Composite Structures 93 (2011) 2646–2654

Contents lists available at ScienceDirect

Composite Structures

journal homepage: www.elsevier .com/locate /compstruct

Investigation into laminate design of open carbon–fibre/epoxy sectionsby quasi–static and dynamic crushing

A. Jackson a,⇑, S. Dutton a, A.J. Gunnion b, D. Kelly a

a School of Mechanical and Manufacturing Engineering, University of New South Wales, 2052, Australiab Cooperative Research Centre for Advanced Composite Structures, 506 Lorimer Street, Fishermans Bend, 3207 Victoria, Australia

a r t i c l e i n f o

Article history:Available online 5 May 2011

Keywords:Energy absorptionCrushingQuasi–static loadingDynamic loadingInterleaving

0263-8223/$ - see front matter � 2011 Elsevier Ltd. Adoi:10.1016/j.compstruct.2011.04.032

⇑ Corresponding author.E-mail address: a.jackson@crc-acs.com.au (A. Jacks

a b s t r a c t

The effect laminate design on the crush performance of carbon–fibre/epoxy ‘‘DLR’’ crush elements hasbeen experimentally investigated. A quasi-isotropic lay-up was found to result in the highest SpecificEnergy Absorption (SEA) for Four-Harness (4HS) reinforced laminates; however a hybrid of unidirectionalweave and 4HS fabric produced the highest SEA of 114 kJ/kg. Interleaving with thin thermoplastic filmsincreased the steady state crushing force, however the increase in laminate density associated with theaddition of the film caused no improvement and in some cases a reduction in SEA depending on materialand lay-up. Dynamic crush testing of selected laminate designs resulted in a reduction in SEA of between6% and 15% compared to the quasi–static case.

� 2011 Elsevier Ltd. All rights reserved.

1. Introduction

Since the 1970s there has been considerable interest in thecrashworthiness of structures manufactured from compositematerials, driven mainly by the increasing use of these materialsin automobiles and aircraft. It is desirable from a crashworthinessperspective for designated areas of structures to fail in a progres-sive manner, thereby reducing the maximum acceleration on occu-pants and eliminating injuries and fatalities in relatively mildimpacts and minimising them in all severe but survivable mishaps[1].

Failure by progressive crushing differs from catastrophic failuremodes in that a stable region of failure exists over which the crush-ing load is approximately constant. Progressive failure of compos-ites is usually achieved by applying a compressive load to asuitably-triggered component. Triggering typically involves a geo-metric gradient feature in the component which causes a localstress concentration. The high local stress initiates microfracturein the triggered region [2], which progresses through the gradientfeature, and eventually forms a stable crush zone.

Much work has been performed on the static and dynamic axialcrushing of open and closed sections made from composite mate-rials, e.g. [3–20]. These studies have focused on the effect of matrixand fibre properties, geometry, loading rates and triggering mech-anisms. Farley and Jones [4] described three main progressive fail-ure modes: (1) transverse shearing, (2) lamina bending and (3)local buckling. These first two failure modes have become more

ll rights reserved.

on).

commonly referred to as fragmentation and splaying modesrespectively [2,10,16,21] and are typically observed only in brittlefibre-reinforced materials.

Of current interest is the effect that laminate design, includingply lay-up and toughening through thermoplastic interleaving,has on the Specific Energy Absorption (SEA) of crush elements de-signed by the Deutsches Zentrum für Luft-und Raumfahrt (DLR;German Aerospace Centre). DLR crush elements are omega-shapedsections and were developed as a standard specimen configurationto characterise the crush performance of materials. DLR crush ele-ments were chosen over other section designs as they are easy tofabricate and provide reproducible crush failures under dynamicand static loads [22]. In this work, specimens with varying plylay-up were manufactured and tested quasi-statically and dynam-ically. Specimens were also manufactured with different thin ther-moplastic films as interleaves between plies in an attempt totoughen the laminate and increase SEA.

There is known to be a trade off between including load carry-ing plies parallel to the crush direction (0�) versus ±45� or 90� pliesto provide hoop reinforcement in the crushing of carbon fibre/epoxy sections [3,16,23,24]. Achieving the optimal balance de-pends on the material properties, and thus it is difficult to predictthe optimal lay-up for new materials.

Interleaving of brittle, fibre reinforced composites with thermo-plastic layers has been shown to increase Mode I and Mode IIfracture toughness as well as Inter-Lamina Shear Strength (ILSS)[25–30] and increasing these properties with through-thicknessreinforcing strategies has been shown to increase SEA in crushingof composite materials [13,31–34]. Only a limited number ofauthors have reported crush tests of carbon–fibre/epoxy specimens

A. Jackson et al. / Composite Structures 93 (2011) 2646–2654 2647

with addition of thermoplastics as interlayers [29,31] with the re-sults being mixed. Warrior et al. [29] added a urethane interleaf atthe mid-plane of both continuous filament random mat and non-crimp-fabric E-glass/polyester laminates, and found an increasedGIC in Double Cantilever Beam testing, but an overall decrease inSEA during quasi–static crush testing of circular tubes. Yuan etal. [31] however reported an increase in SEA of up to 19% fordynamically crushed tubes when adding a tough PET/modifiedepoxy resin interleaf into carbon–fibre/epoxy tubes. This currentstudy used thin polyimide thermoplastic films interleaved everyply and every second ply in carbon–fibre/epoxy laminates withthe aim of improving SEA.

