Effect of fatigue loading rate on lifespan and

International Journal of Fatigue 113 (2018) 253–263

Contents lists available at ScienceDirect

International Journal of Fatigue journal homepage: www.elsevier.com/locate/ijfatigue

Effect of fatigue loading rate on lifespan and temperature of tailored blank C/PPS thermoplastic composite

T

⁎

Roman Růžek , Martin Kadlec, Lucie Petrusová VZLU – Czech Aerospace Research Centre, Strength of Structures Department, Beranových 130, 199 05 Prague, Czech Republic

A R T I C LE I N FO

A B S T R A C T

Keywords: Fatigue Frequency effect Ply drops Fibre-reinforced composites Thermoplastics

This article reports an experimental investigation of the effects of the loading frequency on the temperature change, fatigue behaviour, and failure mechanisms of carbon-fibre-fabric-reinforced polyphenylenesulphide (PPS) laminates, the thicknesses of which were varied by ply drops. The experiments specifically considered two ply drop configurations and fatigue loading frequencies of 0.5–15 Hz. Fractographic examination revealed the presence of loading-frequency-based surface fracture features in the tested specimens. With increasing loading frequency, the local temperature also increases significantly, reaching as high as above 110 °C, accompanied by more than one order of magnitude decrease in the fatigue life. For a surface temperature of up to 38 °C, there was no specific relationship with the fatigue life. However, further increase of the surface temperature up to and beyond 75 °C was accompanied by significant reduction of the fatigue life. An analytic relationship between the load rate and the local temperature was derived and used to define limits for the fatigue testing of tailored blank structures.

1. Introduction Thermoplastic matrix composites (TPMCs) are increasingly being used for the fabrication of airframes in the aerospace industry owing to their superior resistance to impact damage [1] and better formability [2] compared with conventional epoxy-based laminates. The recyclability of TPMCs [3] is another significant advantage from the perspective of environmental conservation. The large-scale use of TPMCs was previously hindered by the complex manufacturing process and high cost. New technologies have, however, been used to significantly overcome these drawbacks, with large amounts of the materials presently ordered by manufacturers. TPMCs are specifically used as constant-thickness materials for the construction of the primary airframe components. A wide range of thermoplastic resins is available for their production, such as polyetheretherketone (PEEK) and polyphenylenesulphide (PPS), which are among the most widely used highperformance thermoplastic (TP) resins [4]. The present study focused on the mechanical behaviour of carbon/PPS (C/PPS) composites. Compared to thermosets, thermoplastic matrices can be produced by different technologies that not only enable better exploitation of the mechanical properties of the matrices, but also significant weight reduction. Typical advanced technologies that are employed in the current fabrication of airframes include hot forming using ‘tailored blanks’. The use of tailored blank technology facilitates the variation of the

⁎

composite thickness, fibre orientation, composition, and shape in accordance with the specific requirements. The technology is particularly used when the composite thickness needs to be varied within a single component to maintain the strength with reduced weight. However, a number of the material and structural parameters required for the numerical evaluation that is necessary for this purpose have not been sufficiently investigated. These included the fatigue parameters, temperature, and ageing of the parts of varying thickness can be ranked among these parameters. 1.1. Effect of temperature At a high temperature, the nonlinear behaviour of fibre-reinforced composites becomes significant [5–11]. A nonlinear response is associated with the shear deformation of the polymer matrix along the reinforcement fibres. The pronouncement of this behaviour at high temperatures is due to the viscoplastic nature of the TP matrix [11–14]. Few studies have considered the behaviour of carbon-fibre-fabric-reinforced PPS laminates or the effect of a high temperature on the behaviour of UD-reinforced PPS matrices [8,14–21]. With regard to the mechanical properties of notched and un-notched laminates, a service temperature higher than the glass transition temperature of a C/PPS composite (i.e., Tg = 98 °C) does not seem to significantly affect the strength or stiffness of quasi-isotropic laminates, whereas it severely

Corresponding author. E-mail address: [email protected] (R. Růžek).

https://doi.org/10.1016/j.ijfatigue.2018.04.023 Received 10 January 2018; Received in revised form 13 April 2018; Accepted 19 April 2018 Available online 22 April 2018 0142-1123/ © 2018 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/BY/4.0/).


R. Růžek et al.

affect the equilibrium temperature include the friction between damaged sites and the heat accumulation due to the poor thermal conductivity of the material. In some cases, the material fails under fatigue long before reaching the equilibrium temperature. Nevertheless, the effect of the loading frequency on the fatigue behaviour of an advanced thermoplastic matrix composite requires further investigation. The pronounced viscoelastic nature of the matrix material and the accompanying thermal effect during fatigue loading significantly impact the reliability and mechanical performance of the composite.

degrades the mechanical properties of angle-ply laminates [14]. It has actually been established that temperature is the environmental conditioning factor that is most detrimental to the fatigue properties of a laminate [17]. It is generally accepted that C/PPS-reinforced thermoplastic materials have high static and fatigue strengths. However, this is not the case under certain conditions of temperature and moisture. Aircrafts operate under different temperature and humidity conditions, and the hygrothermal ageing of a utilised composite may alter the glass transition temperature and lead to a significant decrease in the mechanical strength and ultimate strain [22]. Several researchers [23–26] have investigated the detrimental effects of ageing, and the effect of moisture has also been shown to be less severe than that of temperature [27–30].

