Mechanical properties of a 3D braided carbon/epoxy composite

Transcription

1 Mechanical properties of a 3D braided carbon/epoxy composite Juan Pazmino 1, Valter Carvelli 1, Stepan Lomov 2, Alexander E. Bogdanovich 3, Dmitri D. Mungalov 3, Ignaas Verpoest 2 1 Department of Structural Engineering, Politecnico di Milano, Italy pazmino@stru.polimi.it, valter.carvelli@polimi.it 2 Department of Metallurgy and Materials Engineering, K. U. Leuven, Belgium stepan.lomov@mtm.kuleuven.be, ignaas.verpoest@mtm.kuleuven.be 3 3TEX Inc., Cary NC, USA bogdanovicha@3tex.com, mungalovd@3tex.com Keywords: 3D braided composite, Internal geometry, Mechanical properties, Fatigue life. SUMMARY. The present work describes an experimental study dealing with a 3D braided carbon/epoxy composite produced by 3TEX, Inc. The experimental activities aim to obtain the mechanical properties of the material and involve quasi-static tensile and tensile-tensile fatigue tests in the longitudinal (braiding) direction. Moreover, observations by micro-ct are performed to assess the quality of impregnation and to measure the internal geometry of the reinforcement. 1 INTRODUCTION Three-dimensional (3D) braided unitary fabrics have important advantages in many specific applications over conventional prepreg tape laminates and other types of 2D and 3D fabric used as preforms for composites ([1], [2]). They combine high conformability and drapability, suppression of delamination owed to the presence of through thickness reinforcement, improved damage tolerance, high impact resistance, shear stiffness and strength, and unique torsional rigidity. The advanced, automated computer controlled technology, like the recent novel 3D rotary braiding process and machines developed by 3TEX, Inc. [3], provide efficient and affordable manufacturing means for producing complex shape fabric preforms for composites. The broad range of industrial application of 3D braided composites ([4], [5]) requires an in-depth knowledge of their mechanical properties. Obtaining quasi-static and fatigue tensile properties of one specific 3D braided carbon/epoxy composite is the main topics of the present experimental work. In this paper the following experimental results are presented and discussed: - the observations and measurements of the shape of the yarns inside the composite by micro-ct images. - the pre-fatigue quasi-static tensile tests in the longitudinal (braiding) direction of the reinforcement with acoustic emission monitoring and strain mapping by DIC technique for the damage initiation and development detection; - the tensile-tensile cyclic tests for different stress levels in the longitudinal direction at the ratio R=0.1, providing the fatigue life curve and revealing the effect of fiber volume fraction on the fatigue life; - the quasi-static tensile tests of specimens cyclically loaded for different number of cycles in order to understand the influence on the mechanical properties of the damage imparted during subsequent fatigue loading.

2 2 MATERAIL FEATURES The fabric contains Toho Tenax 12K HTS carbon yarns. The preform material was produced on the 144-horngear, 576-carrier 3D rotary braider at 3TEX, Inc. (Figure 1) [3]. The manufacturing of 3D braided preform, having rectangular cross section, involved the total of 96 fiber carriers and the braiding pattern shown in Figure 2. The 3D braided preform has the fiber architecture depicted in Figure 3. The linear weight of the fabric is g/m. The braided composite samples were produced entirely at room temperature. The preform was first soaked with Epoxy West System 105 resin (209 hardener) and then placed in a special, 3TEX constructed [4], pressure mold with open ends aimed at releasing an excess of resin and air bubbles when the soaked preform is pressurized. The preform was kept in the mold under step-wise increasing pressure for the resin gelling time 50 min. After that the preform stayed in the mold for several hours at constant pressure which allowed the resin to fully cure. Then, fabricated composite in the form of bars having 50 cm length and 25.4 mm width were released from the mold. No post-cure was applied. Figure 1: 3D rotary braiding machine at 3TEX, Inc. with full 576-carrier fiber set up and axial fiber supply from the creel. Figure 2: 3D braiding pattern utilized for the rectangular cross section preform fabrication.

