- Email: [email protected]
Impact and post-impact properties of a carbon fibre non-crimp fabric and a twill weave composite
Composites: Part A 41 (2010) 1019–1026
Contents lists available at ScienceDirect
Composites: Part A journal homepage: www.elsevier.com/locate/compositesa
Impact and post-impact properties of a carbon fibre non-crimp fabric and a twill weave composite Katleen Vallons *, Alex Behaeghe, Stepan V. Lomov, Ignaas Verpoest Dept. of Metallurgy and Materials Engineering, Katholieke Universiteit Leuven, Kasteelpark Arenberg 44 bus 2450, BE 3000 Leuven, Belgium
a r t i c l e
i n f o
Article history: Received 26 June 2009 Received in revised form 24 March 2010 Accepted 9 April 2010
Keywords: B. Impact behaviour B. Fatigue B. Mechanical properties
a b s t r a c t The impact and post-impact static and fatigue tensile properties of a carbon fibre/epoxy NCF composite were determined and compared to those of a carbon fibre/epoxy woven fabric composite, for two impact energies (3.5 and 7 J). The projected damage area after impact was larger for the NCF composite than that for the woven fabric composite for both impact energies. Impacted samples were subjected to static tensile tests and tensile–tensile fatigue tests. It was found that even a relatively low energy impact has already a significant negative influence on the residual properties in both static and fatigue tests, in the fibre direction as well as in the matrix dominated direction. In the matrix dominated directions the post-impact behaviour of the two materials is very similar. In the fibre direction, however, the properties of the non-crimp fabric composite are degraded more by an impact than those of the woven fabric composite. Ó 2010 Elsevier Ltd. All rights reserved.
1. Introduction Composite materials have a high in-plane stiffness- and strength-to-weight ratio, making them attractive materials for use in numerous transport applications, like cars and airplanes. One type of composites in particular, the non-crimp fabric composites, seems quite promising in this field, as non-crimp fabrics (NCF) are nearly as easy to handle as woven fabrics and also have a reasonable drapeability, while lacking the fibre crimp present in woven fabrics. This combination of advantages makes it possible to reduce the production cost and maintain excellent properties. In the recent past, extensive research has been conducted concerning non-crimp fabrics and their composites, and the differences between their behaviour and that of composites based on UD prepregs or woven fabrics. Godbehere et al. [1], for example, compared the tensile properties of NCF composites to those of UD prepreg composites and found a minor reduction in tensile strength but a decrease of about 5% in stiffness. Edgren et al. [2] investigated the influence of the stitching parameters on the fatigue performance of several unidirectional NCF composites, and compared the results to the behaviour of a UD prepreg laminate. They concluded that it appears as if a small stitch gauge in combination with a small stitch length can have a negative effect on the fatigue life. They also remarked, however, that overall, the fatigue life of all tested NCFCs did not differ much from the fatigue life of the prepreg laminate. * Corresponding author. Tel.: +32 1 6321193; fax: +32 1 6321990. E-mail address: [email protected] (K. Vallons). 1359-835X/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.compositesa.2010.04.008
Some other interesting studies concerning the properties of NCF composites have been published lately, e.g. by Edgren et al., who studied the behaviour of NCF composites under a combined compression and shear loading, and developed an approximate analytical constitutive model [3,4]. In [5], they focussed primarily on the formation of damage and the influence of this damage on the composite properties during tensile tests on cross-ply NCF composites. A series of seven papers has recently been published by the present authors, covering a wide variety of properties of both dry NCF fabrics and their composites [6–12]. In parts 1, 2, 3 and 5 of this series [6–8,10], a comprehensive study of the behaviour of several dry carbon fibre non-crimp fabrics can be found, including modelling of the internal geometry of undeformed, compressed and sheared fabrics, and characterisation of the behaviour of the dry fabrics in tension, shear, bending, and compression. Parts 4 and 6 of the series [9,11] treat the general mechanical properties of several kinds of composites made from the fabrics discussed in the previous parts, as well as the damage initiation and development during static tensile testing and the characterisation of the fatigue behaviour at low stresses. In part 7 [12], the properties of composites fabricated from the sheared fabrics are investigated. Most of the research done to this date concerns the in-plane properties of NCF composites. However, it is not unlikely that a composite part will encounter one or more significant out-of-plane loads during its lifetime: a car can be hit by a stone, an airplane can suffer from an impact by birds or hail stones etc. Therefore, it is important to know if, and how, the impact and post-impact properties of NCF composites differ from other types of composites.
1020
K. Vallons et al. / Composites: Part A 41 (2010) 1019–1026
Generally, an impact can cause three main types of damage in a composite material: matrix cracks, delaminations, and, if the impact energy is sufficiently high, fibre breakage. Matrix cracks usually have only a marginal effect on the tensile properties of the composite, while fibre breakage can lead to significant reductions in strength. In [13], Ding et al. investigated the effect of impact events with varying energy on the strength of a plain weave carbon epoxy composite. They concluded that below about 2 J of impact energy (i.e. 1 J/mm thickness of the laminate), there was no significant influence of the impact damage on the tensile strength of the material. As the impact energy was increased above this threshold value, a rather sharp decline of the residual strength was noted. Cantwell et al. [14] found a similar behaviour for a UD cross-ply carbon/epoxy material, although the decline in strength after the threshold energy is less steep. Ding et al. [13] also studied the influence of an impact on the tensile fatigue properties. They found that an impact damaged sample will degrade faster than a non-impacted sample, and that the fatigue life of the former is consequently lower. In [15–18], the present authors have reported the static and fatigue tensile properties of a new type of carbon epoxy non-crimp fabric composite. This composite is based on NCFs prelaminated with an epoxy resin layer and produced by means of a modified resin film infusion process. The present paper studies the interlaminar toughness, impact and post-impact properties of the same NCF composite, and compares the results to those obtained for a carbon fibre/epoxy woven fabric composite. The absorbed energy and the projected damage area after impact have been investigated as a function of the impact energy. The influence of the impact damage on the residual static tensile properties and the residual tensile fatigue properties was determined for two impact energy levels.
2. Materials and methods 2.1. Materials and production A ±45° biaxial carbon fibre NCF and a carbon fibre twill weave were used for this study. The materials are shown in Fig. 1. The NCF is the same as that used in [15]. The fabric parameters (dimensions of the openings, stitching parameters etc.) have been characterised and are listed in Table 1. The NCF fabric has an areal density of 540 g/m2 and is prelaminated with an epoxy resin layer. As in [15], both (+45°, 45°) and ( 45°, +45°) fabrics were used, to allow for the production of a fully symmetrical laminate in which also the direction of the stitching is aligned. The twill weave has an areal density of 380 g/m2 and is supplied in the form of prepregs. The epoxy resin, as well as the carbon fibre type is the same for both types of reinforcement. Composite plates were produced using a modified resin film infusion process. The production process is schematically illus-
MD
BD+
BDCD
Table 1 Internal characteristics of the ±45° balanced biaxial NCF. Fabric property Stitching yarn Stitching type Areal density Fibre type Carbon fibre yarn Opening length (front), mm Opening length (back), mm Opening width (front), mm Stitch gauge, mm Stitch length, mm
Polyester Chain 540 g/m2 (270 2) STS carbon fibre 24 K 9.6 ± 1.6 mm 11.9 ± 1.3 mm 0.26 ± 0.01 mm 5.3 ± 0.1 mm 3.3 ± 0.4 mm
trated in Fig. 2. In a first step, prepregs were made from the prelaminated NCF material. For the production of the woven fabric composite plates, this first step was omitted, as the woven fabric was supplied ready-impregnated. To produce the final laminates, several prepreg layers were stacked symmetrically. The processing conditions for the NCF prepregging step (step 1) and the plate production step (step 2) can be found in Table 2. The set-up used for the two production steps is the same, and is shown in Fig. 3. For the impact and post-impact tests, NCFC plates with an average thickness of 2.1 mm and WFC plates with an average thickness of 1.9 mm were made. To obtain an average fibre volume fraction of 58% in both materials, four prepreg layers were used for the NCF composite (NCFC) plates (lay up: [ 45, +45]2s), and five layers for the woven fabric composite plates (WFC). The nomenclature for the sample directions used for mechanical tests is illustrated in Fig. 1. In the NCFC material, MD and CD stand for machine and cross direction respectively, BD+ is one of the two fibre directions, oriented so that the fibres in the outer layer of the sample are parallel to the loading direction (i.e. the 45° direction), and BD is perpendicular to BD+ (i.e. the +45° direction). FD and BD denote fibre (warp) and bias direction in the WFC. For the impact and post-impact tests in this study, samples were cut in the following directions: BD+ and CD (NCFC) and FD and BD (WFC). The dimensions of the samples were 280 50 mm2. Composite end tabs were glued onto the samples, giving a gauge length of approximately 200 mm. To assure that no significant plate effect is present, a rudimentary FE simulation was done. This indicated that apart from a transition zone of about 5 cm on either side of the sample, the stress state in the samples could be assumed to be uni-axial. For the interlaminar fracture toughness tests, the number of prepregs in a plate was doubled, to obtain thicker samples. A thin aluminium foil coated with a release agent was inserted in the middle to act as a crack/delamination initiator. For the mode I tests, aluminium load transfer blocks were glued to the end of the sample.
