Compression, Flexure and Shear Properties of a Sandwich Composite

COMPOSITE STRUCTURES

ELSEVIER Composite Structures 44 (1999) 263-278

Compression, flexure and shear properties of a sandwich composite containing defects

A.P. Mouritz a,*, R.S. Thomson b a Department of Defence, DSTO, Aeronautical and Maritime Research Laboratory, P. 0. Box 4331, Melbourne, Victoria 3001, Australia b Cooperative Research Centre for Advanced Composite Structures Ltd., 506 Lorimer Street, Fishermens Bend, Victoria 3207, Australia

Abstract

The mechanical properties of a sandwich composite containing interfacial cracks or impact damage are compared when loaded in edgewise compression, flexure or shear. The composite is made from glass fibre reinforced polymer (GFRP) laminate skins over a core of foamed poly vinyl chloride (PVC), and this sandwich material is used in some naval minehunting ships. The properties are reduced with increasing interfacial crack or impact damage length, but only when the defects cause a change in the failure mode, which is dependent on the load state. The principal failure modes under the different load states are compared. The properties are also dependent on the severity of impact damage, with low energy damage to the skin having a smaller effect on stiffness and strength than high energy impacts which damage both the skin and foam core. The implications of these findings on the structural integrity of a minehunting ship made from GFRP/PVC foam sandwich composite is discussed. 0 1999 Elsevier Science Ltd. All rights reserved.

1. Introduction

Sandwich composites made from glass fibre reinforced polymer (GFRP) laminate skins over a poly vinyl chloride (PVC) foam core are steadily gaining popularity as marine construction materials. These composites are used in racing yachts, superstructures for fast pas- senger ferries, control surfaces for submarines, and in internal ship hull stiffeners [ 1,2]. The growing popularity of sandwich composites is due mainly to their low cost, high specific stiffness, buoyancy, and excellent corrosion resistance compared against traditional marine construction materials such as wood, steel and aluminium alloys. One of the most common uses of sandwich composites is in naval minehunting ships because of their low magnetic signature, good sound and vibration damping properties together with reasonable resistance to damage by underwater explosive blasts caused by sea- mines. Examples of minehunters made from sandwich composites are the Swedish Landsort class [1,2] and the Australian Bay class vessels. Figure 1 shows HMAS Rushcutter, one of two Bay class minehunters operated by the Royal Australian Navy, in which the superstructure and catamaran hull are made with over 100

*Corresponding author. Tel.: 00 61 396268276; fax: 00 61 396268999; e-mail: [email protected].

tonnes of sandwich composite consisting of thin GFRP skins over a foam core of closed-cell PVC [3].

One problem with using sandwich composites is that high quality fabrication standards are required, other- wise processing defects may be inadvertently introduced into the minehunter structure. Hall [4] reviewed the major types of defects that can occur during fabrication of the Bay class minehunters when strict fabrication conditions and quality control inspection procedures are not met. Defects within the fibreglass skin include ex- cessive voiding, foreign inclusions, uncured resin, dry fibres and resin-rich regions. Interfacial defects between the skin and core can also occur, such as porosity and foreign inclusions. Fabrication defects such as these can easily develop into interfacial cracks in the minehunter by shear stresses which are generated by waves slamming against the hull. Fabrication defects and interfacial cracks are usually not easily detected using conventional non-destructive inspection techniques such as ultrason- ics or thermography, but they have the potential to seriously degrade the structural integrity of the vessel. Despite this problem, the reduction to the mechanical properties of marine-grade sandwich composites caused by fabrication defects has received little attention. McClave and Goodwin [5] investigated the effect of defects on the tensile and flexural properties of a variety of composites for maritime craft, including GFRP/

0263-8223/99/$ - see front matter 0 1999 Elsevier Science Ltd. All rights reserved. PII:SO263-8223(98)00133-O

264

Fig. 1. HMAS Rushcutter. which is made almost entirely from a GFRP/PVC foam sandwich composite.

polymer foam sandwich materials. In almost all cases, they measured reductions in the tensile and flexural strengths of between 5.6% and ~30% when dry fibres, cracks or foreign inclusions (dirt) were in the fibreglass skin. Other studies indicate that greater reductions in strength occur when an interfacial crack develops between the skin and core, particularly when the composite is loaded in shear. For example, Zenkert [6] measured a large reduction (-70%) in the shear strength of a GFRP/PVC foam sandwich composite due to a short interfacial crack. Similarly, Triantofillou and Gibson [7] and Carlsson et al. [S] report that interfacial cracks can cause significant reductions in the shear strengths of aluminium/PVC foam and GFRP/balsa sandwich composites, respectively. These studies found, however, that the strengths were only reduced when the interfacial crack exceeded a critical length. Thomson et al. [9] recently reported large falls in the fatigue life of a GFRP/PVC composite with increasing interfacial crack length when tested under cyclic shear loading.