1.4. Effect of ply drops With the aim of weight saving and more uniform stress distribution, it is necessary to decrease the thickness of the less loaded parts of the structure. The details of the loading of an integral structure are determined by the specific type of external load, the centre of mass of the structure, and the location of its neutral axis. Thickness optimisation is an ideal approach to decreasing the weight of a structure. The insertion of semipregs at suitable locations of the structure could be used to facilitate a more uniform stress distribution, as well as thickness variation. However, such might induce an undesirable temperature increase, which would affect the structure under fatigue loading. The use of ply drops to vary the thickness may be classified as an external, an internal (longitudinal or transverse), or a mid-plane ply drop construction [42]. However, problems may arise from the use of ply drops (inserted ply ends) because the resultant stress concentrations could induce damage [42,43]. Helmy et al. [44] investigated the tensile fatigue behaviour of internal ply drops and observed that the fatigue cracks were initiated near the ply drops and propagated along the interface to the thicker section through a mode II propagation. According to Wisnom et al. [45], asymmetry does not appear to be a critical factor of this type of delamination. The fatigue life of a ply drop composite is significantly shorter than that of a plain composite mainly because of the initiation and growth of the delamination near a ply drop [46]. To investigate these problems, numerical models based on globallocal approaches have been developed using Timoschnecko beam elements [47] or cohesive zone elements with the Hashin and LaRC failure theories [48]. Nevertheless, to gain a deeper understanding of the fatigue behaviour of a composite component with varying thickness, an extensive experimental program is required, such as that used for the certification of airframes, parts, and materials. Time and cost are the main considerations in such material and structural certification processes, and this necessitates the use of higher loading frequencies to conduct fatigue tests. Thickness variations in a component also constitute stress concentrators that may affect the durability of the component under the increased temperature induced by an applied cyclic load. This paper describes an experimental investigation of an asymmetrically tapered laminate with internal ply drops loaded under tensiontension fatigue conditions. The objective of the study was to acquire data for designing and determining the fatigue strength of a rib demonstrator and to define the load rate limitation that must be considered during the certification of experimental materials and structural parts. Additional data was collected for the development of new models for conducting fatigue degradation simulations.

1.2. Fatigue loading of thermoplastic matrix composites Fatigue loading is one of the main types of loading of structural elements, and it is capable of causing catastrophic failure under certain conditions. However, the fatigue response of C/PPS thermoplastics, especially those with varying thickness and ply drops, has not been sufficiently investigated. Indeed, there is a need for extensive experimental studies in this area. Continuous-fibre-reinforced composites are characterised by the development and accumulation of several types of defects including matrix cracking between the fibres, delamination between adjacent plies, debonding between the matrix and fibres, and fibre fracture. Such defects in a composite system are not isolated but interconnected. The identification of existing crack paths is thus a highly complex task. Moreover, most of the defects occur long before the ultimate failure of the component, and many types of subcritical failures may thus occur [22,31–33]. In the case of matrix-dominated composites, the fatigue behaviour is significantly controlled by the behaviour of the polymer matrix. Polymers do not exhibit the equivalent of the crystallographic stage I crack propagation observed in metals, for which reason crack initiation is the most important stage in the fatigue life of a polymer [34]. Conversely, polymers experience two types of failures, namely, cyclic creep failure and thermal failure, which occur under certain combinations of loading conditions. Cyclic creep occurs under conditions of sufficiently high loads and low frequencies [35], while thermal failure occurs at high frequencies, which prevent the dissipation of the energy loss due to the inherent high damping of the polymer. This results in a significant increase in temperature, which in turn induces thermal softening and loss of property [29,31,36–40]. 1.3. Effect of loading frequency As the role of the matrix becomes more dominant, there is an increase in the viscoelastic effect on the time-dependent properties in the entire composite. For this reason, and considering the nature of fatigue loading, the loading frequency is assumed to be one of the most important factors that affect the fatigue behaviour of such materials. Fatigue failure associated with large-scale hysteretic heating has been observed in polymers [37] and glass-reinforced epoxy [38], with the fatigue life decreasing with increasing loading frequency. The effect of hysteretic heating is less severe in graphite/epoxy composites, in which case the fatigue life increases with increasing loading frequency when temperature increase during the fatigue test is small (temperature rise lower than 10 °C) [41]. A similar trend has been observed in the case of boron-epoxy composites [36]. Several studies on the fatigue behaviour of AS4 PEEK have shown that the effect of hysteretic heating on the composite is much more pronounced than in the case of thermoset composites [29,36,37], and that the fatigue life significantly decreases with increasing loading frequency. Generally, the temperature of the composite tends to stabilise at a certain value after a certain number of cycles, and this equilibrium temperature also increases with increasing load level or loading frequency. The main parameters that

2. Materials and methods 2.1. Material configurations and parameters The stacking and thickness variations of the rib used for the experiments of the present study corresponded to those of a real structure, namely, the rib of a large aerospace structure [49,50]. The laminate thickness varied between 3.4 and 5 mm. The plates were fabricated from carbon-fibre fabric prepreg according to AIMS 05-09-002 [51] 254


R. Růžek et al.

Fig. 1. Schematic of the layup of the tapered section of Configuration A with the nominal dimensions.

with an actual weight of 285 ± 15 g/m2. The fabric had a five-harness satin weave, which is more flexible for producing a curved surface compared to a plain weave. The matrix consisted of polyphenylene sulphide (PPS) with a weight ratio of 43%. The nominal thickness of the lamina was 0.31 mm. Two ply drop configurations were investigated by considering the thickness transition areas of the actual design of the structural component. Configuration A contained two transitions from 16 to 13 layers and from 13 to 11 layers, respectively (Fig. 1). The layup of Configuration A was coded as follows: (0/90, ± 45)∫ (0/90)| ± 45| (0/ 90, ± 45, 0/90)∫ ± 45| ( ± 45, 0/90, ± 45, 0/90)∫ ± 45| (0/90)| ( ± 45, 0/90)∫ , where ∫ denotes continuous plies, and | denotes ply drops (cut lamina ends). The tapered angle of Configuration A was 3.5°. The actual thicknesses of the different sections were 5.10 mm for 16 layers, 4.15 mm for 13 layers, and 3.55 mm for 11 layers. Configuration B contained one transition from 16 to 11 layers (Fig. 2). The layup of the configuration was coded as follows: (0/ 90, ± 45)∫ (0/90, ± 45)| (0/90, ± 45, 0/90)∫ ± 45| ( ± 45, 0/ 90, ± 45, 0/90)∫ ( ± 45, 0/90)| ( ± 45, 0/90)∫ . The tapered angle of Configuration B was 6°. The actual thicknesses of the different sections were 5.05 mm for 16 layers and 3.53 mm for 11 layers. The plates had these transitions on each side; thus, there were 16 layers on each side and 11 layers in the middle. The internal ply drops were not exactly positioned in the columns as was intended, owing to the movement of the layers during the layup and curing processes. Deviations of up to several millimetres were observed in the cross sections at the related ply drop positions in the longitudinal direction. Four plates measuring 400 × 400 mm were fabricated in two batches by hot forming. Two of the plates were fabricated with Configuration A (specimens 2-X and 3-X), and the other two with Configuration B (specimens 1-X and 4-X). Twelve specimens were extracted from each plate using a water jet cutter (non-milled). Specimens 3-X and 4-X were further machined (milling) at their edges for better

surface quality, and were thus expected to exhibit better fatigue characteristics. The specimens were 370 mm long and 25 mm wide.