FD BD
Fig. 1. The materials used in this study (left: NCF, right: twill weave) and the testing directions. For the NCFC, MD and CD represent ±45° fibre orientations, BD+ and BD represent 0°/90° orientations. In the WFC, FD and BD represent 0°/90° and ±45° orientations.
1021
K. Vallons et al. / Composites: Part A 41 (2010) 1019–1026
STEP 1 Prelaminated fabric
Prepreg
STEP 2 Prepregs
Plate
Fig. 2. Schematic illustration of the composite plate production process used in this study.
Table 2 Processing conditions for the production of the composite plates. Step
Time (min)
Temperature (°C)
1
15 5 15 60 90
20 90 20 90 130
2
Vacuum (MPa) 0.095 0.095 0.095 0.095 0.095
2.2. Experimental methods 2.2.1. Interlaminar fracture toughness tests Double cantilever beam (DCB) and end notch flexural (ENF) tests were done to determine the mode I and mode II interlaminar fracture toughness. The test set-up for these two types of tests is illustrated in Fig. 4. As mentioned above, a thin aluminium film (about 15 lm thick) coated with a release agent (PMR 90) was inserted in the middle of the laminate over a length of about 5 cm, to act as a crack initiator. In the NCFC, this foil was placed in such a way that the crack growth would occur in between two 0° oriented fibre layers (i.e. two layers with the fibres in the direction of the length of the specimen). This implies that the obtained GIc value for the ±45° NCFC will be an underestimation for the value for the 0°/90° interface.
The DCB tests were done in accordance with the ASTM D5528 standard. The mode I interlaminar fracture toughness values for initiation and propagation were calculated from the measured forces, crack opening displacements and crack lengths using the modified beam theory described in the standard. For the ENF tests the method outlined by Carlsson in [19] was used. During this type of test there is a high risk of instable crack growth, resulting in only one data point per test. To avoid measuring a combination of initiation and propagation values in such cases, Carlsson recommended that the initiation of a crack from the foil be done before starting the test. This recommendation was followed in this work, so only values for the mode II propagation toughness were obtained.
2.2.2. Impact tests and damage analysis The impact tests were done using an instrumented drop weight impact tester with a square clamping opening of 40 40 mm2. Two different impact energies were used: 3.5 and 7 J, i.e. 1.75 and 3.5 J/mm thickness of the laminate. All tests were done with a hemispherical impact head with a diameter of 13 mm. During each test, the force acting on the impact head was measured with a load cell, and the displacement of the head was registered by means of a reflected-laser beam optical device. From these displacements, the initial speed just before the impact and the final speed just after impact can be determined, and hence the differ-
Fig. 3. Set-up used for the composite plate production.
Fig. 4. Schematic representation of the double cantilever beam (DCB) test (left), and the end notch flexural (ENF) test (right).
1022
K. Vallons et al. / Composites: Part A 41 (2010) 1019–1026
ence in kinetic energy before and after the impact. This difference in energy is referred to as the ‘absorbed energy’. Due to the square clamping opening, the span length of the fibres in the tested material will differ for different fibre orientations in the samples. This is illustrated in Fig. 5. As this span length will influence the deflection behaviour of the sample, this set-up artefact can have an effect on the impact behaviour. To characterize the amount of damage, and more specifically the amount of delamination in the samples, they were investigated with ultrasonic C-scanning. This was also done before the impact tests, to eliminate samples with flaws or initial defects. With the C-scan technique, samples are placed on supports under water and scanned point by point by a ultrasonic transducer (step size 1200 lm). A grey value is assigned to each point in accordance with the intensity of the reflected ultrasonic wave. Undamaged areas will appear black or dark grey, areas with delaminations will be white or light grey. By applying a threshold value for the grey level, the delamination area can be determined. However, it is important to note that in this way not the sum of the delaminations between the different layers of the material is determined, but the area of the projection in the same plane of all the delaminations in the sample.
2.2.3. Post-impact static and fatigue tensile tests Post-impact tensile tests were done on the four types of samples (two directions for each of the two materials), impacted with an energy of 3.5 and 7 J (1.75 and 3.5 J/mm). For the static tests, an Instron 4505 tensile machine was used, and the test speed was 1 mm/min. The tensile–tensile fatigue tests were done in load control mode on a vertical 160 kN Schenck machine. To avoid heating of the samples during the test, a test frequency of 6 Hz was adopted for all tests. The R value (the ratio of minimum applied stress to maximum applied stress) was 0.1. The maximum stress for the fatigue tests was chosen as the stress level corresponding to a fatigue life of about 106 cycles for an undamaged sample.
(a)
Fibre directions
(b)
Fibre directions
Fig. 5. Illustration of the impact test set-up span length difference between ±45° oriented samples (a) and 0°/90° oriented samples (b).
3. Results and discussion 3.1. Interlaminar fracture toughness tests 3.1.1. Mode I interlaminar fracture toughness The mode I interlaminar fracture toughness values (mode I critical energy release rate, GIc) for the NCFC and the WFC are shown in the graph in Fig. 6. Both the initiation and the propagation toughness are given. No significant difference can be found between the initiation values for the two materials. This is not surprising, as initiation of a crack is mainly determined by the resin properties, and not by the reinforcement architecture. There is, however, a pronounced difference in the propagation values of GIc. The toughness of the WFC is about 50% higher than that of the NCFC. This can be explained by the nature of the reinforcement: in the NCFC, the crack follows the interface between the two quasi-straight 0° fibre layers, much like in a UD laminate, and the crack path therefore is approximately straight. It should be noted at this point that in a real impact, delaminations most often occur in between layers with a large orientation difference (e.g. 0°/90°), not in between layers with the same orientation. In the WFC, a crack following the matrix/reinforcement interface is forced to follow a more wavy crack path, as the crack must change its direction to follow the undulations of the fibre bundles in the woven fabric. A wavy crack path has a longer effective crack length than a straight crack, and thus a higher amount of energy is needed to propagate a wavy crack over the same global distance as a straight crack. Another phenomenon increasing the energy for crack growth in this case is the branching of the crack path when it encounters a transversely oriented fibre bundle. Also the changes in direction themselves require an additional amount of energy. 3.1.2. Mode II interlaminar fracture toughness As it was explained in paragraph 2.2.1, only propagation values for the mode II interlaminar fracture toughness (mode II critical energy release rate, GIIc) were obtained. Crack growth during the tests was fairly stable. The mode II toughness results are also shown in Fig. 6. As for the mode I toughness, the mode II toughness of the WFC is higher (about 30%) than that of the NCFC. A similar reasoning as described above for the mode I behaviour can be followed to explain this difference. Especially the mode II fracture toughness is important when considering impact of a composite material, since it is mainly mode II crack propagation which causes the formation and growth of delaminations. As both the mode I and the mode II fracture toughness are higher for the WFC than for the NCFC, it can be concluded that a delamination in the NCFC will grow faster than a similar one in the WFC.
