In many applications fibre reinforced plastics (FRP) and especially carbon fibre reinforced plastics (CFRP) are increasingly used for weight optimization due to their density specific mechanical properties. However, their sudden and brittle failure behaviour and the complex damage mechanisms involved require often conservative design. A forecast when and where damage is initiated is thus very difficult, leading to high safety margins and limiting their potential for lightweight design. The compressive strength is often a design limit of CFRP, as it is significantly lower than the tensile strength. For the brittle fracture behaviour of CFRP under compression, no conclusive and entire theory exists, which describes all relevant mechanisms during failure, especially at stress intensifications.
The first model for predicting compressive strength of composite laminates was presented by Rosen [1
]. He proposed that compressive failure initiates due to fibre microbuckling and distinguished between two modes of microbuckling: in-phase microbuckling (shear mode) for higher and out-of-phase microbuckling (extension mode) for lower fibre volume fractions [1
]. The in-phase microbuckling leads to the formation of a kink-band with increasing load. This is similar to the compressive failure of other fibrous materials like i.e., wood. Moran et al. [2
] and Poulsen et al. [3
] investigated the kinking failure in spruce and compared the damage process with that of CFRP. They identified three stages of compressive kinking. Incipient kinking as the first stage begins on a very small scale and is characterised by localised plastic shearing and buckling of fibres. In the following transient kinking stage the localised incipient kinking areas grow and coalesce to form a single dominant kink-band across the specimen. The last stage is steady state kinking during which the kink-band broadens laterally [2
]. In CFRP these three stages occur as well in a similar process [2
], but they are difficult to clearly identify because of the brittle fracture within a very short amount of time. Incipient kinking occurs when the matrix shear stress in fibre direction reaches a critical value [2
]. The analytical model of Budiansky et al. [5
] describes the initiation and propagation of such a kink-band with the orientation angle
, the kink-band width
and the inclination angle
of the fibres. Gutkin et al. [7
] described the initiation of microbuckling and the following kink-band as shear driven fibre failure with a distinct shearing angle
. Figure 1
sums up the different failure mechanisms of FRP under compressive loading with the nomenclature used in this work.
Initiation of kink-bands is facilitated at defects, e.g., voids [9
] or local fibre misalignment [11
]. Despite these flaws, the laminate properties and stacking sequence play an important role in damage initiation and propagation and the resulting mechanical properties under compressive loading. For composite laminates in general, size effects with regard to scaling of the specimens on the one hand and thickness of the constitutive plies on the other hand should be considered [12
]. Soutis [13
] and Lee and Soutis [14
] investigated the influence of specimen and layer thickness on the strength and failure behaviour for unnotched compression (UNC) [13
] and open-hole compression (OHC) [15
] load cases and compared the compression behaviour of open-hole specimens with that under tensile loading and found a strength increase with decreasing dimensions or ply thickness [17
]. Arteiro et al. [18
] showed with micromechanical modelling and finite element simulation that the in situ effect in tension [19
] exists as well under compression loading : They reported a higher ply strength with decreasing ply thickness and with increasing the stiffness of the surrounding layers [18
]. These ply thickness effects become more and more important with the possibilities of the thin-ply technology by using plies below
of thickness. The damage mechanisms for standard and thin-ply laminates with regard to ply thickness and stacking sequence effects under compression are not yet fully clear [20
] and require further research.
For composite laminates in many applications the influence of stress intensifications such as holes, notches or barely visible impact damage is critical. At holes or the edges of a laminate the free edge effect must be considered [22
]. Due to high peel stresses, delamination between two layers initiates at the edges (resulting from a mathematical stress singularity [24
]). With decreasing ply thickness, a suppression of these edge delaminations is reported [25
]. Under compression fibre microbuckling initiates at the hole boundary followed by delamination and formation of a kink-band [26
]. At free edges, compression failure is hence partly a result of delamination growth. With increasing hole diameter/specimen width ratio, compressive strength decreases [14
]. The OHC failure process depend on interlaminar toughness. A high interlaminar toughness leads to a short crack rest after being initiated at the hole before brittle failure occurs, whereas a weaker interlaminar interface results in sudden failure [28
]. Wang et al. [29
] compared experimental results for open-hole tension and compression to predictions of a numerical finite element method (FEM) analysis and pointed out, that the compressive strength of a lay-up with subsurface 0°-plies is higher than that of a unidirectional (UD) lay-up due to higher stability of the load bearing fibres [29
]. Thus, stacking sequence and optimum support of the 0°-layers carrying the highest load share is critical.