2.2. Test procedures The experiments were performed at room temperature. At first the static tests were conducted according to ASTM D3039M-08 [52] using Instron 5500R1185 test equipment and ± 100 kN load cell. The crosshead displacement rate was 1 mm/min. The fatigue tests were performed according to ASTM D3479M [53] on IST Hydropuls Sinus 100 kN and Schenck 250 kN test benches using frequencies of 0.5–15 Hz. A tension-tension fatigue test using a stress ratio of 0.05 was employed because it is considered to be the most appropriate test for the investigated structural part. The average ultimate static strength of Alternative A was 533 MPa, whereas that of Alternative B was 551 MPa (3% higher than that of Alternative A). However, the difference was not statistically significant. The maximum cyclic stress levels were selected to achieve a minimum fatigue life of 104 cycles. During the fatigue loading, the surface temperature of the selected specimens was measured using an FLIR E60 infrared camera (Fig. 3) with a sensitivity of < 0.07 °C and precision of 2 °C. The temperature was either measured at predefined intervals or continuously recorded to the memory card. The measurements were adjusted such that the warmest point on the specimen surface was recorded. The measured specimens were selected to obtain data under different loading conditions over the entire range of the considered frequencies (f = 0.5–15 Hz). Selected test specimens were tested under various loading conditions (frequency) to demonstrate different stabilised temperature levels; these specimens are not included into the S-N curves.

Fig. 2. Schematic of the layup of the tapered section of Configuration B with the nominal dimensions. 255


R. Růžek et al.

cycling frequency for the non-milled specimens. As can be observed, the increase of the frequency from 0.5 to 4–7 Hz decreases the fatigue life by 10 orders of magnitude when only jet cutting is used. The measured temperature for frequencies of 4–7 Hz was 35 °C. In the present study, the typical fracture was observed to be initiated by a ply drop in the thinnest part of the component. Fig. 5 shows the effect of the cycling frequency for the milled specimens. No degradation of the fatigue life can be observed when the frequency is increased from 0.5 to 5 Hz. However, significant decreases in the fatigue life occur when the frequency is increased to 10 Hz, and then to 15 Hz. The stable temperatures for these frequencies were above 50 °C. The typical fracture was initiated at the 0/90 ply drop near the thinnest part. At the higher frequencies, more fractures were initiated away from the ply drops. A smoother thickness transition (from 11 to 13, and then to 16 plies) does not induce local concentration, with the fatigue failure initiated in a larger area. 3.2. Temperature The highest temperature area always initially occurred at the middle of the specimen, where the thickness was constant (11 plies). This area was also characterised by the highest stress level. During the following fatigue cycling, the area with the highest temperature shifted closer to the thickness transition zone where failure eventually occurred. The global temperature increase was caused by rapid cyclic deformation, with no global microcracking observed in the cross section. The highest temperature increase corresponded to the area of the initiation of fatigue cracking (delamination) and propagation. Examples of the movement of the area of highest temperature from the middle part of the specimen with constant thickness to the failure area are shown in Fig. 6 (test frequency of 10 Hz). In the first stage of the fatigue test, the warmest point at a temperature of 32.2 °C occurred in the middle part of the specimen with constant thickness (Fig. 6a). With continuous loading, the heat produced by rapid deformation accumulated and temperature was stabilised at 64.6 °C (Fig. 6b). This warmest point had also moved toward the thickness transition area (ply drop) where the initial delamination shearing caused an even higher temperature. Fig. 6c shows the area with the highest temperature of 85.6 °C just before failure, while Fig. 6d shows the specimen just after failure, with disconnection of the failed ends. The failure was followed by rapid temperature decrease. The evolutions of the measured temperature during the fatigue testing of different specimens under different load conditions are shown in Figs. 7–12. The initial phase during which the temperature rapidly increases can be recognised for the higher frequencies. After several cycles, the temperature stabilises at a certain value determined by the maximum stress level, test frequency, and layup configuration. The figures reveal no significant effect of the additional milling of the specimen edge after water-jet cutting, with regard to the temperature evolution. The stabilised temperature is achieved between 5000 and 20,000 cycles, with the maximum observed value being 75 °C. This value is relatively close to the glass transition temperature of C/PPS composites (Tg = 98 °C). The fatigue life is significantly decreased in the particular case of this maximum temperature. Further, no surface temperature changes can be observed for a frequency of 0.5 Hz (see Figs. 7 and 8), and the stabilised temperature was also not attained for one specimen loaded with a frequency of 10 Hz (see Fig. 8). When the stabilised surface temperature was lower than 38 °C, no relationship was observed between the temperature and the fatigue life of the milled specimens. Conversely, when the stabilised surface temperature exceeded 75 °C, a significant reduction in fatigue life was observed in all the cases. During continuous fatigue loading, the temperature stabilised when the generated heat was being commensurately conducted away. Subsequently, cracks were initiated near the transition area and the temperature began to increase gradually again. The warmest point

Fig. 3. Fatigue test set-up with focused infrared camera.

2.3. Morphological and fractographic analysis methods Fractographic analyses of the specimens were performed after the fatigue testing. A diamond-coated saw was used to cut the selected specimens, maintaining the entire fracture surface. Pressurised air was used to remove the saw debris from the surface. Each specimen was then mounted on a pin stub by means of a double-sided sticky target and coated with a 10-nm-thick Au/Pt layer using a sputter coater. Microfractographic analyses of the fracture surfaces were conducted using a TESCAN VEGA 3SBU scanning electron microscope (SEM) set to the secondary electron imaging mode. The high vacuum mode with a voltage of 10 kV was used for the evaluation. The adjusted microscopy parameters used to obtain each image were noted below the image. In the case of the edgewise ply drop, to describe the delamination, a precise linear metallographic saw (IsoMet 4000) was used to cut each selected specimen in the longitudinal direction to observe the central plane along the length. A diamond wafering blade designed for composite materials was employed in the saw. The blade was cooled by a cutting fluid during operation at a velocity of 3000 rpm with an automatic feed rate of 3.3 mm/min. The specimens were laid on their cut sides at the bottom of a silicone mould and placed in a metallographic resin (Epoplast) via vacuum impregnation. After curing for 6 h, a common metallographic specimen preparation method involving grinding and polishing was implemented on the specimens using the KOMPAKT 1031 grinder-polisher. Wet-rough and fine grindings using silicon carbide abrasive discs (grades P320-P600-P1200) were initially performed. This was followed by polishing processes using polishing cloths holding a water-based monocrystalline diamond suspension (MetaDi) with grain sizes of 6 and 3 µm, respectively. An aluminium oxide suspension (MasterPrep) was used for the final lapping. The specimens were observed by a metallographic microscope (Olympus GX-51) and subsequently photographed by a digital camera (ARTRAY ARTCAM-300MI) using the QuickPHOTO Industrial software. 3. Results 3.1. Fatigue life The fatigue test results were prepared in the form of graphs of the maximum stress versus the fatigue life. Fig. 4 shows the effect of the 256