Fig. 6. Mode I (left) and mode II (right) interlaminar fracture toughness test results for the non-crimp fabric composite (NCFC) and the woven fabric composite (WFC).
K. Vallons et al. / Composites: Part A 41 (2010) 1019–1026
1023
Therefore, it is to be expected that impact damage in the NCFC will be more severe than in the WFC. 3.2. Impact tests 3.2.1. Absorbed energy From the displacement measurements of the impacter, it can be concluded that the deflection of the two materials, in spite of the slight thickness difference, was virtually the same. Fig. 7 shows the absorbed energy for each direction of both materials as a function of the impact energy. Two impact energies were used: 3.5 and 7 J (i.e. 1.75 and 3.5 J per mm thickness of the plates). It can be seen that in general, for both impact energies, the NCFC absorbed about 10% more of the impact energy than the WFC. A small difference between the two directions in each material was noted: the absorbed energy is slightly higher for the samples with a ±45° orientation than for samples with a 0° orientation. However, due to the large variation on the results, this difference is not significant. From the results in Fig. 7, and taking into account the discussion above about the interlaminar fracture toughness, it can again be expected that for the same impact energy, there will be more delamination damage in the NCFC than in the WFC. 3.2.2. Damage A visual inspection of the samples revealed that there was a very small amount of fibre damage on the bottom side of the samples impacted with an energy of 7 J. Fibres in the bottom layer were broken over a width of approximately 4–5 mm. Compared to the sample width of 50 mm and assuming that fibre failure has only occurred in the bottom layer, this constitutes less than 5% of the total number of 0° oriented fibres. No fibre breakage was seen on the 3.5 J samples. The projected damage area was determined by C-scan as described in paragraph 2.2.2. It is important to keep in mind that this is not the total delamination area, but the area of the projection onto a plane of all the delaminations between the different layers. The results can be seen in Fig. 8. As was predicted by the lower fracture toughness and the higher absorbed energy, the projected delamination area in the NCFC samples is consistently larger than in the WFC. Even taking into account the fact that there are five fabric layers in the WFC and only four in the NCFC, the difference is still significant. The delaminated area, as seen in the C-scans, never reached the sample edges (minimum distance = approximately 5 mm). The width of the clamping opening was 40 mm. Since the projected damage area for the NCFC in ±45° direction (about 1400 mm2) is larger than a circle with a radius of 40 mm (1256 mm2), it is possible that the clamping fixture did affect the
Fig. 7. Absorbed energy as a function of the impact energy for both materials.
Fig. 8. The projected damage area, as measured with C-scan, as a function of the impact energy, for the different types of samples.
damage formation in the ±45° oriented samples to a certain degree. Despite this, some interesting trends can be deduced from the results. From Fig. 8, it becomes apparent that there is a difference in damage area between the two fibre orientations used: for samples which have fibres oriented at ±45° relative to the sample’s length axis, the damage area is significantly larger than for samples with a 0°/90° orientation. However, this again is most likely an artefact of the set-up used: the clamping has a square opening, so the span length of the fibres in the material will differ for the two orientations (see Fig. 5). Due to the shorter span length in the samples with 0°/90° orientation, the maximum strain in the direction of the fibres in the bottom layer will be higher than in the ±45° samples. The fact that the 0°/90° samples absorb approximately the same energy as the ±45° samples, while having a smaller projected damage area, could therefore indicate that in the former, locally, the failure strain of the fibres is reached and some of the absorbed energy is used for fibre failure, and not only for the creation of delaminations. This is consistent with the fact that the difference between the two fibre orientations becomes more apparent for higher impact energies. 3.3. Post-impact tests 3.3.1. Post-impact static tensile tests In the post-impact static tensile tests, the vast majority of samples, both in fibre and matrix dominated directions, failed in the impacted area. The graph in Fig. 9 shows the tensile strength for the intact materials, together with the values obtained after an impact of 3.5 or 7 J (1.75 and 3.5 J/mm thickness). It is clear that in each case, there is a significant influence of the impact on the residual static strength, and also, this influence becomes larger for higher impact energy. An impact of 3, 5 J causes a decrease in tensile strength in the fibre direction of 6% for the NCFC, and 3.5% for the WFC (for the latter only one sample could be tested). After an impact of 7 J, the difference has increased to 19% for the NCFC and 11.5% for the WFC. As stated above, some fibre damage was observed for the higher energy impact, but it was estimated to be only a few percent. The fibre damage can therefore only explain part of the reduction in strength. It is possible that some additional fibre damage is located on the inside of the sample. This will however be limited. If these results are compared to those obtained for a carbon– epoxy WFC by Ding et al. in [13], it can be concluded that the present decrease in tensile strength is much less severe than that observed in [13]: they found that after an impact of 3 J (thickness
1024
K. Vallons et al. / Composites: Part A 41 (2010) 1019–1026 Table 3 Maximum fatigue stresses used for the post-impact fatigue tests. Type
rmax (MPa)
WFC BD (±45°) NCFC CD (±45°) WFC FD (0°/90°) NCFC BD+ (0°/90°)
115 65 680 700
and delaminations) caused by the impact. In this case there is no difference in behaviour between the two materials.
Fig. 9. Intact and post-impact tensile strength results for the different types of samples.
of the plate 2.1 mm) the static tensile strength was reduced by about 50%, which is much more than the currently observed 6% and 3.5% decrease after an impact of 3.5 J (thickness of the plates approximately 2 mm). No specific reason was found for this pronounced difference in observations. It is possible that the epoxy used in the study of Ding was much more brittle than that in the present work, or that the difference is due to a different type of carbon fibres. In general, it can be concluded from the above results that for the residual static tensile properties in the fibre direction, the influence of a relatively low energy impact is significant, and that this influence is more pronounced for the NCFC than for the WFC. In the matrix dominated directions (±45° fibre orientation, NCFC CD and WFC BD), an impact of 3.5 J leads to a decrease in strength of 10% for both the NCFC and the WFC. After a 7 J impact, this has for both materials increased to 16% loss in strength. Also for the matrix dominated directions it can therefore be concluded that there is a significant influence of the damage (matrix cracks
3.3.2. Post-impact tensile–tensile fatigue tests In previous work, the fatigue life curves for the different directions of the two materials were determined. From these curves, the maximum fatigue stress level corresponding to an average fatigue life of 106 could be found. This level was used for the post-impact fatigue tests. As an example, the fatigue life curve for the NCFC in BD+ direction is shown in Fig. 10, and the load level corresponding to a fatigue life of 106 cycles is indicated. The stresses for the different directions of the two materials are listed in Table 3. For NCFC 0°/90° direction samples (NCFC BD+) impacted with an energy of 7 J (3.5 J/mm), another series of tests was done at a slightly lower stress level (600 MPa), to investigate the behaviour of the material more thoroughly. As in the post-impact static tests, most of the samples failed in the impacted area. In Fig. 11, the average fatigue life curve of samples with a ±45° fibre orientation (WFC BD and NCFC CD) is shown, together with the results of the tests after 3.5 J (1.75 J/mm) and 7 J (3.5 J/mm) impact loads. The average post-impact quasi-static strength is also indicated on the graphs. It can be seen that for both impact energies, the fatigue life of WFC samples with a ±45° fibre orientation is significantly lower than for undamaged samples. However, due to the large standard deviation on the results, it is not clear whether a higher impact energy results in a shorter fatigue life. This, however, is clearly the case for the NCFC samples with ±45° fibre orientation. An impact of 3.5 J decreases the fatigue life only marginally, whereas an impact of 7 J causes a reduction in fatigue life by an order of magnitude. Fig. 12 shows the average fatigue life curve for the 0°/90° oriented NCFC (BD+ direction) and the results of the post-impact fati-
Fig. 10. Average fatigue life curve for the NCFC in BD+ (0°/90°) direction and indication of the load level for the post-impact fatigue tests.