While these studies clearly show that long interfacial cracks are detrimental to the shear properties, the in- fluence of other load states such as compression or bending have not been as widely studied.

As well as having potential problems with fabrication defects, minehunters can also be damaged in-service by underwater explosions during sea-mine clearance oper- ations and by accidental collisions with wharves, other

ships and the sea-bed. This impact-type damage is expected to be more detrimental than fabrication defects because it usually consists of extensive delamination cracking within the GFRP skin, debonding of the skin from the core and, under severe impact conditions, shear cracking through the core [ 10,111. Impact damage can seriously degrade compression [12,13], flexural [14] and shear strengths [9]. However, most studies examining the damage tolerance of sandwich composites focussed on aircraft materials such as carbon fibre reinforced epoxy skins over honeycomb cores made of aluminium or NomexB. Virtually no work has been reported on the

A. P. Mouritz, RS Thomson I Composite Structures 44 (1999) 263-278 265

damage tolerance of marine-grade sandwich composites. Furthermore, most studies have only investigated the effect of small, localised areas of impact damage on the mechanical properties, whereas the damage to minehunters caused by an underwater explosion or collision occurs over a much larger area.

Therefore, the aim of this paper is to compare the stiffness and strength of a sandwich composite containing an interfacial crack or impact damage when loaded in edgewise compression, shear or flexure. The size of the interfacial crack and impact damaged area is increased to assess the effect of defect size for the different load states. The composite studied is a thick-section GFRP/PVC foam sandwich material that is similar to that used in the Bay class minehunters. The load states of edgewise compression, flexure and shear are investigated because they are the main load types acting on the minehunter.

2. Materials and experimental techniques

2.1. Sandwich composite

The sandwich composite was made from thin GFRP skins and a thick core of closed-cell PVC foam. The foam (DivinylcellB HT-90) was a rigid cross-linked PVC with a density of 90 kg/m3, shear modulus (GE) of 19.7 MPa and thickness of 30 mm. The foam was sandwiched between two GFRP skins which were each 2.8 mm thick. ’ The fibreglass in the skins was a mixture of chopped strand mat (CSM) with an area1 density of 0.3 kg/m2 and plain woven roving (WR) with a density of 0.6 kg/m2. The CSM and WR were made by Con- solidated Industries and Colan Products, respectively. Each skin consisted of three CSM plies and two WR plies, and these were laminated by hand onto the core in an alternating ply sequence. A cold-curing vinyl ester, produced by the Dow Chemical Company as De- rakane@ 411-45, was used as the resin in the GFRP skins. The GFRP skins were cold-cured under ambient conditions for at least two weeks before mechanical testing. Differential scanning calorimetry measurements on the vinyl ester revealed that cold-curing under these conditions only partially cured the resin, with the po- lymerisation cure reaction being about 85% complete. The resin content of the skin was 45-50% by weight.

Interfacial cracks which occur in poorly fabricated sandwich composites were simulated in test specimens by inserting 90 urn thick Teflon film along the skin/core interface during lamination to stop the GFRP skin from

’ The skins on the Bay class minehunters are about 8 mm thick while the thickness of the PVC used in the hull is 60 mm.

bonding to the core. Interfacial crack lengths of 20, 40, 70, 100 and 150 mm were studied, and all cracks extended across the entire width of the specimens.

The types of impact damage the Bay class minehunter is likely to suffer when involved in a low speed collision or subjected to a underwater explosive blast was simulated by impact testing. Sandwich composite beams were impacted with a 1.3 kg tup dropped from heights of 2 m or 3 m to create low energy (-20 J) or high energy (-30 J) damage, respectively. During testing the beams were rigidly clamped close to the impact site by two steel platens to minimise bending under impact, and as a result a considerable percentage of the absorbed impact energy is expected to dissipate through damage pro- cesses. The hemispherical nose of the tup was 25 mm in diameter, and when it struck the composite a circular damaged area with a diameter of -20 mm was formed. By subjecting the beams to a distributed array of impacts which were spaced 15 mm apart, it was possible to increase the length of the impact damaged zone, as shown schematically in Fig. 2. Inspection of the specimens in cross-section using scanning electron microscopy (SEM) showed that the damage formed at the impact sites was interconnected to produce a relatively even distribution of damage. Impact damage lengths of 25, 58, 87 and 116 mm were studied.

2.2. Mechanical property testing

2.2.1. Edgewise compression testing The edgewise compression properties of the sandwich

composite were measured in accordance with ASTM C364-94 specifications [ 151 using specimens that were 75 mm wide and 35.6 mm thick. The ASTM method rec- ommends that the unsupported specimen gauge length is less than eight times the thickness, which for the sandwich composite means a gauge length less than 285 mm. Because an exact gauge length is not specified by the ASTM, the variation in compression strength of the pristine sandwich composite (i.e. without an interfacial crack or impact damage) was measured over a range of gauge lengths between 50 and 225 mm. Based on these results a gauge length best suited for testing specimens containing an interfacial crack or impact damage can be determined. Four compression tests were performed for each gauge length with the pristine composite specimens, while three compression tests were performed for each interfacial crack length and impact damage length.