R. Růžek et al.

600

Configuration B Configuration A

σ max ( MPa )

500

f = 0.5 Hz

f = 4 - 7 Hz

400

f = 4 - 7 Hz

Run out

300 1 000

R = 0.05

.

10 000

100 000

N ( cycles )

1 000 000

Fig. 4. Effect of the edge cycling frequency for the non-milling specimens. An increase of the frequency from 0.5 to 4–7 Hz decreases the fatigue life by 10 orders of magnitude.

Fig. 5. Effect of the edge cycling frequency for the edge-milling specimens (0.5–15 Hz). An increase of the frequency from 5 to 15 Hz decreases the fatigue life by more than 10 orders of magnitude.

exceeds the glass transition temperature of C/PPS composites. The temperature increase apparently enables the damage detection during the fatigue testing. A good correlation between the measured temperature and the number of cycles to failure was established also in [54]. Thus the temperature is suitable parameter for structural health monitoring which enable to interrupt fatigue test before the total destruction of the specimen.

moved toward the area, where failure occurred. The temperature evolution until failure is shown in Figs. 9–12 with respect to the loading frequency, for comparison with the steady-state temperature. The two specimens (Figs. 9 and 11) tested under various loading conditions were used only for demonstration of stabilised temperature levels. As can be seen from Figs. 9–12, the temperature increase is significant before the failure of the specimen. In addition, the difference between the temperature in the steady-state phase of the fatigue process and the surface specimen temperature just before failure may exceed 30 °C. Fig. 10 shows that configuration B results in a shorter fatigue life than configuration A with the same frequency. It leads to conclusion that the higher stress concentration affects the fatigue life. The S-N curves of the non-milled specimens (Fig. 4) favour this conclusion. The difference is determined by the parameters of the fatigue loading. The maximum temperature just before the specimen fails is higher than 110 °C, which

3.3. Fractography The edge appearance of the specimen was first evaluated by macrophotography. Based on the location, there were two possible types of failure: failure at a ply drop and failure away from a ply drop. Fig. 13 shows macro-photographs of both types of failures in a Configuration B specimen. The ply drops were not exactly positioned in a column but 257


R. Růžek et al.

Fig. 6. Thermo-images of the specimen during fatigue test (frequency of 10 Hz): (a) Initial temperature increase – warmest point in the middle; (b) equilibrium temperature; (c) final increase of temperature, warmest point moves to the stress concentration; (d) final fracture.

Fig. 7. Temperature evolution up to stabilisation during cyclic loading with respect to the frequency (maximum stress of 420–434 MPa).

Fig. 9. Temperature evolution during cyclic loading until failure with respect to the frequency (maximum stress level of 420 MPa, stress ratio of 0.05). 45

Temperature (°C)

40 35 30

f = 4 Hz

25 20

Alternative A; Max stress 434.5 MPa Alternative B; Max stress 445 MPa 0

10 000

20 000

30 000

40 000

N (cycles)

Fig. 8. Temperature evolution up to stabilisation during cyclic loading with respect to the frequency (maximum stress level of 466.5 MPa).

Fig. 10. Temperature evolution during cyclic loading until failure with respect to the ply drop configuration (frequency of 4 Hz, stress ratio of 0.05).

scattered within ± 5 mm. The actual positions of the drops are visible in the cross-sectional images in Figs. 14–16. Observation of the longitudinal cross sections of the selected specimens revealed the actual ply drops as well as the cracks caused by the loading. Delaminations were found near the 0/90 ply drops for both types of ply drop configurations, but none was observed at the ± 45 ply drops. In some specimens, compression failure was observed near the

upper ply drop at the end of the delamination. This was due to compression impact waves created after final specimen fracture in the length wise opposite part of the specimen. It is quite typical for static tension testing where there are two fractures in the gauge section, one in tension and one in compression. It shows how the delamination near the ply drop weakens the composite for compression loading and the

258


R. Růžek et al.

Configuration B specimen with edge milling after 34,000 cycles. Fig. 15b shows the details of the ply drops, with a crack initiated from the 0/90° ply drop. Fig. 16a shows the cross section of a Configuration B specimen without milling, with a fracture in the gauge section far from the ply drops, caused by a compression loading wave from the opposite side of the specimen. The occurrence of this type of loading was confirmed by the observation of microbuckling, which is typical of a compression load and is visible in Fig. 16b. No crack initiation was observed in this specimen. The selected specimens were observed by a scanning electron microscope (SEM) in the secondary electron imaging mode to determine the effect of the load type. Fig. 17 shows the two observed failure types. The left panels of the figure show a fracture that was initiated away from a ply drop where tensile-dominant failure occurred. The right panels show a fracture that was initiated at a ply drop visible in an area where shear (mode II)-dominant failure was observed. Fig. 17a shows a static fracture with the fibres covered by a smooth smeared matrix. Cohesive failure with no damage of the fibre/matrix interface is apparent. Fig. 17c shows a fatigue fracture at 4 Hz with ductile drawing of the forming fibrils of the deformed matrix, smeared across the fracture surface. Fibre/matrix failure and additional matrix deformation (fibrils) are visible, in contrast to the case of static fracture. Fig. 17b shows a fatigue fracture at 0.5 Hz on a ply drop with apparent mode II features. Resin-dominant failure with textured microflow and rounded cusps can be observed in this case. The cusps are typical of mode II failure. The rounded shapes are due to fatigue loading. Fig. 17d shows a fatigue fracture at 15 Hz with the fibrillation caused by significant matrix deformation. The rollers that are typical of mode II fatigue are apparent. The matrix is not smeared, with the bare carbon fibres visible. This is because cyclic loading induces preferential damage at fibre/matrix interfaces.