K. Vallons et al. / Composites: Part A 41 (2010) 1019–1026
Fig. 11. Average fatigue life curve and results of the post-impact fatigue tests for the ±45° oriented samples (NCFC CD and WFC BD).
1025
Fig. 13. Detail of the fatigue results of non-impacted NCFC BD+ (0°/90°) samples and samples after an impact of 3.5 J. The arrow indicates a sample that did not break up to 5 106 cycles.
Fig. 12. Average fatigue life curve for the NCFC in BD+ direction (0°/90° direction), and the results of the post-impact static and fatigue tests.
gue tests. As said above, an additional series of fatigue tests was done up to 600 MPa for samples impacted with 7 J energy. Also the static post-impact results are shown in the graph (fatigue life = 0.5). In Fig. 13, a detail of the 3.5 J results is shown, together with the results obtained for samples without impact. This graph shows that most of the results after 3.5 J lie within the scatter band for the samples without impact. Two samples, however, broke much earlier (at about 40 000 cycles), and seem to indicate that the average fatigue life after 3.5 J might still be slightly reduced. Even though statistically, this cannot be considered a significant decrease in fatigue life, if these results are combined with the small reduction in static tensile strength after 3.5 J impact, it can be expected that the average fatigue life after an impact of 3.5 J will be slightly shorter than for samples without an impact, and that the fatigue limit will be slightly lower. As Fig. 12 shows, after an impact of 7 J, it is clear that the average fatigue life is much shorter (by more than 2 orders of magnitude) than for the non-impacted material. These results are in accordance with the pronounced decrease in static tensile strength after impact, and indicate that also the fatigue limit after 7 J impact will be significantly lower. Fig. 14 shows the results of the post-impact fatigue tests on the WFC material under 0°/90° orientation (WFC FD), and compares them with the average fatigue life curve of the intact material. The static post-impact results are also shown in the graph (fatigue life = 0.5). No significant decrease in fatigue life was found after an
Fig. 14. Average fatigue life curve for the WFC in FD direction (0°/90° orientation), and the results of the post-impact static and fatigue tests.
impact of 3.5 J. After an impact of 7 J, the average fatigue life is again shorter than for the intact material. In this case, however, the decrease is only by about 1 order of magnitude, and is thus much less pronounced than in the NCFC material. From the results, it is to be expected that for the WFC, the decrease in fatigue limit will be less than for the NCFC. It can also be concluded that in fibre direction, the post-impact residual tensile fatigue properties of the WFC are better than those of the NCFC. This difference can be linked to the lower mode II interlaminar fracture toughness and the larger impact damage in the latter material. 4. Conclusion In this study, the interlaminar fracture toughness mode I and II of a NCFC and a WFC were determined, as well as the impact properties and the post-impact static and fatigue tensile behaviour. The initiation value of the mode I interlaminar fracture toughness was the same for both materials, which was explained by the use of the same resin in both materials. The propagation value, however, was for both mode I and mode II tests lower for the NCFC than for the WFC, indicating a lower resistance to delamination
1026
K. Vallons et al. / Composites: Part A 41 (2010) 1019–1026
growth along the fibre direction of the former. The impact tests confirmed this result: the NCFC absorbed a higher percentage of the impact energy, and the projected damage area caused by a same energy impact was consistently larger than for the WFC. A clear influence of the impact damage on the residual static strength was found for both materials in the two testing directions, but the decrease was more pronounced for the NCFC in fibre direction. The fatigue tests showed that a relatively low energy impact can cause a drastic decrease of the fatigue life for all types of test samples. Again the decrease in fibre direction fatigue life after impact was larger for the NCFC than for the WFC. This difference in residual properties is attributed to the lower resistance of the NCFC to delamination growth, as it was shown in this study, and consequently the larger impact damage.
References [1] Godbehere AP, Mills AR, Irving P. Non-crimp fabrics versus prepreg CFRP composites – a comparison of mechanical performance. In: 6th International conference on fibre reinforced composites – FRC’94. Newcastle; 1994. [2] Edgren FL, Asp E, Sjögren A. Effects of stitch pattern on the mechanical properties of non-crimp fabric composites. In: ECCM 11. Rhodos; 2004. [3] Edgren F, Asp LE. Approximate analytical constitutive model for non-crimp fabric composites. Composites Part A 2005;36(2):173–81. [4] Edgren FL, Asp E, Joffe R. Failure of NCF composites subjected to combined compression and shear loading. Compos Sci Technol 2006;66(15):2865–77. [5] Edgren F, Mattsson D, Asp LE, Varna J. Formation of damage and its effects on non-crimp fabric reinforced composites loaded in tension. Compos Sci Technol 2004;64(5):675–92. [6] Lomov SV, Belov EB, Bischoff T, Ghosh SB, Chi TT, Verpoest I. Carbon composites based on multiaxial multiply stitched preforms. Part 1. Geometry of the preform. Composites Part A 2002;33(9):1171–83.
[7] Lomov SV, Verpoest I, Barburski M, Laperre J. Carbon composites based on multiaxial multiply stitched preforms. Part 2. KES-F characterisation of the deformability of the preforms at low loads. Composites Part A 2003;34(4):359–70. [8] Lomov SV, Barburski M, Stoilova T, Verpoest I, Akkerman R, Loendersloot R, et al. Carbon composites based on multiaxial multiply stitched preforms. Part 3: biaxial tension, picture frame and compression tests of the preforms. Composites Part A 2005;36(9):1188–206. [9] Truong TC, Vettori M, Lomov S, Verpoest I. Carbon composites based on multiaxial multi-ply stitched preforms. Part 4. Mechanical properties of composites and damage observation. Composites Part A 2005;36(9):1207–21. [10] Loendersloot R, Lomov SV, Akkerman R, Verpoest I. Carbon composites based on multiaxial multiply stitched preforms. Part V: geometry of sheared biaxial fabrics. Composites Part A 2006;37(1):103–13. [11] Vallons K, Zong M, Lomov SV, Verpoest I. Carbon composites based on multiaxial multi-ply stitched preforms – Part 6. Fatigue behaviour at low loads: stiffness degradation and damage development. Composites Part A 2007;38(7):1633–45. [12] Truong TC, Ivanov DS, Klimshin DV, Lomov SV, Verpoest I. Carbon composites based on multi-axial multi-ply stitched preforms. Part 7: mechanical properties and damage observations in composites with sheared reinforcement. Compos Part A: Appl Sci Manuf 2008;39(9):1380–93. [13] Ding YQ, Yan Y, McIlhagger R. Effect of impact and fatigue loads on the strength of plain weave carbon–epoxy composites. J Mater Process Technol 1995;55(2):58–62. [14] Cantwell WJ, Morton J. The impact resistance of composite materials – a review. Composites 1991;22(5):347–62. [15] Vallons KS, Lomov V, Verpoest I. Fatigue and post-fatigue behaviour of carbon/ epoxy non crimp fabric composites. Composites Part A 2009;40:251–9. [16] Vallons K, Lomov SV, Verpoest I. Fatigue and post-fatigue behaviour of carbon/ epoxy noncrimp fabric composites. In: ICCM 16. Kyoto, Japan; 2007. [17] Vallons K, Lomov SV, Verpoest I. Damage evolution in static and fatigue tensile loading of carbon/epoxy NCF composites. In: ECOMAS thematic conference on mechanical response of composites. Porto; 2007. [18] Vallons K, Lomov SV, Verpoest I. Mechanical properties and damage evolution during static and fatigue loading of carbon–epoxy ncf composites. In: SEICO 2007. Paris, France; 2007. [19] Carlsson LA, Adams DF, Pipes RB. Experimental characterization of advanced composite materials. CRC Press; 1987. p. 167–71.