The specimens were loaded in edgewise compression at a cross-head speed of 0.5 mm/min to failure. Because the Young’s modulus of the GFRP skins (Ef = 12 100 MPa) was much greater than for the foam core (EC = 76 MPa), it is assumed that the axial load is carried by the skins. Therefore, the edgewise compressive stress (0) exerted on the sandwich composite can be calculated from the applied load (P) by the expression

266 A.P. Mouric R.S. Thomson I Composite Structures 44 (1999) 263-278

Fig. 2. Illustration showing the distribution of impacts used to extend damage along and across the sandwich composite

P

a=2btf’ where b is the specimen width and tf is the skin thickness.

2.2.2. Flexural testing

The flexural properties were measured under four- point bending in general accordance with ASTM C393- 94 [16]. The only departure from the test specification was that the specimens were 34 mm wide, rather than being the recommended minimum width of 72 mm. However, this difference is not expected to significantly affect the flexural properties of the sandwich composite. The general lay-out of the flexural test is shown in Fig. 3, with the beams tested at a cross-head speed of 3 mmlmin in quarter-point loading using load and support spans of 400 and 800 mm, respectively. The specimens with the interfacial crack or impact damage area had the crack/damage positioned mid-span between the load points. The maximum flexural stress carried by the surface fibres of the skin for a sandwich composite under l/Cpoint bending is calculated by the expression given in Ref. [ 171:

CT= PL(Q + 03,)

4btft,z ’ (2)

where L is the support span and t, is the core thickness. The specimens were tested so that the interfacial crack or impact damage was near the top surface which experienced bending induced compressive stresses, and in this paper is termed the “bending-compression” load state. In addition, specimens were tested with the crack or impact damage near the bottom surface that experienced a “bending-tension” load state. Three bending- compression and three bending-tension tests were performed on specimens for each interfacial crack length and impact damage length.

2.2.3. Shear testing

The shear properties were also measured using the four-point flexural configuration, but in this case the interfacial crack or impact damage was positioned mid- way between one of the load and support points as shown in Fig. 4. In this position the region of beam containing the crack or impact damage is subject to shear loading. As the skins of the sandwich composite are thin, the shear stress (r) is assumed to be carried predominantly by the core, and is calculated by

P

T=2bt,.

Aluminium Tab---,

Bending-Tension -/ L- Bending-Compressionw

800 mm

Fig. 3. Illustration of the four-point bend test for measuring flexural properties. Tests were performed with the interfacial crack or impact damage

(length of 2a) located mid-span between the load points.

A.P. Mouritz. R.S. Thomson I Composite Structures 44 (1999) 263-278 267

I- ammm j

Fig. 4. Illustration of the four-point bend test for measuring shear properties. Tests were performed with the interfacial crack or impact damage (length of 2a) located mid-span between one of the load and support points.

where &,, is the distance between the skin mid-planes. Three shear tests were performed for each interfacial crack length and impact damage length.

3. Results and discussion

3.1. Compression specimen size eflects

The ASTM method for edgewise compression testing does not specify a specimen gauge length other than it must be less than eight times the beam thickness. Therefore, to assess whether the compression properties of the minehunter composite are affected by gauge length, the strength and failure mechanism were determined for lengths between 50 and 225 mm. Figure 5 shows that the compression strength decreased rapidly with increasing gauge length, with failure occurring by compressive fracture of the GFRP skins for specimens shorter than N 100-120 mm while longer specimens failed by shear crimping of the foam core.

The strong dependence of compression strength on gauge length reveals that short specimens do not give an accurate measure of the compression properties expected of much larger sandwich composite structures, such as the Bay class minehunter. In this study a gauge length of 225 mm was chosen as the most suitable for testing composites with interfacial cracks or impact damage. This is because the core crimping failure mechanism is the same as that expected for large sandwich composite ship panels that have low shear rigidity and/or high flexural rigidity.

3.2. Properties and failure mechanisms for the composite with an interfacial crack

Figures 6-9 show stress against cross-head displacement curves for the sandwich composite with and without an interfacial crack when tested in edgewise compression, shear, bending-compression and bending- tension, respectively. The failure mode of the crack-free

composite was found to be dependent on the load state. Under edgewise compression the crack-free composite failed by shear crimping of the core (Fig. 6). This type of failure is common for sandwich composite beams that are unstable in compression, and the stress at which shear crimping occurs can be predicted using

1171

where G, is the shear modulus of the core. Using this equation it is predicted that crimping failure of the sandwich composite will occur at 126 MPa, which agrees well with the measured value of 150 MPa.