Since conflicting results have been reported as to the effect thatloading rate has upon the energy absorption capability of compos-ite materials, selected designs were also tested under dynamiccrushing conditions. Early studies by Hull [35,36], Thornton [17]and Farley [37] investigated a wide range of materials under staticand dynamic rates up to 15 m/s using both drop towers andhydraulic test machines, and suggested that crushing speed didnot effect energy absorption. However, around this same timeBannerman and Kindervater [38], Kindervater [39], Berry and Hull[40] and Schmueser and Wickliffe [41] all reported results on com-posite tubes and beams of various materials which suggested thatenergy absorption is a function of crushing speed.

Where crushing speed has been shown to have an effect on en-ergy absorption, conflicting results as to the direction of the trendhave been reported. Table 1 lists some examples from the litera-ture in which both the crushing speed has been shown to havean effect on SEA and also where the author(s) have given a hypoth-eses for the change.

Quasi–static testing is simpler and less expensive than dynamictesting and facilities are more readily available. Quasi–static canprovide good qualitative assessment as to the trend of differentvariables upon energy absorption, however for useful design datafor future work, dynamic testing is essential for determining aquantitative measure of energy absorption. The overarching aimof this study was to gain a full understanding of the crushing char-acteristics of the new materials being investigated before furthertesting of more complex elements representative of rotorcraftsub-floor structures. Coupon level testing of the DLR crush seg-ments was undertaken to achieve this aim by:

1. Determining the energy absorbing capability of the carbon–fibre/epoxy pre-preg materials.

2. Optimising the lay-up for the material systems being tested formaximum SEA.

Table 1Examples of literature which cite a dependence of SEA upon crushing speed.

Reference Specimen typea Crushingspeed (m/s)

SEA chang

Farley [42] CF/Epoxy tubes [±75]3 1E-4 to 12 +35%

CF/Epoxy tubes [0, ±45]2 &[0, ±75]2

No change

Tao et al. [15] GF/Polyester unidirectionalrods (5 mm dia.)

2E-6 to 4E-5 +24%

Mamalis and colleagues[43,44]

GF/Epoxy square tubes andsquare cones

2E-4 to 8.0 �15% to +2

Ramakrishna [45] CF/Epoxy tubes (knittedfabric)

0.001 to 12 �20%

Kindervater et al. [22],Kohlgrüber andKamoulakos [46]

CF/Epoxy DLR testspecimens (0/90 and ±45plain weave fabric)

0.0003 to 10 �25%

a CF = Carbon fibre, GF = Glass fibre.

3. Attempting to increase the energy absorbing capability of thecarbon–fibre/epoxy materials by interleaving with thin thermo-plastic films.

4. Determining the effect of crushing speed upon selected lami-nate designs.

2. Experimental methods

2.1. Materials

DLR crush elements were manufactured using three carbon–fi-bre reinforced epoxy pre-pregs as listed in Table 2. The two ther-moplastic films used as interleaves were 0.003’’ Pyralux

�LF0111

and 0.002’’ Pyralux�

LF0110. Pyralux�

films are DuPont Kapton�

Polyimide (PI) film coated with a B-staged modified acrylic onone and both sides respectively for LF0110 and LF0111.

2.2. Specimen manufacture

DLR crush element lengths were laid up by hand using M18/1,MTM44-1FR-468 and MTM44-1FR-756 into female moulds withthe lay-ups shown in Table 3. Since an individual 4HS satin weaveply is not balanced, all lengths were laid up symmetrically, i.e. theply directions mirrored around the laminate mid-plane. Specimenswere also manufactured with thermoplastic film interleaved everyply or every second ply, as shown in Table 4. Specimens were auto-clave cured in the case of the M18/1, whilst MTM44-1FR specimenswere oven (vacuum only) cured, all according to the manufac-turer’s specification. Lengths were then trimmed to the geometryshown in Fig. 1, and a 45� chamfer was machined into one end ofeach specimen using custom-built tooling. Note that the actualspecimen thickness varied from the nominal 2 mm shown inFig. 1 depending upon the material, lay-up and interleaving. Theweight of each specimen was recorded; the average weight permm length for each batch is shown in Tables 3 and 4. The uncham-fered end of the specimens were then either potted in Araldite KitK 106 resin or fixed in an aluminium clamp for testing. Previoustesting had shown no difference in the crush performance betweenthese two restraint methods [47].

2.3. Test method

Quasi–static testing was performed in an Instron 1185 100 kNmachine with a crosshead speed of 5 mm/min for the first 20mm of vertical displacement at which point the rate was increasedto 20 mm/min for the final 30 mm of vertical displacement.

e Hypotheses

Tubes which had fewer fibres in the loading direction were influenced byloading rate. Fracture of laminar bundles (dominated by fibre properties)was more or less insensitive to strain-rate, whilst interlaminar crackgrowth (dominated by matrix properties) was strain-rate dependant

Increase in matrix compressive strength was almost identical over thesame load rate range, indicating that matrix compressive strength (and itsrate dependence) is dominating SEA

0% Mechanical properties (in particular coefficient of friction) of differentmaterials were affected by strain rate in different waysMode I fracture toughness and friction reduces with increasing testing rate

The influence of the crush speed depends on the energy absorbingmechanisms. This seems to be more important in the test wherefragmentation and friction are the primary mechanisms.