Contents lists available at ScienceDirect
Composites: Part A journal homepage: www.elsevier.com/locate/compositesa
Impact and post-impact properties of a carbon fibre non-crimp fabric and a twill weave composite Katleen Vallons *, Alex Behaeghe, Stepan V. Lomov, Ignaas Verpoest Dept. of Metallurgy and Materials Engineering, Katholieke Universiteit Leuven, Kasteelpark Arenberg 44 bus 2450, BE 3000 Leuven, Belgium
a r t i c l e
i n f o
Article history: Received 26 June 2009 Received in revised form 24 March 2010 Accepted 9 April 2010
Keywords: B. Impact behaviour B. Fatigue B. Mechanical properties
a b s t r a c t The impact and post-impact static and fatigue tensile properties of a carbon fibre/epoxy NCF composite were determined and compared to those of a carbon fibre/epoxy woven fabric composite, for two impact energies (3.5 and 7 J). The projected damage area after impact was larger for the NCF composite than that for the woven fabric composite for both impact energies. Impacted samples were subjected to static tensile tests and tensile–tensile fatigue tests. It was found that even a relatively low energy impact has already a significant negative influence on the residual properties in both static and fatigue tests, in the fibre direction as well as in the matrix dominated direction. In the matrix dominated directions the post-impact behaviour of the two materials is very similar. In the fibre direction, however, the properties of the non-crimp fabric composite are degraded more by an impact than those of the woven fabric composite. Ó 2010 Elsevier Ltd. All rights reserved.
1. Introduction Composite materials have a high in-plane stiffness- and strength-to-weight ratio, making them attractive materials for use in numerous transport applications, like cars and airplanes. One type of composites in particular, the non-crimp fabric composites, seems quite promising in this field, as non-crimp fabrics (NCF) are nearly as easy to handle as woven fabrics and also have a reasonable drapeability, while lacking the fibre crimp present in woven fabrics. This combination of advantages makes it possible to reduce the production cost and maintain excellent properties. In the recent past, extensive research has been conducted concerning non-crimp fabrics and their composites, and the differences between their behaviour and that of composites based on UD prepregs or woven fabrics. Godbehere et al. [1], for example, compared the tensile properties of NCF composites to those of UD prepreg composites and found a minor reduction in tensile strength but a decrease of about 5% in stiffness. Edgren et al. [2] investigated the influence of the stitching parameters on the fatigue performance of several unidirectional NCF composites, and compared the results to the behaviour of a UD prepreg laminate. They concluded that it appears as if a small stitch gauge in combination with a small stitch length can have a negative effect on the fatigue life. They also remarked, however, that overall, the fatigue life of all tested NCFCs did not differ much from the fatigue life of the prepreg laminate. * Corresponding author. Tel.: +32 1 6321193; fax: +32 1 6321990. E-mail address: [email protected] (K. Vallons). 1359-835X/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.compositesa.2010.04.008
Some other interesting studies concerning the properties of NCF composites have been published lately, e.g. by Edgren et al., who studied the behaviour of NCF composites under a combined compression and shear loading, and developed an approximate analytical constitutive model [3,4]. In [5], they focussed primarily on the formation of damage and the influence of this damage on the composite properties during tensile tests on cross-ply NCF composites. A series of seven papers has recently been published by the present authors, covering a wide variety of properties of both dry NCF fabrics and their composites [6–12]. In parts 1, 2, 3 and 5 of this series [6–8,10], a comprehensive study of the behaviour of several dry carbon fibre non-crimp fabrics can be found, including modelling of the internal geometry of undeformed, compressed and sheared fabrics, and characterisation of the behaviour of the dry fabrics in tension, shear, bending, and compression. Parts 4 and 6 of the series [9,11] treat the general mechanical properties of several kinds of composites made from the fabrics discussed in the previous parts, as well as the damage initiation and development during static tensile testing and the characterisation of the fatigue behaviour at low stresses. In part 7 [12], the properties of composites fabricated from the sheared fabrics are investigated. Most of the research done to this date concerns the in-plane properties of NCF composites. However, it is not unlikely that a composite part will encounter one or more significant out-of-plane loads during its lifetime: a car can be hit by a stone, an airplane can suffer from an impact by birds or hail stones etc. Therefore, it is important to know if, and how, the impact and post-impact properties of NCF composites differ from other types of composites.
1020
K. Vallons et al. / Composites: Part A 41 (2010) 1019–1026
Generally, an impact can cause three main types of damage in a composite material: matrix cracks, delaminations, and, if the impact energy is sufficiently high, fibre breakage. Matrix cracks usually have only a marginal effect on the tensile properties of the composite, while fibre breakage can lead to significant reductions in strength. In [13], Ding et al. investigated the effect of impact events with varying energy on the strength of a plain weave carbon epoxy composite. They concluded that below about 2 J of impact energy (i.e. 1 J/mm thickness of the laminate), there was no significant influence of the impact damage on the tensile strength of the material. As the impact energy was increased above this threshold value, a rather sharp decline of the residual strength was noted. Cantwell et al. [14] found a similar behaviour for a UD cross-ply carbon/epoxy material, although the decline in strength after the threshold energy is less steep. Ding et al. [13] also studied the influence of an impact on the tensile fatigue properties. They found that an impact damaged sample will degrade faster than a non-impacted sample, and that the fatigue life of the former is consequently lower. In [15–18], the present authors have reported the static and fatigue tensile properties of a new type of carbon epoxy non-crimp fabric composite. This composite is based on NCFs prelaminated with an epoxy resin layer and produced by means of a modified resin film infusion process. The present paper studies the interlaminar toughness, impact and post-impact properties of the same NCF composite, and compares the results to those obtained for a carbon fibre/epoxy woven fabric composite. The absorbed energy and the projected damage area after impact have been investigated as a function of the impact energy. The influence of the impact damage on the residual static tensile properties and the residual tensile fatigue properties was determined for two impact energy levels.
2. Materials and methods 2.1. Materials and production A ±45° biaxial carbon fibre NCF and a carbon fibre twill weave were used for this study. The materials are shown in Fig. 1. The NCF is the same as that used in [15]. The fabric parameters (dimensions of the openings, stitching parameters etc.) have been characterised and are listed in Table 1. The NCF fabric has an areal density of 540 g/m2 and is prelaminated with an epoxy resin layer. As in [15], both (+45°, 45°) and ( 45°, +45°) fabrics were used, to allow for the production of a fully symmetrical laminate in which also the direction of the stitching is aligned. The twill weave has an areal density of 380 g/m2 and is supplied in the form of prepregs. The epoxy resin, as well as the carbon fibre type is the same for both types of reinforcement. Composite plates were produced using a modified resin film infusion process. The production process is schematically illus-
MD
BD+
BDCD
Table 1 Internal characteristics of the ±45° balanced biaxial NCF. Fabric property Stitching yarn Stitching type Areal density Fibre type Carbon fibre yarn Opening length (front), mm Opening length (back), mm Opening width (front), mm Stitch gauge, mm Stitch length, mm
Polyester Chain 540 g/m2 (270 2) STS carbon fibre 24 K 9.6 ± 1.6 mm 11.9 ± 1.3 mm 0.26 ± 0.01 mm 5.3 ± 0.1 mm 3.3 ± 0.4 mm
trated in Fig. 2. In a first step, prepregs were made from the prelaminated NCF material. For the production of the woven fabric composite plates, this first step was omitted, as the woven fabric was supplied ready-impregnated. To produce the final laminates, several prepreg layers were stacked symmetrically. The processing conditions for the NCF prepregging step (step 1) and the plate production step (step 2) can be found in Table 2. The set-up used for the two production steps is the same, and is shown in Fig. 3. For the impact and post-impact tests, NCFC plates with an average thickness of 2.1 mm and WFC plates with an average thickness of 1.9 mm were made. To obtain an average fibre volume fraction of 58% in both materials, four prepreg layers were used for the NCF composite (NCFC) plates (lay up: [ 45, +45]2s), and five layers for the woven fabric composite plates (WFC). The nomenclature for the sample directions used for mechanical tests is illustrated in Fig. 1. In the NCFC material, MD and CD stand for machine and cross direction respectively, BD+ is one of the two fibre directions, oriented so that the fibres in the outer layer of the sample are parallel to the loading direction (i.e. the 45° direction), and BD is perpendicular to BD+ (i.e. the +45° direction). FD and BD denote fibre (warp) and bias direction in the WFC. For the impact and post-impact tests in this study, samples were cut in the following directions: BD+ and CD (NCFC) and FD and BD (WFC). The dimensions of the samples were 280 50 mm2. Composite end tabs were glued onto the samples, giving a gauge length of approximately 200 mm. To assure that no significant plate effect is present, a rudimentary FE simulation was done. This indicated that apart from a transition zone of about 5 cm on either side of the sample, the stress state in the samples could be assumed to be uni-axial. For the interlaminar fracture toughness tests, the number of prepregs in a plate was doubled, to obtain thicker samples. A thin aluminium foil coated with a release agent was inserted in the middle to act as a crack/delamination initiator. For the mode I tests, aluminium load transfer blocks were glued to the end of the sample.