In contrast, testing the defect-free composite in the three other load states using the four-point bend method caused the specimen to fail gradually under one or both upper load points by skin wrinkling, despite the use of aluminium tabs at the contact points between the loading rods and specimen. This failure mode involved the localised wrinkling of the skin as it was pressed by the loading rods into the core, and the shear and flexural tests were terminated when the buckling became exces- sive. The stress required to cause skin wrinkling is calculated by method described in Ref. [17]:

~wrinkling = QVGWZ, (5) where Ef and E, are the Young’s moduli of the skin and core, respectively, and Q is a factor based on the initial skin waviness. Caprino and Teti [17] report that sandwich composites typically have a surface waviness of about 0.5 mm, which is equivalent to a Q value of 0.4. Using this value, a skin wrinkling failure stress of 118 MPa for the sandwich composite is calculated using Eq. (5). This value is in excellent agreement with the measured flexural failure stress of about 105 MPa.

Skin wrinkling is a common failure mode in defect- free sandwich composites [7,17,18], and is difficult to avoid when testing materials with a compliant core using the four-point bend method. Because the defect- free sandwich composite did not fail by a shear or bending dominated process, this indicates that the

268 A. P. Mouritz. R.S. Thomson I Composite Srructures 44 11999) 263-278

225

125

Core Crimping F

\

‘hilure

0 50 100 150 200 250

Gauge Length (mm)

Fig. 5. Variation in edgewise compression strength with specimen gauge length. The vertical line shows the gauge length at which the failure

mechanism changes from compressive fracture of the skins to shear crimping of the core. The scatter bars show the standard deviation from four

strength measurements.

respective shear and flexural strength values of 1.1 and 105 MPa shown in Figs. 7-9 are lower than the true shear and flexural strengths. It is worth noting, however, that when Lingard [18] tested a variety of GFRP/PVC foam sandwich composites under four-point bending it was found that the failure load needed to cause skin wrinkling under the load points was about of the same order as the load needed to cause true bending failure, which was observed to consist of skin wrinkling and core compression failure at the mid-span between the load points. It appears, therefore, that the failure strengths shown in Figs. 8 and 9 may be similar in value to the true flexural strength of the sandwich composite.

The stress-displacement curve was affected by an interfacial crack only when the sandwich composite was loaded in edgewise compression or shear but not in bending. Figure 6 shows that under edgewise compression a reduction in stiffness occurred at 50 MPa because of the sudden outward buckling of the skin covering the crack. This skin buckling stress was found to decrease as

the interfacial crack length increased, as shown in Fig. 10. This trend has been observed in a variety of monolithic composites, where the compression strength fell with increasing delamination size due to sub-laminate buckling [19]. It has been shown that the compression stress needed to cause this buckling in monolithic laminates can be predicted using the Euler equation for elastic instability of a long slender beam, which in its simplest form is given by

where C is the coefficient of constraint, which for a beam with clamped ends equals 4, E is the elastic modulus, 2a

is the interfacial crack length, Z is the second moment of area and A is the cross-section area of the beam. The theoretical skin buckling stress determined using Eq. (6) for the sandwich composite is compared against the measured stresses in Fig. 10. For short interfacial crack lengths, the buckling stress is over-predicted by the

A.P. Mouritz, R.S. Thomson I Composite Structures 44 (1999) 263-278 269

150 t

125

3

P ; 100

!!

3j g 75 ._

8 ? 50

s 25

0

Crack Flee -

40 mm Interfacial Crack

0.0 0.5 1.0 1.5 2.0 2.5

Cross-head Displacement (mm)

3.0 3.5

Fig. 6. Edgewise compression stress against cross-head displacement curves for the composite without a crack and with a 40 mm long interfacial crack. The photographs show the failure mechanisms of the composite while under load.

/ Sk/n Buckhg klhmm

equation as the assumption of a long, slender beam is violated. While the agreement improves for longer interfacial crack lengths, the difference that exist is probably due to the assumed isotropic properties of the skin and the nature of the support provided by the foam core.

loading grew into the foam until it reached the opposing skin, at which point the composite suddenly failed. Zenkert [6] also observed this type of shear failure in polymer foam sandwich composites containing interfacial cracks. Zenkert [6], Triantofillou and Gibson [7] and Thomson et al. [9] have shown that the load needed to cause the onset of shear cracking can be predicted with good accuracy using analytical or finite element models based on Mode II fracture mechanics theory applied to

Figure 7 shows that in shear loading the composite containing the interfacial crack failed at a lower load than the defect-free specimen. In this case a shear crack initiated near the interfacial crack tip, and upon further a layered anisotropic material.

z SO mm Interfacial Crack

1.0

q 0.8

c !

10 20

Displacement (mm)

30 40

Fig. 7. Shear stress against cross-head displacement curves and photographs of the failure mechanism for the composite without a crack and with a 90 mm long interfacial crack.

Shar Core PaEwe

270 A. P. Mowit;. R. S. Thomson I Composite Structures 44 i 1999J 263-278

120

!