Table 2Carbon–fibre epoxy pre-pregs used in manufacture of crush specimens.

Designation Manufacturer Architecture Areal weight(g/m2)

Nominal fibrevolume fraction

Nominal cured plythickness (mm)

M18/1/43%/G939 Hexcel (Four harness) 4HS 220 0.55 0.227MTM44-1FR-468 ACG 4HS 220 0.55 0.227MTM44-1FR-756 ACG Unidirectional (Uni) weave (10% transverse fibres) 160 0.55 0.165

Table 3DLR crush element specimens (lay-up comparison).

Batch ID Material Lay-up Weight (g/mm) No. of specimens

Quasi–static Dynamic

CR-M18-LC-01 M18/1 [(0, 90)2, 0, (90, 0)2] 0.277 3CR-M18-LC-02 M18/1 [0, 90, +45, 0, 90, 0, +45, 90, 0] 0.297 3CR-M18-LC-03 M18/1 [0, +45, 90, �45, 0, �45, 90, +45, 0] 0.293 3CR-M18-LC-04 M18/1 [+45, �45, 0, 90, 0, 90, 0,�45, +45] 0.294 3CR-M18-LC-05 M18/1 [+45, �45, 0, +45, �45, +45, 0,�45, +45] 0.296 3CR-M18-LC-06 M18/1 [(+45, �45)2, +45, (�45, +45)2] 0.291 3CR-MTM468-LC-07 MTM44-1FR-468 [0, 90]2S 0.260 3CR-MTM468-LC-08 MTM44-1FR-468 [0, +45, 90, �45]S 0.255 3 3CR-MTM468-LC-09 MTM44-1FR-468 [+45, �45]2S 0.260 3CR-MTM756-LC-10 MTM44-1FR-756 [0, 90]3S 0.294 3CR-MTM756-LC-11 MTM44-1FR-756 [+45, �45]3S 0.288 3CR-MTMHYB-LC-12 MTM44-1FR-468/-756 [0, +45, 0, �45, 0]S Highlighted plies are �756 Uni-weave fabric 0.283 3 3

Table 4DLR crush element specimens (interleaving trials).

Batch ID Material/lay-up Thermoplastic Weight (g/mm) No. of specimens

Quasi–static Dynamic

CR-M18-LF0111-I M18/1 [0, +45, 90, �45]S LF0111, interleave every ply 0.318 3CR-M18-LF0111-II LF0111, interleave every 2nd ply 0.278 3CR-M18-LF0110-I LF0110, interleave every ply 0.301 3CR-M18-LF0110-II LF0110, interleave every 2nd ply 0.271 3CR-MTM468-LF0111-I MTM44-1FR-468 [0, +45, 90, �45]S LF0111, interleave every ply 0.324 1 1CR-MTMHYB-LF0111-I MTM44-1FR-468/�756 [0, +45, 0, �45, 0]S

highlighted plies are �756 Uni-weave fabricLF0111, interleaved every plyexcept between 3rd/4th and 7th/8th plies in order to have seveninterleaves

0.349 1 1

Fig. 1. DLR crush element geometry (dimensions in mm).

2648 A. Jackson et al. / Composite Structures 93 (2011) 2646–2654

Dynamic testing was performed in an Instron VHS 100/20 highstrain rate test machine with a 100 kN load cell as described byDavid [48]. Impact velocity was 8.5 m/s for all dynamic tests.

3. Results and discussion

3.1. Lay-up comparison

Stable, progressive crushing was initiated and sustained for allspecimens. The load typically ramped up linearly over the first2–2.5 mm of vertical displacement (corresponding with the

chamfer height) to a maximum value, after which the crushingload stabilised until the end of the test. Typical curves for thequasi–static and dynamic crushing response are shown in Fig. 2.

Typical fragmentation behaviour of a quasi-statically testedspecimen is shown in Fig. 3 and a dynamically tested specimenin Fig. 4. The quasi-statically tested crush specimens failed in thesplaying crush failure mode, with internal and external frondsforming which were progressively bent and deformed as the plat-ens came together. Some frond fragments broke off completelyfrom the crushed specimen, and smaller pieces of debris (majordimension typically <3 mm) were also produced. A small debris

Fig. 2. Typical quasi–static and dynamic crushing response.

Fig. 5. Variation in SEA with relative ±45� ply content.

A. Jackson et al. / Composite Structures 93 (2011) 2646–2654 2649

wedge was present where the main central wall crack had beenpropagating. With increasing ±45 ply content, it was found thatless axial splits (and therefore fronds) were formed (as shown inFig. 6–8), the crushed portion of the specimen was less cohesive,and the crushing force showed greater instability. The dynamicallytested MTM44-1FR specimens failed in a similar manner to thosetested quasi-statically; however more debris was produced dueto material being ejected from the specimen rather than formingfronds. It has been previously shown that energy absorption in thisfailure mode is primarily through crack growth (longitudinal andinterlaminar), friction between separated lamina layers and be-tween the specimen and the loading surface and bending/fractureof fibres in the fronds [4].