Fig. 11. Temperature evolution during cyclic loading until failure with respect to the frequency (maximum stress level of 466.5 MPa, stress ratio of 0.05).

Max stress 434.5 MPa

Temperature (°C)

100

80

f = 15 Hz 60

40

Configuration A 20 0

10 000

20 000

30 000

40 000

N (cycles) Fig. 12. Temperature evolution during cyclic loading at a frequency of 15 Hz until failure (maximum stress level of 434.5 MPa, stress ratio of 0.05).

4. Discussion The above observations reveal a strong correlation between the loading parameters and the material temperature, and that the temperature affects the fatigue life of a component fabricated from the material. Shifting of a local ply drop causes friction between the plies, resulting in a temperature increase beyond acceptable levels. Vieille et al. [14] and Franco et al. [17] noted that the mechanical properties of notched and un-notched angle-ply laminates subjected to temperatures higher than the glass transition temperature of C/PPS composites (Tg = 98 °C) may be severely degraded. However, the results of the present study indicate that, in the event of ply drop decomposition, the critical temperature that causes degradation of the fatigue properties is significantly lower. The critical temperature is about 40 °C, as determined from the effect of frequency on the fatigue life and the temperature evolution trend in the present study. Actual temperature value is significantly determined by the loading parameters. Fig. 18 shows a linear relationship between the temperature and the load rate. The load rate in MPa·s−1 is defined as the ratio of the stress amplitude to the loading frequency. By linear regression, the following equation of the stabilised temperature was obtained:

Fig. 13. Typical macro-photographs of failures (a) away from a ply drop (Configuration A, without milling of the specimen) and (b) at a ply drop (Configuration B, with milling of the specimen edge).

fatigue loaded ply drop becomes the weakest link of the structure. In the case of the Configuration A specimens with two tapers, crack initiation was observed at both 0/90 ply drops near the gauge section, with subsequent delamination and intralaminar growth occurring toward the thick section in both the milled and un-milled specimens (see Fig. 14). This can be explained by the higher stress in the thinner tapered parts, as well as the critical condition engendered by the high stiffness of the 0/90 ply drop. The cracks grew both between the carbon tows (intralaminar cracks) and between the layers along the 0 tows (interlaminar crack also called delaminations). In the case of a Configuration B specimen with one taper, the ply drops that were intended to be in a single column were highly scattered. This can be observed from Fig. 15a, which shows the cross section of a

(1)

T = 0.0091v + 21.6, −1

where T is the temperature, v is the load rate [MPa·s ], and 21.6 °C is the ambient temperature. The equation affords an important tool for the design and control of fatigue experiments performed on composites with ply drops. The present results indicate that temperatures higher than 40 °C are attained for loading frequencies of 4–7 Hz, or load rates above 2000 MPa·s−1. Such extreme conditions significantly impact the fatigue life, and the foregoing critical parameters can be adopted as criteria for designing fatigue tests using specimens with thicknesses and tapering configurations similar to those of the specimens employed in the present study. Machining of the specimens was also observed to have a 259


R. Růžek et al.

Fig. 14. Cross section of a Configuration A specimen after 62,000 cycles: (a) overview of ply drops, (b) delamination/intralaminar growth from the upper 0/90 ply drop toward the thick section and compression microbuckling failure near the delamination due to the energy released after tensile failure on the other side of the specimen, and (c) crack initiated from the bottom 0/90 ply drop.

Fig. 15. (a) Cross section of a Configuration B specimen with milling after 34,000 cycles, showing the scattered ply drops, and (b) details of a delamination in the specimen.

Fig. 16. Cross section of (a) a Configuration A specimen without milling, showing a fracture in the gauge section far from the ply drops caused by a compression wave load from the opposite side of the specimen. No crack initiation was observed. (b) Details of the microbuckling typically caused by a compression load.

260


R. Růžek et al.

Fig. 17. SEM images of the fracture surfaces for failure away from a ply drop showing (a) the cohesive smooth failure under static loading, and (c) ductile drawing under 4-Hz fatigue failure; and ply drop interface showing (b) rounded cusps under micro-textured 0.5-Hz fatigue failure, and (d) matrix rollers and debris under 15Hz fatigue failure.

movement of the ply drops during moulding. Nevertheless, the positions of the ply drops affected the fatigue initiation points because the ply drops near the gauge section were more susceptible to damage. Fatigue cracking was observed to be initiated at only the 0/90 ply drops, with no delamination observed at the ± 45 ply drops in the specimens with a second taper. In the case of the Configuration A specimens (two tapers), cracking was initiated only in the parts of the tapers near the gauge section. This crack initiation near the thin region agrees with the observations in [55], where fatigue failure is considered using R = −1, and the first two ply drops close to the thin section are noted to be the most critical for damage initiation. In the case of laminates with ply drops in multiple steps [56], delamination begins at the terminal ply group closest to the thinnest (gauge) region. An optimal approach would be to end the 0° plies in the thicker section, and the 45° plies in the thinner section where the stress is higher. The absence of river lines in a thermoplastic makes it difficult to detect crack growth from SEM images of a fracture surface. No beach marks or striations were observed in the SEM images obtained in the present study. The failure position was, however, determined to affect the fracture surface. A smooth smeared matrix was observed in the case of tensile failure away from a ply drop, and mode II (in-plane shear) failure features were also apparent in the case of fracturing initiated at the ply drops. The direction of the crack growth was driven by the shear stress present in this region (Fig. 19). The shear stress direction caused crack opening in diagonal direction perpendicular to major stress as

Fig. 18. Relationship between the temperature (steady-state phase) and the load rate.

significant effect. The surfaces of the un-milled specimens are of lower quality compared to milled specimens, and the resultant lower loading rate of the former should be taken into consideration. Based on the data in Fig. 4, a maximum load rate of 1500 MPa·s−1 can be adopted for unmilled specimens. Significant scatter was observed in the horizontal positioning of the ply drops. This may be due to errors in the manual positioning or 261


R. Růžek et al.

Fig. 19. (a) Cross section of a specimen after 34,000 cycles, showing fatigue crack growth initiated on 0/90 ply drop and (b) the fracture mechanism explained by the shear forces and local diagonal forces causing the crack opening.