Crack-free

9 I 100 mm Interfacial Crack I

0 10 20 30

Displacement (mm)

40 50 ov 1 I I I

Fig. 8. Bending-compression stress against cross-head displacement curves for the composite without a crack and with a 100 mm long interfacial

crack. The upper photograph shows the failure mechanism of localised skin wrinkling beneath the load point.

Interfacial cracks did not affect the stressdisplace- ment curve when the composite was loaded in bending- compression or bending-tension, as shown in Figs. 8 and 9. In both load states failure occurred under the load points by the same skin wrinkling process which caused the crack-free composite to fail. It is worth noting, however, that soon after skin wrinkling had occurred in bending-compression, the skin over the interfacial crack usually buckled upwards away from the

core by a similar skin buckling process to that observed in edgewise compression.

The interfacial cracks did not alter the stiffness of the composite, however they had a significant affect on the strength when tested in edgewise compression or shear. The effect of interfacial crack length on the normalised strengths in the different load states are compared in Fig. 11. The strength of the composite containing the interfacial crack has been normalised against the

Skin Whklhg Failure

0 0 10 20 30 40 50

Displacement (mm)

Fig. 9. Bending-tension stress against cross-head displacement curves for the composite without a crack and with a 100 mm long interfacial crack. The photograph shows the failure mechanism of localised skin wrinkling beneath the load point.


80

I I I I I I

40 60 80 100 120 140 160

lnterhcial Crack Length (mm)

Fig. 10. Comparison of measured and predicted skin buckling stresses for the sandwich composite containing an interfacial crack loaded in edgewise compression.

strength of the crack-free composite so the different load states can be compared directly. Both the edgewise compression and shear strengths fell rapidly when the interfacial crack exceeded a length of -30 mm. At this point the failure mechanism changed suddenly from core crimping to skin buckling under edgewise compression and from skin wrinkling to core cracking under shear. In comparison, the failure strength was unaffected by an interfacial crack in the composite loaded in bending-compression or bending-tension because the failure mechanism of local skin wrinkling remained unchanged.

3.3. Properties and failure mechanisms of the impact damaged composite

Optical microscopy and scanning electron microscopy (SEM) were used to observe the main types of damage suffered by the sandwich composite from impact loading. Damage produced by the low energy impacts was confined mostly to the impacted GFRP skin. SEM examination revealed that the damage consisted of delamination cracking between the fibreglass plies and shear cracks through the resin matrix of the skin, as shown in Fig. 12. The underlying foam showed a small

0.8

f

SJ b

0.6

I! z 0.4

P

0.2

0.0

A 4

A Ccmpressive Bending

J

0 50 loo 150 200 250

Interfacial Crack LengUt (mm)

Fig. 11. Effect of interfacial crack length on the normalised strength for the four load states.

amount of compaction, but this was considered negli- gible. There was no evidence of broken fibres or interfacial cracks between the skin and core despite the widespread delamination cracking within the skin. The good impact damage tolerance of the interface was probably caused by the reasonably high surface rough- ness of the foam, which promoted good mechanical bonding to the skin.

The high energy impacts caused more severe damage. The damage to the impacted skin consisted of delaminations and resin cracks similar to those shown in Fig. 12, but in some locations the skin was completely broken, as shown in Fig. 13. The foam below the impacted skin was also damaged; with tears occurring directly beneath the rupture sites in the skin and many of the foam cells were partially or completely crushed. This crushing resulted in the compaction of the foam, which caused a shallow depression to form across the impacted surface. The depth of this depression crater increased towards the centre of the impact damaged zone where the depth reached about 2 mm. Despite the severe damage to the skin and core, interfacial cracks were only observed in those localised regions where the skin was broken. Nemes and Simmons [l l] examined the low-velocity impact response of foam core sandwich composites using a combination of computational and experimental

272 A. I’. Mouritz, R. S. Thomson I Compositc~ Structures 44 (1999) 263-278

200 pm

Fig. 12. Scanning electron micrograph showing low energy impact damage to the GFRP skin. The damage consisted of shear cracks through the

resin-rich regions and delaminations between the fibreglass plies.

techniques, and concluded that high shear forces generated under impact are responsible for most of the damage, which can include interfacial cracks and core shear cracks. While dynamic shear forces were generated in the GFRP/PVC foam sandwich composite studied here because the specimens were rigidly clamped close to the impact site, interfacial cracks and shear cracks through the core were not produced because the impact energy was too low. It is worth noting that Hall [lo] found in the underwater explosion blast testing of large sandwich panels made from the Bay class minehunter composite

that damage usually consisted of skin fracture and core crushing. This damage was similar to that produced by the high energy impacts. However, in cases when a large explosion was used by Hall to produce a high pressure underwater shock wave the core failed by shear cracking. This type of failure was not generated by the impact test where the core failed by tearing. From this comparison of impact and underwater shock damage it appears that impact testing reproduced medium level blast damage with good accuracy, but could not generate high level shock damage.