An open or filled hole is usually a design feature and can thus be considered for when designing a composite part. Impact events may occur during lifetime of a composite part and result in a stress intensification that is difficult to account for in the design process. Impact damage may be barely visible at the surface of a FRP laminate but result in severe damage such as matrix cracking, fibre breakage and delamination. Hence, the introduction of impact damage and compression after impact (CAI) properties are design limits. In addition, stacking sequence and scaling of the constitutive layers have an influence on resistance against impact damage [30
]. A comparison between ply-block scaled and sublaminate scaled laminates reveals that the increase of interfaces available for delamination in the distributed plies of sublaminate scaled laminates results in more, but less large delaminations. If the number of interfaces available for delamination is reduced, larger delaminations occur [25
]. This may be beneficial for thin-ply laminates with distributed plies, because of the high number of interfaces. Impact tests with different types of thin-ply laminates exhibit equal [20
] or larger [21
] delamination areas after impact with decreasing ply thickness. CAI strength is slightly improved with a significant decrease in ply thickness [30
], with the delamination being less severe [20
Delamination size is critical under compression as the constitutive layers are not supported at the delaminated areas. Fractography investigations from Greenhalgh et al. [35
] about delamination growth and migration revealed that migration between different interfaces is important, as it is the slower propagating mechanisms, resulting in a smaller projected damage area. When delamination driving force direction and fibre orientation of the adjacent layer are in the same axis this results in fast delamination growth with larger damage areas [35
]. In compression, delamination growth is preferably at interfaces with plies transverse to the loading direction, thus the 90°-layers are most critical for delamination propagation.
The subject of this work is an experimental investigation on the influence of ply thickness and stacking sequence in quasi-isotropic (QI) CFRP laminates containing stress intensifications under compression loading. The aim is to identify and discuss the different effects that influence the compression failure and the role the stacking sequence has on damage development and the resulting compressive strength. The influence of stress intensifications is investigated in detail at a hole in OHC tests, because with this type of stress intensification the exact shape as well as the stress state is well known and an interrupted test approach allows to identify the mechanisms of damage initiation and propagation from the free edge of the hole. CAI tests are executed in order to compare the OHC results to a different type of stress intensifications. UNC tests are carried out for comparison as a reference. We present an approach to use open-hole specimens for causing a distinct damage state and examine it at a precise instant of time during fracture process. Focus is thus set on failure initiation and propagation as well as the resulting final failure mode in OHC. With this method, a more detailed description of the failure mechanisms during the brittle compression failure of CFRP is achieved. In addition, the influence of the 0°-layer position with regard to damage initiation and the resulting mechanical properties is examined. The results contribute to a better understanding of the complex failure mechanisms under compression loading and may help to design composite parts accordingly.