Total energy absorption (U) was found by determining the areabeneath the load–displacement curve for each test.

U ¼Z S

0P ds

Fig. 3. Quasi-statically crushed C

Fig. 4. Dynamically crushed CR

where P is the crushing load and S is the total crush distance. SEAwas found by dividing the total energy absorbed (U) by the masscorresponding to the length of the specimen crushed (mc) whichwas determined by multiplying the specimen mass-per-unit-length(M/L) by the total crush distance for each test.

SEA ¼ Umc¼ UðM=LÞS

Steady State Crush Force (SSCF) was calculated as the mean loadfor the test period from 3 mm vertical displacement to the end ofthe test. Steady State Crush Stress (SSCS) for a given test is calcu-lated as the SSCF divided by the measured pre-test cross-sectionalarea, and is a useful means for comparing specimens with differentnumbers of plies.

Crush test results for the specimens with varying ply lay-up areshown in Table 5 and Fig. 5. For the M18/1 specimens, the variationin SEA from the orthogonal 0/90 to ±45 lay-up was relatively small.Maximum SEA was obtained from the quasi-isotropic configura-tion (CR-M18-LC03), although interestingly when the ±45 plies

R-MTM468-LC-08 specimen.

-MTM468-LC-08 specimen.

Fig. 6. Crushed M18/1 specimens: (L) CR-M18-LC-01, (R): CR-M18-LC-06.

Fig. 7. Crushed MTM44-1FR-468 specimen: (L) CR-MTM468-LC-07, (R): CR-MTM468-LC-09.

Fig. 8. Crushed MTM44-1FR-756 specimens: (L) CR-MTM756-LC-10, (R) CR-MTM756-LC-11.

Table 5Lay-up comparison results.

Batch ID Static/dynamic Peak load (kN) [std dev] SSCF (kN) [std dev] SSCS (MPa) [std dev] SEA (kJ/kg) [std dev]

CR-M18-LC-01 Static 28.0 [1.0] 25.0 [0.61] 130 [2.7] 87.2 [3.0]CR-M18-LC-02 Static 32.3 [0.13] 28.0 [0.66] 137 [3.6] 88.3 [3.5]CR-M18-LC-03 Static 33.8 [1.4] 28.5 [1.3] 140 [6.7] 90.6 [5.0]CR-M18-LC-04 Static 31.0 [0.52] 26.6 [0.67] 126 [2.4] 86.5 [1.6]CR-M18-LC-05 Static 32.2 [4.4] 27.8 [1.9] 129 [8.3] 89.6 [5.8]CR-M18-LC-06 Static 29.1 [0.97] 24.0 [0.48] 113 [1.6] 79.4 [3.6]CR-MTM468-LC-07 Static 30.6 [0.57] 27.2 [0.46] 150 [2.3] 101 [0.79]CR-MTM468-LC-08 Static 31.6 [1.3] 27.8 [1.2] 165 [6.8] 104 [3.3]

Dynamic 30.2 [1.1] 25.0 [1.1] 148 [6.7] 88.4 [3.0]CR-MTM468-LC-09 Static 25.2 [0.56] 17.5 [0.55] 96.4 [2.4] 71.2 [2.5]CR-MTM756-LC-10 Static 29.1 [0.70] 23.6 [0.39] 115 [3.3] 77.9 [0.46]CR-MTM756-LC-11 Static 25.7 [0.86] 16.3 [2.3] 82.7 [14] 57.4 [11]CR-MTMHYB-LC-12 Static 41.1 [0.91] 34.4 [0.93] 184 [4.2] 114 [3.9]

Dynamic 38.4 [0.87] 33.1 [2.81] 177 [15] 108 [7.4]

2650 A. Jackson et al. / Composite Structures 93 (2011) 2646–2654

A. Jackson et al. / Composite Structures 93 (2011) 2646–2654 2651

were biased towards the outside of the specimen with the sameoverall content of 0/90 and ±45 plies (CR-M18-LC04), the SEAwas lower. This result is similar to what Fleming and Nicot [24]found with carbon–fibre/epoxy X-shaped specimens; they showedthat a [(±45, 0)3]S lay-up had an average SEA of around 175% thanthat with a [±453, 03]S lay-up. With more interfaces between dis-creet plies of differing angles, the energy to fracture these inter-faces is increased [34,49] which also results in a smaller scale ofdelaminations.

MTM44-1FR specimens exhibited larger variations between thetwo extremes of orthogonal 0/90 and ±45 lay-ups. MTM44-1FR-468 (4HS) specimens had higher SEA than M18/1 for 0/90 and qua-si-isotropic lay-ups, but lower for the ±45 arrangement. The frag-mentation behaviour of MTM44-1FR-468 was similar to that ofM18/1, although the ±45 lay-up (CR-MTM468-LC09) had quitelarge fractured pieces in addition to the usual smaller debris (seeFig. 7) which caused higher crush force instability and lower SEA.The quasi-isotropic [0/+45/90/�45]S 4HS specimens showed thehighest SEA (104 kJ/kg), displaying the same trend observed inthe M18/1 specimens shown in Fig. 5.

