 216 45 98 0 733 1.0E+04 137 146 286 51 220 5 845 1.5E+04 70 106 323 64 250 21 834 3.0E+04 33 59 362 98 302 34 888 6.0E+04 9 51 380 105 348 70 963 1.0E+05 5 48 385 107 360 89 994 37 Table 4. Damages detected in the S1 CP-30 sample at 0.3Fu after each increment of cyclic load (N). N ID I-CD GTD LD DT DL Damages 1.0E+00 1 0 0 0 0 0 1 5.0E+00 1 1 0 0 0 0 2 2.0E+01 1 5 0 0 0 0 6 5.0E+01 1 6 0 0 0 0 7 1.0E+02 3 10 0 0 0 0 13 3.0E+02 6 12 0 0 0 0 18 1.0E+03 10 17 2 0 0 0 29 3.0E+03 24 36 5 0 0 0 65 2.0E+04 36 58 18 15 3 0 130 1.0E+05 13 81 41 39 6 2 182 Table 5. Damages detected in the S2 CP-30 sample at 0.3Fu after each increment of cyclic load (N). N ID I-CD GTD LD DT DL Damages 1.0E+00 0 0 0 0 0 0 0 1.0E+02 0 6 0 0 0 0 6 1.0E+03 1 15 0 0 0 0 16 3.0E+03 1 20 2 0 0 0 23 1.0E+04 2 26 7 7 0 0 42 1.5E+04 7 37 11 18 2 0 75 3.0E+04 12 56 22 34 3 0 127 5.0E+04 9 64 24 37 3 0 137 7.0E+04 6 69 29 40 4 0 148 1.0E+05 4 78 38 43 5 1 169 38 Table 6. Damages detected in the CP-30 sample at 0.2Fu after each increment of cyclic load (N). N ID I-CD GTD LD DT DL Damages 1.00E+00 0 0 0 0 0 0 0 5.00E+01 0 0 0 0 0 0 0 4.00E+02 0 0 0 0 0 0 0 3.00E+03 0 0 0 0 0 0 0 1.00E+04 0 0 0 0 0 0 0 3.00E+04 0 0 0 0 0 0 0 6.00E+04 0 4 0 0 0 0 0 1.00E+05 1 10 0 11 0 0 22 1.50E+05 8 24 2 22 0 0 56 3.00E+05 11 35 5 29 2 0 82 3.72E+05 12 43 6 36 2 0 99 4.50E+05 10 48 6 40 3 0 107 5.50E+05 8 50 7 44 3 0 112 7.00E+05 7 52 8 46 3 0 116 1.00E+06 3 54 10 46 4 0 117