FD BD
Fig. 1. The materials used in this study (left: NCF, right: twill weave) and the testing directions. For the NCFC, MD and CD represent ±45° fibre orientations, BD+ and BD represent 0°/90° orientations. In the WFC, FD and BD represent 0°/90° and ±45° orientations.
1021
K. Vallons et al. / Composites: Part A 41 (2010) 1019–1026
STEP 1 Prelaminated fabric
Prepreg
STEP 2 Prepregs
Plate
Fig. 2. Schematic illustration of the composite plate production process used in this study.
Table 2 Processing conditions for the production of the composite plates. Step
Time (min)
Temperature (°C)
1
15 5 15 60 90
20 90 20 90 130
2
Vacuum (MPa) 0.095 0.095 0.095 0.095 0.095
2.2. Experimental methods 2.2.1. Interlaminar fracture toughness tests Double cantilever beam (DCB) and end notch flexural (ENF) tests were done to determine the mode I and mode II interlaminar fracture toughness. The test set-up for these two types of tests is illustrated in Fig. 4. As mentioned above, a thin aluminium film (about 15 lm thick) coated with a release agent (PMR 90) was inserted in the middle of the laminate over a length of about 5 cm, to act as a crack initiator. In the NCFC, this foil was placed in such a way that the crack growth would occur in between two 0° oriented fibre layers (i.e. two layers with the fibres in the direction of the length of the specimen). This implies that the obtained GIc value for the ±45° NCFC will be an underestimation for the value for the 0°/90° interface.
The DCB tests were done in accordance with the ASTM D5528 standard. The mode I interlaminar fracture toughness values for initiation and propagation were calculated from the measured forces, crack opening displacements and crack lengths using the modified beam theory described in the standard. For the ENF tests the method outlined by Carlsson in [19] was used. During this type of test there is a high risk of instable crack growth, resulting in only one data point per test. To avoid measuring a combination of initiation and propagation values in such cases, Carlsson recommended that the initiation of a crack from the foil be done before starting the test. This recommendation was followed in this work, so only values for the mode II propagation toughness were obtained.
2.2.2. Impact tests and damage analysis The impact tests were done using an instrumented drop weight impact tester with a square clamping opening of 40 40 mm2. Two different impact energies were used: 3.5 and 7 J, i.e. 1.75 and 3.5 J/mm thickness of the laminate. All tests were done with a hemispherical impact head with a diameter of 13 mm. During each test, the force acting on the impact head was measured with a load cell, and the displacement of the head was registered by means of a reflected-laser beam optical device. From these displacements, the initial speed just before the impact and the final speed just after impact can be determined, and hence the differ-
Fig. 3. Set-up used for the composite plate production.
Fig. 4. Schematic representation of the double cantilever beam (DCB) test (left), and the end notch flexural (ENF) test (right).
1022
K. Vallons et al. / Composites: Part A 41 (2010) 1019–1026
ence in kinetic energy before and after the impact. This difference in energy is referred to as the ‘absorbed energy’. Due to the square clamping opening, the span length of the fibres in the tested material will differ for different fibre orientations in the samples. This is illustrated in Fig. 5. As this span length will influence the deflection behaviour of the sample, this set-up artefact can have an effect on the impact behaviour. To characterize the amount of damage, and more specifically the amount of delamination in the samples, they were investigated with ultrasonic C-scanning. This was also done before the impact tests, to eliminate samples with flaws or initial defects. With the C-scan technique, samples are placed on supports under water and scanned point by point by a ultrasonic transducer (step size 1200 lm). A grey value is assigned to each point in accordance with the intensity of the reflected ultrasonic wave. Undamaged areas will appear black or dark grey, areas with delaminations will be white or light grey. By applying a threshold value for the grey level, the delamination area can be determined. However, it is important to note that in this way not the sum of the delaminations between the different layers of the material is determined, but the area of the projection in the same plane of all the delaminations in the sample.
2.2.3. Post-impact static and fatigue tensile tests Post-impact tensile tests were done on the four types of samples (two directions for each of the two materials), impacted with an energy of 3.5 and 7 J (1.75 and 3.5 J/mm). For the static tests, an Instron 4505 tensile machine was used, and the test speed was 1 mm/min. The tensile–tensile fatigue tests were done in load control mode on a vertical 160 kN Schenck machine. To avoid heating of the samples during the test, a test frequency of 6 Hz was adopted for all tests. The R value (the ratio of minimum applied stress to maximum applied stress) was 0.1. The maximum stress for the fatigue tests was chosen as the stress level corresponding to a fatigue life of about 106 cycles for an undamaged sample.
(a)
Fibre directions
(b)
Fibre directions
Fig. 5. Illustration of the impact test set-up span length difference between ±45° oriented samples (a) and 0°/90° oriented samples (b).
3. Results and discussion 3.1. Interlaminar fracture toughness tests 3.1.1. Mode I interlaminar fracture toughness The mode I interlaminar fracture toughness values (mode I critical energy release rate, GIc) for the NCFC and the WFC are shown in the graph in Fig. 6. Both the initiation and the propagation toughness are given. No significant difference can be found between the initiation values for the two materials. This is not surprising, as initiation of a crack is mainly determined by the resin properties, and not by the reinforcement architecture. There is, however, a pronounced difference in the propagation values of GIc. The toughness of the WFC is about 50% higher than that of the NCFC. This can be explained by the nature of the reinforcement: in the NCFC, the crack follows the interface between the two quasi-straight 0° fibre layers, much like in a UD laminate, and the crack path therefore is approximately straight. It should be noted at this point that in a real impact, delaminations most often occur in between layers with a large orientation difference (e.g. 0°/90°), not in between layers with the same orientation. In the WFC, a crack following the matrix/reinforcement interface is forced to follow a more wavy crack path, as the crack must change its direction to follow the undulations of the fibre bundles in the woven fabric. A wavy crack path has a longer effective crack length than a straight crack, and thus a higher amount of energy is needed to propagate a wavy crack over the same global distance as a straight crack. Another phenomenon increasing the energy for crack growth in this case is the branching of the crack path when it encounters a transversely oriented fibre bundle. Also the changes in direction themselves require an additional amount of energy. 3.1.2. Mode II interlaminar fracture toughness As it was explained in paragraph 2.2.1, only propagation values for the mode II interlaminar fracture toughness (mode II critical energy release rate, GIIc) were obtained. Crack growth during the tests was fairly stable. The mode II toughness results are also shown in Fig. 6. As for the mode I toughness, the mode II toughness of the WFC is higher (about 30%) than that of the NCFC. A similar reasoning as described above for the mode I behaviour can be followed to explain this difference. Especially the mode II fracture toughness is important when considering impact of a composite material, since it is mainly mode II crack propagation which causes the formation and growth of delaminations. As both the mode I and the mode II fracture toughness are higher for the WFC than for the NCFC, it can be concluded that a delamination in the NCFC will grow faster than a similar one in the WFC.