specific relationship with the fatigue life. However, further increase of the surface temperature up to beyond 75 °C was accompanied by significant reduction of the fatigue life. Based on the findings, the criteria for designing a fatigue test and implementing certain thicknesses and tapering configurations were defined. Temperature increase was identified as a factor that affected damage detection during a fatigue test, and the parameter can thus be used to appropriately terminate a fatigue test before total destruction of the test specimen. This can also be applied to structural health monitoring. Further, the matrix was observed to experience only small plastic deformations at a low loading frequency, with apparent textured microflow. At a high frequency, larger matrix failure features such as rollers and debris were found. The fibrils that were also produced by the large matrix deformation could be attributed to the higher temperature generated by a higher loading frequency.

indicated in Fig. 19b. Therefore, the crack propagated by intralaminar cracking between fibres to the upper surface where it followed the nearest 0° ply driven by the in-plane shear (mode II) loading. The matrix fracture of a thermoplastic may be characterised by extensive wear and plasticity, in addition to being rate-dependent [57]. Further, the load rate of the present specimens were observed to affect the specimen temperature owing to the occurrence of wear and cyclic deformation. Under a quasi-static load, fracturing occurred with the fibres covered by a smooth smeared matrix, indicating cohesive failure, but there was no apparent damage of the fibre/matrix interface. Conversely, for low cyclic loading frequencies, the matrix experienced small plastic deformations and textured microflow was apparent. Moreover, the cross sections of the specimens revealed that the fatigue cracks preferentially grew between the carbon tows. At high frequencies, larger matrix failure features such as rollers and debris were found, with the rounded cusps observed at low frequencies altered into matrix rollers. The observed fibrils produced by large matrix deformation may be attributed to the higher temperature induced by higher loading frequencies. Ref. [45] stated that fatigue failure mechanisms are very similar to those under static loading. It corresponds with the presented results where smeared matrix was observed on the fracture surfaces. However, higher degree of cohesive fracture was found for static loading than for fatigue loading It is also believed that the main driving force of the failure process is the straightening of the kinked (not straight) cover plies in the tapering region. However, the layers of the present specimens were pressed into each other with no significant kink near the ply drop. Straightening of the cover plies was thus not a relevant issue for the ply drop failure region. Nevertheless, for the fracture away from ply drop, the straitening in the tapering region could cause out of plane stress to some degree especially for configuration B with more steep transition.

Acknowledgements This study was funded by an institutional support from the Ministry of Industry and Trade of the Czech Republic. References [1] Reyes G, Sharma U. Modeling and damage repair of woven thermoplastic composites subjected to low velocity impact. Compos Struct 2010;92(2):523–31. http:// dx.doi.org/10.1016/j.compstruct.2009.08.038. [2] Maron B, Garthaus Ch, Hornig A, Lenz F, Hübner M, Gude M. Forming of carbon fiber reinforced thermoplastic composite tubes – experimental and numerical approaches. J Manuf Sci Technol, Available online 7 October 2016, ISSN 1755–5817. http://dx.doi.org/10.1016/j.cirpj.2016.09.004. [3] Boria S, Scattina A, Belingardi G. Experimental evaluation of a fully recyclable thermoplastic composite. Compos Struct 2016;140:21–35. http://dx.doi.org/10. 1016/j.compstruct.2015.12.049. [4] Vodicka R. Thermoplastics for Airframe Applications: A Review of the Properties and Repair Methods for Thermoplastic Composites, Defense Science and Technology Organization, Australian Government of Defense, Report n◦ DSTOTR0424; 2006. < https://www.abbottaerospace.com/wpdm-package/dsto-tr-0424thermoplastics-for-airframe-applications-a-review-of-the-properties-and-repairmethods-for-thermoplastic-composites > . [5] Myers FA. Advances in Thermoplastic Matrix Composite Materials. ASTM STP 1044; 1989. p. 154–62, ISBN-EB: 978-0-8031-5099-7, http://dx.doi.org/10.1520/ STP1044-EB. [6] Gabrion X, Placet V, Trivaudey F, Boubakar L. About the thermo-mechanical behaviour of a carbon fibre reinforced high-temperature thermoplastic composite. Compos Part B Eng 2016;95:386–94. http://dx.doi.org/10.1016/j.compositesb. 2016.03.068. [7] Jen MHR, Tseng YC, Kung HK, Huang JC. Fatigue response of APC-2 composite laminates at elevated temperatures. Compos Part B Eng 2008;39(7–8):1142–6. http://dx.doi.org/10.1016/j.compositesb.2008.03.005. [8] Mahieux CA, Scheurer C. Elevated temperature bending stress rupture behavior AS4/APC-2 and comparison with AS4/PPS literature data. Compos Part A Appl Sci Manuf 2002;33(7):935–8. http://dx.doi.org/10.1016/S1359-835X(02)00039-8. [9] Parlevliet PP, Bersee HEN, Beukers A. Residual stresses in thermoplastic compositesa study of the literature. Part III: effects of thermal residual stresses. Compos Part A Appl Sci Manuf 2007:38(6);1581–1596. http://dx.doi.org/10.1016/j.compositesa. 2006.12.005. [10] Colombo C, Libonati F, Vergani L. Fatigue damage in GFRP. Int J Struct Integrity 2012;3(4):424–40. http://dx.doi.org/10.1108/17579861211281218. [11] Kawai M, Yajima S, Hachinohe A, Kawase Y. High-temperature off-axis fatigue behavior of unidirectional carbon-fibre-reinforced composites with different resin matrices. Compos Sci Technol 2001;61(9):1285–302. http://dx.doi.org/10.1016/

5. Conclusion This study considered the effects of the cyclic loading frequency on the temperature, fatigue behaviour, and failure mechanisms of carbonfibre-fabric-reinforced PPS laminates, the thickness of which was varied by ply drops. The main objective was to evaluate the material characteristics and parameters that predominantly affected the fatigue characteristics of the material, especially with reference to the loading frequency. The effects of the two ply drop configurations under fatigue loading frequencies of 0.5–15 Hz were also investigated. The loading frequency was found to significantly affect the fatigue life of the specimens, with the fatigue life decreasing by more than one order of magnitude with increasing loading frequency. The local temperature also increased significantly, reaching as high as above 110 °C. The temperature initially increased rapidly, but later stabilised at a certain level determined by the load parameters. For a loading frequency of 0.5 Hz, no surface temperature change was observed. Moreover, for a surface temperature of up to 38 °C, there was no 262