Fig. 13. A sandwich composite in cross-section showing high energy impact damage. The impacted GFRP skin was broken and the underlying foam

core was torn and compressed.


A comparison of the stress against cross-head displacement curve for the undamaged sandwich composite against curves for the impact damaged composite are presented in Figs. 14-17 for each of the four load states. The curves shown for the damaged composite were measured after impact loading over a length of 116 mm at low and high energies.

Figure 14 shows that when the impacted composite was tested in edgewise compression it suffered large reductions in stiffness and strength, particularly when damaged by high energy impacts. As reported earlier, the undamaged composite failed in edgewise compression by shear crimping of the foam core, however none of the impacted specimens failed in this way. Instead failure occurred by skin wrinkling because the damaged skin collapsed into the foam due to the depression crater formed by the compaction and crushing of the foam under impact. The skin wrinkling caused by the high energy impacts was always centred where the skin had suffered most damage and the underlying foam was torn, which is the region of lowest compressive stiffness.

The stress-displacement curve and failure mechanism of the sandwich composite in shear was also affected by impact (Fig. 15). The strength was reduced about 10% by low energy impact damage and about 50% by high energy damage. The impact damage also caused a complete change in the failure mechanism: the undamaged composite failed gradually by skin wrinkling under the load points whereas the dominant failure mode for the impacted specimens was skin buckling. This buckling was characterised by the separation of delaminated plies in the damaged skin under shear, which ultimately caused the plies to fold and buckle at the peak load. In a

few cases the impact damaged composite failed instead by compressive fracture of the skin, as shown in Fig. 18. This mode of failure occurred mostly in specimens with a skin that had ruptured or suffered extensive fibre damage as a result of the impact.

When the impacted sandwich composite was tested in bending-compression it suffered substantial reductions in strength and stiffness due to failure occurring by wrinkling of the damaged skin over the entire impacted area (Fig. 16). In comparison, when tested in bending- tension the properties were largely unaffected by the impact damage, although the composite failed suddenly by tensile rupture of the damaged skin (Fig. 17). This reveals that despite extensive and severe damage to the skin on the tensile side of the flexural specimen, the beam is still able to support significant bending stresses.

Figures 19 and 20 compare the effect of impact damage length on the normalised stiffness and strength of the sandwich composite when tested in the different load states. The stiffness and strength values of the impacted composite were normalised against the properties of the undamaged composite tested in the same load state. The properties decreased rapidly with increasing impact damage length and impact energy except when the composite was tested in bending-tension. The greatest reductions in strength were experienced when the composite was loaded in edgewise compression or bending-compression. In these load states failure occurred by skin wrinkling, indicating that this failure mode is extremely detrimental to the strength of damaged sandwich composites. In comparison, when the composite failed in shear by skin buckling the degra- dation to the properties was not as severe and when

140 -

- 120 -

8 - loo - tll

f 80 -

s

‘! 60 -

a

40 -

Cora Sham &ln@ng

1.0 1.5 2.0 2.5

Cross-Head Displacmmt (mm)

Fig. 14. Edgewise compression stress against cross-head displacement curves for the undamaged composite and with impact damage over a length of 116 mm. The photographs show the failure mechanisms of the composite while under load.

274 A. P. Mowit:, R.S. Thomson I Composite Structures 44 (1999) 263-278

1.c

a 0.8

8

3

P

0.6

j 0.4

v)

0.2

0.0

I-

I -

Low Impact Energy

- High Impact Energy

Sldn Budding Failure

10 20 30

Cross-head Displacement (mm)

40

Fig. 15. Shear stress against cross-head displacement curves and photographs of the failure mechanism for the undamaged composite and with

impact damage over a length of 116 mm.

failure in bending-tension properties were unchanged.

4. Further discussion

occurred by rupture the

The mechanical properties of the sandwich composite used on the Bay class minehunter appear to be highly sensitive to interfacial cracks (which can result from poor quality fabrication) and impact damage

(from an accidental collision or underwater explosive blast), but only when the load state causes a change in the failure mechanism. Table 1 summarises the dominant failure mechanisms, and it is apparent that the failure process is complicated because it is controlled by the type and size of the defect as well as by the load state. It is expected that the failure mode will also be determined by other parameters which were not studied in this paper, including skin and core thickness, skin and core modulus, skin and core strength, and core

LowEnergyhnpact High Energy Impact

10 20 30 40

Cross-head Dlspiacemsnt (mm)

Fig. 16. Bending-compression stress against cross-head displacement curves and photographs of the failure mechanism for the undamaged composite

and with impact damage over a length of 116 mm.