Compression failure in FRP is dominated by the matrix properties. For the UNC, OHC and CAI tests, the same matrix system is used, to exclude this influence. The different matrix systems in the interrupted tests are very similar, so that the influence of the matrix properties for these tests is assumed to be small. Regarding the resistance against a low velocity impact, the smaller projected damage area of the sublaminate scaled laminates can be explained by the higher number of interfaces available for delamination. Induced energy at the impact is dissipated by matrix cracking and delamination damage. For these laminates, the delamination damage is distributed over 22 interfaces in the pine-tree shape, typical for impact damage in FRP [33
]. For the ply-block scaled laminates only six interfaces between layers of different fibre orientation are available for delamination. As the induced energy is in the same range, it is dissipated by larger delaminations in contrast to a higher number of smaller size delaminations with increasing layer thickness. Due to the measuring principle, only the projected damage area and not the sum over all delaminated areas can be measured in the US c-scan analysis. This results in the significantly higher projected delamination area for the ply-block scaled specimens. The total summation of delamination area over all interfaces is assumed to be of equal size for both configurations. Similar findings are reported for thin-ply laminates compared to laminates with conventional ply thickness [21
]. The size of the respective delamination areas and the fibre orientation next to the largest delamination is critical in the compression after impact test. A larger delamination area results in a weakening of the laminate against compression in general and a smaller support of the 0°-layers in particular due to the delamination crack opening being mode I dominated in compression. This explains the larger buckling, measured with the DIC, for ply-block scaled laminates with outer 0°-plies. The influence of the 0°-layer position, with inner 0°-layer exhibiting higher CAI strength, becomes clear, when considering the pine-tree delamination shape. Outer 0°-layers at the back sublaminate from the impact position are not supported over a comparable larger delamination area than inner 0°-layers. The unsupported length of the sublaminate increases due to delamination growth [33
]. Consequently, the specimens have a lower resistance against global buckling, resulting in lower CAI strength with increasing delamination areas between the outer 0°-layers.
The damage onset at lower strains with increasing ply thickness observed in UNC and OHC tests is in accordance with results from literature for compressive [14
] and tensile behaviour [20
]. The higher UNC strength for sublaminate scaled specimens with inner 0°-layers can be attributed to a better support of the load bearing 0°-layers by adjacent plies, leading to a delay of the onset of microbuckling and the resulting kink-band initiation and rupture. This implies, that out-of-plane microbuckling is the dominating mechanisms for inner 0°-layers. This effect is more pronounced when a stress intensification is present. Here, the stacking sequence has a higher effect on strength than the layer thickness. Accordingly, the OHC and CAI strength is higher for inner 0°-layers layers. The highest CAI strength and comparably high OHC strength is measured for the configuration with one central 0°-layer consisting of six plies and surrounded by all other layers. Further investigations with thin-ply laminates might be necessary to verify these findings with thinner layers as a positive influence of a ply thickness below
is reported in literature [20
]. However, regarding the results in Figure 7
, decreasing the ply thickness down to thin-ply laminates (
) may not be the optimum in laminates with stress intensifications under pure compression loading. An open hole or impact damage reduces the bending resistance and has therefore a negative influence on global buckling. Central
-layers increase the bending resistance under compression and show higher OHC strength. For CAI properties, central
-layers are advantageous because of the conic delamination damage shape on an impact with the largest delaminations at the backside from the impact point. The delamination area at the load carrying
-layers should be preferably small so that these layers should be arranged in the centre of a laminate. However, for bending load cases, outer 0°-layers are an optimum. Regarding the 0°-layer position, a trade-off between bending and compression strength has to be made.
OHC failure process depends on ply thickness, whereas the stacking sequence has no significant influence. Failure originates at the free edge of the hole in all tests and is perpendicular to loading direction in the 90°-direction for the ply-block scaled and in 0°-direction parallel to loading direction for the sublaminate scaled laminates. This can be explained by the difference in transverse contraction. Sublaminate scaled laminates with thinner layers fail brittle at comparable high strains whereas ply-block scaled laminates exhibit a damage process that initiates at lower strains and is more continuous and thus less brittle. Less brittle materials have a lower notch sensitivity, because early 0°-layer fibre matrix splitting in the ply-block scaled laminates may act as a blunting mechanism at the free edge [27
]. Although a more progressive failure process might be advantageous in some materials, as it may result in the possibility to take measures for repair of replacement of the damaged part, a damage initiation at lower strains is mostly no acceptable for FRP as first ply failure is often a design criterion.