R. Růžek et al.

S0266-3538(01)00027-6. [12] Kawai M, Masuko Y, Kawase Y, Negishi R. Micromechanical analysis of the off-axis rate-dependent inelastic behavior of unidirectional AS4/PEEK at high temperature. J Mech Sci 2001:43;2069–2090, ISSN: 0020–7403. http://dx.doi.org/10.1016/ S0020-7403(01)00029-7. [13] Uematsu Y, Kitamura T, Ohtani R. Delamination behavior of a carbon-fiber-reinforced thermoplastic polymer at high temperatures. Compos Sci Technol 1995;53(3):333–41. http://dx.doi.org/10.1016/0266-3538(95)00005-4. [14] Vieille B, Aucher J, Taleb L. Influence of temperature on the behavior of carbon. Mater Sci Eng, A 2009;517(1–2):51–60. http://dx.doi.org/10.1016/j.msea.2009. 03.038. [15] De Baere I, Van Paepegem W, Degrieck J. Modelling the nonlinear shear stress–strain behavior of a carbon fabric reinforced polyphenylene sulphide from rail shear and [(45°−45°)]4s tensile test. Polym Compos 2009;30(7):1016–26. http://dx.doi. org/10.1002/pc.20650. [16] Vieille B, Aucher J, Taleb L. Comparative study on the behavior of woven-ply reinforced thermoplastic or thermosetting laminates under severe environmental conditions. Mater Des 2012;35:707–19. http://dx.doi.org/10.1016/j.matdes.2011. 10.037. [17] Franco LAL, Graca MLA, Silva FS. Fractography analysis and fatigue of thermoplastic composite laminates at different environmental conditions. Mater Sci Eng, A 2008;488(1–2):505–13. http://dx.doi.org/10.1016/j.msea.2007.11.053. [18] Reifsnider K, Case S. Strength-based predictive methods for large-scale combined degradation effects. In: Proceedings of 39th AIAA/ASME/ASCE/AHS/ASC structures, structural dynamics, and materials conference and exhibit, Long Beach, CA, Apr. 20–23, 1998. http://dx.doi.org/10.2514/6.1998-1739. [19] Tserpes KI, Ruzek R, Pantelakis Sp. Strength of Pi shaped non-crimp fabric adhesively bonded joints. Plast Rubber Compos 2012:41(2). http://dx.doi.org/10.1179/ 1743289810Y.0000000036. [20] Loverich JS, Russel BE, Case SW, Reifnider KL. Life prediction of PPS composites subjected to cyclic loading at elevated temperatures, ASTM STP 1357; 2000. p. 310–17. ISBN-EB: 978–0-8031-5417-9, http://dx.doi.org/10.1520/STP15840S. [21] Mahieux CA, Scheurer C. Elevated temperature bending stress rupture behavior AS4/APC-2 and comparison with AS4/PPS literature data. Composites Part A 2002;33:935–8. http://dx.doi.org/10.1016/S1359-835X(02)00039-8. [22] Bellenger V, Decelle J, Huet N. Ageing of a carbon epoxy composite for aeronautic applications. Compos B 2005;36:189–94. http://dx.doi.org/10.1016/j.compositesb. 2004.04.016. [23] Selzer R, Friedrich K. Mechanical properties and failure behaviour of carbon fibrereinforced polymer composites under the influence of moisture. Composites Part A 1997;28(6):595–604. http://dx.doi.org/10.1016/S1359-835X(96)00154-6. [24] Ma CCM, Yur SW. Environmental effects on the water absorption and mechanical properties of carbon fiber reinforced PPS and PEEK composites. Part II, Polym Eng Sci 1991;31(1):34–9. http://dx.doi.org/10.1002/pen.760310107. [25] John NASt, Brown JR. Flexural and interlaminar shear properties of glass-reinforced phenolic composites. Composites Part A 1998;29(8):939–46. http://dx.doi. org/10.1016/S1359-835X(98)00019-0. [26] Earl JS, Shenoi RA. Hygrothermal ageing effects on FRP laminate and structural foam materials. Composites Part A 2004;35(11):1237–47. http://dx.doi.org/10. 1016/j.compositesa.2004.04.007. [27] Shen CH, Springer GS, editors. Environmental effects on composite materials. Westport, CT: Technomic Pub. Co., Inc; 1981. p. 15–33, ISBN: 0-87762-300-7. [28] Imaz J, Rodriguez JL, Rubio A, Mondragon I. Hydrothemal environment influence on water diffusion and mechanical behaviour of carbon fibre/epoxy laminatesm. J Mater Sci Lett 1991;10(11):662–5. http://dx.doi.org/10.1007/BF00723372. [29] Wolff EG. In: Jones FR, editor. Handbook of polymer-fibre composites. Harlow; Longman Publishing Group; 1994. p. 362–6, ISBN-13: 978–0582065543. [30] Collings TA, Harvey RJ, Dalziel AW. The use of elevated temperature in the structural testing of FRP components for simulating the effects of hot and wet environmental exposure. Composites 1993;24(8):625–34. http://dx.doi.org/10.1016/ 0010-4361(93)90125-R. [31] Talreja R. Fatigue of composite materials. In: Altenbach H, Becker W, editors. Modern trends in composite laminates mechanics. International Centre for Mechanical Sciences (Courses and Lectures), vol. 448. Vienna: Springer; 1976, ISBN 978-3-7091-2544-1, http://dx.doi.org/10.1007/978-3-7091-2544-1_6. [32] Dan-Jumbo E, Zhou SG, Sun CT. Load-frequency effect on fatigue life of IMP6/APC2 thermoplastic composite laminates. In: Newaz GM, editor. Advances in thermoplastic matrix composite materials, ASTM STP 1044; 1989. p. 113–32, ISBN-EB: 978–0-8031-5099-7, http://dx.doi.org/10.1520/STP24598S. [33] Franco LAL. Fractography of composite materials. In: Luigi N, editor. Encyclopedia of composites. Wiley & Sons; 2011. p. 1–16. http://dx.doi.org/10.1002/ 9781118097298.weoc094. [34] van den Oever M, Peijs T. Continuous-glass-fibre-reinforced polypropylene composites II. Influence of maleic-anhydride modified polypropylene on fatigue behaviour. Composites Part A 1998;29(3):227–39. http://dx.doi.org/10.1016/S1359835X(97)00089-4. [35] Costa J, Turon A, Trias D, Blanco N, Mayugo JA. A progressive damage model for unidirectional fibre-reinforced composites based on fibre fragmentation. Part II:

[36]

[37]

[38]

[39] [40]

[41]

[42]

[43] [44]

[45]

[46]

[47]

[48]

[49]

[50]

[51]

[52]

[53]

[54]

[55]

[56]

[57]

263

Stiffness reduction in environment sensitive fibres under fatigue. Compos Sci Technol 2005;65(14):2269–75. http://dx.doi.org/10.1016/j.compscitech.2005.05. 011. Reifsnider KL, Stinchomb WW, O'Brien TK. Frequency effect on a stiffness-based fatigue failure criterion in Bawed composite specimens. In: Reifsnider KL, Lauraitis KN, editors. Fatigue of filarnentary composite materials, ASTM STP 636; 1977. p. 171–84. http://dx.doi.org/10.1520/STP27977S. Curtis DC, Moore DR, Slatter B, Zahlan N. Fatigue testing of multi angle laminates of CF/PEEK. Composites 1988;19(6):446–52. http://dx.doi.org/10.1016/00104361(88)90702-. Dally JW, Broutman LJ. Frequency effect on the fatigue of glass reinforced plastics. J Compos Mater 1967;1:424–42. http://dx.doi.org/10.1177/ 002199836700100410. Rui Y, Sun CT. Cyclic plasticity in AS4PEEK composite laminates. J Thermoplastic Compos Mater 1993:6;312–22. Kujawski D, Ellyin F. Rate/frequency-dependent behavior of fiberglass/epoxy laminates in tensile and cyclic loading. Composites 1995;26(10):719–23. http://dx. doi.org/10.1016/0010-4361(95)91139-V. Sun CT, Chan WS. Frequency effect on the fatigue life of a laminated composite. In: Tsai SW, editors. Composite materials; testing and design (fifth conference). ASTM STP 674; 1979. p. 418–30. http://dx.doi.org/10.1520/STP36921S. He K, Hoa S, Ganesan R. The study of tapered laminated composite structures: a review. Compos Sci Technol 2000;60(14):2643–57. http://dx.doi.org/10.1016/ S0266-3538(00)00138-X. Hoa SV, Daoust J, Du BL, Vu-Khanh T. Interlaminar stresses in tapered laminates. Polym Compos 1988;9:337–44. http://dx.doi.org/10.1002/pc.750090506. Helmy S, Hoa SV. Tensile fatigue behavior of tapered glass fiber reinforced epoxy composites containing nanoclay. Compos Sci Technol 2014;102:10–9. http://dx. doi.org/10.1016/j.compscitech.2014.05.038. Wisnom MR, Dixon R, Hill G. Delamination in asymmetrically tapered composites loaded in tension. Compos Struct 1996;35(3):309–22. http://dx.doi.org/10.1016/ 0263-8223(96)00044-X. Thawre MM, Verma KK, Jagannathan N, Peshwe DR, Paretkar RK, Manjunatha CM. Effect of ply-drop on fatigue life of a carbon fiber composite under a fighter aircraft spectrum load sequence. Compos Part B Eng 2016;86:120–5. http://dx.doi.org/10. 1016/j.compositesb.2015.10.002. Gan KW, Allegri G, Hallett SR. A simplified layered beam approach for predicting ply drop delamination in thick composite laminates. Mater Des 2016;108:570–80. http://dx.doi.org/10.1016/j.matdes.2016.06.105. Šedek J, Hron R, Kadlec M. Bond joint analysis of thermoplastic composite made from stacked tailored blanks. Appl Mech Mater, 2016:827;161–168. ISSN 16627482. 10.4028/www.scientific.net/AMM.827.161, https://www.scientific.net/ AMM.827.161. Šedek J. Optimization of composite airframe rib using tailoring blank technology. In: 54th international conference on experimental stress analysis (EAN 2016), May 30–June 2, 2016, Srní, Czech Republic, Czech Society for Mechanics (CSM), Curran Associates, Inc. (Aug 2016). p. 401–404, ISBN: 978-1-5108-2670-0, http://www. proceedings.com/31161.html. Růžek R, Šedek J, Kadlec M, Kucharský P. Mechanical behavior of thermoplastic rib under loading representing real structure conditions. In: 54th international conference on experimental stress analysis (EAN 2016), May 30–June 2, 2016, Srni; Czech Republic; Code 123122, Czech Society for Mechanics (CSM), Curran Associates, Inc. (Aug 2016). p. 356–359, ISBN: 978-1-5108-2670-0, http://www. proceedings.com/31161.html. AIMS 05-09-002. Airbus material specification/carbon fiber reinforced thermoplastic materials. Airbus Industrie Engineering Directorate 31707 Blachnac Cedex France: Airbus Industrie; 1998. ASTM D 3039/D3039M-08. Standard test method for tensile properties of polymer matrix composite materials, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428–2959, United States: ASTM International; 2008. ASTM D 3479/D3479M-12. Standard test method for tension-tension fatigue of polymer matrix composite materials, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA 19428-2959, United States: ASTM International; 2007. http:// dx.doi.org/10.1520/D3479_D3479M-12. Montesano J, Fawaz Y, Bougherara H. Use of infrared thermography to investigate the fatigue behaviour of a carbon fiber reinforced polymer composite. Compos Struct 2013;97:76–83. http://dx.doi.org/10.1016/j.compstruct.2012.09.046. Weiss A, Trabelsi W, Michel L, Barrau JJ, Mahdi S. Influence of ply-drop location on the fatigue behaviour of tapered composites laminates. Procedia Eng 2010;2(1):1105–14. http://dx.doi.org/10.1016/j.proeng.2010.03.119. Shim DJ. Role of delamination and interlaminar fatigue in the failure of laminates with ply drop offs. Massachusetts Institute of Technology, Dept. of Aeronautics and Astronautics, Thesis (Ph. D.), Cambridge, MA; 2002. http://hdl.handle.net/1721.1/ 29245. Greengalgh ES. Failure analysis and fractography of polymer composites. 1st ed. Cambridge: Woodhead Publishing; 2009. p. 608, ISBN: 1845696816, 9781845696818.