Relatively few studies were found in the literature which havecompared Carbon Fibre/Epoxy specimens made with woven-fabricreinforced architecture with a variety of lay-ups. Bolukbasi andLaananen [23] experimented with lay-up design for Carbon Fibre/Epoxy (plain weave fabric) open section stiffeners (flat, L and C sec-tions) in axial crush. The inclusion of 0� plain-weave plies (in loaddirection) in a 10 ply laminate significantly increase SEA comparedto a [±45] lay-up, with the best being [452/02/45]S, a similar resultto that reported in this paper. Hadavinia and Ghasemnejad [34]also showed a very similar trend to that shown in Fig. 5 for CarbonFibre/Epoxy (twill weave) square tubes, with the lay-ups of [0/45]2,[0]4 and [45]4 being ranked in that order with respect to SEA.

MTM44-1FR-756 (unidirectional reinforcement) specimensshowed different fragmentation behaviour and lower SEA com-pared to the 4HS fabrics. Whilst they still failed in the splayingmode the [0/90]3S specimens disintegrated more (Fig. 8), withthe splayed fronds springing back up after the test, indicating thatless fibre fracture had occurred. MTM44-1FR-756 [+45/�45]3S

specimens failed in sections along the 45� axis, resulting in largepieces of the specimen breaking off and not being crushed(Fig. 8). The crushing force therefore varied significantly duringeach test, and the SEA was relatively low.

The hybrid MTM44-1FR-468/756 specimens were tested in anattempt to harness the greater load carrying capability of 0� ori-ented UNI-weave plies combined with the stability and reinforce-ment of 4HS plies. These specimens resulted in the highest SEA ofall specimen sets (114 kJ/kg), failing in splaying mode in a mannercombining the different fragmentation behaviours already de-

Fig. 9. Crushed hybrid CR-MTMHYB-LC-12 s

scribed. The central crack occurred down the mid-plane of thespecimen, leaving 0� UNI-weave plies exposed, which sprang backup slightly after the load was removed (Fig. 9). The underlyingcrushed portion of the specimen was less cohesive and similar tothat of the ±45 4HS specimens. Increasing the relative content of0� fibres by reducing the number of 0/90 4HS plies and including0� UNI-weave plies had the effect of increasing the compressivestiffness and strength of the specimen. The greater load carryingcapacity combined with the remaining hoop reinforcement in-creased the crushing load and SEA. The SEA of 114 kJ/kg is at theupper limits of values reported in literature for Carbon Fibre/Epoxymaterials, and only some of the tests performed by Farley andJones [3] using AS4/5245 and AS6/5245 fibre/resin combinationswere found to give similar or higher SEA. The materials used inthe current study have similar mechanical properties to those usedby Farley and Jones [3], however fibre architecture differs.

The two best performing designs (CR-MTM468-LC-08 and CR-MTMHYB-LC-12 batches) were selected for dynamic testing. Therewas an overall reduction in crushing performance from the quasi-static case in both batches. The reduction in SEA was 15.2% and5.5% respectively for the CR-MTM468-LC-08 and CR-MTMHYB-LC-12 batches respectively. More debris was ejected from the spec-imen during the test, however the final crushed specimen stillexhibited the same evidence of the splaying failure mode (i.e.fronds and debris wedge) as can be seen in Figs. 4 and 9 (R). Dueto the large amount of debris being ejected obscuring the view ofthe high speed camera, it is difficult to draw further conclusionsas to the nature of the failure mode during the test other thanthe observations of the final failed specimen. Alongside the sus-pected reduction in friction [16,45] for the dynamic case, it isthought that since the hybrid 4HS/unidirectional specimens havemore fibres oriented in the loading direction than the MTM4684HS specimens the latter showed a greater reduction than thedue to the fibres being less sensitive to loading rate then the matrixmaterial.

3.2. Thermoplastic interleaving

The results of the crush testing of specimens interleaved withthin thermoplastic films are shown in Table 6. Also included is areference to the closest non-interleaved specimen from Table 5as a baseline for comparison.

Addition of the thermoplastic films increased the SSCF in allcases where the film was interleaved in between every ply.Fragmentation behaviour of the quasi-statically tested specimenswas typical of that already described above, and residual thermo-plastic was visible on the fractured surfaces (see Fig. 10). However,the increase in SSCF was not large enough to offset the increase in

pecimen: (L) quasi–static, (R) dynamic.

Table 6Interleaving trials results.