A.P. Mouritz, R.S. Thomson I Composite Structures 44 (1999) 263-278

Shin olmphg Ftilun impact Damaged

I

10 20 30 40

Cross-head Dispkcem~t (mm)

215

Fig. 17. Bending-tension stress against cross-head displacement curves and photographs of the failure mechanism for the undamaged composite and with impact damage over a length of 116 mm.

density. This complex interaction of material, structural crack was long. However the model by Triantofillou geometry and damage parameters makes it difficult to and Gibson became increasingly less accurate with develop models which accurately determine the prop- shorter crack lengths. Predicting the properties of an erties of a sandwich composite containing an interfacial impacted sandwich composite is more difficult because crack and, in particular, impact damage. Some success of the complex types and distribution of damage, has been made in modelling the compression and shear particularly under high energy impacts where damage strengths of sandwich composites with an interfacial to the core becomes an important factor. A model for crack. For example, Zenkert [6] and Triantofillou and estimating the residual edgewise compression strength Gibson [7] proposed models for predicting the shear of the impact damaged composite has recently been failure load, and Thomson et al. [9] recently tested proposed by Thomson and Mouritz [20], however fur- these models against the minehunter composite and ther refinement is still required to improve the accu- found good accuracy (within 25%) when the interfacial racy.

T

276 A.P. Mow-it;. R.S. Thomson I Composite Structurrs 44 (1999) 263-278

1.a

0.8

a g 0.6

to

f L

E & 0.4

9

0.2

0.0

64

I.(

0.E

1 0.6

8

L 1 E P o.4

0.2

. 0 LIXV L Hgh Energy Da-e

in Edgewise Conpmssion l 0 Low& H&h Energy Dmnap in Shear

0 20 40 60 00 100 120

Impact Damage Length (mm)

0.0 - 0

A A A A

A \ A

A A Low B High Energy Damage in BendingCompnssIon

4 0 Low a High Ensrgy Dsmmgs in Bending-lenelon

I I I I I I

20 40 60 a0 100

impact Dmagr Lenglh (mm)

120

Fig. 19. Effect of impact damage length on normalised stiffness when Fig. 20. Eflect of impact damage length on normalised strength when

loaded in (a) edgewise compression and shear and (b) bending-com- loaded in (a) edgewise compression and shear and (b) bending-com-

pression and bending-tension. pression and bending-tension.

0.e I -

- 0 Lcw 6 High Energy Damage

In Edgewise CompressIon . 0 Low& Huh Ensrgy Damsgs in Shear

I I I I I I

u 20 40 60 80 100 120

(a) Impact Damage Length (mm)

1.0

0.8

B e!

0.6

tj

1

i 0.4

z

0.2

0.c

A A Lcw & Hgh Energy Damage

in BendhgCompnsrion l 0 Lw a Hgh Energy Damage

in Bending-Tension

A

)- 0 20 40 60 00 100 120

(b) Impact Damage Length (mm)

A. P. Mouritz, R.S. Thomson I Composite Structures 44 (1999) 263-278 211

Table 1 Summary of failure modes

Load state Interfacial crack Low energy impact High energy impact

Edgewise compression Skin buckling* Skin wrinkling Skin wrinkling Shear Core shear cracking * Skin buckling/skin compressive fracture Skin buckling/skin compressive fracture Bending-compression No effect Skin wrinkling Skin wrinkling Bending-tension No effect Tensile rupture Tensile rupture

* Only when the crack exceeds a critical length (e.g. 20-30 mm).

The results indicate that the structural integrity of a minehunter containing cracks or impact damage will be determined to a large extent on the load state. In normal sea states, however, minehunter hulls are subjected to a complicated mixture of flexural and shear fatigue loads due to waves slamming against the ship combined with static compression and shear loads exerted under the weight of the superstructure. As a result, it is difficult to translate the results from single load states to multiple load conditions.

Another important consideration when using the results and observations from this study to understand the damage tolerance of the minehunter is that the tests were performed on small coupons in which the interfacial cracks and impact damage extended across the entire specimen width. Any defects within the minehunter caused by poor fabrication or in-service damage are likely to be isolated, and therefore edge effects may not be important. As a result the damage tolerance of the minehunter will be greater than that suggested by the small coupon tests. Mouritz and Thomson [20,21] have shown, for example, that the edgewise compression strength of the minehunter composite was not degraded by an interfacial crack unless it extended more than 70% across the panel. Similarly, the compression strength of the impact damaged composite was not reduced until the damage covered more than -20% of the panel width. The minehunter is a large composite structure, with an overall length of 3 1 .O m and maximum beam of 9.0 m. As a result, it is expected that a very large interfacial crack or impact damage area would be required to significantly degrade the structural integrity.

5. Conclusions

This study has investigated the edgewise compression, flexure and shear properties of small GFRP/PVC foam sandwich composite specimens to gain an insight into the damage tolerance of a large minehunter ship. It was found that determining the edgewise compression properties of a large sandwich structure (such as a minehunting ship) using small specimens is difficult because the strength and failure mechanism are dependent on the gauge length. The compression strength decreases rapidly with increasing gauge length, and the failure

mechanism changes suddenly from compressive fracture of the skins to shear crimping of the core when the specimen length reaches -100 mm. Because of this sensitivity, size effects become an important factor when scaling-up the results from small coupon tests to a much larger structure such as a minehunter.