3 Figure 3: Schematic of the fiber architecture of the produced 3D braided preform. The first batch of material produced for this study had V f (fiber volume fraction) within the composite samples of 55.6% 0.4%. The void content was determined to be in the range between 0.8% and 2.3% with average 1.5%. The angle between the braiding yarns has been measured on the surface of the composite samples, and it appeared to be 10.0 ± 0.5. This angle is different from the true braid angle inside the preform which is defined as the angle between the braided fiber direction and the braid formation direction. The true braid angle was evaluated as 14. Other batches of the same material have been produced, using the above described process. In those samples, only the thickness of the composite (the V f value, respectively) has been varied. This allowed for investigating the influence of the fiber volume fraction on the mechanical properties of the composite. Table 1 presents the V f values of the investigated materials along with the measured average thickness and porosity. The fabricated 3D braided composite bars were cut along the length into test specimens for tension and fatigue testing; each specimen was 25 cm long. V f Thickness Porosity % [mm] % Table 1: Fiber volume fraction (V f ), thickness and porosity of the braided composite bars. 3 OBSERVATIONS AND MEASUREMENTS OF THE INTERNAL STRUCTURE The internal structure of the 3D braided carbon/epoxy resin composite was observed and

4 measured with micro computed tomography (micro-ct). The purpose was to experimentally characterize the internal geometry of the reinforcement, and thus obtain an important knowledge for numerical modeling [6] and gather information on the quality of the 3D braided fabric impregnation process. A Philips HOMX 161 X-ray system (Philips X-ray, Germany) with the AEA Tomohawk upgrade (AEA Technology, UK) was used for micro-ct images acquisition. Its detector system has an image intensifier TH 9428HX and a CCD camera (1024x1024 pixels) 12 bit dynamic range. A tube voltage of 65kV, a current of 0.54mA and an angular increment of 0.3 degree were adopted. The NRecon software of SkyScan NV (Kontich, Belgium) was used for the reconstruction of the micro-ct images. The observed portions of material (12 mm wide and 3.2 mm thick) allowed achieving a resolution of µm/pix and µm/pix for the specimens having 9 and 18 mm length in the braiding direction, respectively. The voxel size, which is determined by the distance between the X-ray source and the sample, was of 14.1 µm and 19.8 µm for the specimens of 9 and 18 mm length, respectively. The three-dimensional measurements on micro-ct images were performed with Data Viewer Software (SkyScan NV). Figure 4 shows some micro-ct pictures of the cross-section in three positions having 1 mm distance between each other in the braiding direction; z is the reference axis in this direction. These pictures allow to locate a yarn, to measure the yarn cross-section and to follow its path in the observed material portion. The assemblies of the measured cross-sections of the yarn in the specimens of 9 mm and of 18 mm length are shown in Figure 5 and Figure 6, respectively. These pictures illustrate the relevant variation of the yarn cross-section due to the compaction of the preform during resin infusion. The knowledge of the real shape of the yarns in the composite is an important input for a reliable simulation and prediction of the mechanical behaviour with finite element method. The quality of impregnation process was evaluated by means of micro-ct analysis. Voids inside the composite were detected in certain positions, then three-dimensionally reconstructed and measured. Voids inside the samples were identified with CTAn Software using binarized micro-ct images. The analysed portions of the specimens were 7.36 mm and mm long. In Figure 7, some voids detected in a material portion of length 7.36 mm are evident. Some measurements of the length of the regions, which were not filled with resin, are listed in Table 2. Figure 4: Some micro-ct images of the specimen cross-section at different position in the braiding direction.

5 (a) (b) Figure 5: (a) A yarn shape reconstruction for the 9 mm long specimen. (b) Position of the yarn (red circle) in the unit cell. (a) (b) Figure 6: (a) A yarn shape reconstruction for the 18 mm long specimen. (b) Position of the yarn (red circle) in the unit cell Figure 7: Identification of voids in the portion of length 7.36 mm. Position of the voids in the cross section (upper pictures) and 3D visualization (lower pictures). Void ID Estimated length [mm] Table 2: Length of some voids (see Figure 7).