Batch ID Static/dynamic Peak load (kN)[std dev]

SSCF (kN)[std dev]

SSCS (MPa)[std dev]

SEA (kJ/kg)[std dev]

Comparable baseline(see Table 4)

CR-M18-LF0111-I Static 37.1 [1.54] 31.3 [0.92] 138 [3.8] 92.0 [2.9] CR-M18-LC-03a

CR-M18-LF0111-II Static 31.5 [0.86] 26.3 [0.82] 131 [1.4] 89.4 [0.95]CR-M18-LF0110-I Static 35.8 [0.74] 29.1 [0.47] 134 [2.4] 90.9 [2.6]CR-M18-LF0110-II Static 30.6 [0.81] 25.5 [0.38] 129 [1.9] 89.3 [1.0]CR-MTM468-LF0111-I Static 39.6 33.7 155 97.7 CR-MTM44-LC-08

Dynamic 38.7 30.3 139 86.3CR-MTMHYB-LF0111-I Static 42.0 36.6 154 99.1 CR-MTMHYB-LC-12

Dynamic 38.7 34.2 143 92.7

a CR-M18-LC-03 batch had one extra 0/90 ply at the mid-plane compared to the interleaved specimens.

Fig. 10. Crushed interleaved specimen; (L) CR-M18-LF0111-I, quasi–static, (R) CR-MTM468-LF0111-I dynamic.

Fig. 11. Crushed interleaved CR-MTMHYB-LF0111-I specimens; (L) quasi–static, (R) dynamic.

2652 A. Jackson et al. / Composite Structures 93 (2011) 2646–2654

specimen density due to the addition of the thermoplastic film, andthe SEA was approximately equivalent to the baseline specimens inthe case of the M18-1 specimens, and less than the baseline for theMTM44 specimens. The difference between the performance of theLF0110 and LF0111 interleaved specimens was negligible andwithin the experimental scatter, suggesting the acrylic adhesiveon the films does not influence the failure mechanisms.

The dynamically tested interleaved MTM44 specimens showeda similar knockdown from the quasi–static performance to thatseen in the non-interleaved specimens. In this case the reductionin SEA was 11.7% and 6.5% respectively for the quasi-isotropiclay-up and the hybrid lay-up. Also of note was the comparativefragmentation behaviour of the interleaved and non-interleavedspecimens under the different loading conditions. The dynamicallytested interleaved specimens held together much better comparedto the non-interleaved batches; see Figs. 10 and 11 in comparisonto Figs. 3, 4 and 9.

The small improvements in SSCF in this work indicate that thistoughening can increase energy absorption through increasing the

energies associated with crack propagation. Current estimates[43,44,49] put the energy associated with crack propagation duringcrushing of composite sections at between 5% and 20%, so unless atoughening strategy improves crack propagation energies by a sub-stantial amount there will be no significant increase in total energyabsorption. In the current case, a small improvement in total en-ergy absorption was found with interleaving every ply with PI film;however the increase did not translate to an improvement in SEAdue to the increase in density. Furthermore, both M18-1 andMTM44 are toughened epoxy pre-pregs, so addition of thermoplas-tic interleaves into untoughened systems may yield greaterimprovements.

4. Conclusion

Quasi–static and dynamic crush testing of DLR crush elementshave been performed to investigate the effect that laminate designhas on crush performance. Specimens were manufactured using

A. Jackson et al. / Composite Structures 93 (2011) 2646–2654 2653

three carbon–fibre/epoxy pre-preg materials with varying lay-upsin order to maximise the SEA for these specific materials, sinceprevious testing has shown that crush performance is sensitiveto material properties [3,16,23,24] and predicting the optimallay-up for new materials is difficult. Some specimens were alsomanufactured with PI film interleaved between plies in an attemptto increase the fracture toughness properties and therefore crush-ing performance; a technique which has only been studied in alimited manner previously [29,31].

Steady state crushing was initiated in all specimens tested via amachined chamfer at the top edge of the specimen, with testsbeing performed at quasi–static speeds and dynamically with animpact velocity of 8.5 m/s.

The response to changes in lay-up depended upon the materialused. The M18/1 4HS material showed a small variation in SEAwith lay-up, while the MTM44-1FR-468 4HS material had highervariation in SEA between orthogonal 0/90, orthogonal ±45 andquasi-isotropic lay-ups. In both cases a quasi-isotropic lay-up gavethe highest SEA results, a similar trend to other results fromliterature using carbon–fibre/epoxy pre-pregs [23,34]. MTM44-1FR-756 unidirectional material did not perform as well as the4HS fabrics, however, a hybrid MTM44-1FR-468 and �756 lay-upproduced the highest SEA (114 kJ/kg) of all the configurationstested. This is amongst the highest SEA results that have beenreported for carbon–fibre/epoxy materials.

Dynamic testing of selected designs found reductions in SEA of6–15% and a subtle change in fragmentation mode with more deb-ris ejected from the specimen. Similar to other variables, changesin loading rate has been shown to cause a variety of responses tocrush performance depending on the materials being tested. Themoderate reductions reported in this study are within the boundsof that found in literature [15,16,40–45]. Reduced friction [16,45]and lower matrix material properties for the dynamic case arethought to cause this difference.

Thermoplastic interleaving every ply with Pyralux�

PI film in-creased the SSCF, however this increase was not large enough tooffset the increase in specimen density from adding the thermo-plastic material, and SEA was found to remain the same as fornon-interleaved specimens. Dynamic testing again resulted in areduction in SEA of 7–12%, however the failure mode was moresimilar to the quasi–static cases than comparable non-interleavedspecimens due to the thermoplastic holding the specimen togetherbetter during the crushing process.