Fig. 6. Mode I (left) and mode II (right) interlaminar fracture toughness test results for the non-crimp fabric composite (NCFC) and the woven fabric composite (WFC).
K. Vallons et al. / Composites: Part A 41 (2010) 1019–1026
1023
Therefore, it is to be expected that impact damage in the NCFC will be more severe than in the WFC. 3.2. Impact tests 3.2.1. Absorbed energy From the displacement measurements of the impacter, it can be concluded that the deflection of the two materials, in spite of the slight thickness difference, was virtually the same. Fig. 7 shows the absorbed energy for each direction of both materials as a function of the impact energy. Two impact energies were used: 3.5 and 7 J (i.e. 1.75 and 3.5 J per mm thickness of the plates). It can be seen that in general, for both impact energies, the NCFC absorbed about 10% more of the impact energy than the WFC. A small difference between the two directions in each material was noted: the absorbed energy is slightly higher for the samples with a ±45° orientation than for samples with a 0° orientation. However, due to the large variation on the results, this difference is not significant. From the results in Fig. 7, and taking into account the discussion above about the interlaminar fracture toughness, it can again be expected that for the same impact energy, there will be more delamination damage in the NCFC than in the WFC. 3.2.2. Damage A visual inspection of the samples revealed that there was a very small amount of fibre damage on the bottom side of the samples impacted with an energy of 7 J. Fibres in the bottom layer were broken over a width of approximately 4–5 mm. Compared to the sample width of 50 mm and assuming that fibre failure has only occurred in the bottom layer, this constitutes less than 5% of the total number of 0° oriented fibres. No fibre breakage was seen on the 3.5 J samples. The projected damage area was determined by C-scan as described in paragraph 2.2.2. It is important to keep in mind that this is not the total delamination area, but the area of the projection onto a plane of all the delaminations between the different layers. The results can be seen in Fig. 8. As was predicted by the lower fracture toughness and the higher absorbed energy, the projected delamination area in the NCFC samples is consistently larger than in the WFC. Even taking into account the fact that there are five fabric layers in the WFC and only four in the NCFC, the difference is still significant. The delaminated area, as seen in the C-scans, never reached the sample edges (minimum distance = approximately 5 mm). The width of the clamping opening was 40 mm. Since the projected damage area for the NCFC in ±45° direction (about 1400 mm2) is larger than a circle with a radius of 40 mm (1256 mm2), it is possible that the clamping fixture did affect the
Fig. 7. Absorbed energy as a function of the impact energy for both materials.
Fig. 8. The projected damage area, as measured with C-scan, as a function of the impact energy, for the different types of samples.
damage formation in the ±45° oriented samples to a certain degree. Despite this, some interesting trends can be deduced from the results. From Fig. 8, it becomes apparent that there is a difference in damage area between the two fibre orientations used: for samples which have fibres oriented at ±45° relative to the sample’s length axis, the damage area is significantly larger than for samples with a 0°/90° orientation. However, this again is most likely an artefact of the set-up used: the clamping has a square opening, so the span length of the fibres in the material will differ for the two orientations (see Fig. 5). Due to the shorter span length in the samples with 0°/90° orientation, the maximum strain in the direction of the fibres in the bottom layer will be higher than in the ±45° samples. The fact that the 0°/90° samples absorb approximately the same energy as the ±45° samples, while having a smaller projected damage area, could therefore indicate that in the former, locally, the failure strain of the fibres is reached and some of the absorbed energy is used for fibre failure, and not only for the creation of delaminations. This is consistent with the fact that the difference between the two fibre orientations becomes more apparent for higher impact energies. 3.3. Post-impact tests 3.3.1. Post-impact static tensile tests In the post-impact static tensile tests, the vast majority of samples, both in fibre and matrix dominated directions, failed in the impacted area. The graph in Fig. 9 shows the tensile strength for the intact materials, together with the values obtained after an impact of 3.5 or 7 J (1.75 and 3.5 J/mm thickness). It is clear that in each case, there is a significant influence of the impact on the residual static strength, and also, this influence becomes larger for higher impact energy. An impact of 3, 5 J causes a decrease in tensile strength in the fibre direction of 6% for the NCFC, and 3.5% for the WFC (for the latter only one sample could be tested). After an impact of 7 J, the difference has increased to 19% for the NCFC and 11.5% for the WFC. As stated above, some fibre damage was observed for the higher energy impact, but it was estimated to be only a few percent. The fibre damage can therefore only explain part of the reduction in strength. It is possible that some additional fibre damage is located on the inside of the sample. This will however be limited. If these results are compared to those obtained for a carbon– epoxy WFC by Ding et al. in [13], it can be concluded that the present decrease in tensile strength is much less severe than that observed in [13]: they found that after an impact of 3 J (thickness
1024
K. Vallons et al. / Composites: Part A 41 (2010) 1019–1026 Table 3 Maximum fatigue stresses used for the post-impact fatigue tests. Type
rmax (MPa)
WFC BD (±45°) NCFC CD (±45°) WFC FD (0°/90°) NCFC BD+ (0°/90°)
115 65 680 700
and delaminations) caused by the impact. In this case there is no difference in behaviour between the two materials.
Fig. 9. Intact and post-impact tensile strength results for the different types of samples.
of the plate 2.1 mm) the static tensile strength was reduced by about 50%, which is much more than the currently observed 6% and 3.5% decrease after an impact of 3.5 J (thickness of the plates approximately 2 mm). No specific reason was found for this pronounced difference in observations. It is possible that the epoxy used in the study of Ding was much more brittle than that in the present work, or that the difference is due to a different type of carbon fibres. In general, it can be concluded from the above results that for the residual static tensile properties in the fibre direction, the influence of a relatively low energy impact is significant, and that this influence is more pronounced for the NCFC than for the WFC. In the matrix dominated directions (±45° fibre orientation, NCFC CD and WFC BD), an impact of 3.5 J leads to a decrease in strength of 10% for both the NCFC and the WFC. After a 7 J impact, this has for both materials increased to 16% loss in strength. Also for the matrix dominated directions it can therefore be concluded that there is a significant influence of the damage (matrix cracks
3.3.2. Post-impact tensile–tensile fatigue tests In previous work, the fatigue life curves for the different directions of the two materials were determined. From these curves, the maximum fatigue stress level corresponding to an average fatigue life of 106 could be found. This level was used for the post-impact fatigue tests. As an example, the fatigue life curve for the NCFC in BD+ direction is shown in Fig. 10, and the load level corresponding to a fatigue life of 106 cycles is indicated. The stresses for the different directions of the two materials are listed in Table 3. For NCFC 0°/90° direction samples (NCFC BD+) impacted with an energy of 7 J (3.5 J/mm), another series of tests was done at a slightly lower stress level (600 MPa), to investigate the behaviour of the material more thoroughly. As in the post-impact static tests, most of the samples failed in the impacted area. In Fig. 11, the average fatigue life curve of samples with a ±45° fibre orientation (WFC BD and NCFC CD) is shown, together with the results of the tests after 3.5 J (1.75 J/mm) and 7 J (3.5 J/mm) impact loads. The average post-impact quasi-static strength is also indicated on the graphs. It can be seen that for both impact energies, the fatigue life of WFC samples with a ±45° fibre orientation is significantly lower than for undamaged samples. However, due to the large standard deviation on the results, it is not clear whether a higher impact energy results in a shorter fatigue life. This, however, is clearly the case for the NCFC samples with ±45° fibre orientation. An impact of 3.5 J decreases the fatigue life only marginally, whereas an impact of 7 J causes a reduction in fatigue life by an order of magnitude. Fig. 12 shows the average fatigue life curve for the 0°/90° oriented NCFC (BD+ direction) and the results of the post-impact fati-
Fig. 10. Average fatigue life curve for the NCFC in BD+ (0°/90°) direction and indication of the load level for the post-impact fatigue tests.