6 4 FATIGUE BEHAVIOUR The response to fatigue loading of the 3D braided carbon/epoxy composite was investigated by means of tension-tension cyclic tests. The aim of the fatigue tests were: (i) to obtain the stress level at which complete failure did not occur during 5 million cycle loading ( 5m ); (ii) to generate the fatigue life curve ( -N) of the material with V f = 55.6%; (iii) to compare the fatigue life of 3D braided composites having different fiber volume fractions. Aluminum tabs, mm long and 2 mm thick, were used to grip the specimen in two hydraulic MTS testing machines. Fatigue tests were performed under constant stress amplitude, sinusoidal wave-form tensile-tensile loading with the ratio R = 0.1 (ratio of the minimum to the maximum stress in the cycle). The stress was evaluated by using the sample cross-sectional area averaged from three measurements. The frequency was set to 6 Hz. The fatigue tests of 3D braided composite with V f = 55.6% involved different maximum stress ( max ) levels in the cycle, ranging from 950 MPa (note that the respective average tensile static strength is 1351 MPa) to the stress level at which complete failure did not occur during 5 million cycle loading. The complete specimen failure was stated as the separation of the specimen into two parts during cyclic loading (as illustrated in Figure 8). At least three valid tests have been performed for each stress level. A cyclic test was considered valid if the specimen did not break in the tabs or too close to the tabs. Figure 8: Fatigue failure of a specimen. The response to cyclic tensile loading at different stress levels allowed us to determine the stress level 5m as 800 MPa for V f = 55.6%. The fatigue tests performed in the considered stress range, enabled for depicting a fatigue life curve (Wöhler-type diagram) for the braided composite with V f = 55.6%. In Figure 9a, the valid tests of the composite are represented by the maximum stress in the cycle ( max ) vs. the number of cycles to failure N. In the diagram, the average static tensile strength is also indicated in its correspondence to the extrapolated lowest number of cycles, N=1. Appropriate fitting equation was applied to the obtained experimental fatigue data. It provides reliable predictions of the fatigue life corresponding to the other stress levels, which were not investigated in the present experimental study. The diagram in Figure 9a shows the fitting of obtained experimental results (run-outs are not included) by means of a linear semi-logarithmic function (red curve). The results of tensile-tensile fatigue tests can be compared with those presented in [7] for a carbon fibre/epoxy UD composite having fiber volume fraction 55%. In order to perform a consistent comparison, the data from [7] and the data for the 3D braided composite were normalized to their respective static ultimate tensile strengths u (which are 3000 MPa and 1351 MPa). The comparison depicted in Figure 9b shows a similar fatigue behaviour of the carbon 3D braided and the unidirectional carbon fiber composites.

7 cycles to failure N run-out run-out (a) (b) Figure 9: (a) Fatigue life curve of the 3D braided carbon fiber composite (V f = 55.6%). (b) Comparison to the UD carbon fiber composite studied in [7]. 5x10 6 4x10 6 3x10 6 V f = 47.9% V f = 51.6% V f = 55.6% V f = 58.4% V f = 63.2% 2x10 6 1x10 6 0x max [MPa] Figure 10: Fatigue life of the 3D braided carbon fiber composite with different V f values. The effect of the fiber volume fraction on the fatigue life of the 3D braided composite materials was investigated on test specimens having five different V f values (see Table 1). A reduced number of specimens have been produced for composites having V f higher and lower than 55.6%. This set of cyclic tests was not performed for all those maximum stress levels which were used for the specimens with V f = 55.6%. The results of the cyclic tests of the composite materials with different V f are collected in the diagram of Figure 10 as the average number of cycles to failure (N) vs. the maximum stress in the cycle ( max ). The comparison in Figure 10 shows that when the V f value increases, the stress level 5m and the fatigue life for the considered maximum stress level in the cycle also increase.