Acknowledgements

This work is part of the research program of the Cooperative Re-search Centre for Advanced Composite Structures established andsupported under the Australian Government Cooperative ResearchCentres Program, and was conducted under the Helicopter Re-search Program project, Design for Crashworthiness. The authorswish to acknowledge the support and guidance of the project teamfrom the CRC-ACS, University of New South Wales, Pacific ESI,Defence Science and Technology Organisation, AustralianAerospace and the Deutsches Zentrum für Luft-und Raumfahrt(DLR). The support of the DLR Institute for Structures and Designis greatly appreciated; in particular the information on the DLRcrush element and test methodology provided by Mr. Christof Kin-dervater and Dr. Alastair Johnson, and the expertise of MatthewDavid and Hussam Abuel-Hija in conducting the dynamic testing.The first author also acknowledges financial support providedthrough the Australian Postgraduate Award and the CRC-ACS.The authors also wish to thank the Advanced Composites GroupLimited (UK) for providing materials for use in the work presentedin this paper.

References

[1] JSSG-2010-7, Department of defence joint service specification guide, crewsystems crash protection handbook. Defence USDo; 1998.

[2] Ramakrishna S. Microstructural design of composite materials for crashworthystructural applications. Mater Des 1997;18(3):167–73.

[3] Farley GL, Jones RM. Energy absorption capability of composite tubes andbeams. NASA Technical Memorandum 101634 1989, Hampton: US ArmyAviation Research and Technology Activity – AVSCOM.

[4] Farley GL, Jones RM. Crushing characteristics of continuous fiber-reinforcedcomposite tubes. J Compos Mater 1992;26(1):36–50.

[5] Hamada H, Ramakrishna S. Scaling effects in the energy absorption of carbon-fiber/PEEK composite tubes. Compos Sci Technol 1995;55(3):211–21.

[6] Hamada H, Ramakrishna S, Sato H. Effect of fiber orientation on the energyabsorption capability of carbon fiber/PEEK composite tubes. J Compos Mater1996;30(8):947–63.

[7] Mahdi E, Hamouda AMS, Mokhtar AS, Majid DL. Many aspects to improvedamage tolerance of collapsible composite energy absorber devices. ComposStruct 2005;67(2):175–87.

[8] Mamalis AG, Manolakos DE, Ioannidis MB, Papapostolou DP. On the responseof thin-walled CFRP composite tubular components subjected to static anddynamic axial compressive loading: experimental. Compos Struct2005;69(4):407–20.

[9] Melo JDD, Silva ALS, Villena JEN. The effect of processing conditions on theenergy absorption capability of composite tubes. Compos Struct2008;82(4):622–8.

[10] Ramakrishna S, Hull D. Energy absorption capability of epoxy composite tubeswith knitted carbon fibre fabric reinforcement. Compos Sci Technol1993;49(4):349–56.

[11] Sigalas I, Kumosa M, Hull D. Trigger mechanisms in energy-absorbing glasscloth/epoxy tubes. Compos Sci Technol 1991;40(3):265–87.

[12] Solaimurugan S, Velmurugan R. Influence of fibre orientation and stackingsequence on petalling of glass/polyester composite cylindrical shells underaxial compression. Int J Solids Struct 2007;44(21):6999–7020.

[13] Solaimurugan S, Velmurugan R. Progressive crushing of stitched glass/polyester composite cylindrical shells. Compos Sci Technol 2007;67(3–4):422–37.

[14] Song H-W, Du X-W, Zhao G-F. Energy absorption behavior of double-chamfertriggered glass/epoxy circular tubes. J Compos Mater 2002;36(18):2183–98.

[15] Tao WH, Robertson RE, Thornton PH. Effects of material properties and crushconditions on the crush energy absorption of fiber composite rods. Compos SciTechnol 1993;47(4):405–18.

[16] Hull D. A unified approach to progressive crushing of fibre-reinforcedcomposite tubes. Compos Sci Technol 1991;40(4):377–421.

[17] Thornton PH. Energy absorption in composite structures. J Compos Mater1979;13(3):247–62.

[18] Thornton PH. The crush behavior of glass fiber reinforced plastic sections.Compos Sci Technol 1986;27(3):199–223.

[19] Thuis HGSJ, Metz VH. Influence of trigger configurations and laminate lay-upon the failure mode of composite crush cylinders. Compos Struct 1993;25(1-4):37–43.

[20] Mamalis AG, Robinson M, Manolakos DE, Demosthenous GA, Ioannidis MB,Carruthers J. Crashworthy capability of composite material structures. ComposStruct 1997;37(2):109–34.

[21] Ramakrishna S, Hamada H. Energy absorption characteristics of crash worthystructural composite materials. Key Eng Mater 1998;141-143(2):585–620.

[22] Kindervater CM, Johnson AF, Kohlgrüber D, Lützenburger M, Pentecote N.Crash and impact simulation of aircraft structures – hybrid and FE basedapproaches. In: european congress on computational methods in appliedsciences and engineering. Barcelona (Spain): ECCOMAS; 2000.

[23] Bolukbasi AO, Laananen DH. Energy absorption in composite stiffeners.Composites 1995;26(4):291–301.