The polished micrograph sections from interrupted tests of representative specimens give insight on the influence of ply thickness on damage initiation and propagation at a free edge. The observed initiation of kink-band failure by shear driven damage mechanisms confirms the results from Gutkin et al. [7
] for both, ply blocked and sublaminate scaled specimens. First damage of fibres occurs always in 0°-plies. Ply thickness and ply position have significant influence on the failure mechanisms. In ply-blocked specimens with thicker plies, a shear failure crack originates at IFF parallel to loading direction next to the free edge. The three stages of kinking [2
] can be clearly identified in the SEM images. The matrix failure between the fibres supports fibre microbuckling and leads to shear driven failure and subsequently to localised incipient kinking of fibres (stage 1). The transient kinking region (stage 2) as the transition to a single dominant kink-band across the specimen is very small, leading to steady state kinking (stage 3) in close proximity to the hole. This kink-band is established almost instantly after damage initiation. With reduced ply thickness (sublaminate scaling) damage initiates and propagates as shear failure and a single kink-band is established at some distance away from the hole in a transition from the shear failure. This transition from incipient to transient and finally steady state kinking is shifted to higher compressive strains with decreasing ply thickness. Thinner plies exhibit a longer region of and a shear failure and more stable fracture process, which is less less prone to global buckling. This results in splitting final failure type observed in the UNC and OHC test. With increasing ply thickness, IFF in loading direction promotes formation of a single, stable kink-band and delaminations, leading to global buckling at lower strains. The observations regarding damage type and kink-band geometry are summarized in Table 4
The experimental investigation on the influence of ply thickness and the position of the 0°-layers in QI CFRP laminates with a detailed analysis of the brittle failure process initiating at a stress concentration show the influence of the FRP lay-up on mechanical properties and damage propagation. This works thus may help to select an appropriate stacking sequence in the design of FRP parts with stress intensifications or where impact damage cannot be excluded regarding compression behaviour. When regarding free edges or an impact damage as delamination inducing stress intensifications within a laminate, the position of the 0°-layer is critical for stability under compression and is thus more important than the ply thickness. Central 0°-layers show best results for OHC and CAI strength due to higher bending stiffness and better supporting effect of the adjacent layers. Nonetheless, open-hole and CAI strength are higher for thinner layers, when regarding laminates with distributed 0°-plies. This is due to a reduced delamination area resulting an a shorter unsupported length of the load bearing sublaminates. The statistical defect distribution and the increased in situ strength lead to a delayed damage initiation with decreasing ply thickness. Thus, the unnotched compressive strength increases with decreasing ply thickness. With increasing ply thickness, damage initiation in the form of IFF is at lower strains. This reduces the stress concentration factor at the stress intensification and leads to a change from brittle to progressive delamination failure. In laminates with blocked plies, final failure is transverse to loading direction with an orientation along the -layers around the stress intensification that leads to a step in the fracture plane, whereas in laminates with distributes plies, final fracture occurs as one straight splitted crack transverse to loading direction.
Damage initiation and propagation at stress intensifications, such as the free edges in open-hole specimens and the impact damage in the case of CAI, is analysed with AE analysis, X-ray and polished micrograph sections in SEM of representative specimens. AE signals are used in interrupted OHC tests for causing a distinct damage state and examine it at a precise instant of time during fracture process. Furthermore, characteristic plots for different types of damage propagation highlight the potential of the AE-analysis for monitoring failure in composites.
First damage of fibres occurs always in a 0°-ply. Fibre shear failure leads to local microbuckling and the formation and growth of a kink-band as final failure mechanisms. Ply thickness and position of ply influence the failure mechanism. The formation of a kink-band and finally steady state kinking is shifted to higher compressive strains with decreasing ply thickness. Final failure mode in laminates with stress intensification depends on ply thickness. In thick or inner plies, damage initiates as shear failure with an in-plane inclination angle and fibre buckling into the drilled hole (free edge). The kink-band orientation angle is changing with increasing strain. In outer or thin plies we observed shear failure of single fibres as first damage and an out-of-plane inclination angle . The kink-band orientation angle is constant until final failure. Further investigations should focus on thin-ply laminates that offer improved mechanical properties due to a dramatic reduction in ply thickness and allow a broader variety of lay-up configurations.