8 5 RESIDUAL MECHANICAL PROPERTIES The residual fatigue strength and stiffness of the two composites were evaluated comparing pre- and post-fatigue traction tests. The post-fatigue tests were performed on the samples undergone a cyclic loading at the stress level 5m (800 MPa) for different number of cycles. The tension tests, accompanied with the monitoring of the damage initiation by means of acoustic emission and with optical measurement of the surface strain field of the sample, were performed in accordance with experimental methodology described in [8]. The damage initiation thresholds were determined by an acoustic emission system. Two AE sensors were situated at the boundaries of the gauge length region. AE signals were registered and processed by AE system AMSY-5 (Vallen Systems Gmbh). An optical extensometry was used to measure the true values of the strain in the centre of the sample (the gauge region was about 30 mm long). The Vic2D software (LIMESS Messtechnik und Software GmbH) was used for the evaluation of the full-field strain. The influence of the fiber content V f on the mechanical properties of the 3D braided composite, which was not previously fatigued, is presented in Figure 11. The initial elastic modulus (E), as well as the ultimate tensile strength ( u ), show their linear variation with the V f value (see Figure 11). The reduced number of specimens produced for composites having V f higher and lower than 55.6%, did not allow to get a statistically meaningful number of tests. For those V f, the standard deviation was not detailed in Figure 11. The stress vs. strain curves shown in Figure 12 provide a comparison of tensile test results for unfatigued and fatigued samples. The latter group contains samples subjected to cyclic loading with maximum stress 5m for a number of cycles of 1, 3 and 5 million. Some of the mechanical features extracted from these curves are compared in Figure 13. The initial elastic modulus (E) has a reduction with respect to the unfatigued material of 15% after the application of the cyclic loading for 5 million cycles. At the same time, the ultimate tensile strength ( u ) does not have a significant influence of the damage imparted during fatigue loading (the variation is within the experimental scatter band of the pre-fatigue tests). The registrations of acoustic emissions during some of the pre- and post-fatigue quasi-static tensile tests are summarized in Figure 14. The AE events of unfatigued samples were registered very early. The damage initiation threshold (shown by the left, narrow grey rectangular shadow in Figure 14, strain range %) is low in comparison with some other types of carbon/epoxy textile composites. For comparison, in [9] damage initiation and development was studied for triaxial 2D braided carbon composite laminates. In those tests, the initial AE events were registered at the strain level of %. The samples not cyclically loaded, at the applied stress levels of MPa, show another abrupt slope variation of the cumulative AE energy curves, i.e. another damage threshold (the right, wide grey rectangular shadow in Figure 14). This threshold is associated with the arrival of highenergy events which may be caused by a local splitting of the impregnated yarns and formation of the cracks located at the yarn boundaries. The graph in Figure 14 presents the cumulative AE energy curve for a specimen fatigued for 3 million cycles with a maximum stress of 5m. The curve shows significant variation of the cumulative AE energy when the stress is getting close to the maximum stress set during fatigue tests ( 5m = 800 MPa). The damage detected during quasi-static tensile tests below the stress level of 5m was not recorded, because it was fully imparted during the cyclic tests. The large part of the AE events appears in the present quasi-static tensile loading for the stress levels above the maximum stress applied during fatigue.

9 E [GPa] u [MPa] stress [MPa] E [GPa] u [MPa] V f [%] V f [%] (a) (b) Figure 11: Pre-fatigue mechanical properties vs. V f : (a) initial elastic modulus (E); (b) ultimate tensile strength ( u ) pre-fatigue after 1x10 6 cycles after 3x10 6 cycles after 5x10 6 cycles strain Figure 12: Stress vs. strain curves for the pre- and post-fatigue quasi-static tensile tests x x10 6 2x10 6 3x10 6 4x10 6 5x10 6 0x x10 6 2x10 6 3x10 6 4x10 6 5x10 6 number of cycles number of cycles (a) (b) Figure 13: Post-fatigue mechanical properties vs. the number of cycles: (a) initial elastic modulus (E); (b) ultimate tensile strength ( u ). 0

10 cumulative AE [a.u.] pre-fatigue after 3x10 6 cycles stress [MPa] Figure 14: Cumulative acoustic energy vs. applied stress. References [1] Tong L., Mouritz A.P., Bannister M.K., 3D fibre reinforced polymer composites, Elsevier, (2002). [2] Bogdanovich E., Mohamed M.H., Three-Dimensional Reinforcements for Composites, SAMPE Journal, 45, 8-28 (2009). [3] Mungalov D., Bogdanovich A., Automated 3-D Braiding Machine and Method, US Patent No. 6,439,096 B1, (2002). [4] Mungalov D., Bogdanovich A., Complex Shape 3-D Braided Composite Preforms: Structural Shapes for Marine and Aerospace. SAMPE Journal, 40, 7-20 (2004). [5] Mungalov D., Duke P., Bogdanovich A., High Performance 3-D Braided Fiber Preforms: Design and Manufacturing Advancements for Complex Composite Structures, SAMPE Journal, 43, (2007). [6] Lomov S. V., Ivanov D. S., Verpoest I., Zako M., Kurashiki T., Nakai H., Hirosawa S., Meso-FE modelling of textile composites: Road map, data flow and algorithms, Composites Science and Technology, 67, (2007). [7] Sims D., Fatigue test methods, problems and standards, in Fatigue in composites: Science and technology of the fatigue response of fibre-reinforced plastics, Harris B. Ed., Woodhead Publishing Ltd., Cambridge (2003). [8] Lomov S. V., Ivanov D. S., Truong Chi T., Verpoest I., Baudry F., Vanden Bosche K., Xie H., Experimental methodology of study of damage initiation and development in textile composites in uniaxial tensile test, Composites Science and Technology, 68, (2008). [9] Ivanov D.S., Baudry F., Van Den Broucke B., Lomov S.V., Xie H., Verpoest I., Failure analysis of triaxial braided composite, Composites Science and Technology, 69, (2009).