[24] Fleming DC, Nicot F. Crushing of flat graphite/epoxy laminates using a columnspecimen. J Compos Mater 2003;37(24):2225–39.

[25] Aksoy A, Carlsson LA. Interlaminar shear fracture of interleaved graphite/epoxy composites. Compos Sci Technol 1992;43(1):55–69.

[26] Chen SF, Jang BZ. Fracture behaviour of interleaved fiber-resin composites.Compos Sci Technol 1991;41(1):77–97.

[27] Gao F, Jiao G, Lu Z, Ning R. Mode II delamination and damage resistance ofcarbon/epoxy composite laminates interleaved with thermoplastic particles. JCompos Mater 2007;41(1):111–23.

[28] Li L, Lee-Sullivan P, Liew KM. The influence of thermoplastic film interleavingon the interlaminar shear strength and mode I fracture of laminatedcomposites. J Eng Mater Technol 1996;118(3):302–9.

[29] Warrior NA, Turner TA, Robitaille F, Rudd CD. The effect of interlaminartoughening strategies on the energy absorption of composite tubes. ComposPart A 2004;35(4):431–7.

[30] Sela N, Ishai O. Interlaminar fracture toughness and toughening of laminatedcomposite materials: a review. Composites 1989;20(5):423–35.

[31] Yuan Q, Kerth S, Karger-Kocsis J, Friedrich K. Crash and energy absorptionbehaviour of interleaved carbon–fibre-reinforced epoxy tubes. J Mater SciLetters 1997;16(22):1793–6.

[32] Cauchi Savona S, Hogg PJ. Effect of fracture toughness properties on thecrushing of flat composite plates. Compos Sci Technol 2006;66(13):2317–28.

2654 A. Jackson et al. / Composite Structures 93 (2011) 2646–2654

[33] Daniel L, Hogg PJ, Curtis PT. The relative effects of through-thicknessproperties and fibre orientation on energy absorption by continuous fibrecomposites. Compos Part B 1999;30(3):257–66.

[34] Hadavinia H, Ghasemnejad H. Effects of mode-I and mode-II interlaminarfracture toughness on the energy absorption of CFRP twill/weave compositebox sections. Compos Struct 2009;89(2):303–14.

[35] Hull D. Energy absorption of composite materials under crash conditions. In:Proceedings of the fourth international conference on compositematerials. Tokyo (Japan): Japan Society for Composite Materials; 1982. p. 861–70.

[36] Hull D. Research on composite materials at Liverpool University, part 2:absorbing composite materials. Phys Technol 1983:99–112.

[37] Farley GL. Energy absorption of composite materials. J Compos Mater1983:267–79.

[38] Bannerman DC, Kindervater CM. Crash energy absorption properties ofcomposite structural elements. In: Proceedings of the fourth internationalconference SAMPE European chapter. Bordeaux, France; 1983. p. 155–67.

[39] Kindervater CM. Compression and crush energy absorption behavior ofcomposite laminates. In: Proceedings of the E/MRS conference. Strasbourg,Germany; 1985.

[40] Berry J, Hull D. Effect of speed on progressive crushing of epoxy-glass clothtubes. In: Proceedings of the institute physics conference, vol. 70; 1984. p.463–70.

[41] Schmueser DW, Wickliffe LE. Impact energy absorption of continuous fibercomposite tubes. J Eng Mater Technol 1987;109:72–7.

[42] Farley GL. Effects of crushing speed on the energy-absorption capability ofcomposite tubes. J Compos Mater 1991;25(10):1314–29.

[43] Mamalis AG, Manolakos DE, Demosthenous GA, Ioannidis MB. The static anddynamic axial crumbling of thin-walled fibreglass composite square tubes.Compos Part B 1997;28(4):439–51.

[44] Mamalis AG, Manolakos DE, Demosthenous GA, Ioannidis MB. Energyabsorption capability of fibreglass composite square frusta subjectedto static and dynamic axial collapse. Thin Wall Struct 1996;25(4):269–95.

[45] Ramakrishna S. Energy absorption characteristics of knitted fabric reinforcedepoxy composite tubes. J Reinf Plast Comp 1995;14(10):1121–41.

[46] Kohlgrüber D, Kamoulakos A. Validation of numerical simulation of compositehelicopter sub-floor structures under crash loading. In: Annual forumproceedings – american helicopter society, vol. 1. Washington (DC, USA):AHS, Alexandria (VA, USA); 1998. p. 340–9.

[47] Jackson A, Dutton S, Gunnion AJ, Kelly D. Effect of manufacture and laminatedesign on energy absorption of open carbon–fibre/epoxy sections. In: 17thinternational conference on composite materials. Edinburgh, UK: IOMCommunications Ltd. (CD-ROM); 2009.

[48] David M. Analysis of crushing response in crashworthy structures. In: 10thONERA-DLR aerospace symposium. Berlin, Germany; 2009.

[49] Ghasemnejad H, Blackman BRK, Hadavinia H, Sudall B. Experimental studieson fracture characterisation and energy absorption of GFRP composite boxstructures. Compos Struct 2009;88(2):253–61.