K. Vallons et al. / Composites: Part A 41 (2010) 1019–1026
Fig. 11. Average fatigue life curve and results of the post-impact fatigue tests for the ±45° oriented samples (NCFC CD and WFC BD).
1025
Fig. 13. Detail of the fatigue results of non-impacted NCFC BD+ (0°/90°) samples and samples after an impact of 3.5 J. The arrow indicates a sample that did not break up to 5 106 cycles.
Fig. 12. Average fatigue life curve for the NCFC in BD+ direction (0°/90° direction), and the results of the post-impact static and fatigue tests.
gue tests. As said above, an additional series of fatigue tests was done up to 600 MPa for samples impacted with 7 J energy. Also the static post-impact results are shown in the graph (fatigue life = 0.5). In Fig. 13, a detail of the 3.5 J results is shown, together with the results obtained for samples without impact. This graph shows that most of the results after 3.5 J lie within the scatter band for the samples without impact. Two samples, however, broke much earlier (at about 40 000 cycles), and seem to indicate that the average fatigue life after 3.5 J might still be slightly reduced. Even though statistically, this cannot be considered a significant decrease in fatigue life, if these results are combined with the small reduction in static tensile strength after 3.5 J impact, it can be expected that the average fatigue life after an impact of 3.5 J will be slightly shorter than for samples without an impact, and that the fatigue limit will be slightly lower. As Fig. 12 shows, after an impact of 7 J, it is clear that the average fatigue life is much shorter (by more than 2 orders of magnitude) than for the non-impacted material. These results are in accordance with the pronounced decrease in static tensile strength after impact, and indicate that also the fatigue limit after 7 J impact will be significantly lower. Fig. 14 shows the results of the post-impact fatigue tests on the WFC material under 0°/90° orientation (WFC FD), and compares them with the average fatigue life curve of the intact material. The static post-impact results are also shown in the graph (fatigue life = 0.5). No significant decrease in fatigue life was found after an
Fig. 14. Average fatigue life curve for the WFC in FD direction (0°/90° orientation), and the results of the post-impact static and fatigue tests.
impact of 3.5 J. After an impact of 7 J, the average fatigue life is again shorter than for the intact material. In this case, however, the decrease is only by about 1 order of magnitude, and is thus much less pronounced than in the NCFC material. From the results, it is to be expected that for the WFC, the decrease in fatigue limit will be less than for the NCFC. It can also be concluded that in fibre direction, the post-impact residual tensile fatigue properties of the WFC are better than those of the NCFC. This difference can be linked to the lower mode II interlaminar fracture toughness and the larger impact damage in the latter material. 4. Conclusion In this study, the interlaminar fracture toughness mode I and II of a NCFC and a WFC were determined, as well as the impact properties and the post-impact static and fatigue tensile behaviour. The initiation value of the mode I interlaminar fracture toughness was the same for both materials, which was explained by the use of the same resin in both materials. The propagation value, however, was for both mode I and mode II tests lower for the NCFC than for the WFC, indicating a lower resistance to delamination
1026
K. Vallons et al. / Composites: Part A 41 (2010) 1019–1026
growth along the fibre direction of the former. The impact tests confirmed this result: the NCFC absorbed a higher percentage of the impact energy, and the projected damage area caused by a same energy impact was consistently larger than for the WFC. A clear influence of the impact damage on the residual static strength was found for both materials in the two testing directions, but the decrease was more pronounced for the NCFC in fibre direction. The fatigue tests showed that a relatively low energy impact can cause a drastic decrease of the fatigue life for all types of test samples. Again the decrease in fibre direction fatigue life after impact was larger for the NCFC than for the WFC. This difference in residual properties is attributed to the lower resistance of the NCFC to delamination growth, as it was shown in this study, and consequently the larger impact damage.
References [1] Godbehere AP, Mills AR, Irving P. Non-crimp fabrics versus prepreg CFRP composites – a comparison of mechanical performance. In: 6th International conference on fibre reinforced composites – FRC’94. Newcastle; 1994. [2] Edgren FL, Asp E, Sjögren A. Effects of stitch pattern on the mechanical properties of non-crimp fabric composites. In: ECCM 11. Rhodos; 2004. [3] Edgren F, Asp LE. Approximate analytical constitutive model for non-crimp fabric composites. Composites Part A 2005;36(2):173–81. [4] Edgren FL, Asp E, Joffe R. Failure of NCF composites subjected to combined compression and shear loading. Compos Sci Technol 2006;66(15):2865–77. [5] Edgren F, Mattsson D, Asp LE, Varna J. Formation of damage and its effects on non-crimp fabric reinforced composites loaded in tension. Compos Sci Technol 2004;64(5):675–92. [6] Lomov SV, Belov EB, Bischoff T, Ghosh SB, Chi TT, Verpoest I. Carbon composites based on multiaxial multiply stitched preforms. Part 1. Geometry of the preform. Composites Part A 2002;33(9):1171–83.
[7] Lomov SV, Verpoest I, Barburski M, Laperre J. Carbon composites based on multiaxial multiply stitched preforms. Part 2. KES-F characterisation of the deformability of the preforms at low loads. Composites Part A 2003;34(4):359–70. [8] Lomov SV, Barburski M, Stoilova T, Verpoest I, Akkerman R, Loendersloot R, et al. Carbon composites based on multiaxial multiply stitched preforms. Part 3: biaxial tension, picture frame and compression tests of the preforms. Composites Part A 2005;36(9):1188–206. [9] Truong TC, Vettori M, Lomov S, Verpoest I. Carbon composites based on multiaxial multi-ply stitched preforms. Part 4. Mechanical properties of composites and damage observation. Composites Part A 2005;36(9):1207–21. [10] Loendersloot R, Lomov SV, Akkerman R, Verpoest I. Carbon composites based on multiaxial multiply stitched preforms. Part V: geometry of sheared biaxial fabrics. Composites Part A 2006;37(1):103–13. [11] Vallons K, Zong M, Lomov SV, Verpoest I. Carbon composites based on multiaxial multi-ply stitched preforms – Part 6. Fatigue behaviour at low loads: stiffness degradation and damage development. Composites Part A 2007;38(7):1633–45. [12] Truong TC, Ivanov DS, Klimshin DV, Lomov SV, Verpoest I. Carbon composites based on multi-axial multi-ply stitched preforms. Part 7: mechanical properties and damage observations in composites with sheared reinforcement. Compos Part A: Appl Sci Manuf 2008;39(9):1380–93. [13] Ding YQ, Yan Y, McIlhagger R. Effect of impact and fatigue loads on the strength of plain weave carbon–epoxy composites. J Mater Process Technol 1995;55(2):58–62. [14] Cantwell WJ, Morton J. The impact resistance of composite materials – a review. Composites 1991;22(5):347–62. [15] Vallons KS, Lomov V, Verpoest I. Fatigue and post-fatigue behaviour of carbon/ epoxy non crimp fabric composites. Composites Part A 2009;40:251–9. [16] Vallons K, Lomov SV, Verpoest I. Fatigue and post-fatigue behaviour of carbon/ epoxy noncrimp fabric composites. In: ICCM 16. Kyoto, Japan; 2007. [17] Vallons K, Lomov SV, Verpoest I. Damage evolution in static and fatigue tensile loading of carbon/epoxy NCF composites. In: ECOMAS thematic conference on mechanical response of composites. Porto; 2007. [18] Vallons K, Lomov SV, Verpoest I. Mechanical properties and damage evolution during static and fatigue loading of carbon–epoxy ncf composites. In: SEICO 2007. Paris, France; 2007. [19] Carlsson LA, Adams DF, Pipes RB. Experimental characterization of advanced composite materials. CRC Press; 1987. p. 167–71.