Interfacial cracks only affect the strength of the sandwich composite when they cause a change in the failure mechanism. Edgewise compression loading causes the failure mechanism to change from core shear crimping to skin buckling while shear loading changes the mechanism from localised skin buckling to core shear cracking. As a result, the strength (but not the stiffness) decrease rapidly with increasing crack length above 30 mm. In contrast, the properties are not affected under flexural loading because the failure mechanism remains unchanged.

The stiffness and strength of the sandwich composite decrease with increasing impact energy and impact damage area except when the composite is loaded in bending-tension. The properties are most severely degraded when the depression in the impacted surface causes the composite to fail by skin wrinkling, which occurs in edgewise compression and bending-compression.

In conclusion, using these results and observations from small coupon tests to determine the damage tolerance of a minehunter is difficult because the vessels are exposed to a complex mixture of structural and wave slamming loads and also because most defects will be isolated within a large superstructure or hull panel. Further testing on large structural components repre- sentative of the minehunter should be conducted to verify the results.

References

111

121

[31

[41

Smith CS. Design of marine structures in composite materials. London: Elsevier Applied Science, 1990. Karlsson KF, Astrom BT. Manufacturing and applications of structural sandwich components. Comp 1997;28A:97-111. Hall DJ, Robson BL. A review of the design and materials evaluation programme for the GRP/foam sandwich composite hull of the RAN minehunter. Comp. 1984;15:26676. Hall D. Types, causes and effect of defects in GFRP used for marine defence applications. Non-Destructive Testing - Australia 1986;23(2):347.

278 A. P. Mourit:. R.S. Thomson I Composite Structures 44 (1999) 263-278

[5] McClave EF. Goodwin MJ. Development of a test program to

evaluate structural defects in glass-reinforced plastic (GRP).

United States Coast Guard Report No. vol. 1, CG-D-02A-93.

1992.

[6] Zenkert D. Strength of sandwich beams with interface debond-

ings. Comp. Struct. 1991:17:331--50.

[7] Triantofillou TC, Gibson LJ. Debonding in foam-core sandwich

panels. Mat. and Struct. 1989;22:64-9.

[S] Carlsson LA, Sendlein LS, Merry SL. Characterization of face

sheet/core shear fracture of composite sandwich beams. J. Comp.

Mat. 1991;25:101~16.

[9] Thomson RS, Shah Khan Z, Mouritz AP. Shear properties of a

sandwich composite containing defects, Comp. Struct..

1998;42:107718.

[lo] Hall DJ. Examination of the effects of underwater blasts on

sandwich composite structures. Comp. Struct. 1989;ll: 101-20.

[ll] Nemes JA, Simmonds KE. Low-velocity impact response of

foam-core sandwich composites. J. Comp. Mat. 1992;26:500-

19.

[12] Kwon YW. Fuller LB. Compressive failure of unbalanced

sandwich composites after impact loading. In: Recent Advances

in Structural Mechanics, vol. 295, American Society for Mechan-

ical Engineers, PVP. ASME, New York. 1994:53363.

[13] Kassapoglou C. Jonas PJ, Abbott R. Compressive strength of

composite sandwich panels after impact damage: an experimental

and analytical study. J. Comp. Tech. and Res. 1988;10:65573.

[I41

[I51

1161

[I71

[181

[191

WI

1211

Auerkari P, Pankakoski PH, Kauppinen P. Effect of impact face

damage on the strength of sandwich composites. In: Williams JG,

Pavan A. (Ed). Impact and Dynamic Fracture of Polymers and

Composites, ESIS 19. London: Mechanical Engineering Publica-

tions, 1995:423-3 I. Standard test method for edgewise compressive strength of

sandwich constructions. Annual Book of ASTM Standards, vol

15.03. American Society for Testing and Materials, PA, 1996: 13

4.

Standard test method for flexural properties of sandwich

constructions. Annual Book of ASTM Standards, vol. 15.03,

American Society for Testing and Materials, PA, 1996:21 4.

Caprino G, Teti R. Sandwich Structures Handbook, I1 Prato.

Padua, 1989.

Lingard JR. Optimising foam sandwich composites for ship

superstructure applications. Proc. of Shipshape 2000. Tenth

International Maritime and Shipping Symposium. 8-9 Nov

1993:653-70.

Cable CW. The effect of defects in glass-reinforced plastic (GRP).

Marine Tech. 1991;28:91l8.

Thomson RS, Mouritz AP. Skin wrinkling of impact damaged

sandwich composite. Sandwich J. Struct. and Mat. submitted.

Mouritr AP. Thomson RS. Compressive properties of a sandwich

composite containing fabrication defects or impact damage. The

Second Australasian Congress on Applied Mechanics (ACAM

99). IO-12 February 1999, Canberra. in press.

Documents

Compression, Flexure and Shear Properties of a Sandwich Composite